diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,164331 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 23470,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.521516829995739e-05,
+      "grad_norm": 18.5,
+      "learning_rate": 2.5000000000000002e-08,
+      "loss": 1.21,
+      "step": 1
+    },
+    {
+      "epoch": 0.00017043033659991478,
+      "grad_norm": 34.25,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 1.5951,
+      "step": 2
+    },
+    {
+      "epoch": 0.00025564550489987217,
+      "grad_norm": 32.25,
+      "learning_rate": 7.500000000000001e-08,
+      "loss": 1.4702,
+      "step": 3
+    },
+    {
+      "epoch": 0.00034086067319982955,
+      "grad_norm": 27.5,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 1.3454,
+      "step": 4
+    },
+    {
+      "epoch": 0.00042607584149978694,
+      "grad_norm": 28.125,
+      "learning_rate": 1.2500000000000002e-07,
+      "loss": 1.16,
+      "step": 5
+    },
+    {
+      "epoch": 0.0005112910097997443,
+      "grad_norm": 27.75,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 1.6235,
+      "step": 6
+    },
+    {
+      "epoch": 0.0005965061780997018,
+      "grad_norm": 27.0,
+      "learning_rate": 1.7500000000000002e-07,
+      "loss": 1.3127,
+      "step": 7
+    },
+    {
+      "epoch": 0.0006817213463996591,
+      "grad_norm": 35.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 1.4014,
+      "step": 8
+    },
+    {
+      "epoch": 0.0007669365146996165,
+      "grad_norm": 19.375,
+      "learning_rate": 2.2500000000000002e-07,
+      "loss": 1.1322,
+      "step": 9
+    },
+    {
+      "epoch": 0.0008521516829995739,
+      "grad_norm": 28.25,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.2969,
+      "step": 10
+    },
+    {
+      "epoch": 0.0009373668512995313,
+      "grad_norm": 43.0,
+      "learning_rate": 2.75e-07,
+      "loss": 1.4659,
+      "step": 11
+    },
+    {
+      "epoch": 0.0010225820195994887,
+      "grad_norm": 14.875,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.8019,
+      "step": 12
+    },
+    {
+      "epoch": 0.001107797187899446,
+      "grad_norm": 45.75,
+      "learning_rate": 3.25e-07,
+      "loss": 1.4666,
+      "step": 13
+    },
+    {
+      "epoch": 0.0011930123561994035,
+      "grad_norm": 52.0,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 2.0725,
+      "step": 14
+    },
+    {
+      "epoch": 0.0012782275244993609,
+      "grad_norm": 91.5,
+      "learning_rate": 3.75e-07,
+      "loss": 1.517,
+      "step": 15
+    },
+    {
+      "epoch": 0.0013634426927993182,
+      "grad_norm": 53.5,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.6968,
+      "step": 16
+    },
+    {
+      "epoch": 0.0014486578610992758,
+      "grad_norm": 68.0,
+      "learning_rate": 4.2500000000000006e-07,
+      "loss": 1.3316,
+      "step": 17
+    },
+    {
+      "epoch": 0.001533873029399233,
+      "grad_norm": 17.875,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 1.1111,
+      "step": 18
+    },
+    {
+      "epoch": 0.0016190881976991904,
+      "grad_norm": 36.75,
+      "learning_rate": 4.7500000000000006e-07,
+      "loss": 1.5615,
+      "step": 19
+    },
+    {
+      "epoch": 0.0017043033659991478,
+      "grad_norm": 55.75,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.5168,
+      "step": 20
+    },
+    {
+      "epoch": 0.0017895185342991053,
+      "grad_norm": 23.625,
+      "learning_rate": 5.250000000000001e-07,
+      "loss": 1.3197,
+      "step": 21
+    },
+    {
+      "epoch": 0.0018747337025990627,
+      "grad_norm": 31.5,
+      "learning_rate": 5.5e-07,
+      "loss": 1.3818,
+      "step": 22
+    },
+    {
+      "epoch": 0.00195994887089902,
+      "grad_norm": 24.75,
+      "learning_rate": 5.750000000000001e-07,
+      "loss": 1.4715,
+      "step": 23
+    },
+    {
+      "epoch": 0.0020451640391989773,
+      "grad_norm": 26.25,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.2183,
+      "step": 24
+    },
+    {
+      "epoch": 0.0021303792074989347,
+      "grad_norm": 52.5,
+      "learning_rate": 6.25e-07,
+      "loss": 1.5819,
+      "step": 25
+    },
+    {
+      "epoch": 0.002215594375798892,
+      "grad_norm": 18.875,
+      "learning_rate": 6.5e-07,
+      "loss": 1.224,
+      "step": 26
+    },
+    {
+      "epoch": 0.0023008095440988498,
+      "grad_norm": 47.75,
+      "learning_rate": 6.750000000000001e-07,
+      "loss": 1.0704,
+      "step": 27
+    },
+    {
+      "epoch": 0.002386024712398807,
+      "grad_norm": 174.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 1.2073,
+      "step": 28
+    },
+    {
+      "epoch": 0.0024712398806987644,
+      "grad_norm": 37.75,
+      "learning_rate": 7.25e-07,
+      "loss": 1.6614,
+      "step": 29
+    },
+    {
+      "epoch": 0.0025564550489987218,
+      "grad_norm": 47.25,
+      "learning_rate": 7.5e-07,
+      "loss": 1.1309,
+      "step": 30
+    },
+    {
+      "epoch": 0.002641670217298679,
+      "grad_norm": 24.625,
+      "learning_rate": 7.750000000000001e-07,
+      "loss": 1.4704,
+      "step": 31
+    },
+    {
+      "epoch": 0.0027268853855986364,
+      "grad_norm": 53.25,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.93,
+      "step": 32
+    },
+    {
+      "epoch": 0.0028121005538985938,
+      "grad_norm": 33.25,
+      "learning_rate": 8.250000000000001e-07,
+      "loss": 1.2446,
+      "step": 33
+    },
+    {
+      "epoch": 0.0028973157221985515,
+      "grad_norm": 25.75,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 1.4553,
+      "step": 34
+    },
+    {
+      "epoch": 0.002982530890498509,
+      "grad_norm": 17.0,
+      "learning_rate": 8.75e-07,
+      "loss": 1.0626,
+      "step": 35
+    },
+    {
+      "epoch": 0.003067746058798466,
+      "grad_norm": 39.75,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 1.3777,
+      "step": 36
+    },
+    {
+      "epoch": 0.0031529612270984235,
+      "grad_norm": 22.875,
+      "learning_rate": 9.25e-07,
+      "loss": 1.2808,
+      "step": 37
+    },
+    {
+      "epoch": 0.003238176395398381,
+      "grad_norm": 15.125,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 1.0063,
+      "step": 38
+    },
+    {
+      "epoch": 0.003323391563698338,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.750000000000002e-07,
+      "loss": 1.2981,
+      "step": 39
+    },
+    {
+      "epoch": 0.0034086067319982955,
+      "grad_norm": 15.75,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.8237,
+      "step": 40
+    },
+    {
+      "epoch": 0.0034938219002982533,
+      "grad_norm": 21.375,
+      "learning_rate": 1.025e-06,
+      "loss": 1.3627,
+      "step": 41
+    },
+    {
+      "epoch": 0.0035790370685982106,
+      "grad_norm": 24.125,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 1.4088,
+      "step": 42
+    },
+    {
+      "epoch": 0.003664252236898168,
+      "grad_norm": 18.75,
+      "learning_rate": 1.075e-06,
+      "loss": 1.1094,
+      "step": 43
+    },
+    {
+      "epoch": 0.0037494674051981253,
+      "grad_norm": 20.875,
+      "learning_rate": 1.1e-06,
+      "loss": 1.3761,
+      "step": 44
+    },
+    {
+      "epoch": 0.0038346825734980826,
+      "grad_norm": 30.5,
+      "learning_rate": 1.125e-06,
+      "loss": 1.1231,
+      "step": 45
+    },
+    {
+      "epoch": 0.00391989774179804,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 1.0932,
+      "step": 46
+    },
+    {
+      "epoch": 0.004005112910097998,
+      "grad_norm": 34.0,
+      "learning_rate": 1.175e-06,
+      "loss": 1.5352,
+      "step": 47
+    },
+    {
+      "epoch": 0.004090328078397955,
+      "grad_norm": 34.25,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.4602,
+      "step": 48
+    },
+    {
+      "epoch": 0.004175543246697912,
+      "grad_norm": 27.25,
+      "learning_rate": 1.2250000000000001e-06,
+      "loss": 1.0401,
+      "step": 49
+    },
+    {
+      "epoch": 0.004260758414997869,
+      "grad_norm": 22.625,
+      "learning_rate": 1.25e-06,
+      "loss": 1.3274,
+      "step": 50
+    },
+    {
+      "epoch": 0.004345973583297827,
+      "grad_norm": 27.0,
+      "learning_rate": 1.275e-06,
+      "loss": 1.5387,
+      "step": 51
+    },
+    {
+      "epoch": 0.004431188751597784,
+      "grad_norm": 26.0,
+      "learning_rate": 1.3e-06,
+      "loss": 1.1358,
+      "step": 52
+    },
+    {
+      "epoch": 0.004516403919897742,
+      "grad_norm": 27.375,
+      "learning_rate": 1.3250000000000002e-06,
+      "loss": 1.2844,
+      "step": 53
+    },
+    {
+      "epoch": 0.0046016190881976995,
+      "grad_norm": 23.25,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 1.1243,
+      "step": 54
+    },
+    {
+      "epoch": 0.004686834256497656,
+      "grad_norm": 49.25,
+      "learning_rate": 1.3750000000000002e-06,
+      "loss": 1.3856,
+      "step": 55
+    },
+    {
+      "epoch": 0.004772049424797614,
+      "grad_norm": 44.5,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 1.4186,
+      "step": 56
+    },
+    {
+      "epoch": 0.004857264593097571,
+      "grad_norm": 26.75,
+      "learning_rate": 1.425e-06,
+      "loss": 1.0897,
+      "step": 57
+    },
+    {
+      "epoch": 0.004942479761397529,
+      "grad_norm": 17.0,
+      "learning_rate": 1.45e-06,
+      "loss": 1.0632,
+      "step": 58
+    },
+    {
+      "epoch": 0.005027694929697486,
+      "grad_norm": 98.5,
+      "learning_rate": 1.475e-06,
+      "loss": 1.0083,
+      "step": 59
+    },
+    {
+      "epoch": 0.0051129100979974435,
+      "grad_norm": 62.25,
+      "learning_rate": 1.5e-06,
+      "loss": 0.9524,
+      "step": 60
+    },
+    {
+      "epoch": 0.005198125266297401,
+      "grad_norm": 74.0,
+      "learning_rate": 1.525e-06,
+      "loss": 1.3847,
+      "step": 61
+    },
+    {
+      "epoch": 0.005283340434597358,
+      "grad_norm": 21.0,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 0.7665,
+      "step": 62
+    },
+    {
+      "epoch": 0.005368555602897316,
+      "grad_norm": 27.625,
+      "learning_rate": 1.5750000000000002e-06,
+      "loss": 0.9051,
+      "step": 63
+    },
+    {
+      "epoch": 0.005453770771197273,
+      "grad_norm": 35.75,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.1105,
+      "step": 64
+    },
+    {
+      "epoch": 0.005538985939497231,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6250000000000001e-06,
+      "loss": 0.9974,
+      "step": 65
+    },
+    {
+      "epoch": 0.0056242011077971875,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 0.9864,
+      "step": 66
+    },
+    {
+      "epoch": 0.005709416276097145,
+      "grad_norm": 10.375,
+      "learning_rate": 1.6750000000000003e-06,
+      "loss": 0.5889,
+      "step": 67
+    },
+    {
+      "epoch": 0.005794631444397103,
+      "grad_norm": 23.875,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 1.0158,
+      "step": 68
+    },
+    {
+      "epoch": 0.00587984661269706,
+      "grad_norm": 15.875,
+      "learning_rate": 1.725e-06,
+      "loss": 0.6812,
+      "step": 69
+    },
+    {
+      "epoch": 0.005965061780997018,
+      "grad_norm": 35.75,
+      "learning_rate": 1.75e-06,
+      "loss": 1.4199,
+      "step": 70
+    },
+    {
+      "epoch": 0.006050276949296975,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.7750000000000002e-06,
+      "loss": 1.2153,
+      "step": 71
+    },
+    {
+      "epoch": 0.006135492117596932,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.8568,
+      "step": 72
+    },
+    {
+      "epoch": 0.006220707285896889,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.825e-06,
+      "loss": 1.2163,
+      "step": 73
+    },
+    {
+      "epoch": 0.006305922454196847,
+      "grad_norm": 22.5,
+      "learning_rate": 1.85e-06,
+      "loss": 1.0211,
+      "step": 74
+    },
+    {
+      "epoch": 0.006391137622496805,
+      "grad_norm": 41.25,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 1.7824,
+      "step": 75
+    },
+    {
+      "epoch": 0.006476352790796762,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.9872,
+      "step": 76
+    },
+    {
+      "epoch": 0.0065615679590967195,
+      "grad_norm": 14.75,
+      "learning_rate": 1.925e-06,
+      "loss": 1.1802,
+      "step": 77
+    },
+    {
+      "epoch": 0.006646783127396676,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 1.0842,
+      "step": 78
+    },
+    {
+      "epoch": 0.006731998295696634,
+      "grad_norm": 17.75,
+      "learning_rate": 1.975e-06,
+      "loss": 0.6949,
+      "step": 79
+    },
+    {
+      "epoch": 0.006817213463996591,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.0766,
+      "step": 80
+    },
+    {
+      "epoch": 0.006902428632296549,
+      "grad_norm": 23.0,
+      "learning_rate": 2.025e-06,
+      "loss": 1.2883,
+      "step": 81
+    },
+    {
+      "epoch": 0.006987643800596507,
+      "grad_norm": 14.0,
+      "learning_rate": 2.05e-06,
+      "loss": 0.9718,
+      "step": 82
+    },
+    {
+      "epoch": 0.0070728589688964635,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.075e-06,
+      "loss": 1.0055,
+      "step": 83
+    },
+    {
+      "epoch": 0.007158074137196421,
+      "grad_norm": 24.25,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.0122,
+      "step": 84
+    },
+    {
+      "epoch": 0.007243289305496378,
+      "grad_norm": 13.375,
+      "learning_rate": 2.125e-06,
+      "loss": 0.6408,
+      "step": 85
+    },
+    {
+      "epoch": 0.007328504473796336,
+      "grad_norm": 12.25,
+      "learning_rate": 2.15e-06,
+      "loss": 0.9325,
+      "step": 86
+    },
+    {
+      "epoch": 0.007413719642096293,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1750000000000004e-06,
+      "loss": 1.1798,
+      "step": 87
+    },
+    {
+      "epoch": 0.007498934810396251,
+      "grad_norm": 14.875,
+      "learning_rate": 2.2e-06,
+      "loss": 0.9925,
+      "step": 88
+    },
+    {
+      "epoch": 0.0075841499786962075,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2250000000000003e-06,
+      "loss": 0.8689,
+      "step": 89
+    },
+    {
+      "epoch": 0.007669365146996165,
+      "grad_norm": 19.25,
+      "learning_rate": 2.25e-06,
+      "loss": 1.1733,
+      "step": 90
+    },
+    {
+      "epoch": 0.007754580315296123,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.2750000000000002e-06,
+      "loss": 0.8938,
+      "step": 91
+    },
+    {
+      "epoch": 0.00783979548359608,
+      "grad_norm": 15.25,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.7673,
+      "step": 92
+    },
+    {
+      "epoch": 0.007925010651896037,
+      "grad_norm": 17.5,
+      "learning_rate": 2.325e-06,
+      "loss": 0.8134,
+      "step": 93
+    },
+    {
+      "epoch": 0.008010225820195995,
+      "grad_norm": 11.625,
+      "learning_rate": 2.35e-06,
+      "loss": 0.7227,
+      "step": 94
+    },
+    {
+      "epoch": 0.008095440988495952,
+      "grad_norm": 20.375,
+      "learning_rate": 2.375e-06,
+      "loss": 0.8756,
+      "step": 95
+    },
+    {
+      "epoch": 0.00818065615679591,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.0683,
+      "step": 96
+    },
+    {
+      "epoch": 0.008265871325095868,
+      "grad_norm": 20.125,
+      "learning_rate": 2.425e-06,
+      "loss": 1.3241,
+      "step": 97
+    },
+    {
+      "epoch": 0.008351086493395825,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.7591,
+      "step": 98
+    },
+    {
+      "epoch": 0.008436301661695782,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.475e-06,
+      "loss": 0.8464,
+      "step": 99
+    },
+    {
+      "epoch": 0.008521516829995739,
+      "grad_norm": 20.75,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0074,
+      "step": 100
+    },
+    {
+      "epoch": 0.008606731998295697,
+      "grad_norm": 25.125,
+      "learning_rate": 2.499999988705622e-06,
+      "loss": 1.1472,
+      "step": 101
+    },
+    {
+      "epoch": 0.008691947166595654,
+      "grad_norm": 21.375,
+      "learning_rate": 2.4999999548224866e-06,
+      "loss": 1.5212,
+      "step": 102
+    },
+    {
+      "epoch": 0.008777162334895611,
+      "grad_norm": 13.25,
+      "learning_rate": 2.499999898350595e-06,
+      "loss": 0.6892,
+      "step": 103
+    },
+    {
+      "epoch": 0.008862377503195568,
+      "grad_norm": 22.25,
+      "learning_rate": 2.4999998192899483e-06,
+      "loss": 1.219,
+      "step": 104
+    },
+    {
+      "epoch": 0.008947592671495527,
+      "grad_norm": 19.875,
+      "learning_rate": 2.4999997176405483e-06,
+      "loss": 1.0446,
+      "step": 105
+    },
+    {
+      "epoch": 0.009032807839795483,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4999995934023964e-06,
+      "loss": 0.8379,
+      "step": 106
+    },
+    {
+      "epoch": 0.00911802300809544,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4999994465754946e-06,
+      "loss": 0.9465,
+      "step": 107
+    },
+    {
+      "epoch": 0.009203238176395399,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4999992771598457e-06,
+      "loss": 0.952,
+      "step": 108
+    },
+    {
+      "epoch": 0.009288453344695356,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4999990851554527e-06,
+      "loss": 0.8594,
+      "step": 109
+    },
+    {
+      "epoch": 0.009373668512995313,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4999988705623203e-06,
+      "loss": 0.9269,
+      "step": 110
+    },
+    {
+      "epoch": 0.00945888368129527,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4999986333804506e-06,
+      "loss": 0.7822,
+      "step": 111
+    },
+    {
+      "epoch": 0.009544098849595228,
+      "grad_norm": 10.625,
+      "learning_rate": 2.4999983736098486e-06,
+      "loss": 0.5126,
+      "step": 112
+    },
+    {
+      "epoch": 0.009629314017895185,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.499998091250519e-06,
+      "loss": 0.9473,
+      "step": 113
+    },
+    {
+      "epoch": 0.009714529186195142,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.499997786302467e-06,
+      "loss": 0.7369,
+      "step": 114
+    },
+    {
+      "epoch": 0.0097997443544951,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.499997458765698e-06,
+      "loss": 0.5523,
+      "step": 115
+    },
+    {
+      "epoch": 0.009884959522795058,
+      "grad_norm": 15.875,
+      "learning_rate": 2.499997108640218e-06,
+      "loss": 0.8407,
+      "step": 116
+    },
+    {
+      "epoch": 0.009970174691095015,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4999967359260335e-06,
+      "loss": 0.8085,
+      "step": 117
+    },
+    {
+      "epoch": 0.010055389859394971,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4999963406231508e-06,
+      "loss": 0.923,
+      "step": 118
+    },
+    {
+      "epoch": 0.01014060502769493,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4999959227315773e-06,
+      "loss": 1.1125,
+      "step": 119
+    },
+    {
+      "epoch": 0.010225820195994887,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4999954822513203e-06,
+      "loss": 1.0749,
+      "step": 120
+    },
+    {
+      "epoch": 0.010311035364294844,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.499995019182388e-06,
+      "loss": 0.9686,
+      "step": 121
+    },
+    {
+      "epoch": 0.010396250532594803,
+      "grad_norm": 21.125,
+      "learning_rate": 2.4999945335247888e-06,
+      "loss": 1.3533,
+      "step": 122
+    },
+    {
+      "epoch": 0.01048146570089476,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4999940252785316e-06,
+      "loss": 0.8111,
+      "step": 123
+    },
+    {
+      "epoch": 0.010566680869194716,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4999934944436256e-06,
+      "loss": 0.6665,
+      "step": 124
+    },
+    {
+      "epoch": 0.010651896037494673,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4999929410200797e-06,
+      "loss": 0.7747,
+      "step": 125
+    },
+    {
+      "epoch": 0.010737111205794632,
+      "grad_norm": 22.25,
+      "learning_rate": 2.4999923650079044e-06,
+      "loss": 1.0984,
+      "step": 126
+    },
+    {
+      "epoch": 0.010822326374094589,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4999917664071105e-06,
+      "loss": 1.0494,
+      "step": 127
+    },
+    {
+      "epoch": 0.010907541542394546,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.499991145217708e-06,
+      "loss": 0.9469,
+      "step": 128
+    },
+    {
+      "epoch": 0.010992756710694504,
+      "grad_norm": 10.75,
+      "learning_rate": 2.4999905014397087e-06,
+      "loss": 0.5802,
+      "step": 129
+    },
+    {
+      "epoch": 0.011077971878994461,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4999898350731242e-06,
+      "loss": 0.9716,
+      "step": 130
+    },
+    {
+      "epoch": 0.011163187047294418,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4999891461179667e-06,
+      "loss": 1.0489,
+      "step": 131
+    },
+    {
+      "epoch": 0.011248402215594375,
+      "grad_norm": 21.0,
+      "learning_rate": 2.4999884345742476e-06,
+      "loss": 1.3434,
+      "step": 132
+    },
+    {
+      "epoch": 0.011333617383894334,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.499987700441981e-06,
+      "loss": 0.6333,
+      "step": 133
+    },
+    {
+      "epoch": 0.01141883255219429,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4999869437211795e-06,
+      "loss": 0.8497,
+      "step": 134
+    },
+    {
+      "epoch": 0.011504047720494247,
+      "grad_norm": 22.875,
+      "learning_rate": 2.499986164411857e-06,
+      "loss": 1.2663,
+      "step": 135
+    },
+    {
+      "epoch": 0.011589262888794206,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4999853625140277e-06,
+      "loss": 1.0296,
+      "step": 136
+    },
+    {
+      "epoch": 0.011674478057094163,
+      "grad_norm": 18.25,
+      "learning_rate": 2.499984538027706e-06,
+      "loss": 0.7126,
+      "step": 137
+    },
+    {
+      "epoch": 0.01175969322539412,
+      "grad_norm": 25.375,
+      "learning_rate": 2.4999836909529067e-06,
+      "loss": 1.3332,
+      "step": 138
+    },
+    {
+      "epoch": 0.011844908393694077,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.499982821289645e-06,
+      "loss": 0.7095,
+      "step": 139
+    },
+    {
+      "epoch": 0.011930123561994035,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4999819290379367e-06,
+      "loss": 1.0405,
+      "step": 140
+    },
+    {
+      "epoch": 0.012015338730293992,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.499981014197798e-06,
+      "loss": 0.537,
+      "step": 141
+    },
+    {
+      "epoch": 0.01210055389859395,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4999800767692456e-06,
+      "loss": 0.6729,
+      "step": 142
+    },
+    {
+      "epoch": 0.012185769066893908,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4999791167522964e-06,
+      "loss": 1.2783,
+      "step": 143
+    },
+    {
+      "epoch": 0.012270984235193865,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4999781341469673e-06,
+      "loss": 1.2955,
+      "step": 144
+    },
+    {
+      "epoch": 0.012356199403493822,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4999771289532765e-06,
+      "loss": 0.6357,
+      "step": 145
+    },
+    {
+      "epoch": 0.012441414571793779,
+      "grad_norm": 15.375,
+      "learning_rate": 2.499976101171242e-06,
+      "loss": 1.0036,
+      "step": 146
+    },
+    {
+      "epoch": 0.012526629740093737,
+      "grad_norm": 24.125,
+      "learning_rate": 2.4999750508008825e-06,
+      "loss": 1.1779,
+      "step": 147
+    },
+    {
+      "epoch": 0.012611844908393694,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4999739778422164e-06,
+      "loss": 0.8735,
+      "step": 148
+    },
+    {
+      "epoch": 0.012697060076693651,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.499972882295264e-06,
+      "loss": 0.7951,
+      "step": 149
+    },
+    {
+      "epoch": 0.01278227524499361,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4999717641600446e-06,
+      "loss": 0.8998,
+      "step": 150
+    },
+    {
+      "epoch": 0.012867490413293567,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4999706234365786e-06,
+      "loss": 1.4487,
+      "step": 151
+    },
+    {
+      "epoch": 0.012952705581593523,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4999694601248865e-06,
+      "loss": 1.0404,
+      "step": 152
+    },
+    {
+      "epoch": 0.01303792074989348,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4999682742249894e-06,
+      "loss": 0.8277,
+      "step": 153
+    },
+    {
+      "epoch": 0.013123135918193439,
+      "grad_norm": 18.875,
+      "learning_rate": 2.499967065736908e-06,
+      "loss": 0.9591,
+      "step": 154
+    },
+    {
+      "epoch": 0.013208351086493396,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4999658346606655e-06,
+      "loss": 0.7346,
+      "step": 155
+    },
+    {
+      "epoch": 0.013293566254793353,
+      "grad_norm": 19.25,
+      "learning_rate": 2.499964580996283e-06,
+      "loss": 0.8684,
+      "step": 156
+    },
+    {
+      "epoch": 0.013378781423093311,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.499963304743784e-06,
+      "loss": 0.8248,
+      "step": 157
+    },
+    {
+      "epoch": 0.013463996591393268,
+      "grad_norm": 15.375,
+      "learning_rate": 2.499962005903191e-06,
+      "loss": 1.0007,
+      "step": 158
+    },
+    {
+      "epoch": 0.013549211759693225,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4999606844745272e-06,
+      "loss": 0.8773,
+      "step": 159
+    },
+    {
+      "epoch": 0.013634426927993182,
+      "grad_norm": 12.625,
+      "learning_rate": 2.499959340457817e-06,
+      "loss": 0.8744,
+      "step": 160
+    },
+    {
+      "epoch": 0.01371964209629314,
+      "grad_norm": 20.25,
+      "learning_rate": 2.499957973853085e-06,
+      "loss": 1.0151,
+      "step": 161
+    },
+    {
+      "epoch": 0.013804857264593098,
+      "grad_norm": 21.75,
+      "learning_rate": 2.4999565846603556e-06,
+      "loss": 1.0634,
+      "step": 162
+    },
+    {
+      "epoch": 0.013890072432893055,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4999551728796533e-06,
+      "loss": 0.9191,
+      "step": 163
+    },
+    {
+      "epoch": 0.013975287601193013,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4999537385110047e-06,
+      "loss": 0.7709,
+      "step": 164
+    },
+    {
+      "epoch": 0.01406050276949297,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4999522815544345e-06,
+      "loss": 0.6242,
+      "step": 165
+    },
+    {
+      "epoch": 0.014145717937792927,
+      "grad_norm": 12.375,
+      "learning_rate": 2.49995080200997e-06,
+      "loss": 0.6428,
+      "step": 166
+    },
+    {
+      "epoch": 0.014230933106092884,
+      "grad_norm": 18.125,
+      "learning_rate": 2.499949299877637e-06,
+      "loss": 0.9351,
+      "step": 167
+    },
+    {
+      "epoch": 0.014316148274392843,
+      "grad_norm": 27.75,
+      "learning_rate": 2.499947775157464e-06,
+      "loss": 1.3185,
+      "step": 168
+    },
+    {
+      "epoch": 0.0144013634426928,
+      "grad_norm": 19.375,
+      "learning_rate": 2.4999462278494776e-06,
+      "loss": 0.847,
+      "step": 169
+    },
+    {
+      "epoch": 0.014486578610992756,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4999446579537054e-06,
+      "loss": 1.1577,
+      "step": 170
+    },
+    {
+      "epoch": 0.014571793779292713,
+      "grad_norm": 10.25,
+      "learning_rate": 2.499943065470177e-06,
+      "loss": 0.4071,
+      "step": 171
+    },
+    {
+      "epoch": 0.014657008947592672,
+      "grad_norm": 11.25,
+      "learning_rate": 2.49994145039892e-06,
+      "loss": 0.5795,
+      "step": 172
+    },
+    {
+      "epoch": 0.014742224115892629,
+      "grad_norm": 14.875,
+      "learning_rate": 2.499939812739964e-06,
+      "loss": 0.7708,
+      "step": 173
+    },
+    {
+      "epoch": 0.014827439284192586,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.499938152493339e-06,
+      "loss": 0.9726,
+      "step": 174
+    },
+    {
+      "epoch": 0.014912654452492544,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4999364696590747e-06,
+      "loss": 0.9742,
+      "step": 175
+    },
+    {
+      "epoch": 0.014997869620792501,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4999347642372017e-06,
+      "loss": 0.7613,
+      "step": 176
+    },
+    {
+      "epoch": 0.015083084789092458,
+      "grad_norm": 18.375,
+      "learning_rate": 2.49993303622775e-06,
+      "loss": 1.0638,
+      "step": 177
+    },
+    {
+      "epoch": 0.015168299957392415,
+      "grad_norm": 16.25,
+      "learning_rate": 2.499931285630752e-06,
+      "loss": 0.7163,
+      "step": 178
+    },
+    {
+      "epoch": 0.015253515125692374,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4999295124462384e-06,
+      "loss": 0.958,
+      "step": 179
+    },
+    {
+      "epoch": 0.01533873029399233,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4999277166742418e-06,
+      "loss": 0.8928,
+      "step": 180
+    },
+    {
+      "epoch": 0.015423945462292287,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4999258983147944e-06,
+      "loss": 0.836,
+      "step": 181
+    },
+    {
+      "epoch": 0.015509160630592246,
+      "grad_norm": 14.0,
+      "learning_rate": 2.499924057367929e-06,
+      "loss": 0.883,
+      "step": 182
+    },
+    {
+      "epoch": 0.015594375798892203,
+      "grad_norm": 15.375,
+      "learning_rate": 2.499922193833679e-06,
+      "loss": 0.9203,
+      "step": 183
+    },
+    {
+      "epoch": 0.01567959096719216,
+      "grad_norm": 10.375,
+      "learning_rate": 2.4999203077120782e-06,
+      "loss": 0.4823,
+      "step": 184
+    },
+    {
+      "epoch": 0.015764806135492117,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4999183990031606e-06,
+      "loss": 1.1846,
+      "step": 185
+    },
+    {
+      "epoch": 0.015850021303792074,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4999164677069605e-06,
+      "loss": 0.7984,
+      "step": 186
+    },
+    {
+      "epoch": 0.015935236472092034,
+      "grad_norm": 15.375,
+      "learning_rate": 2.499914513823513e-06,
+      "loss": 0.7548,
+      "step": 187
+    },
+    {
+      "epoch": 0.01602045164039199,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4999125373528532e-06,
+      "loss": 0.8622,
+      "step": 188
+    },
+    {
+      "epoch": 0.016105666808691948,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.499910538295017e-06,
+      "loss": 0.4636,
+      "step": 189
+    },
+    {
+      "epoch": 0.016190881976991905,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.4999085166500406e-06,
+      "loss": 0.5639,
+      "step": 190
+    },
+    {
+      "epoch": 0.01627609714529186,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4999064724179605e-06,
+      "loss": 0.7793,
+      "step": 191
+    },
+    {
+      "epoch": 0.01636131231359182,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4999044055988135e-06,
+      "loss": 0.8171,
+      "step": 192
+    },
+    {
+      "epoch": 0.016446527481891775,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.499902316192637e-06,
+      "loss": 0.6888,
+      "step": 193
+    },
+    {
+      "epoch": 0.016531742650191736,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4999002041994685e-06,
+      "loss": 0.7249,
+      "step": 194
+    },
+    {
+      "epoch": 0.016616957818491693,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.499898069619347e-06,
+      "loss": 0.7304,
+      "step": 195
+    },
+    {
+      "epoch": 0.01670217298679165,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.49989591245231e-06,
+      "loss": 0.863,
+      "step": 196
+    },
+    {
+      "epoch": 0.016787388155091607,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4998937326983974e-06,
+      "loss": 0.7757,
+      "step": 197
+    },
+    {
+      "epoch": 0.016872603323391563,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4998915303576478e-06,
+      "loss": 0.8415,
+      "step": 198
+    },
+    {
+      "epoch": 0.01695781849169152,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4998893054301014e-06,
+      "loss": 1.3295,
+      "step": 199
+    },
+    {
+      "epoch": 0.017043033659991477,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4998870579157984e-06,
+      "loss": 0.9835,
+      "step": 200
+    },
+    {
+      "epoch": 0.017128248828291438,
+      "grad_norm": 21.75,
+      "learning_rate": 2.4998847878147796e-06,
+      "loss": 1.1561,
+      "step": 201
+    },
+    {
+      "epoch": 0.017213463996591395,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.499882495127086e-06,
+      "loss": 0.5615,
+      "step": 202
+    },
+    {
+      "epoch": 0.01729867916489135,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.499880179852759e-06,
+      "loss": 0.8957,
+      "step": 203
+    },
+    {
+      "epoch": 0.01738389433319131,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4998778419918395e-06,
+      "loss": 0.6773,
+      "step": 204
+    },
+    {
+      "epoch": 0.017469109501491265,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4998754815443714e-06,
+      "loss": 0.9909,
+      "step": 205
+    },
+    {
+      "epoch": 0.017554324669791222,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.499873098510396e-06,
+      "loss": 0.8327,
+      "step": 206
+    },
+    {
+      "epoch": 0.01763953983809118,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4998706928899567e-06,
+      "loss": 0.7118,
+      "step": 207
+    },
+    {
+      "epoch": 0.017724755006391136,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4998682646830974e-06,
+      "loss": 0.6673,
+      "step": 208
+    },
+    {
+      "epoch": 0.017809970174691096,
+      "grad_norm": 19.75,
+      "learning_rate": 2.499865813889862e-06,
+      "loss": 0.6474,
+      "step": 209
+    },
+    {
+      "epoch": 0.017895185342991053,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.499863340510294e-06,
+      "loss": 0.6667,
+      "step": 210
+    },
+    {
+      "epoch": 0.01798040051129101,
+      "grad_norm": 35.5,
+      "learning_rate": 2.4998608445444387e-06,
+      "loss": 0.7577,
+      "step": 211
+    },
+    {
+      "epoch": 0.018065615679590967,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4998583259923413e-06,
+      "loss": 0.5801,
+      "step": 212
+    },
+    {
+      "epoch": 0.018150830847890924,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4998557848540473e-06,
+      "loss": 0.6952,
+      "step": 213
+    },
+    {
+      "epoch": 0.01823604601619088,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4998532211296024e-06,
+      "loss": 0.9168,
+      "step": 214
+    },
+    {
+      "epoch": 0.018321261184490838,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4998506348190523e-06,
+      "loss": 0.7822,
+      "step": 215
+    },
+    {
+      "epoch": 0.018406476352790798,
+      "grad_norm": 19.25,
+      "learning_rate": 2.499848025922445e-06,
+      "loss": 0.7619,
+      "step": 216
+    },
+    {
+      "epoch": 0.018491691521090755,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4998453944398273e-06,
+      "loss": 0.83,
+      "step": 217
+    },
+    {
+      "epoch": 0.018576906689390712,
+      "grad_norm": 22.375,
+      "learning_rate": 2.499842740371246e-06,
+      "loss": 1.004,
+      "step": 218
+    },
+    {
+      "epoch": 0.01866212185769067,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4998400637167495e-06,
+      "loss": 0.6864,
+      "step": 219
+    },
+    {
+      "epoch": 0.018747337025990626,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4998373644763864e-06,
+      "loss": 0.765,
+      "step": 220
+    },
+    {
+      "epoch": 0.018832552194290583,
+      "grad_norm": 23.5,
+      "learning_rate": 2.4998346426502052e-06,
+      "loss": 1.162,
+      "step": 221
+    },
+    {
+      "epoch": 0.01891776736259054,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4998318982382556e-06,
+      "loss": 0.6497,
+      "step": 222
+    },
+    {
+      "epoch": 0.0190029825308905,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4998291312405867e-06,
+      "loss": 0.556,
+      "step": 223
+    },
+    {
+      "epoch": 0.019088197699190457,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4998263416572484e-06,
+      "loss": 0.8032,
+      "step": 224
+    },
+    {
+      "epoch": 0.019173412867490414,
+      "grad_norm": 18.75,
+      "learning_rate": 2.499823529488291e-06,
+      "loss": 0.7872,
+      "step": 225
+    },
+    {
+      "epoch": 0.01925862803579037,
+      "grad_norm": 16.75,
+      "learning_rate": 2.499820694733766e-06,
+      "loss": 0.7131,
+      "step": 226
+    },
+    {
+      "epoch": 0.019343843204090327,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.499817837393724e-06,
+      "loss": 0.7218,
+      "step": 227
+    },
+    {
+      "epoch": 0.019429058372390284,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4998149574682175e-06,
+      "loss": 0.8036,
+      "step": 228
+    },
+    {
+      "epoch": 0.01951427354069024,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4998120549572972e-06,
+      "loss": 0.9664,
+      "step": 229
+    },
+    {
+      "epoch": 0.0195994887089902,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4998091298610168e-06,
+      "loss": 0.8357,
+      "step": 230
+    },
+    {
+      "epoch": 0.01968470387729016,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4998061821794282e-06,
+      "loss": 0.7416,
+      "step": 231
+    },
+    {
+      "epoch": 0.019769919045590115,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4998032119125858e-06,
+      "loss": 0.5895,
+      "step": 232
+    },
+    {
+      "epoch": 0.019855134213890072,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4998002190605423e-06,
+      "loss": 0.8778,
+      "step": 233
+    },
+    {
+      "epoch": 0.01994034938219003,
+      "grad_norm": 18.625,
+      "learning_rate": 2.499797203623352e-06,
+      "loss": 0.5428,
+      "step": 234
+    },
+    {
+      "epoch": 0.020025564550489986,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4997941656010695e-06,
+      "loss": 1.2933,
+      "step": 235
+    },
+    {
+      "epoch": 0.020110779718789943,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4997911049937493e-06,
+      "loss": 0.7574,
+      "step": 236
+    },
+    {
+      "epoch": 0.020195994887089903,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4997880218014476e-06,
+      "loss": 1.1304,
+      "step": 237
+    },
+    {
+      "epoch": 0.02028121005538986,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.499784916024219e-06,
+      "loss": 0.7651,
+      "step": 238
+    },
+    {
+      "epoch": 0.020366425223689817,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4997817876621204e-06,
+      "loss": 0.8226,
+      "step": 239
+    },
+    {
+      "epoch": 0.020451640391989774,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4997786367152086e-06,
+      "loss": 0.9215,
+      "step": 240
+    },
+    {
+      "epoch": 0.02053685556028973,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4997754631835396e-06,
+      "loss": 1.1801,
+      "step": 241
+    },
+    {
+      "epoch": 0.020622070728589688,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4997722670671715e-06,
+      "loss": 0.8299,
+      "step": 242
+    },
+    {
+      "epoch": 0.020707285896889645,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4997690483661615e-06,
+      "loss": 0.9563,
+      "step": 243
+    },
+    {
+      "epoch": 0.020792501065189605,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.4997658070805684e-06,
+      "loss": 0.5569,
+      "step": 244
+    },
+    {
+      "epoch": 0.020877716233489562,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4997625432104504e-06,
+      "loss": 0.9559,
+      "step": 245
+    },
+    {
+      "epoch": 0.02096293140178952,
+      "grad_norm": 45.75,
+      "learning_rate": 2.499759256755866e-06,
+      "loss": 0.9006,
+      "step": 246
+    },
+    {
+      "epoch": 0.021048146570089476,
+      "grad_norm": 18.875,
+      "learning_rate": 2.499755947716876e-06,
+      "loss": 1.1924,
+      "step": 247
+    },
+    {
+      "epoch": 0.021133361738389433,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4997526160935385e-06,
+      "loss": 0.8332,
+      "step": 248
+    },
+    {
+      "epoch": 0.02121857690668939,
+      "grad_norm": 19.375,
+      "learning_rate": 2.499749261885915e-06,
+      "loss": 1.0387,
+      "step": 249
+    },
+    {
+      "epoch": 0.021303792074989347,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4997458850940654e-06,
+      "loss": 1.1405,
+      "step": 250
+    },
+    {
+      "epoch": 0.021389007243289307,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.499742485718051e-06,
+      "loss": 0.5173,
+      "step": 251
+    },
+    {
+      "epoch": 0.021474222411589264,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4997390637579326e-06,
+      "loss": 0.9637,
+      "step": 252
+    },
+    {
+      "epoch": 0.02155943757988922,
+      "grad_norm": 21.75,
+      "learning_rate": 2.4997356192137734e-06,
+      "loss": 1.2523,
+      "step": 253
+    },
+    {
+      "epoch": 0.021644652748189178,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4997321520856347e-06,
+      "loss": 1.0835,
+      "step": 254
+    },
+    {
+      "epoch": 0.021729867916489135,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4997286623735788e-06,
+      "loss": 0.8121,
+      "step": 255
+    },
+    {
+      "epoch": 0.02181508308478909,
+      "grad_norm": 17.125,
+      "learning_rate": 2.49972515007767e-06,
+      "loss": 1.1225,
+      "step": 256
+    },
+    {
+      "epoch": 0.02190029825308905,
+      "grad_norm": 9.75,
+      "learning_rate": 2.499721615197971e-06,
+      "loss": 0.3624,
+      "step": 257
+    },
+    {
+      "epoch": 0.02198551342138901,
+      "grad_norm": 22.125,
+      "learning_rate": 2.499718057734545e-06,
+      "loss": 1.3517,
+      "step": 258
+    },
+    {
+      "epoch": 0.022070728589688966,
+      "grad_norm": 23.625,
+      "learning_rate": 2.4997144776874577e-06,
+      "loss": 0.867,
+      "step": 259
+    },
+    {
+      "epoch": 0.022155943757988922,
+      "grad_norm": 21.75,
+      "learning_rate": 2.499710875056773e-06,
+      "loss": 1.0509,
+      "step": 260
+    },
+    {
+      "epoch": 0.02224115892628888,
+      "grad_norm": 39.75,
+      "learning_rate": 2.4997072498425554e-06,
+      "loss": 1.6689,
+      "step": 261
+    },
+    {
+      "epoch": 0.022326374094588836,
+      "grad_norm": 36.0,
+      "learning_rate": 2.4997036020448717e-06,
+      "loss": 1.2132,
+      "step": 262
+    },
+    {
+      "epoch": 0.022411589262888793,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4996999316637874e-06,
+      "loss": 0.7928,
+      "step": 263
+    },
+    {
+      "epoch": 0.02249680443118875,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4996962386993685e-06,
+      "loss": 0.8249,
+      "step": 264
+    },
+    {
+      "epoch": 0.02258201959948871,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4996925231516816e-06,
+      "loss": 0.744,
+      "step": 265
+    },
+    {
+      "epoch": 0.022667234767788667,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4996887850207945e-06,
+      "loss": 0.7313,
+      "step": 266
+    },
+    {
+      "epoch": 0.022752449936088624,
+      "grad_norm": 9.0625,
+      "learning_rate": 2.4996850243067744e-06,
+      "loss": 0.3756,
+      "step": 267
+    },
+    {
+      "epoch": 0.02283766510438858,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4996812410096892e-06,
+      "loss": 0.5983,
+      "step": 268
+    },
+    {
+      "epoch": 0.022922880272688538,
+      "grad_norm": 11.875,
+      "learning_rate": 2.499677435129607e-06,
+      "loss": 0.4525,
+      "step": 269
+    },
+    {
+      "epoch": 0.023008095440988495,
+      "grad_norm": 43.5,
+      "learning_rate": 2.4996736066665973e-06,
+      "loss": 1.4105,
+      "step": 270
+    },
+    {
+      "epoch": 0.023093310609288452,
+      "grad_norm": 12.375,
+      "learning_rate": 2.499669755620729e-06,
+      "loss": 1.0042,
+      "step": 271
+    },
+    {
+      "epoch": 0.023178525777588412,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.499665881992071e-06,
+      "loss": 1.0983,
+      "step": 272
+    },
+    {
+      "epoch": 0.02326374094588837,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4996619857806946e-06,
+      "loss": 1.0489,
+      "step": 273
+    },
+    {
+      "epoch": 0.023348956114188326,
+      "grad_norm": 14.25,
+      "learning_rate": 2.499658066986669e-06,
+      "loss": 0.9086,
+      "step": 274
+    },
+    {
+      "epoch": 0.023434171282488283,
+      "grad_norm": 29.5,
+      "learning_rate": 2.499654125610065e-06,
+      "loss": 1.3386,
+      "step": 275
+    },
+    {
+      "epoch": 0.02351938645078824,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4996501616509553e-06,
+      "loss": 1.0422,
+      "step": 276
+    },
+    {
+      "epoch": 0.023604601619088197,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4996461751094105e-06,
+      "loss": 0.8694,
+      "step": 277
+    },
+    {
+      "epoch": 0.023689816787388154,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4996421659855022e-06,
+      "loss": 0.9185,
+      "step": 278
+    },
+    {
+      "epoch": 0.023775031955688114,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4996381342793034e-06,
+      "loss": 0.7571,
+      "step": 279
+    },
+    {
+      "epoch": 0.02386024712398807,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4996340799908873e-06,
+      "loss": 0.8722,
+      "step": 280
+    },
+    {
+      "epoch": 0.023945462292288028,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4996300031203264e-06,
+      "loss": 0.6721,
+      "step": 281
+    },
+    {
+      "epoch": 0.024030677460587985,
+      "grad_norm": 21.25,
+      "learning_rate": 2.499625903667695e-06,
+      "loss": 0.7462,
+      "step": 282
+    },
+    {
+      "epoch": 0.02411589262888794,
+      "grad_norm": 11.5,
+      "learning_rate": 2.499621781633067e-06,
+      "loss": 0.5317,
+      "step": 283
+    },
+    {
+      "epoch": 0.0242011077971879,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4996176370165164e-06,
+      "loss": 0.7925,
+      "step": 284
+    },
+    {
+      "epoch": 0.024286322965487855,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4996134698181184e-06,
+      "loss": 1.1314,
+      "step": 285
+    },
+    {
+      "epoch": 0.024371538133787816,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.499609280037949e-06,
+      "loss": 0.8216,
+      "step": 286
+    },
+    {
+      "epoch": 0.024456753302087773,
+      "grad_norm": 19.75,
+      "learning_rate": 2.499605067676083e-06,
+      "loss": 1.0857,
+      "step": 287
+    },
+    {
+      "epoch": 0.02454196847038773,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.499600832732597e-06,
+      "loss": 0.6202,
+      "step": 288
+    },
+    {
+      "epoch": 0.024627183638687686,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4995965752075673e-06,
+      "loss": 1.033,
+      "step": 289
+    },
+    {
+      "epoch": 0.024712398806987643,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.499592295101071e-06,
+      "loss": 0.648,
+      "step": 290
+    },
+    {
+      "epoch": 0.0247976139752876,
+      "grad_norm": 18.75,
+      "learning_rate": 2.499587992413185e-06,
+      "loss": 1.016,
+      "step": 291
+    },
+    {
+      "epoch": 0.024882829143587557,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.499583667143988e-06,
+      "loss": 0.5957,
+      "step": 292
+    },
+    {
+      "epoch": 0.024968044311887518,
+      "grad_norm": 21.875,
+      "learning_rate": 2.499579319293557e-06,
+      "loss": 0.9271,
+      "step": 293
+    },
+    {
+      "epoch": 0.025053259480187474,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4995749488619716e-06,
+      "loss": 0.7341,
+      "step": 294
+    },
+    {
+      "epoch": 0.02513847464848743,
+      "grad_norm": 14.375,
+      "learning_rate": 2.49957055584931e-06,
+      "loss": 0.8839,
+      "step": 295
+    },
+    {
+      "epoch": 0.025223689816787388,
+      "grad_norm": 22.25,
+      "learning_rate": 2.499566140255652e-06,
+      "loss": 1.6852,
+      "step": 296
+    },
+    {
+      "epoch": 0.025308904985087345,
+      "grad_norm": 26.5,
+      "learning_rate": 2.4995617020810775e-06,
+      "loss": 0.7103,
+      "step": 297
+    },
+    {
+      "epoch": 0.025394120153387302,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4995572413256663e-06,
+      "loss": 0.8754,
+      "step": 298
+    },
+    {
+      "epoch": 0.02547933532168726,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4995527579894994e-06,
+      "loss": 0.9294,
+      "step": 299
+    },
+    {
+      "epoch": 0.02556455048998722,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4995482520726573e-06,
+      "loss": 0.6675,
+      "step": 300
+    },
+    {
+      "epoch": 0.025649765658287176,
+      "grad_norm": 18.375,
+      "learning_rate": 2.499543723575222e-06,
+      "loss": 0.9184,
+      "step": 301
+    },
+    {
+      "epoch": 0.025734980826587133,
+      "grad_norm": 21.125,
+      "learning_rate": 2.499539172497275e-06,
+      "loss": 0.9095,
+      "step": 302
+    },
+    {
+      "epoch": 0.02582019599488709,
+      "grad_norm": 13.625,
+      "learning_rate": 2.499534598838899e-06,
+      "loss": 0.9908,
+      "step": 303
+    },
+    {
+      "epoch": 0.025905411163187047,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4995300026001764e-06,
+      "loss": 1.0894,
+      "step": 304
+    },
+    {
+      "epoch": 0.025990626331487004,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4995253837811895e-06,
+      "loss": 0.7298,
+      "step": 305
+    },
+    {
+      "epoch": 0.02607584149978696,
+      "grad_norm": 16.5,
+      "learning_rate": 2.499520742382023e-06,
+      "loss": 0.8227,
+      "step": 306
+    },
+    {
+      "epoch": 0.02616105666808692,
+      "grad_norm": 22.375,
+      "learning_rate": 2.4995160784027596e-06,
+      "loss": 1.1159,
+      "step": 307
+    },
+    {
+      "epoch": 0.026246271836386878,
+      "grad_norm": 12.75,
+      "learning_rate": 2.499511391843484e-06,
+      "loss": 0.5054,
+      "step": 308
+    },
+    {
+      "epoch": 0.026331487004686835,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4995066827042823e-06,
+      "loss": 0.8779,
+      "step": 309
+    },
+    {
+      "epoch": 0.026416702172986792,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4995019509852374e-06,
+      "loss": 0.5208,
+      "step": 310
+    },
+    {
+      "epoch": 0.02650191734128675,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4994971966864363e-06,
+      "loss": 0.9356,
+      "step": 311
+    },
+    {
+      "epoch": 0.026587132509586706,
+      "grad_norm": 21.125,
+      "learning_rate": 2.499492419807964e-06,
+      "loss": 0.8451,
+      "step": 312
+    },
+    {
+      "epoch": 0.026672347677886662,
+      "grad_norm": 15.375,
+      "learning_rate": 2.499487620349907e-06,
+      "loss": 0.9195,
+      "step": 313
+    },
+    {
+      "epoch": 0.026757562846186623,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4994827983123527e-06,
+      "loss": 0.8463,
+      "step": 314
+    },
+    {
+      "epoch": 0.02684277801448658,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4994779536953877e-06,
+      "loss": 0.8846,
+      "step": 315
+    },
+    {
+      "epoch": 0.026927993182786537,
+      "grad_norm": 23.625,
+      "learning_rate": 2.4994730864990997e-06,
+      "loss": 1.1352,
+      "step": 316
+    },
+    {
+      "epoch": 0.027013208351086494,
+      "grad_norm": 27.5,
+      "learning_rate": 2.4994681967235764e-06,
+      "loss": 1.2751,
+      "step": 317
+    },
+    {
+      "epoch": 0.02709842351938645,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4994632843689063e-06,
+      "loss": 0.6847,
+      "step": 318
+    },
+    {
+      "epoch": 0.027183638687686407,
+      "grad_norm": 29.25,
+      "learning_rate": 2.4994583494351784e-06,
+      "loss": 1.1252,
+      "step": 319
+    },
+    {
+      "epoch": 0.027268853855986364,
+      "grad_norm": 24.625,
+      "learning_rate": 2.4994533919224815e-06,
+      "loss": 0.9692,
+      "step": 320
+    },
+    {
+      "epoch": 0.027354069024286325,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4994484118309056e-06,
+      "loss": 0.6664,
+      "step": 321
+    },
+    {
+      "epoch": 0.02743928419258628,
+      "grad_norm": 20.375,
+      "learning_rate": 2.4994434091605402e-06,
+      "loss": 0.9731,
+      "step": 322
+    },
+    {
+      "epoch": 0.02752449936088624,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4994383839114762e-06,
+      "loss": 0.8914,
+      "step": 323
+    },
+    {
+      "epoch": 0.027609714529186195,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.499433336083804e-06,
+      "loss": 0.7852,
+      "step": 324
+    },
+    {
+      "epoch": 0.027694929697486152,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4994282656776154e-06,
+      "loss": 0.5759,
+      "step": 325
+    },
+    {
+      "epoch": 0.02778014486578611,
+      "grad_norm": 21.5,
+      "learning_rate": 2.4994231726930012e-06,
+      "loss": 0.8376,
+      "step": 326
+    },
+    {
+      "epoch": 0.027865360034086066,
+      "grad_norm": 21.125,
+      "learning_rate": 2.4994180571300546e-06,
+      "loss": 1.1522,
+      "step": 327
+    },
+    {
+      "epoch": 0.027950575202386026,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4994129189888667e-06,
+      "loss": 0.6931,
+      "step": 328
+    },
+    {
+      "epoch": 0.028035790370685983,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4994077582695307e-06,
+      "loss": 0.9413,
+      "step": 329
+    },
+    {
+      "epoch": 0.02812100553898594,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.4994025749721405e-06,
+      "loss": 0.7517,
+      "step": 330
+    },
+    {
+      "epoch": 0.028206220707285897,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.49939736909679e-06,
+      "loss": 0.622,
+      "step": 331
+    },
+    {
+      "epoch": 0.028291435875585854,
+      "grad_norm": 20.625,
+      "learning_rate": 2.499392140643572e-06,
+      "loss": 1.185,
+      "step": 332
+    },
+    {
+      "epoch": 0.02837665104388581,
+      "grad_norm": 22.5,
+      "learning_rate": 2.4993868896125818e-06,
+      "loss": 1.1348,
+      "step": 333
+    },
+    {
+      "epoch": 0.028461866212185768,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4993816160039137e-06,
+      "loss": 0.5254,
+      "step": 334
+    },
+    {
+      "epoch": 0.028547081380485728,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.499376319817664e-06,
+      "loss": 1.1253,
+      "step": 335
+    },
+    {
+      "epoch": 0.028632296548785685,
+      "grad_norm": 17.625,
+      "learning_rate": 2.499371001053928e-06,
+      "loss": 1.0347,
+      "step": 336
+    },
+    {
+      "epoch": 0.028717511717085642,
+      "grad_norm": 11.25,
+      "learning_rate": 2.499365659712801e-06,
+      "loss": 0.5701,
+      "step": 337
+    },
+    {
+      "epoch": 0.0288027268853856,
+      "grad_norm": 24.875,
+      "learning_rate": 2.4993602957943808e-06,
+      "loss": 1.1916,
+      "step": 338
+    },
+    {
+      "epoch": 0.028887942053685556,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4993549092987633e-06,
+      "loss": 0.6434,
+      "step": 339
+    },
+    {
+      "epoch": 0.028973157221985513,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4993495002260463e-06,
+      "loss": 0.8653,
+      "step": 340
+    },
+    {
+      "epoch": 0.02905837239028547,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4993440685763277e-06,
+      "loss": 0.5834,
+      "step": 341
+    },
+    {
+      "epoch": 0.029143587558585426,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.4993386143497056e-06,
+      "loss": 0.5358,
+      "step": 342
+    },
+    {
+      "epoch": 0.029228802726885387,
+      "grad_norm": 19.75,
+      "learning_rate": 2.499333137546278e-06,
+      "loss": 0.9414,
+      "step": 343
+    },
+    {
+      "epoch": 0.029314017895185344,
+      "grad_norm": 21.75,
+      "learning_rate": 2.499327638166145e-06,
+      "loss": 1.1704,
+      "step": 344
+    },
+    {
+      "epoch": 0.0293992330634853,
+      "grad_norm": 22.875,
+      "learning_rate": 2.499322116209405e-06,
+      "loss": 1.2015,
+      "step": 345
+    },
+    {
+      "epoch": 0.029484448231785258,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4993165716761576e-06,
+      "loss": 0.8069,
+      "step": 346
+    },
+    {
+      "epoch": 0.029569663400085214,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.4993110045665037e-06,
+      "loss": 0.4925,
+      "step": 347
+    },
+    {
+      "epoch": 0.02965487856838517,
+      "grad_norm": 17.0,
+      "learning_rate": 2.499305414880544e-06,
+      "loss": 0.8036,
+      "step": 348
+    },
+    {
+      "epoch": 0.029740093736685128,
+      "grad_norm": 13.125,
+      "learning_rate": 2.499299802618379e-06,
+      "loss": 0.6162,
+      "step": 349
+    },
+    {
+      "epoch": 0.02982530890498509,
+      "grad_norm": 16.5,
+      "learning_rate": 2.49929416778011e-06,
+      "loss": 0.7732,
+      "step": 350
+    },
+    {
+      "epoch": 0.029910524073285046,
+      "grad_norm": 23.0,
+      "learning_rate": 2.4992885103658394e-06,
+      "loss": 1.1847,
+      "step": 351
+    },
+    {
+      "epoch": 0.029995739241585002,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4992828303756694e-06,
+      "loss": 0.6363,
+      "step": 352
+    },
+    {
+      "epoch": 0.03008095440988496,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4992771278097017e-06,
+      "loss": 0.486,
+      "step": 353
+    },
+    {
+      "epoch": 0.030166169578184916,
+      "grad_norm": 14.125,
+      "learning_rate": 2.499271402668041e-06,
+      "loss": 1.2398,
+      "step": 354
+    },
+    {
+      "epoch": 0.030251384746484873,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.499265654950789e-06,
+      "loss": 0.8022,
+      "step": 355
+    },
+    {
+      "epoch": 0.03033659991478483,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4992598846580507e-06,
+      "loss": 0.9641,
+      "step": 356
+    },
+    {
+      "epoch": 0.03042181508308479,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4992540917899298e-06,
+      "loss": 0.7756,
+      "step": 357
+    },
+    {
+      "epoch": 0.030507030251384747,
+      "grad_norm": 16.125,
+      "learning_rate": 2.499248276346532e-06,
+      "loss": 1.0406,
+      "step": 358
+    },
+    {
+      "epoch": 0.030592245419684704,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.499242438327961e-06,
+      "loss": 0.6911,
+      "step": 359
+    },
+    {
+      "epoch": 0.03067746058798466,
+      "grad_norm": 18.125,
+      "learning_rate": 2.499236577734323e-06,
+      "loss": 0.5916,
+      "step": 360
+    },
+    {
+      "epoch": 0.030762675756284618,
+      "grad_norm": 16.0,
+      "learning_rate": 2.499230694565724e-06,
+      "loss": 0.9698,
+      "step": 361
+    },
+    {
+      "epoch": 0.030847890924584575,
+      "grad_norm": 15.5,
+      "learning_rate": 2.49922478882227e-06,
+      "loss": 1.1002,
+      "step": 362
+    },
+    {
+      "epoch": 0.030933106092884532,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.499218860504068e-06,
+      "loss": 0.8531,
+      "step": 363
+    },
+    {
+      "epoch": 0.031018321261184492,
+      "grad_norm": 22.5,
+      "learning_rate": 2.4992129096112248e-06,
+      "loss": 1.0199,
+      "step": 364
+    },
+    {
+      "epoch": 0.03110353642948445,
+      "grad_norm": 15.375,
+      "learning_rate": 2.499206936143848e-06,
+      "loss": 0.65,
+      "step": 365
+    },
+    {
+      "epoch": 0.031188751597784406,
+      "grad_norm": 19.5,
+      "learning_rate": 2.499200940102046e-06,
+      "loss": 0.8134,
+      "step": 366
+    },
+    {
+      "epoch": 0.03127396676608436,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.499194921485927e-06,
+      "loss": 0.9027,
+      "step": 367
+    },
+    {
+      "epoch": 0.03135918193438432,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4991888802955993e-06,
+      "loss": 0.5978,
+      "step": 368
+    },
+    {
+      "epoch": 0.03144439710268428,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4991828165311723e-06,
+      "loss": 0.5651,
+      "step": 369
+    },
+    {
+      "epoch": 0.031529612270984234,
+      "grad_norm": 20.625,
+      "learning_rate": 2.499176730192756e-06,
+      "loss": 1.0047,
+      "step": 370
+    },
+    {
+      "epoch": 0.03161482743928419,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4991706212804596e-06,
+      "loss": 0.9238,
+      "step": 371
+    },
+    {
+      "epoch": 0.03170004260758415,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4991644897943945e-06,
+      "loss": 0.6923,
+      "step": 372
+    },
+    {
+      "epoch": 0.031785257775884104,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4991583357346705e-06,
+      "loss": 0.7106,
+      "step": 373
+    },
+    {
+      "epoch": 0.03187047294418407,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4991521591013995e-06,
+      "loss": 0.4967,
+      "step": 374
+    },
+    {
+      "epoch": 0.031955688112484025,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.499145959894693e-06,
+      "loss": 0.9125,
+      "step": 375
+    },
+    {
+      "epoch": 0.03204090328078398,
+      "grad_norm": 17.0,
+      "learning_rate": 2.499139738114662e-06,
+      "loss": 0.9366,
+      "step": 376
+    },
+    {
+      "epoch": 0.03212611844908394,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4991334937614206e-06,
+      "loss": 1.1503,
+      "step": 377
+    },
+    {
+      "epoch": 0.032211333617383896,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4991272268350807e-06,
+      "loss": 0.7203,
+      "step": 378
+    },
+    {
+      "epoch": 0.03229654878568385,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4991209373357558e-06,
+      "loss": 0.6034,
+      "step": 379
+    },
+    {
+      "epoch": 0.03238176395398381,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4991146252635593e-06,
+      "loss": 0.5002,
+      "step": 380
+    },
+    {
+      "epoch": 0.032466979122283766,
+      "grad_norm": 32.25,
+      "learning_rate": 2.4991082906186057e-06,
+      "loss": 1.7323,
+      "step": 381
+    },
+    {
+      "epoch": 0.03255219429058372,
+      "grad_norm": 25.25,
+      "learning_rate": 2.499101933401009e-06,
+      "loss": 1.4725,
+      "step": 382
+    },
+    {
+      "epoch": 0.03263740945888368,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4990955536108846e-06,
+      "loss": 0.7112,
+      "step": 383
+    },
+    {
+      "epoch": 0.03272262462718364,
+      "grad_norm": 20.75,
+      "learning_rate": 2.499089151248347e-06,
+      "loss": 0.9326,
+      "step": 384
+    },
+    {
+      "epoch": 0.032807839795483594,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4990827263135125e-06,
+      "loss": 1.218,
+      "step": 385
+    },
+    {
+      "epoch": 0.03289305496378355,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.499076278806498e-06,
+      "loss": 0.8987,
+      "step": 386
+    },
+    {
+      "epoch": 0.03297827013208351,
+      "grad_norm": 13.375,
+      "learning_rate": 2.499069808727418e-06,
+      "loss": 0.5982,
+      "step": 387
+    },
+    {
+      "epoch": 0.03306348530038347,
+      "grad_norm": 21.375,
+      "learning_rate": 2.4990633160763904e-06,
+      "loss": 1.0252,
+      "step": 388
+    },
+    {
+      "epoch": 0.03314870046868343,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.499056800853533e-06,
+      "loss": 0.869,
+      "step": 389
+    },
+    {
+      "epoch": 0.033233915636983385,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4990502630589637e-06,
+      "loss": 0.5836,
+      "step": 390
+    },
+    {
+      "epoch": 0.03331913080528334,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4990437026927993e-06,
+      "loss": 1.0526,
+      "step": 391
+    },
+    {
+      "epoch": 0.0334043459735833,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.49903711975516e-06,
+      "loss": 0.5121,
+      "step": 392
+    },
+    {
+      "epoch": 0.033489561141883256,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4990305142461635e-06,
+      "loss": 0.7733,
+      "step": 393
+    },
+    {
+      "epoch": 0.03357477631018321,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4990238861659293e-06,
+      "loss": 0.6252,
+      "step": 394
+    },
+    {
+      "epoch": 0.03365999147848317,
+      "grad_norm": 13.5,
+      "learning_rate": 2.499017235514578e-06,
+      "loss": 0.7226,
+      "step": 395
+    },
+    {
+      "epoch": 0.03374520664678313,
+      "grad_norm": 20.875,
+      "learning_rate": 2.499010562292229e-06,
+      "loss": 1.1063,
+      "step": 396
+    },
+    {
+      "epoch": 0.033830421815083084,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4990038664990036e-06,
+      "loss": 0.523,
+      "step": 397
+    },
+    {
+      "epoch": 0.03391563698338304,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4989971481350217e-06,
+      "loss": 1.1988,
+      "step": 398
+    },
+    {
+      "epoch": 0.034000852151683,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4989904072004056e-06,
+      "loss": 0.6412,
+      "step": 399
+    },
+    {
+      "epoch": 0.034086067319982954,
+      "grad_norm": 21.5,
+      "learning_rate": 2.498983643695277e-06,
+      "loss": 1.3073,
+      "step": 400
+    },
+    {
+      "epoch": 0.03417128248828291,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4989768576197583e-06,
+      "loss": 0.8451,
+      "step": 401
+    },
+    {
+      "epoch": 0.034256497656582875,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4989700489739713e-06,
+      "loss": 0.6877,
+      "step": 402
+    },
+    {
+      "epoch": 0.03434171282488283,
+      "grad_norm": 10.75,
+      "learning_rate": 2.49896321775804e-06,
+      "loss": 0.4754,
+      "step": 403
+    },
+    {
+      "epoch": 0.03442692799318279,
+      "grad_norm": 17.875,
+      "learning_rate": 2.498956363972087e-06,
+      "loss": 0.9302,
+      "step": 404
+    },
+    {
+      "epoch": 0.034512143161482746,
+      "grad_norm": 31.25,
+      "learning_rate": 2.4989494876162368e-06,
+      "loss": 1.2954,
+      "step": 405
+    },
+    {
+      "epoch": 0.0345973583297827,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4989425886906133e-06,
+      "loss": 0.5317,
+      "step": 406
+    },
+    {
+      "epoch": 0.03468257349808266,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4989356671953417e-06,
+      "loss": 0.4213,
+      "step": 407
+    },
+    {
+      "epoch": 0.03476778866638262,
+      "grad_norm": 20.375,
+      "learning_rate": 2.4989287231305464e-06,
+      "loss": 0.9238,
+      "step": 408
+    },
+    {
+      "epoch": 0.034853003834682574,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.498921756496353e-06,
+      "loss": 0.5157,
+      "step": 409
+    },
+    {
+      "epoch": 0.03493821900298253,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4989147672928877e-06,
+      "loss": 1.048,
+      "step": 410
+    },
+    {
+      "epoch": 0.03502343417128249,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4989077555202766e-06,
+      "loss": 0.9328,
+      "step": 411
+    },
+    {
+      "epoch": 0.035108649339582444,
+      "grad_norm": 17.75,
+      "learning_rate": 2.498900721178647e-06,
+      "loss": 0.8179,
+      "step": 412
+    },
+    {
+      "epoch": 0.0351938645078824,
+      "grad_norm": 18.75,
+      "learning_rate": 2.498893664268125e-06,
+      "loss": 1.1035,
+      "step": 413
+    },
+    {
+      "epoch": 0.03527907967618236,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4988865847888388e-06,
+      "loss": 0.7153,
+      "step": 414
+    },
+    {
+      "epoch": 0.035364294844482315,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4988794827409165e-06,
+      "loss": 0.9215,
+      "step": 415
+    },
+    {
+      "epoch": 0.03544951001278227,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4988723581244857e-06,
+      "loss": 0.5237,
+      "step": 416
+    },
+    {
+      "epoch": 0.035534725181082236,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4988652109396756e-06,
+      "loss": 0.8331,
+      "step": 417
+    },
+    {
+      "epoch": 0.03561994034938219,
+      "grad_norm": 16.0,
+      "learning_rate": 2.498858041186615e-06,
+      "loss": 1.0079,
+      "step": 418
+    },
+    {
+      "epoch": 0.03570515551768215,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4988508488654342e-06,
+      "loss": 0.6557,
+      "step": 419
+    },
+    {
+      "epoch": 0.035790370685982106,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4988436339762625e-06,
+      "loss": 1.0191,
+      "step": 420
+    },
+    {
+      "epoch": 0.03587558585428206,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4988363965192308e-06,
+      "loss": 0.5937,
+      "step": 421
+    },
+    {
+      "epoch": 0.03596080102258202,
+      "grad_norm": 18.0,
+      "learning_rate": 2.4988291364944694e-06,
+      "loss": 1.2424,
+      "step": 422
+    },
+    {
+      "epoch": 0.03604601619088198,
+      "grad_norm": 21.625,
+      "learning_rate": 2.4988218539021097e-06,
+      "loss": 0.9282,
+      "step": 423
+    },
+    {
+      "epoch": 0.036131231359181934,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4988145487422833e-06,
+      "loss": 0.4837,
+      "step": 424
+    },
+    {
+      "epoch": 0.03621644652748189,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4988072210151224e-06,
+      "loss": 0.8311,
+      "step": 425
+    },
+    {
+      "epoch": 0.03630166169578185,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4987998707207592e-06,
+      "loss": 0.9997,
+      "step": 426
+    },
+    {
+      "epoch": 0.036386876864081805,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4987924978593263e-06,
+      "loss": 0.8165,
+      "step": 427
+    },
+    {
+      "epoch": 0.03647209203238176,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4987851024309577e-06,
+      "loss": 0.5936,
+      "step": 428
+    },
+    {
+      "epoch": 0.03655730720068172,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4987776844357862e-06,
+      "loss": 0.7518,
+      "step": 429
+    },
+    {
+      "epoch": 0.036642522368981675,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4987702438739464e-06,
+      "loss": 0.8977,
+      "step": 430
+    },
+    {
+      "epoch": 0.03672773753728164,
+      "grad_norm": 38.75,
+      "learning_rate": 2.498762780745572e-06,
+      "loss": 1.6056,
+      "step": 431
+    },
+    {
+      "epoch": 0.036812952705581596,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4987552950507994e-06,
+      "loss": 1.0696,
+      "step": 432
+    },
+    {
+      "epoch": 0.03689816787388155,
+      "grad_norm": 17.75,
+      "learning_rate": 2.498747786789762e-06,
+      "loss": 1.1169,
+      "step": 433
+    },
+    {
+      "epoch": 0.03698338304218151,
+      "grad_norm": 10.375,
+      "learning_rate": 2.4987402559625965e-06,
+      "loss": 0.4596,
+      "step": 434
+    },
+    {
+      "epoch": 0.03706859821048147,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4987327025694392e-06,
+      "loss": 0.9313,
+      "step": 435
+    },
+    {
+      "epoch": 0.037153813378781424,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4987251266104263e-06,
+      "loss": 0.7626,
+      "step": 436
+    },
+    {
+      "epoch": 0.03723902854708138,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4987175280856947e-06,
+      "loss": 0.5739,
+      "step": 437
+    },
+    {
+      "epoch": 0.03732424371538134,
+      "grad_norm": 30.625,
+      "learning_rate": 2.4987099069953815e-06,
+      "loss": 0.6925,
+      "step": 438
+    },
+    {
+      "epoch": 0.037409458883681294,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.498702263339625e-06,
+      "loss": 0.3508,
+      "step": 439
+    },
+    {
+      "epoch": 0.03749467405198125,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4986945971185623e-06,
+      "loss": 0.9168,
+      "step": 440
+    },
+    {
+      "epoch": 0.03757988922028121,
+      "grad_norm": 27.125,
+      "learning_rate": 2.4986869083323324e-06,
+      "loss": 0.8307,
+      "step": 441
+    },
+    {
+      "epoch": 0.037665104388581165,
+      "grad_norm": 13.375,
+      "learning_rate": 2.498679196981075e-06,
+      "loss": 0.7669,
+      "step": 442
+    },
+    {
+      "epoch": 0.03775031955688112,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4986714630649287e-06,
+      "loss": 1.4351,
+      "step": 443
+    },
+    {
+      "epoch": 0.03783553472518108,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4986637065840336e-06,
+      "loss": 1.2245,
+      "step": 444
+    },
+    {
+      "epoch": 0.03792074989348104,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4986559275385294e-06,
+      "loss": 0.7966,
+      "step": 445
+    },
+    {
+      "epoch": 0.038005965061781,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.4986481259285574e-06,
+      "loss": 0.6735,
+      "step": 446
+    },
+    {
+      "epoch": 0.03809118023008096,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4986403017542575e-06,
+      "loss": 0.8942,
+      "step": 447
+    },
+    {
+      "epoch": 0.03817639539838091,
+      "grad_norm": 14.75,
+      "learning_rate": 2.498632455015772e-06,
+      "loss": 0.8208,
+      "step": 448
+    },
+    {
+      "epoch": 0.03826161056668087,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4986245857132426e-06,
+      "loss": 0.6493,
+      "step": 449
+    },
+    {
+      "epoch": 0.03834682573498083,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4986166938468114e-06,
+      "loss": 0.9837,
+      "step": 450
+    },
+    {
+      "epoch": 0.038432040903280784,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4986087794166206e-06,
+      "loss": 0.8642,
+      "step": 451
+    },
+    {
+      "epoch": 0.03851725607158074,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4986008424228143e-06,
+      "loss": 0.9541,
+      "step": 452
+    },
+    {
+      "epoch": 0.0386024712398807,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4985928828655346e-06,
+      "loss": 0.7419,
+      "step": 453
+    },
+    {
+      "epoch": 0.038687686408180655,
+      "grad_norm": 17.375,
+      "learning_rate": 2.498584900744926e-06,
+      "loss": 0.8758,
+      "step": 454
+    },
+    {
+      "epoch": 0.03877290157648061,
+      "grad_norm": 13.875,
+      "learning_rate": 2.498576896061133e-06,
+      "loss": 0.7497,
+      "step": 455
+    },
+    {
+      "epoch": 0.03885811674478057,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4985688688142996e-06,
+      "loss": 0.9892,
+      "step": 456
+    },
+    {
+      "epoch": 0.038943331913080526,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4985608190045714e-06,
+      "loss": 0.6137,
+      "step": 457
+    },
+    {
+      "epoch": 0.03902854708138048,
+      "grad_norm": 10.625,
+      "learning_rate": 2.4985527466320937e-06,
+      "loss": 0.4528,
+      "step": 458
+    },
+    {
+      "epoch": 0.039113762249680446,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.4985446516970117e-06,
+      "loss": 0.5635,
+      "step": 459
+    },
+    {
+      "epoch": 0.0391989774179804,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4985365341994734e-06,
+      "loss": 0.9851,
+      "step": 460
+    },
+    {
+      "epoch": 0.03928419258628036,
+      "grad_norm": 24.625,
+      "learning_rate": 2.498528394139624e-06,
+      "loss": 1.373,
+      "step": 461
+    },
+    {
+      "epoch": 0.03936940775458032,
+      "grad_norm": 17.875,
+      "learning_rate": 2.498520231517611e-06,
+      "loss": 0.8514,
+      "step": 462
+    },
+    {
+      "epoch": 0.039454622922880274,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4985120463335816e-06,
+      "loss": 1.1372,
+      "step": 463
+    },
+    {
+      "epoch": 0.03953983809118023,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.498503838587684e-06,
+      "loss": 0.4535,
+      "step": 464
+    },
+    {
+      "epoch": 0.03962505325948019,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4984956082800673e-06,
+      "loss": 0.8392,
+      "step": 465
+    },
+    {
+      "epoch": 0.039710268427780145,
+      "grad_norm": 18.75,
+      "learning_rate": 2.498487355410879e-06,
+      "loss": 1.1626,
+      "step": 466
+    },
+    {
+      "epoch": 0.0397954835960801,
+      "grad_norm": 16.5,
+      "learning_rate": 2.498479079980269e-06,
+      "loss": 1.2061,
+      "step": 467
+    },
+    {
+      "epoch": 0.03988069876438006,
+      "grad_norm": 11.75,
+      "learning_rate": 2.4984707819883863e-06,
+      "loss": 0.6979,
+      "step": 468
+    },
+    {
+      "epoch": 0.039965913932680015,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4984624614353813e-06,
+      "loss": 0.9957,
+      "step": 469
+    },
+    {
+      "epoch": 0.04005112910097997,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4984541183214038e-06,
+      "loss": 0.6148,
+      "step": 470
+    },
+    {
+      "epoch": 0.04013634426927993,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4984457526466055e-06,
+      "loss": 0.614,
+      "step": 471
+    },
+    {
+      "epoch": 0.040221559437579886,
+      "grad_norm": 13.5,
+      "learning_rate": 2.498437364411137e-06,
+      "loss": 0.7254,
+      "step": 472
+    },
+    {
+      "epoch": 0.04030677460587985,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4984289536151496e-06,
+      "loss": 0.7426,
+      "step": 473
+    },
+    {
+      "epoch": 0.04039198977417981,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4984205202587957e-06,
+      "loss": 0.6586,
+      "step": 474
+    },
+    {
+      "epoch": 0.040477204942479764,
+      "grad_norm": 16.375,
+      "learning_rate": 2.498412064342228e-06,
+      "loss": 0.7571,
+      "step": 475
+    },
+    {
+      "epoch": 0.04056242011077972,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4984035858655985e-06,
+      "loss": 0.8438,
+      "step": 476
+    },
+    {
+      "epoch": 0.04064763527907968,
+      "grad_norm": 14.5,
+      "learning_rate": 2.498395084829061e-06,
+      "loss": 0.7932,
+      "step": 477
+    },
+    {
+      "epoch": 0.040732850447379634,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.498386561232769e-06,
+      "loss": 0.5459,
+      "step": 478
+    },
+    {
+      "epoch": 0.04081806561567959,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4983780150768763e-06,
+      "loss": 0.529,
+      "step": 479
+    },
+    {
+      "epoch": 0.04090328078397955,
+      "grad_norm": 13.625,
+      "learning_rate": 2.498369446361538e-06,
+      "loss": 0.6605,
+      "step": 480
+    },
+    {
+      "epoch": 0.040988495952279505,
+      "grad_norm": 10.125,
+      "learning_rate": 2.4983608550869078e-06,
+      "loss": 0.4761,
+      "step": 481
+    },
+    {
+      "epoch": 0.04107371112057946,
+      "grad_norm": 11.375,
+      "learning_rate": 2.498352241253142e-06,
+      "loss": 0.607,
+      "step": 482
+    },
+    {
+      "epoch": 0.04115892628887942,
+      "grad_norm": 21.75,
+      "learning_rate": 2.498343604860396e-06,
+      "loss": 0.7981,
+      "step": 483
+    },
+    {
+      "epoch": 0.041244141457179376,
+      "grad_norm": 16.0,
+      "learning_rate": 2.498334945908825e-06,
+      "loss": 1.15,
+      "step": 484
+    },
+    {
+      "epoch": 0.04132935662547933,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4983262643985873e-06,
+      "loss": 1.2257,
+      "step": 485
+    },
+    {
+      "epoch": 0.04141457179377929,
+      "grad_norm": 32.5,
+      "learning_rate": 2.498317560329838e-06,
+      "loss": 1.0988,
+      "step": 486
+    },
+    {
+      "epoch": 0.04149978696207925,
+      "grad_norm": 33.75,
+      "learning_rate": 2.4983088337027354e-06,
+      "loss": 1.1718,
+      "step": 487
+    },
+    {
+      "epoch": 0.04158500213037921,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4983000845174365e-06,
+      "loss": 0.9357,
+      "step": 488
+    },
+    {
+      "epoch": 0.04167021729867917,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4982913127741006e-06,
+      "loss": 1.2299,
+      "step": 489
+    },
+    {
+      "epoch": 0.041755432466979124,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4982825184728847e-06,
+      "loss": 0.6561,
+      "step": 490
+    },
+    {
+      "epoch": 0.04184064763527908,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4982737016139486e-06,
+      "loss": 1.0078,
+      "step": 491
+    },
+    {
+      "epoch": 0.04192586280357904,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.498264862197452e-06,
+      "loss": 0.7485,
+      "step": 492
+    },
+    {
+      "epoch": 0.042011077971878995,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4982560002235535e-06,
+      "loss": 1.1153,
+      "step": 493
+    },
+    {
+      "epoch": 0.04209629314017895,
+      "grad_norm": 19.25,
+      "learning_rate": 2.498247115692414e-06,
+      "loss": 1.0774,
+      "step": 494
+    },
+    {
+      "epoch": 0.04218150830847891,
+      "grad_norm": 14.5,
+      "learning_rate": 2.498238208604194e-06,
+      "loss": 0.6382,
+      "step": 495
+    },
+    {
+      "epoch": 0.042266723476778865,
+      "grad_norm": 12.625,
+      "learning_rate": 2.498229278959054e-06,
+      "loss": 0.72,
+      "step": 496
+    },
+    {
+      "epoch": 0.04235193864507882,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4982203267571563e-06,
+      "loss": 0.8994,
+      "step": 497
+    },
+    {
+      "epoch": 0.04243715381337878,
+      "grad_norm": 15.0,
+      "learning_rate": 2.498211351998662e-06,
+      "loss": 0.7694,
+      "step": 498
+    },
+    {
+      "epoch": 0.042522368981678736,
+      "grad_norm": 13.125,
+      "learning_rate": 2.498202354683733e-06,
+      "loss": 0.7086,
+      "step": 499
+    },
+    {
+      "epoch": 0.04260758414997869,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4981933348125327e-06,
+      "loss": 1.0981,
+      "step": 500
+    },
+    {
+      "epoch": 0.04269279931827866,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4981842923852232e-06,
+      "loss": 0.716,
+      "step": 501
+    },
+    {
+      "epoch": 0.042778014486578614,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.498175227401969e-06,
+      "loss": 0.8802,
+      "step": 502
+    },
+    {
+      "epoch": 0.04286322965487857,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.498166139862933e-06,
+      "loss": 0.613,
+      "step": 503
+    },
+    {
+      "epoch": 0.04294844482317853,
+      "grad_norm": 10.875,
+      "learning_rate": 2.49815702976828e-06,
+      "loss": 0.4915,
+      "step": 504
+    },
+    {
+      "epoch": 0.043033659991478485,
+      "grad_norm": 14.125,
+      "learning_rate": 2.498147897118174e-06,
+      "loss": 0.9676,
+      "step": 505
+    },
+    {
+      "epoch": 0.04311887515977844,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4981387419127806e-06,
+      "loss": 0.8427,
+      "step": 506
+    },
+    {
+      "epoch": 0.0432040903280784,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4981295641522647e-06,
+      "loss": 1.1045,
+      "step": 507
+    },
+    {
+      "epoch": 0.043289305496378355,
+      "grad_norm": 17.25,
+      "learning_rate": 2.498120363836793e-06,
+      "loss": 1.0553,
+      "step": 508
+    },
+    {
+      "epoch": 0.04337452066467831,
+      "grad_norm": 17.0,
+      "learning_rate": 2.498111140966531e-06,
+      "loss": 1.1626,
+      "step": 509
+    },
+    {
+      "epoch": 0.04345973583297827,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4981018955416454e-06,
+      "loss": 0.7456,
+      "step": 510
+    },
+    {
+      "epoch": 0.043544951001278226,
+      "grad_norm": 20.875,
+      "learning_rate": 2.498092627562304e-06,
+      "loss": 0.9689,
+      "step": 511
+    },
+    {
+      "epoch": 0.04363016616957818,
+      "grad_norm": 17.625,
+      "learning_rate": 2.498083337028673e-06,
+      "loss": 0.8882,
+      "step": 512
+    },
+    {
+      "epoch": 0.04371538133787814,
+      "grad_norm": 10.5,
+      "learning_rate": 2.4980740239409216e-06,
+      "loss": 0.5084,
+      "step": 513
+    },
+    {
+      "epoch": 0.0438005965061781,
+      "grad_norm": 13.875,
+      "learning_rate": 2.498064688299218e-06,
+      "loss": 0.8218,
+      "step": 514
+    },
+    {
+      "epoch": 0.04388581167447806,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4980553301037296e-06,
+      "loss": 0.4602,
+      "step": 515
+    },
+    {
+      "epoch": 0.04397102684277802,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4980459493546266e-06,
+      "loss": 1.0326,
+      "step": 516
+    },
+    {
+      "epoch": 0.044056242011077974,
+      "grad_norm": 31.25,
+      "learning_rate": 2.4980365460520784e-06,
+      "loss": 1.1643,
+      "step": 517
+    },
+    {
+      "epoch": 0.04414145717937793,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4980271201962547e-06,
+      "loss": 0.7769,
+      "step": 518
+    },
+    {
+      "epoch": 0.04422667234767789,
+      "grad_norm": 20.375,
+      "learning_rate": 2.498017671787326e-06,
+      "loss": 1.2848,
+      "step": 519
+    },
+    {
+      "epoch": 0.044311887515977845,
+      "grad_norm": 17.125,
+      "learning_rate": 2.498008200825463e-06,
+      "loss": 0.9698,
+      "step": 520
+    },
+    {
+      "epoch": 0.0443971026842778,
+      "grad_norm": 12.625,
+      "learning_rate": 2.497998707310837e-06,
+      "loss": 0.4529,
+      "step": 521
+    },
+    {
+      "epoch": 0.04448231785257776,
+      "grad_norm": 16.125,
+      "learning_rate": 2.497989191243619e-06,
+      "loss": 1.1708,
+      "step": 522
+    },
+    {
+      "epoch": 0.044567533020877716,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4979796526239816e-06,
+      "loss": 0.9436,
+      "step": 523
+    },
+    {
+      "epoch": 0.04465274818917767,
+      "grad_norm": 17.125,
+      "learning_rate": 2.497970091452097e-06,
+      "loss": 0.7137,
+      "step": 524
+    },
+    {
+      "epoch": 0.04473796335747763,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4979605077281375e-06,
+      "loss": 0.9704,
+      "step": 525
+    },
+    {
+      "epoch": 0.044823178525777586,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.497950901452277e-06,
+      "loss": 0.4724,
+      "step": 526
+    },
+    {
+      "epoch": 0.04490839369407754,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4979412726246894e-06,
+      "loss": 1.0186,
+      "step": 527
+    },
+    {
+      "epoch": 0.0449936088623775,
+      "grad_norm": 22.25,
+      "learning_rate": 2.4979316212455473e-06,
+      "loss": 1.0909,
+      "step": 528
+    },
+    {
+      "epoch": 0.045078824030677464,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4979219473150263e-06,
+      "loss": 0.5592,
+      "step": 529
+    },
+    {
+      "epoch": 0.04516403919897742,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4979122508333004e-06,
+      "loss": 0.678,
+      "step": 530
+    },
+    {
+      "epoch": 0.04524925436727738,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.497902531800546e-06,
+      "loss": 1.113,
+      "step": 531
+    },
+    {
+      "epoch": 0.045334469535577335,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.4978927902169376e-06,
+      "loss": 0.477,
+      "step": 532
+    },
+    {
+      "epoch": 0.04541968470387729,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4978830260826512e-06,
+      "loss": 1.0077,
+      "step": 533
+    },
+    {
+      "epoch": 0.04550489987217725,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4978732393978644e-06,
+      "loss": 1.0396,
+      "step": 534
+    },
+    {
+      "epoch": 0.045590115040477205,
+      "grad_norm": 15.875,
+      "learning_rate": 2.497863430162753e-06,
+      "loss": 1.0904,
+      "step": 535
+    },
+    {
+      "epoch": 0.04567533020877716,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4978535983774945e-06,
+      "loss": 0.5543,
+      "step": 536
+    },
+    {
+      "epoch": 0.04576054537707712,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4978437440422664e-06,
+      "loss": 0.5372,
+      "step": 537
+    },
+    {
+      "epoch": 0.045845760545377076,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4978338671572476e-06,
+      "loss": 0.6376,
+      "step": 538
+    },
+    {
+      "epoch": 0.04593097571367703,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4978239677226155e-06,
+      "loss": 0.9596,
+      "step": 539
+    },
+    {
+      "epoch": 0.04601619088197699,
+      "grad_norm": 28.875,
+      "learning_rate": 2.49781404573855e-06,
+      "loss": 1.2914,
+      "step": 540
+    },
+    {
+      "epoch": 0.04610140605027695,
+      "grad_norm": 16.0,
+      "learning_rate": 2.49780410120523e-06,
+      "loss": 0.8702,
+      "step": 541
+    },
+    {
+      "epoch": 0.046186621218576904,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4977941341228347e-06,
+      "loss": 0.6548,
+      "step": 542
+    },
+    {
+      "epoch": 0.04627183638687686,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4977841444915448e-06,
+      "loss": 0.8117,
+      "step": 543
+    },
+    {
+      "epoch": 0.046357051555176824,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4977741323115406e-06,
+      "loss": 0.4431,
+      "step": 544
+    },
+    {
+      "epoch": 0.04644226672347678,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.497764097583003e-06,
+      "loss": 1.0855,
+      "step": 545
+    },
+    {
+      "epoch": 0.04652748189177674,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.497754040306114e-06,
+      "loss": 0.892,
+      "step": 546
+    },
+    {
+      "epoch": 0.046612697060076695,
+      "grad_norm": 12.375,
+      "learning_rate": 2.497743960481054e-06,
+      "loss": 0.6311,
+      "step": 547
+    },
+    {
+      "epoch": 0.04669791222837665,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.497733858108007e-06,
+      "loss": 0.9285,
+      "step": 548
+    },
+    {
+      "epoch": 0.04678312739667661,
+      "grad_norm": 21.25,
+      "learning_rate": 2.4977237331871535e-06,
+      "loss": 0.9391,
+      "step": 549
+    },
+    {
+      "epoch": 0.046868342564976566,
+      "grad_norm": 15.375,
+      "learning_rate": 2.497713585718678e-06,
+      "loss": 0.8202,
+      "step": 550
+    },
+    {
+      "epoch": 0.04695355773327652,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.497703415702763e-06,
+      "loss": 0.9054,
+      "step": 551
+    },
+    {
+      "epoch": 0.04703877290157648,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4976932231395926e-06,
+      "loss": 0.5876,
+      "step": 552
+    },
+    {
+      "epoch": 0.04712398806987644,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4976830080293514e-06,
+      "loss": 0.5749,
+      "step": 553
+    },
+    {
+      "epoch": 0.04720920323817639,
+      "grad_norm": 16.75,
+      "learning_rate": 2.497672770372223e-06,
+      "loss": 1.0638,
+      "step": 554
+    },
+    {
+      "epoch": 0.04729441840647635,
+      "grad_norm": 19.0,
+      "learning_rate": 2.497662510168393e-06,
+      "loss": 1.0446,
+      "step": 555
+    },
+    {
+      "epoch": 0.04737963357477631,
+      "grad_norm": 28.25,
+      "learning_rate": 2.4976522274180475e-06,
+      "loss": 1.5739,
+      "step": 556
+    },
+    {
+      "epoch": 0.047464848743076264,
+      "grad_norm": 22.125,
+      "learning_rate": 2.497641922121371e-06,
+      "loss": 1.1316,
+      "step": 557
+    },
+    {
+      "epoch": 0.04755006391137623,
+      "grad_norm": 14.75,
+      "learning_rate": 2.497631594278551e-06,
+      "loss": 1.0194,
+      "step": 558
+    },
+    {
+      "epoch": 0.047635279079676185,
+      "grad_norm": 12.25,
+      "learning_rate": 2.497621243889773e-06,
+      "loss": 0.5275,
+      "step": 559
+    },
+    {
+      "epoch": 0.04772049424797614,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4976108709552247e-06,
+      "loss": 0.5929,
+      "step": 560
+    },
+    {
+      "epoch": 0.0478057094162761,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4976004754750933e-06,
+      "loss": 0.9813,
+      "step": 561
+    },
+    {
+      "epoch": 0.047890924584576056,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4975900574495672e-06,
+      "loss": 0.3754,
+      "step": 562
+    },
+    {
+      "epoch": 0.04797613975287601,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4975796168788337e-06,
+      "loss": 0.7562,
+      "step": 563
+    },
+    {
+      "epoch": 0.04806135492117597,
+      "grad_norm": 18.0,
+      "learning_rate": 2.497569153763082e-06,
+      "loss": 1.0602,
+      "step": 564
+    },
+    {
+      "epoch": 0.048146570089475926,
+      "grad_norm": 21.375,
+      "learning_rate": 2.4975586681025016e-06,
+      "loss": 0.6363,
+      "step": 565
+    },
+    {
+      "epoch": 0.04823178525777588,
+      "grad_norm": 23.625,
+      "learning_rate": 2.497548159897281e-06,
+      "loss": 0.9576,
+      "step": 566
+    },
+    {
+      "epoch": 0.04831700042607584,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.497537629147611e-06,
+      "loss": 0.901,
+      "step": 567
+    },
+    {
+      "epoch": 0.0484022155943758,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.497527075853681e-06,
+      "loss": 0.7384,
+      "step": 568
+    },
+    {
+      "epoch": 0.048487430762675754,
+      "grad_norm": 15.875,
+      "learning_rate": 2.497516500015683e-06,
+      "loss": 0.8411,
+      "step": 569
+    },
+    {
+      "epoch": 0.04857264593097571,
+      "grad_norm": 16.625,
+      "learning_rate": 2.497505901633807e-06,
+      "loss": 0.9076,
+      "step": 570
+    },
+    {
+      "epoch": 0.04865786109927567,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4974952807082454e-06,
+      "loss": 0.6311,
+      "step": 571
+    },
+    {
+      "epoch": 0.04874307626757563,
+      "grad_norm": 15.25,
+      "learning_rate": 2.497484637239189e-06,
+      "loss": 0.5999,
+      "step": 572
+    },
+    {
+      "epoch": 0.04882829143587559,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4974739712268307e-06,
+      "loss": 0.6612,
+      "step": 573
+    },
+    {
+      "epoch": 0.048913506604175545,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4974632826713636e-06,
+      "loss": 0.6295,
+      "step": 574
+    },
+    {
+      "epoch": 0.0489987217724755,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4974525715729808e-06,
+      "loss": 0.4852,
+      "step": 575
+    },
+    {
+      "epoch": 0.04908393694077546,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.497441837931875e-06,
+      "loss": 0.5265,
+      "step": 576
+    },
+    {
+      "epoch": 0.049169152109075416,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4974310817482414e-06,
+      "loss": 0.891,
+      "step": 577
+    },
+    {
+      "epoch": 0.04925436727737537,
+      "grad_norm": 21.125,
+      "learning_rate": 2.4974203030222732e-06,
+      "loss": 1.0509,
+      "step": 578
+    },
+    {
+      "epoch": 0.04933958244567533,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.497409501754166e-06,
+      "loss": 0.8276,
+      "step": 579
+    },
+    {
+      "epoch": 0.04942479761397529,
+      "grad_norm": 17.0,
+      "learning_rate": 2.497398677944115e-06,
+      "loss": 0.9166,
+      "step": 580
+    },
+    {
+      "epoch": 0.049510012782275244,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4973878315923153e-06,
+      "loss": 1.0019,
+      "step": 581
+    },
+    {
+      "epoch": 0.0495952279505752,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.497376962698963e-06,
+      "loss": 1.0199,
+      "step": 582
+    },
+    {
+      "epoch": 0.04968044311887516,
+      "grad_norm": 14.625,
+      "learning_rate": 2.497366071264255e-06,
+      "loss": 1.015,
+      "step": 583
+    },
+    {
+      "epoch": 0.049765658287175114,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4973551572883875e-06,
+      "loss": 0.9148,
+      "step": 584
+    },
+    {
+      "epoch": 0.04985087345547507,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.497344220771558e-06,
+      "loss": 1.0273,
+      "step": 585
+    },
+    {
+      "epoch": 0.049936088623775035,
+      "grad_norm": 23.125,
+      "learning_rate": 2.497333261713964e-06,
+      "loss": 1.0809,
+      "step": 586
+    },
+    {
+      "epoch": 0.05002130379207499,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4973222801158044e-06,
+      "loss": 1.0111,
+      "step": 587
+    },
+    {
+      "epoch": 0.05010651896037495,
+      "grad_norm": 10.5,
+      "learning_rate": 2.4973112759772763e-06,
+      "loss": 0.5785,
+      "step": 588
+    },
+    {
+      "epoch": 0.050191734128674906,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.497300249298579e-06,
+      "loss": 0.9292,
+      "step": 589
+    },
+    {
+      "epoch": 0.05027694929697486,
+      "grad_norm": 18.625,
+      "learning_rate": 2.497289200079912e-06,
+      "loss": 1.1965,
+      "step": 590
+    },
+    {
+      "epoch": 0.05036216446527482,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4972781283214753e-06,
+      "loss": 0.7628,
+      "step": 591
+    },
+    {
+      "epoch": 0.050447379633574777,
+      "grad_norm": 18.625,
+      "learning_rate": 2.497267034023468e-06,
+      "loss": 1.1404,
+      "step": 592
+    },
+    {
+      "epoch": 0.05053259480187473,
+      "grad_norm": 23.0,
+      "learning_rate": 2.4972559171860917e-06,
+      "loss": 0.9371,
+      "step": 593
+    },
+    {
+      "epoch": 0.05061780997017469,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4972447778095468e-06,
+      "loss": 1.2122,
+      "step": 594
+    },
+    {
+      "epoch": 0.05070302513847465,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4972336158940344e-06,
+      "loss": 1.4592,
+      "step": 595
+    },
+    {
+      "epoch": 0.050788240306774604,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4972224314397564e-06,
+      "loss": 0.9688,
+      "step": 596
+    },
+    {
+      "epoch": 0.05087345547507456,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.497211224446915e-06,
+      "loss": 0.7272,
+      "step": 597
+    },
+    {
+      "epoch": 0.05095867064337452,
+      "grad_norm": 21.375,
+      "learning_rate": 2.4971999949157123e-06,
+      "loss": 0.7135,
+      "step": 598
+    },
+    {
+      "epoch": 0.051043885811674475,
+      "grad_norm": 21.0,
+      "learning_rate": 2.4971887428463515e-06,
+      "loss": 1.1397,
+      "step": 599
+    },
+    {
+      "epoch": 0.05112910097997444,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4971774682390365e-06,
+      "loss": 0.8959,
+      "step": 600
+    },
+    {
+      "epoch": 0.051214316148274396,
+      "grad_norm": 25.625,
+      "learning_rate": 2.49716617109397e-06,
+      "loss": 1.153,
+      "step": 601
+    },
+    {
+      "epoch": 0.05129953131657435,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.497154851411357e-06,
+      "loss": 0.4679,
+      "step": 602
+    },
+    {
+      "epoch": 0.05138474648487431,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4971435091914014e-06,
+      "loss": 0.9237,
+      "step": 603
+    },
+    {
+      "epoch": 0.051469961653174266,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4971321444343084e-06,
+      "loss": 0.6365,
+      "step": 604
+    },
+    {
+      "epoch": 0.05155517682147422,
+      "grad_norm": 17.75,
+      "learning_rate": 2.497120757140284e-06,
+      "loss": 1.0036,
+      "step": 605
+    },
+    {
+      "epoch": 0.05164039198977418,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4971093473095333e-06,
+      "loss": 0.5024,
+      "step": 606
+    },
+    {
+      "epoch": 0.05172560715807414,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4970979149422624e-06,
+      "loss": 0.5912,
+      "step": 607
+    },
+    {
+      "epoch": 0.051810822326374094,
+      "grad_norm": 40.5,
+      "learning_rate": 2.497086460038678e-06,
+      "loss": 0.3844,
+      "step": 608
+    },
+    {
+      "epoch": 0.05189603749467405,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4970749825989877e-06,
+      "loss": 0.9913,
+      "step": 609
+    },
+    {
+      "epoch": 0.05198125266297401,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.497063482623398e-06,
+      "loss": 0.56,
+      "step": 610
+    },
+    {
+      "epoch": 0.052066467831273965,
+      "grad_norm": 27.875,
+      "learning_rate": 2.4970519601121175e-06,
+      "loss": 0.9598,
+      "step": 611
+    },
+    {
+      "epoch": 0.05215168299957392,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4970404150653534e-06,
+      "loss": 0.6372,
+      "step": 612
+    },
+    {
+      "epoch": 0.05223689816787388,
+      "grad_norm": 16.375,
+      "learning_rate": 2.497028847483316e-06,
+      "loss": 1.0319,
+      "step": 613
+    },
+    {
+      "epoch": 0.05232211333617384,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4970172573662124e-06,
+      "loss": 1.1488,
+      "step": 614
+    },
+    {
+      "epoch": 0.0524073285044738,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4970056447142535e-06,
+      "loss": 0.7887,
+      "step": 615
+    },
+    {
+      "epoch": 0.052492543672773756,
+      "grad_norm": 25.5,
+      "learning_rate": 2.4969940095276483e-06,
+      "loss": 1.1172,
+      "step": 616
+    },
+    {
+      "epoch": 0.05257775884107371,
+      "grad_norm": 17.75,
+      "learning_rate": 2.496982351806608e-06,
+      "loss": 0.7054,
+      "step": 617
+    },
+    {
+      "epoch": 0.05266297400937367,
+      "grad_norm": 18.125,
+      "learning_rate": 2.496970671551342e-06,
+      "loss": 1.325,
+      "step": 618
+    },
+    {
+      "epoch": 0.05274818917767363,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4969589687620626e-06,
+      "loss": 0.9802,
+      "step": 619
+    },
+    {
+      "epoch": 0.052833404345973584,
+      "grad_norm": 15.25,
+      "learning_rate": 2.49694724343898e-06,
+      "loss": 1.0539,
+      "step": 620
+    },
+    {
+      "epoch": 0.05291861951427354,
+      "grad_norm": 17.125,
+      "learning_rate": 2.496935495582308e-06,
+      "loss": 0.5991,
+      "step": 621
+    },
+    {
+      "epoch": 0.0530038346825735,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4969237251922567e-06,
+      "loss": 0.5742,
+      "step": 622
+    },
+    {
+      "epoch": 0.053089049850873454,
+      "grad_norm": 20.0,
+      "learning_rate": 2.49691193226904e-06,
+      "loss": 1.1053,
+      "step": 623
+    },
+    {
+      "epoch": 0.05317426501917341,
+      "grad_norm": 17.375,
+      "learning_rate": 2.496900116812871e-06,
+      "loss": 0.9491,
+      "step": 624
+    },
+    {
+      "epoch": 0.05325948018747337,
+      "grad_norm": 20.625,
+      "learning_rate": 2.496888278823963e-06,
+      "loss": 1.0348,
+      "step": 625
+    },
+    {
+      "epoch": 0.053344695355773325,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4968764183025297e-06,
+      "loss": 0.7143,
+      "step": 626
+    },
+    {
+      "epoch": 0.05342991052407328,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.496864535248786e-06,
+      "loss": 0.8898,
+      "step": 627
+    },
+    {
+      "epoch": 0.053515125692373246,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.496852629662946e-06,
+      "loss": 1.0079,
+      "step": 628
+    },
+    {
+      "epoch": 0.0536003408606732,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4968407015452255e-06,
+      "loss": 0.9257,
+      "step": 629
+    },
+    {
+      "epoch": 0.05368555602897316,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4968287508958397e-06,
+      "loss": 0.8194,
+      "step": 630
+    },
+    {
+      "epoch": 0.053770771197273116,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4968167777150044e-06,
+      "loss": 0.7869,
+      "step": 631
+    },
+    {
+      "epoch": 0.05385598636557307,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4968047820029364e-06,
+      "loss": 0.4579,
+      "step": 632
+    },
+    {
+      "epoch": 0.05394120153387303,
+      "grad_norm": 21.125,
+      "learning_rate": 2.496792763759852e-06,
+      "loss": 0.8788,
+      "step": 633
+    },
+    {
+      "epoch": 0.05402641670217299,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4967807229859685e-06,
+      "loss": 1.048,
+      "step": 634
+    },
+    {
+      "epoch": 0.054111631870472944,
+      "grad_norm": 13.5,
+      "learning_rate": 2.496768659681504e-06,
+      "loss": 0.7781,
+      "step": 635
+    },
+    {
+      "epoch": 0.0541968470387729,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.496756573846676e-06,
+      "loss": 0.7897,
+      "step": 636
+    },
+    {
+      "epoch": 0.05428206220707286,
+      "grad_norm": 14.25,
+      "learning_rate": 2.496744465481703e-06,
+      "loss": 0.7913,
+      "step": 637
+    },
+    {
+      "epoch": 0.054367277375372815,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4967323345868036e-06,
+      "loss": 0.3933,
+      "step": 638
+    },
+    {
+      "epoch": 0.05445249254367277,
+      "grad_norm": 13.5,
+      "learning_rate": 2.496720181162197e-06,
+      "loss": 0.8064,
+      "step": 639
+    },
+    {
+      "epoch": 0.05453770771197273,
+      "grad_norm": 14.875,
+      "learning_rate": 2.496708005208104e-06,
+      "loss": 0.5314,
+      "step": 640
+    },
+    {
+      "epoch": 0.054622922880272685,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4966958067247426e-06,
+      "loss": 0.864,
+      "step": 641
+    },
+    {
+      "epoch": 0.05470813804857265,
+      "grad_norm": 13.25,
+      "learning_rate": 2.496683585712335e-06,
+      "loss": 0.8099,
+      "step": 642
+    },
+    {
+      "epoch": 0.054793353216872606,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4966713421711013e-06,
+      "loss": 0.3473,
+      "step": 643
+    },
+    {
+      "epoch": 0.05487856838517256,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4966590761012627e-06,
+      "loss": 0.7167,
+      "step": 644
+    },
+    {
+      "epoch": 0.05496378355347252,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.496646787503041e-06,
+      "loss": 0.4618,
+      "step": 645
+    },
+    {
+      "epoch": 0.05504899872177248,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4966344763766583e-06,
+      "loss": 0.8458,
+      "step": 646
+    },
+    {
+      "epoch": 0.055134213890072434,
+      "grad_norm": 15.5,
+      "learning_rate": 2.496622142722337e-06,
+      "loss": 1.1295,
+      "step": 647
+    },
+    {
+      "epoch": 0.05521942905837239,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4966097865402995e-06,
+      "loss": 0.6643,
+      "step": 648
+    },
+    {
+      "epoch": 0.05530464422667235,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4965974078307703e-06,
+      "loss": 0.6109,
+      "step": 649
+    },
+    {
+      "epoch": 0.055389859394972304,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.496585006593972e-06,
+      "loss": 0.6909,
+      "step": 650
+    },
+    {
+      "epoch": 0.05547507456327226,
+      "grad_norm": 20.375,
+      "learning_rate": 2.4965725828301294e-06,
+      "loss": 0.9731,
+      "step": 651
+    },
+    {
+      "epoch": 0.05556028973157222,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4965601365394666e-06,
+      "loss": 0.9019,
+      "step": 652
+    },
+    {
+      "epoch": 0.055645504899872175,
+      "grad_norm": 19.75,
+      "learning_rate": 2.496547667722208e-06,
+      "loss": 0.9211,
+      "step": 653
+    },
+    {
+      "epoch": 0.05573072006817213,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4965351763785804e-06,
+      "loss": 0.7265,
+      "step": 654
+    },
+    {
+      "epoch": 0.05581593523647209,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4965226625088084e-06,
+      "loss": 0.6484,
+      "step": 655
+    },
+    {
+      "epoch": 0.05590115040477205,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4965101261131185e-06,
+      "loss": 1.1539,
+      "step": 656
+    },
+    {
+      "epoch": 0.05598636557307201,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.496497567191737e-06,
+      "loss": 0.7442,
+      "step": 657
+    },
+    {
+      "epoch": 0.05607158074137197,
+      "grad_norm": 45.0,
+      "learning_rate": 2.4964849857448907e-06,
+      "loss": 0.923,
+      "step": 658
+    },
+    {
+      "epoch": 0.056156795909671924,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.496472381772808e-06,
+      "loss": 0.5966,
+      "step": 659
+    },
+    {
+      "epoch": 0.05624201107797188,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4964597552757155e-06,
+      "loss": 0.6612,
+      "step": 660
+    },
+    {
+      "epoch": 0.05632722624627184,
+      "grad_norm": 19.75,
+      "learning_rate": 2.496447106253842e-06,
+      "loss": 1.2901,
+      "step": 661
+    },
+    {
+      "epoch": 0.056412441414571794,
+      "grad_norm": 21.125,
+      "learning_rate": 2.4964344347074158e-06,
+      "loss": 0.8304,
+      "step": 662
+    },
+    {
+      "epoch": 0.05649765658287175,
+      "grad_norm": 16.125,
+      "learning_rate": 2.496421740636666e-06,
+      "loss": 1.0131,
+      "step": 663
+    },
+    {
+      "epoch": 0.05658287175117171,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4964090240418215e-06,
+      "loss": 0.9323,
+      "step": 664
+    },
+    {
+      "epoch": 0.056668086919471665,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4963962849231135e-06,
+      "loss": 0.7873,
+      "step": 665
+    },
+    {
+      "epoch": 0.05675330208777162,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4963835232807702e-06,
+      "loss": 0.7644,
+      "step": 666
+    },
+    {
+      "epoch": 0.05683851725607158,
+      "grad_norm": 16.875,
+      "learning_rate": 2.496370739115024e-06,
+      "loss": 1.2728,
+      "step": 667
+    },
+    {
+      "epoch": 0.056923732424371536,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4963579324261052e-06,
+      "loss": 0.6924,
+      "step": 668
+    },
+    {
+      "epoch": 0.05700894759267149,
+      "grad_norm": 18.125,
+      "learning_rate": 2.496345103214245e-06,
+      "loss": 0.9334,
+      "step": 669
+    },
+    {
+      "epoch": 0.057094162760971456,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4963322514796755e-06,
+      "loss": 0.6154,
+      "step": 670
+    },
+    {
+      "epoch": 0.05717937792927141,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4963193772226295e-06,
+      "loss": 1.0111,
+      "step": 671
+    },
+    {
+      "epoch": 0.05726459309757137,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4963064804433382e-06,
+      "loss": 0.755,
+      "step": 672
+    },
+    {
+      "epoch": 0.05734980826587133,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4962935611420364e-06,
+      "loss": 1.1122,
+      "step": 673
+    },
+    {
+      "epoch": 0.057435023434171284,
+      "grad_norm": 17.625,
+      "learning_rate": 2.496280619318956e-06,
+      "loss": 0.6942,
+      "step": 674
+    },
+    {
+      "epoch": 0.05752023860247124,
+      "grad_norm": 20.25,
+      "learning_rate": 2.496267654974332e-06,
+      "loss": 0.9186,
+      "step": 675
+    },
+    {
+      "epoch": 0.0576054537707712,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4962546681083982e-06,
+      "loss": 0.7839,
+      "step": 676
+    },
+    {
+      "epoch": 0.057690668939071155,
+      "grad_norm": 14.375,
+      "learning_rate": 2.496241658721389e-06,
+      "loss": 0.769,
+      "step": 677
+    },
+    {
+      "epoch": 0.05777588410737111,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4962286268135406e-06,
+      "loss": 0.8106,
+      "step": 678
+    },
+    {
+      "epoch": 0.05786109927567107,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4962155723850867e-06,
+      "loss": 0.7286,
+      "step": 679
+    },
+    {
+      "epoch": 0.057946314443971025,
+      "grad_norm": 18.5,
+      "learning_rate": 2.4962024954362653e-06,
+      "loss": 0.6561,
+      "step": 680
+    },
+    {
+      "epoch": 0.05803152961227098,
+      "grad_norm": 19.875,
+      "learning_rate": 2.496189395967311e-06,
+      "loss": 0.9052,
+      "step": 681
+    },
+    {
+      "epoch": 0.05811674478057094,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4961762739784613e-06,
+      "loss": 1.5005,
+      "step": 682
+    },
+    {
+      "epoch": 0.058201959948870896,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.496163129469953e-06,
+      "loss": 0.4962,
+      "step": 683
+    },
+    {
+      "epoch": 0.05828717511717085,
+      "grad_norm": 16.125,
+      "learning_rate": 2.496149962442024e-06,
+      "loss": 0.8615,
+      "step": 684
+    },
+    {
+      "epoch": 0.05837239028547082,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.496136772894912e-06,
+      "loss": 0.6755,
+      "step": 685
+    },
+    {
+      "epoch": 0.058457605453770774,
+      "grad_norm": 20.125,
+      "learning_rate": 2.496123560828856e-06,
+      "loss": 1.1005,
+      "step": 686
+    },
+    {
+      "epoch": 0.05854282062207073,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4961103262440934e-06,
+      "loss": 1.0075,
+      "step": 687
+    },
+    {
+      "epoch": 0.05862803579037069,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4960970691408645e-06,
+      "loss": 0.6836,
+      "step": 688
+    },
+    {
+      "epoch": 0.058713250958670644,
+      "grad_norm": 11.0,
+      "learning_rate": 2.4960837895194083e-06,
+      "loss": 0.7389,
+      "step": 689
+    },
+    {
+      "epoch": 0.0587984661269706,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4960704873799653e-06,
+      "loss": 0.7842,
+      "step": 690
+    },
+    {
+      "epoch": 0.05888368129527056,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4960571627227753e-06,
+      "loss": 0.2997,
+      "step": 691
+    },
+    {
+      "epoch": 0.058968896463570515,
+      "grad_norm": 12.625,
+      "learning_rate": 2.49604381554808e-06,
+      "loss": 0.5738,
+      "step": 692
+    },
+    {
+      "epoch": 0.05905411163187047,
+      "grad_norm": 11.5,
+      "learning_rate": 2.496030445856119e-06,
+      "loss": 0.4964,
+      "step": 693
+    },
+    {
+      "epoch": 0.05913932680017043,
+      "grad_norm": 23.375,
+      "learning_rate": 2.4960170536471355e-06,
+      "loss": 0.8448,
+      "step": 694
+    },
+    {
+      "epoch": 0.059224541968470386,
+      "grad_norm": 17.25,
+      "learning_rate": 2.496003638921371e-06,
+      "loss": 0.7069,
+      "step": 695
+    },
+    {
+      "epoch": 0.05930975713677034,
+      "grad_norm": 17.375,
+      "learning_rate": 2.495990201679067e-06,
+      "loss": 0.7296,
+      "step": 696
+    },
+    {
+      "epoch": 0.0593949723050703,
+      "grad_norm": 14.625,
+      "learning_rate": 2.495976741920468e-06,
+      "loss": 1.0706,
+      "step": 697
+    },
+    {
+      "epoch": 0.059480187473370257,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4959632596458153e-06,
+      "loss": 0.6849,
+      "step": 698
+    },
+    {
+      "epoch": 0.05956540264167022,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4959497548553546e-06,
+      "loss": 0.9897,
+      "step": 699
+    },
+    {
+      "epoch": 0.05965061780997018,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4959362275493287e-06,
+      "loss": 1.1302,
+      "step": 700
+    },
+    {
+      "epoch": 0.059735832978270134,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.495922677727982e-06,
+      "loss": 0.8003,
+      "step": 701
+    },
+    {
+      "epoch": 0.05982104814657009,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.49590910539156e-06,
+      "loss": 0.4348,
+      "step": 702
+    },
+    {
+      "epoch": 0.05990626331487005,
+      "grad_norm": 19.125,
+      "learning_rate": 2.495895510540307e-06,
+      "loss": 0.8835,
+      "step": 703
+    },
+    {
+      "epoch": 0.059991478483170005,
+      "grad_norm": 20.25,
+      "learning_rate": 2.49588189317447e-06,
+      "loss": 1.1262,
+      "step": 704
+    },
+    {
+      "epoch": 0.06007669365146996,
+      "grad_norm": 13.25,
+      "learning_rate": 2.495868253294294e-06,
+      "loss": 0.6083,
+      "step": 705
+    },
+    {
+      "epoch": 0.06016190881976992,
+      "grad_norm": 18.25,
+      "learning_rate": 2.495854590900026e-06,
+      "loss": 1.07,
+      "step": 706
+    },
+    {
+      "epoch": 0.060247123988069876,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4958409059919124e-06,
+      "loss": 1.0939,
+      "step": 707
+    },
+    {
+      "epoch": 0.06033233915636983,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.495827198570201e-06,
+      "loss": 1.0319,
+      "step": 708
+    },
+    {
+      "epoch": 0.06041755432466979,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4958134686351396e-06,
+      "loss": 0.9677,
+      "step": 709
+    },
+    {
+      "epoch": 0.060502769492969746,
+      "grad_norm": 9.5,
+      "learning_rate": 2.495799716186976e-06,
+      "loss": 0.2821,
+      "step": 710
+    },
+    {
+      "epoch": 0.0605879846612697,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4957859412259584e-06,
+      "loss": 0.4358,
+      "step": 711
+    },
+    {
+      "epoch": 0.06067319982956966,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4957721437523365e-06,
+      "loss": 1.0668,
+      "step": 712
+    },
+    {
+      "epoch": 0.060758414997869624,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.495758323766359e-06,
+      "loss": 0.9974,
+      "step": 713
+    },
+    {
+      "epoch": 0.06084363016616958,
+      "grad_norm": 17.375,
+      "learning_rate": 2.495744481268276e-06,
+      "loss": 1.0701,
+      "step": 714
+    },
+    {
+      "epoch": 0.06092884533446954,
+      "grad_norm": 10.375,
+      "learning_rate": 2.4957306162583376e-06,
+      "loss": 0.3589,
+      "step": 715
+    },
+    {
+      "epoch": 0.061014060502769495,
+      "grad_norm": 13.375,
+      "learning_rate": 2.495716728736794e-06,
+      "loss": 0.6527,
+      "step": 716
+    },
+    {
+      "epoch": 0.06109927567106945,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4957028187038966e-06,
+      "loss": 0.8977,
+      "step": 717
+    },
+    {
+      "epoch": 0.06118449083936941,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4956888861598968e-06,
+      "loss": 0.5937,
+      "step": 718
+    },
+    {
+      "epoch": 0.061269706007669365,
+      "grad_norm": 30.625,
+      "learning_rate": 2.495674931105046e-06,
+      "loss": 1.0829,
+      "step": 719
+    },
+    {
+      "epoch": 0.06135492117596932,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4956609535395966e-06,
+      "loss": 1.0039,
+      "step": 720
+    },
+    {
+      "epoch": 0.06144013634426928,
+      "grad_norm": 18.0,
+      "learning_rate": 2.495646953463801e-06,
+      "loss": 1.0935,
+      "step": 721
+    },
+    {
+      "epoch": 0.061525351512569236,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4956329308779125e-06,
+      "loss": 1.0674,
+      "step": 722
+    },
+    {
+      "epoch": 0.06161056668086919,
+      "grad_norm": 19.375,
+      "learning_rate": 2.4956188857821846e-06,
+      "loss": 0.6899,
+      "step": 723
+    },
+    {
+      "epoch": 0.06169578184916915,
+      "grad_norm": 16.75,
+      "learning_rate": 2.495604818176871e-06,
+      "loss": 0.7459,
+      "step": 724
+    },
+    {
+      "epoch": 0.06178099701746911,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4955907280622253e-06,
+      "loss": 0.6505,
+      "step": 725
+    },
+    {
+      "epoch": 0.061866212185769064,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4955766154385025e-06,
+      "loss": 1.2594,
+      "step": 726
+    },
+    {
+      "epoch": 0.06195142735406903,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.495562480305958e-06,
+      "loss": 0.8341,
+      "step": 727
+    },
+    {
+      "epoch": 0.062036642522368984,
+      "grad_norm": 16.25,
+      "learning_rate": 2.495548322664847e-06,
+      "loss": 0.582,
+      "step": 728
+    },
+    {
+      "epoch": 0.06212185769066894,
+      "grad_norm": 17.625,
+      "learning_rate": 2.495534142515425e-06,
+      "loss": 0.8811,
+      "step": 729
+    },
+    {
+      "epoch": 0.0622070728589689,
+      "grad_norm": 16.75,
+      "learning_rate": 2.495519939857949e-06,
+      "loss": 0.984,
+      "step": 730
+    },
+    {
+      "epoch": 0.062292288027268855,
+      "grad_norm": 23.75,
+      "learning_rate": 2.4955057146926754e-06,
+      "loss": 1.0588,
+      "step": 731
+    },
+    {
+      "epoch": 0.06237750319556881,
+      "grad_norm": 32.75,
+      "learning_rate": 2.4954914670198605e-06,
+      "loss": 1.0906,
+      "step": 732
+    },
+    {
+      "epoch": 0.06246271836386877,
+      "grad_norm": 39.5,
+      "learning_rate": 2.4954771968397624e-06,
+      "loss": 1.5367,
+      "step": 733
+    },
+    {
+      "epoch": 0.06254793353216873,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4954629041526393e-06,
+      "loss": 0.6287,
+      "step": 734
+    },
+    {
+      "epoch": 0.06263314870046868,
+      "grad_norm": 10.375,
+      "learning_rate": 2.495448588958749e-06,
+      "loss": 0.3451,
+      "step": 735
+    },
+    {
+      "epoch": 0.06271836386876864,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4954342512583503e-06,
+      "loss": 0.789,
+      "step": 736
+    },
+    {
+      "epoch": 0.0628035790370686,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.495419891051702e-06,
+      "loss": 0.4972,
+      "step": 737
+    },
+    {
+      "epoch": 0.06288879420536855,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4954055083390643e-06,
+      "loss": 0.7407,
+      "step": 738
+    },
+    {
+      "epoch": 0.06297400937366851,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4953911031206966e-06,
+      "loss": 1.0121,
+      "step": 739
+    },
+    {
+      "epoch": 0.06305922454196847,
+      "grad_norm": 17.375,
+      "learning_rate": 2.495376675396859e-06,
+      "loss": 0.5802,
+      "step": 740
+    },
+    {
+      "epoch": 0.06314443971026842,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4953622251678133e-06,
+      "loss": 0.5409,
+      "step": 741
+    },
+    {
+      "epoch": 0.06322965487856838,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4953477524338194e-06,
+      "loss": 0.6038,
+      "step": 742
+    },
+    {
+      "epoch": 0.06331487004686834,
+      "grad_norm": 13.0,
+      "learning_rate": 2.495333257195139e-06,
+      "loss": 0.5836,
+      "step": 743
+    },
+    {
+      "epoch": 0.0634000852151683,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4953187394520348e-06,
+      "loss": 0.9286,
+      "step": 744
+    },
+    {
+      "epoch": 0.06348530038346825,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4953041992047687e-06,
+      "loss": 0.8986,
+      "step": 745
+    },
+    {
+      "epoch": 0.06357051555176821,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4952896364536032e-06,
+      "loss": 0.4778,
+      "step": 746
+    },
+    {
+      "epoch": 0.06365573072006817,
+      "grad_norm": 20.25,
+      "learning_rate": 2.495275051198802e-06,
+      "loss": 1.0113,
+      "step": 747
+    },
+    {
+      "epoch": 0.06374094588836814,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4952604434406285e-06,
+      "loss": 0.9292,
+      "step": 748
+    },
+    {
+      "epoch": 0.0638261610566681,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.495245813179346e-06,
+      "loss": 0.8457,
+      "step": 749
+    },
+    {
+      "epoch": 0.06391137622496805,
+      "grad_norm": 15.75,
+      "learning_rate": 2.49523116041522e-06,
+      "loss": 0.9048,
+      "step": 750
+    },
+    {
+      "epoch": 0.06399659139326801,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4952164851485143e-06,
+      "loss": 0.5674,
+      "step": 751
+    },
+    {
+      "epoch": 0.06408180656156796,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.495201787379495e-06,
+      "loss": 0.9348,
+      "step": 752
+    },
+    {
+      "epoch": 0.06416702172986792,
+      "grad_norm": 15.125,
+      "learning_rate": 2.495187067108427e-06,
+      "loss": 1.1715,
+      "step": 753
+    },
+    {
+      "epoch": 0.06425223689816788,
+      "grad_norm": 14.5,
+      "learning_rate": 2.495172324335577e-06,
+      "loss": 0.5497,
+      "step": 754
+    },
+    {
+      "epoch": 0.06433745206646783,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4951575590612106e-06,
+      "loss": 0.503,
+      "step": 755
+    },
+    {
+      "epoch": 0.06442266723476779,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4951427712855952e-06,
+      "loss": 1.2421,
+      "step": 756
+    },
+    {
+      "epoch": 0.06450788240306775,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4951279610089978e-06,
+      "loss": 0.5308,
+      "step": 757
+    },
+    {
+      "epoch": 0.0645930975713677,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4951131282316858e-06,
+      "loss": 1.1092,
+      "step": 758
+    },
+    {
+      "epoch": 0.06467831273966766,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4950982729539274e-06,
+      "loss": 0.8727,
+      "step": 759
+    },
+    {
+      "epoch": 0.06476352790796762,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.495083395175992e-06,
+      "loss": 0.8149,
+      "step": 760
+    },
+    {
+      "epoch": 0.06484874307626758,
+      "grad_norm": 13.875,
+      "learning_rate": 2.495068494898147e-06,
+      "loss": 0.6574,
+      "step": 761
+    },
+    {
+      "epoch": 0.06493395824456753,
+      "grad_norm": 16.0,
+      "learning_rate": 2.495053572120662e-06,
+      "loss": 0.8122,
+      "step": 762
+    },
+    {
+      "epoch": 0.06501917341286749,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4950386268438074e-06,
+      "loss": 0.6436,
+      "step": 763
+    },
+    {
+      "epoch": 0.06510438858116745,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.4950236590678526e-06,
+      "loss": 0.3646,
+      "step": 764
+    },
+    {
+      "epoch": 0.0651896037494674,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4950086687930684e-06,
+      "loss": 0.7299,
+      "step": 765
+    },
+    {
+      "epoch": 0.06527481891776736,
+      "grad_norm": 14.5,
+      "learning_rate": 2.494993656019726e-06,
+      "loss": 0.8499,
+      "step": 766
+    },
+    {
+      "epoch": 0.06536003408606732,
+      "grad_norm": 18.375,
+      "learning_rate": 2.494978620748096e-06,
+      "loss": 1.2044,
+      "step": 767
+    },
+    {
+      "epoch": 0.06544524925436727,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.49496356297845e-06,
+      "loss": 0.4475,
+      "step": 768
+    },
+    {
+      "epoch": 0.06553046442266723,
+      "grad_norm": 11.25,
+      "learning_rate": 2.494948482711061e-06,
+      "loss": 0.5462,
+      "step": 769
+    },
+    {
+      "epoch": 0.06561567959096719,
+      "grad_norm": 15.125,
+      "learning_rate": 2.494933379946201e-06,
+      "loss": 0.9137,
+      "step": 770
+    },
+    {
+      "epoch": 0.06570089475926714,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4949182546841426e-06,
+      "loss": 1.0257,
+      "step": 771
+    },
+    {
+      "epoch": 0.0657861099275671,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.49490310692516e-06,
+      "loss": 0.9288,
+      "step": 772
+    },
+    {
+      "epoch": 0.06587132509586706,
+      "grad_norm": 13.625,
+      "learning_rate": 2.494887936669526e-06,
+      "loss": 0.7343,
+      "step": 773
+    },
+    {
+      "epoch": 0.06595654026416702,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4948727439175153e-06,
+      "loss": 0.9281,
+      "step": 774
+    },
+    {
+      "epoch": 0.06604175543246697,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4948575286694022e-06,
+      "loss": 0.632,
+      "step": 775
+    },
+    {
+      "epoch": 0.06612697060076694,
+      "grad_norm": 17.25,
+      "learning_rate": 2.494842290925462e-06,
+      "loss": 0.7674,
+      "step": 776
+    },
+    {
+      "epoch": 0.0662121857690669,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4948270306859695e-06,
+      "loss": 0.5834,
+      "step": 777
+    },
+    {
+      "epoch": 0.06629740093736686,
+      "grad_norm": 16.5,
+      "learning_rate": 2.494811747951201e-06,
+      "loss": 0.7233,
+      "step": 778
+    },
+    {
+      "epoch": 0.06638261610566681,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4947964427214326e-06,
+      "loss": 0.75,
+      "step": 779
+    },
+    {
+      "epoch": 0.06646783127396677,
+      "grad_norm": 14.625,
+      "learning_rate": 2.49478111499694e-06,
+      "loss": 0.591,
+      "step": 780
+    },
+    {
+      "epoch": 0.06655304644226673,
+      "grad_norm": 16.5,
+      "learning_rate": 2.494765764778002e-06,
+      "loss": 0.8904,
+      "step": 781
+    },
+    {
+      "epoch": 0.06663826161056668,
+      "grad_norm": 16.75,
+      "learning_rate": 2.494750392064894e-06,
+      "loss": 0.9807,
+      "step": 782
+    },
+    {
+      "epoch": 0.06672347677886664,
+      "grad_norm": 34.5,
+      "learning_rate": 2.4947349968578956e-06,
+      "loss": 0.9434,
+      "step": 783
+    },
+    {
+      "epoch": 0.0668086919471666,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.494719579157283e-06,
+      "loss": 0.5366,
+      "step": 784
+    },
+    {
+      "epoch": 0.06689390711546656,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.494704138963337e-06,
+      "loss": 0.648,
+      "step": 785
+    },
+    {
+      "epoch": 0.06697912228376651,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.494688676276335e-06,
+      "loss": 1.1552,
+      "step": 786
+    },
+    {
+      "epoch": 0.06706433745206647,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.494673191096557e-06,
+      "loss": 1.0506,
+      "step": 787
+    },
+    {
+      "epoch": 0.06714955262036643,
+      "grad_norm": 11.0,
+      "learning_rate": 2.494657683424283e-06,
+      "loss": 0.4094,
+      "step": 788
+    },
+    {
+      "epoch": 0.06723476778866638,
+      "grad_norm": 17.0,
+      "learning_rate": 2.494642153259793e-06,
+      "loss": 0.9257,
+      "step": 789
+    },
+    {
+      "epoch": 0.06731998295696634,
+      "grad_norm": 24.375,
+      "learning_rate": 2.494626600603368e-06,
+      "loss": 0.6784,
+      "step": 790
+    },
+    {
+      "epoch": 0.0674051981252663,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4946110254552886e-06,
+      "loss": 0.6307,
+      "step": 791
+    },
+    {
+      "epoch": 0.06749041329356625,
+      "grad_norm": 18.75,
+      "learning_rate": 2.494595427815836e-06,
+      "loss": 1.0362,
+      "step": 792
+    },
+    {
+      "epoch": 0.06757562846186621,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4945798076852928e-06,
+      "loss": 0.6927,
+      "step": 793
+    },
+    {
+      "epoch": 0.06766084363016617,
+      "grad_norm": 15.0,
+      "learning_rate": 2.494564165063941e-06,
+      "loss": 1.057,
+      "step": 794
+    },
+    {
+      "epoch": 0.06774605879846612,
+      "grad_norm": 16.125,
+      "learning_rate": 2.494548499952063e-06,
+      "loss": 0.5934,
+      "step": 795
+    },
+    {
+      "epoch": 0.06783127396676608,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4945328123499424e-06,
+      "loss": 0.7026,
+      "step": 796
+    },
+    {
+      "epoch": 0.06791648913506604,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4945171022578622e-06,
+      "loss": 0.9534,
+      "step": 797
+    },
+    {
+      "epoch": 0.068001704303366,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4945013696761066e-06,
+      "loss": 0.6558,
+      "step": 798
+    },
+    {
+      "epoch": 0.06808691947166595,
+      "grad_norm": 35.25,
+      "learning_rate": 2.49448561460496e-06,
+      "loss": 1.0942,
+      "step": 799
+    },
+    {
+      "epoch": 0.06817213463996591,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4944698370447066e-06,
+      "loss": 0.7719,
+      "step": 800
+    },
+    {
+      "epoch": 0.06825734980826587,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.494454036995632e-06,
+      "loss": 0.7603,
+      "step": 801
+    },
+    {
+      "epoch": 0.06834256497656582,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4944382144580213e-06,
+      "loss": 0.7476,
+      "step": 802
+    },
+    {
+      "epoch": 0.06842778014486578,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.494422369432161e-06,
+      "loss": 0.3921,
+      "step": 803
+    },
+    {
+      "epoch": 0.06851299531316575,
+      "grad_norm": 15.75,
+      "learning_rate": 2.494406501918337e-06,
+      "loss": 1.0187,
+      "step": 804
+    },
+    {
+      "epoch": 0.06859821048146571,
+      "grad_norm": 13.625,
+      "learning_rate": 2.494390611916836e-06,
+      "loss": 0.6273,
+      "step": 805
+    },
+    {
+      "epoch": 0.06868342564976566,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4943746994279456e-06,
+      "loss": 0.7789,
+      "step": 806
+    },
+    {
+      "epoch": 0.06876864081806562,
+      "grad_norm": 23.75,
+      "learning_rate": 2.4943587644519528e-06,
+      "loss": 1.0404,
+      "step": 807
+    },
+    {
+      "epoch": 0.06885385598636558,
+      "grad_norm": 9.25,
+      "learning_rate": 2.494342806989146e-06,
+      "loss": 0.3221,
+      "step": 808
+    },
+    {
+      "epoch": 0.06893907115466553,
+      "grad_norm": 21.75,
+      "learning_rate": 2.4943268270398135e-06,
+      "loss": 1.2056,
+      "step": 809
+    },
+    {
+      "epoch": 0.06902428632296549,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.494310824604244e-06,
+      "loss": 0.5997,
+      "step": 810
+    },
+    {
+      "epoch": 0.06910950149126545,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.494294799682726e-06,
+      "loss": 0.7444,
+      "step": 811
+    },
+    {
+      "epoch": 0.0691947166595654,
+      "grad_norm": 24.25,
+      "learning_rate": 2.494278752275551e-06,
+      "loss": 0.7059,
+      "step": 812
+    },
+    {
+      "epoch": 0.06927993182786536,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.494262682383006e-06,
+      "loss": 0.9566,
+      "step": 813
+    },
+    {
+      "epoch": 0.06936514699616532,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.4942465900053846e-06,
+      "loss": 0.5052,
+      "step": 814
+    },
+    {
+      "epoch": 0.06945036216446528,
+      "grad_norm": 6.25,
+      "learning_rate": 2.4942304751429757e-06,
+      "loss": 0.1778,
+      "step": 815
+    },
+    {
+      "epoch": 0.06953557733276523,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4942143377960708e-06,
+      "loss": 0.6495,
+      "step": 816
+    },
+    {
+      "epoch": 0.06962079250106519,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.4941981779649614e-06,
+      "loss": 0.3381,
+      "step": 817
+    },
+    {
+      "epoch": 0.06970600766936515,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4941819956499404e-06,
+      "loss": 0.8703,
+      "step": 818
+    },
+    {
+      "epoch": 0.0697912228376651,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4941657908512994e-06,
+      "loss": 0.6306,
+      "step": 819
+    },
+    {
+      "epoch": 0.06987643800596506,
+      "grad_norm": 12.375,
+      "learning_rate": 2.494149563569331e-06,
+      "loss": 0.4808,
+      "step": 820
+    },
+    {
+      "epoch": 0.06996165317426502,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4941333138043294e-06,
+      "loss": 0.7736,
+      "step": 821
+    },
+    {
+      "epoch": 0.07004686834256497,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4941170415565873e-06,
+      "loss": 0.6722,
+      "step": 822
+    },
+    {
+      "epoch": 0.07013208351086493,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4941007468264e-06,
+      "loss": 0.5138,
+      "step": 823
+    },
+    {
+      "epoch": 0.07021729867916489,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4940844296140605e-06,
+      "loss": 0.6698,
+      "step": 824
+    },
+    {
+      "epoch": 0.07030251384746485,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4940680899198644e-06,
+      "loss": 0.8847,
+      "step": 825
+    },
+    {
+      "epoch": 0.0703877290157648,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4940517277441073e-06,
+      "loss": 0.841,
+      "step": 826
+    },
+    {
+      "epoch": 0.07047294418406476,
+      "grad_norm": 12.25,
+      "learning_rate": 2.494035343087084e-06,
+      "loss": 0.4975,
+      "step": 827
+    },
+    {
+      "epoch": 0.07055815935236472,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4940189359490915e-06,
+      "loss": 0.6772,
+      "step": 828
+    },
+    {
+      "epoch": 0.07064337452066467,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4940025063304256e-06,
+      "loss": 0.7616,
+      "step": 829
+    },
+    {
+      "epoch": 0.07072858968896463,
+      "grad_norm": 35.25,
+      "learning_rate": 2.4939860542313833e-06,
+      "loss": 1.4091,
+      "step": 830
+    },
+    {
+      "epoch": 0.07081380485726459,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4939695796522624e-06,
+      "loss": 0.5019,
+      "step": 831
+    },
+    {
+      "epoch": 0.07089902002556454,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.49395308259336e-06,
+      "loss": 0.656,
+      "step": 832
+    },
+    {
+      "epoch": 0.07098423519386451,
+      "grad_norm": 18.375,
+      "learning_rate": 2.493936563054975e-06,
+      "loss": 0.8033,
+      "step": 833
+    },
+    {
+      "epoch": 0.07106945036216447,
+      "grad_norm": 16.5,
+      "learning_rate": 2.493920021037405e-06,
+      "loss": 1.0798,
+      "step": 834
+    },
+    {
+      "epoch": 0.07115466553046443,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4939034565409493e-06,
+      "loss": 0.5937,
+      "step": 835
+    },
+    {
+      "epoch": 0.07123988069876439,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.493886869565908e-06,
+      "loss": 0.661,
+      "step": 836
+    },
+    {
+      "epoch": 0.07132509586706434,
+      "grad_norm": 27.25,
+      "learning_rate": 2.4938702601125792e-06,
+      "loss": 1.5583,
+      "step": 837
+    },
+    {
+      "epoch": 0.0714103110353643,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4938536281812645e-06,
+      "loss": 0.5201,
+      "step": 838
+    },
+    {
+      "epoch": 0.07149552620366426,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.493836973772264e-06,
+      "loss": 1.0613,
+      "step": 839
+    },
+    {
+      "epoch": 0.07158074137196421,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4938202968858784e-06,
+      "loss": 0.6031,
+      "step": 840
+    },
+    {
+      "epoch": 0.07166595654026417,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4938035975224092e-06,
+      "loss": 0.6495,
+      "step": 841
+    },
+    {
+      "epoch": 0.07175117170856413,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4937868756821585e-06,
+      "loss": 0.77,
+      "step": 842
+    },
+    {
+      "epoch": 0.07183638687686408,
+      "grad_norm": 11.875,
+      "learning_rate": 2.493770131365428e-06,
+      "loss": 0.5273,
+      "step": 843
+    },
+    {
+      "epoch": 0.07192160204516404,
+      "grad_norm": 27.0,
+      "learning_rate": 2.493753364572521e-06,
+      "loss": 1.2015,
+      "step": 844
+    },
+    {
+      "epoch": 0.072006817213464,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4937365753037394e-06,
+      "loss": 0.7516,
+      "step": 845
+    },
+    {
+      "epoch": 0.07209203238176395,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4937197635593875e-06,
+      "loss": 1.0393,
+      "step": 846
+    },
+    {
+      "epoch": 0.07217724755006391,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4937029293397685e-06,
+      "loss": 1.0639,
+      "step": 847
+    },
+    {
+      "epoch": 0.07226246271836387,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4936860726451874e-06,
+      "loss": 0.6474,
+      "step": 848
+    },
+    {
+      "epoch": 0.07234767788666382,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.493669193475948e-06,
+      "loss": 0.8118,
+      "step": 849
+    },
+    {
+      "epoch": 0.07243289305496378,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.493652291832356e-06,
+      "loss": 0.8965,
+      "step": 850
+    },
+    {
+      "epoch": 0.07251810822326374,
+      "grad_norm": 9.8125,
+      "learning_rate": 2.4936353677147165e-06,
+      "loss": 0.3174,
+      "step": 851
+    },
+    {
+      "epoch": 0.0726033233915637,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.493618421123335e-06,
+      "loss": 1.0894,
+      "step": 852
+    },
+    {
+      "epoch": 0.07268853855986365,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4936014520585185e-06,
+      "loss": 0.7397,
+      "step": 853
+    },
+    {
+      "epoch": 0.07277375372816361,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.493584460520573e-06,
+      "loss": 0.711,
+      "step": 854
+    },
+    {
+      "epoch": 0.07285896889646357,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4935674465098054e-06,
+      "loss": 0.7559,
+      "step": 855
+    },
+    {
+      "epoch": 0.07294418406476352,
+      "grad_norm": 16.5,
+      "learning_rate": 2.493550410026524e-06,
+      "loss": 0.669,
+      "step": 856
+    },
+    {
+      "epoch": 0.07302939923306348,
+      "grad_norm": 16.5,
+      "learning_rate": 2.493533351071036e-06,
+      "loss": 1.0556,
+      "step": 857
+    },
+    {
+      "epoch": 0.07311461440136344,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4935162696436497e-06,
+      "loss": 0.7872,
+      "step": 858
+    },
+    {
+      "epoch": 0.0731998295696634,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4934991657446743e-06,
+      "loss": 0.8275,
+      "step": 859
+    },
+    {
+      "epoch": 0.07328504473796335,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.4934820393744187e-06,
+      "loss": 0.6826,
+      "step": 860
+    },
+    {
+      "epoch": 0.07337025990626332,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4934648905331923e-06,
+      "loss": 0.8516,
+      "step": 861
+    },
+    {
+      "epoch": 0.07345547507456328,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.493447719221304e-06,
+      "loss": 0.9051,
+      "step": 862
+    },
+    {
+      "epoch": 0.07354069024286324,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4934305254390657e-06,
+      "loss": 0.9234,
+      "step": 863
+    },
+    {
+      "epoch": 0.07362590541116319,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4934133091867873e-06,
+      "loss": 0.6714,
+      "step": 864
+    },
+    {
+      "epoch": 0.07371112057946315,
+      "grad_norm": 13.75,
+      "learning_rate": 2.49339607046478e-06,
+      "loss": 0.6635,
+      "step": 865
+    },
+    {
+      "epoch": 0.0737963357477631,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4933788092733556e-06,
+      "loss": 0.9683,
+      "step": 866
+    },
+    {
+      "epoch": 0.07388155091606306,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4933615256128256e-06,
+      "loss": 1.2642,
+      "step": 867
+    },
+    {
+      "epoch": 0.07396676608436302,
+      "grad_norm": 15.125,
+      "learning_rate": 2.493344219483503e-06,
+      "loss": 1.0422,
+      "step": 868
+    },
+    {
+      "epoch": 0.07405198125266298,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4933268908857e-06,
+      "loss": 0.8265,
+      "step": 869
+    },
+    {
+      "epoch": 0.07413719642096293,
+      "grad_norm": 13.25,
+      "learning_rate": 2.493309539819729e-06,
+      "loss": 0.8263,
+      "step": 870
+    },
+    {
+      "epoch": 0.07422241158926289,
+      "grad_norm": 15.75,
+      "learning_rate": 2.493292166285905e-06,
+      "loss": 0.7975,
+      "step": 871
+    },
+    {
+      "epoch": 0.07430762675756285,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4932747702845413e-06,
+      "loss": 0.5741,
+      "step": 872
+    },
+    {
+      "epoch": 0.0743928419258628,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4932573518159522e-06,
+      "loss": 0.9232,
+      "step": 873
+    },
+    {
+      "epoch": 0.07447805709416276,
+      "grad_norm": 12.5,
+      "learning_rate": 2.493239910880453e-06,
+      "loss": 0.5227,
+      "step": 874
+    },
+    {
+      "epoch": 0.07456327226246272,
+      "grad_norm": 14.125,
+      "learning_rate": 2.493222447478358e-06,
+      "loss": 0.6978,
+      "step": 875
+    },
+    {
+      "epoch": 0.07464848743076268,
+      "grad_norm": 15.875,
+      "learning_rate": 2.493204961609983e-06,
+      "loss": 1.008,
+      "step": 876
+    },
+    {
+      "epoch": 0.07473370259906263,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4931874532756447e-06,
+      "loss": 0.8061,
+      "step": 877
+    },
+    {
+      "epoch": 0.07481891776736259,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.493169922475658e-06,
+      "loss": 0.7792,
+      "step": 878
+    },
+    {
+      "epoch": 0.07490413293566255,
+      "grad_norm": 17.75,
+      "learning_rate": 2.493152369210342e-06,
+      "loss": 1.2115,
+      "step": 879
+    },
+    {
+      "epoch": 0.0749893481039625,
+      "grad_norm": 15.625,
+      "learning_rate": 2.493134793480012e-06,
+      "loss": 0.8823,
+      "step": 880
+    },
+    {
+      "epoch": 0.07507456327226246,
+      "grad_norm": 24.875,
+      "learning_rate": 2.493117195284986e-06,
+      "loss": 1.1535,
+      "step": 881
+    },
+    {
+      "epoch": 0.07515977844056242,
+      "grad_norm": 22.25,
+      "learning_rate": 2.493099574625583e-06,
+      "loss": 1.3755,
+      "step": 882
+    },
+    {
+      "epoch": 0.07524499360886237,
+      "grad_norm": 13.75,
+      "learning_rate": 2.49308193150212e-06,
+      "loss": 0.935,
+      "step": 883
+    },
+    {
+      "epoch": 0.07533020877716233,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.493064265914916e-06,
+      "loss": 1.0063,
+      "step": 884
+    },
+    {
+      "epoch": 0.07541542394546229,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4930465778642916e-06,
+      "loss": 0.5326,
+      "step": 885
+    },
+    {
+      "epoch": 0.07550063911376224,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4930288673505652e-06,
+      "loss": 0.6159,
+      "step": 886
+    },
+    {
+      "epoch": 0.0755858542820622,
+      "grad_norm": 21.5,
+      "learning_rate": 2.4930111343740574e-06,
+      "loss": 0.887,
+      "step": 887
+    },
+    {
+      "epoch": 0.07567106945036216,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4929933789350884e-06,
+      "loss": 0.9039,
+      "step": 888
+    },
+    {
+      "epoch": 0.07575628461866213,
+      "grad_norm": 18.625,
+      "learning_rate": 2.492975601033979e-06,
+      "loss": 0.9532,
+      "step": 889
+    },
+    {
+      "epoch": 0.07584149978696209,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4929578006710508e-06,
+      "loss": 0.6734,
+      "step": 890
+    },
+    {
+      "epoch": 0.07592671495526204,
+      "grad_norm": 18.25,
+      "learning_rate": 2.492939977846625e-06,
+      "loss": 0.8263,
+      "step": 891
+    },
+    {
+      "epoch": 0.076011930123562,
+      "grad_norm": 21.875,
+      "learning_rate": 2.4929221325610243e-06,
+      "loss": 1.0999,
+      "step": 892
+    },
+    {
+      "epoch": 0.07609714529186196,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4929042648145705e-06,
+      "loss": 0.7115,
+      "step": 893
+    },
+    {
+      "epoch": 0.07618236046016191,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4928863746075867e-06,
+      "loss": 0.6361,
+      "step": 894
+    },
+    {
+      "epoch": 0.07626757562846187,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4928684619403964e-06,
+      "loss": 1.11,
+      "step": 895
+    },
+    {
+      "epoch": 0.07635279079676183,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4928505268133234e-06,
+      "loss": 0.8011,
+      "step": 896
+    },
+    {
+      "epoch": 0.07643800596506178,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4928325692266914e-06,
+      "loss": 0.3879,
+      "step": 897
+    },
+    {
+      "epoch": 0.07652322113336174,
+      "grad_norm": 18.75,
+      "learning_rate": 2.492814589180825e-06,
+      "loss": 1.1808,
+      "step": 898
+    },
+    {
+      "epoch": 0.0766084363016617,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4927965866760494e-06,
+      "loss": 1.0175,
+      "step": 899
+    },
+    {
+      "epoch": 0.07669365146996165,
+      "grad_norm": 18.25,
+      "learning_rate": 2.49277856171269e-06,
+      "loss": 1.067,
+      "step": 900
+    },
+    {
+      "epoch": 0.07677886663826161,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4927605142910716e-06,
+      "loss": 0.6143,
+      "step": 901
+    },
+    {
+      "epoch": 0.07686408180656157,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4927424444115217e-06,
+      "loss": 0.7555,
+      "step": 902
+    },
+    {
+      "epoch": 0.07694929697486153,
+      "grad_norm": 18.375,
+      "learning_rate": 2.492724352074366e-06,
+      "loss": 0.511,
+      "step": 903
+    },
+    {
+      "epoch": 0.07703451214316148,
+      "grad_norm": 12.25,
+      "learning_rate": 2.492706237279931e-06,
+      "loss": 0.3614,
+      "step": 904
+    },
+    {
+      "epoch": 0.07711972731146144,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4926881000285453e-06,
+      "loss": 0.9427,
+      "step": 905
+    },
+    {
+      "epoch": 0.0772049424797614,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4926699403205356e-06,
+      "loss": 0.8772,
+      "step": 906
+    },
+    {
+      "epoch": 0.07729015764806135,
+      "grad_norm": 13.5,
+      "learning_rate": 2.492651758156231e-06,
+      "loss": 0.6097,
+      "step": 907
+    },
+    {
+      "epoch": 0.07737537281636131,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.492633553535959e-06,
+      "loss": 1.0343,
+      "step": 908
+    },
+    {
+      "epoch": 0.07746058798466127,
+      "grad_norm": 25.375,
+      "learning_rate": 2.4926153264600494e-06,
+      "loss": 0.4935,
+      "step": 909
+    },
+    {
+      "epoch": 0.07754580315296122,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4925970769288314e-06,
+      "loss": 0.5237,
+      "step": 910
+    },
+    {
+      "epoch": 0.07763101832126118,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.492578804942635e-06,
+      "loss": 0.8587,
+      "step": 911
+    },
+    {
+      "epoch": 0.07771623348956114,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4925605105017893e-06,
+      "loss": 0.9437,
+      "step": 912
+    },
+    {
+      "epoch": 0.0778014486578611,
+      "grad_norm": 15.625,
+      "learning_rate": 2.492542193606626e-06,
+      "loss": 0.8953,
+      "step": 913
+    },
+    {
+      "epoch": 0.07788666382616105,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4925238542574758e-06,
+      "loss": 0.9435,
+      "step": 914
+    },
+    {
+      "epoch": 0.07797187899446101,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4925054924546702e-06,
+      "loss": 1.2316,
+      "step": 915
+    },
+    {
+      "epoch": 0.07805709416276096,
+      "grad_norm": 23.875,
+      "learning_rate": 2.492487108198541e-06,
+      "loss": 1.1427,
+      "step": 916
+    },
+    {
+      "epoch": 0.07814230933106094,
+      "grad_norm": 12.125,
+      "learning_rate": 2.49246870148942e-06,
+      "loss": 0.6263,
+      "step": 917
+    },
+    {
+      "epoch": 0.07822752449936089,
+      "grad_norm": 10.5,
+      "learning_rate": 2.4924502723276403e-06,
+      "loss": 0.4651,
+      "step": 918
+    },
+    {
+      "epoch": 0.07831273966766085,
+      "grad_norm": 13.25,
+      "learning_rate": 2.492431820713535e-06,
+      "loss": 0.3829,
+      "step": 919
+    },
+    {
+      "epoch": 0.0783979548359608,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4924133466474372e-06,
+      "loss": 0.4553,
+      "step": 920
+    },
+    {
+      "epoch": 0.07848317000426076,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4923948501296805e-06,
+      "loss": 0.7863,
+      "step": 921
+    },
+    {
+      "epoch": 0.07856838517256072,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4923763311606003e-06,
+      "loss": 0.8902,
+      "step": 922
+    },
+    {
+      "epoch": 0.07865360034086068,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4923577897405297e-06,
+      "loss": 0.8082,
+      "step": 923
+    },
+    {
+      "epoch": 0.07873881550916063,
+      "grad_norm": 16.875,
+      "learning_rate": 2.492339225869805e-06,
+      "loss": 0.9792,
+      "step": 924
+    },
+    {
+      "epoch": 0.07882403067746059,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4923206395487616e-06,
+      "loss": 1.0222,
+      "step": 925
+    },
+    {
+      "epoch": 0.07890924584576055,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4923020307777344e-06,
+      "loss": 0.6075,
+      "step": 926
+    },
+    {
+      "epoch": 0.0789944610140605,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4922833995570604e-06,
+      "loss": 0.7074,
+      "step": 927
+    },
+    {
+      "epoch": 0.07907967618236046,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.492264745887076e-06,
+      "loss": 0.783,
+      "step": 928
+    },
+    {
+      "epoch": 0.07916489135066042,
+      "grad_norm": 16.875,
+      "learning_rate": 2.492246069768119e-06,
+      "loss": 0.7242,
+      "step": 929
+    },
+    {
+      "epoch": 0.07925010651896038,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4922273712005256e-06,
+      "loss": 0.4453,
+      "step": 930
+    },
+    {
+      "epoch": 0.07933532168726033,
+      "grad_norm": 22.625,
+      "learning_rate": 2.492208650184635e-06,
+      "loss": 1.0361,
+      "step": 931
+    },
+    {
+      "epoch": 0.07942053685556029,
+      "grad_norm": 14.875,
+      "learning_rate": 2.492189906720785e-06,
+      "loss": 0.9549,
+      "step": 932
+    },
+    {
+      "epoch": 0.07950575202386025,
+      "grad_norm": 10.625,
+      "learning_rate": 2.4921711408093137e-06,
+      "loss": 0.4851,
+      "step": 933
+    },
+    {
+      "epoch": 0.0795909671921602,
+      "grad_norm": 10.875,
+      "learning_rate": 2.492152352450561e-06,
+      "loss": 0.3842,
+      "step": 934
+    },
+    {
+      "epoch": 0.07967618236046016,
+      "grad_norm": 18.5,
+      "learning_rate": 2.4921335416448668e-06,
+      "loss": 1.4054,
+      "step": 935
+    },
+    {
+      "epoch": 0.07976139752876012,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4921147083925695e-06,
+      "loss": 0.7233,
+      "step": 936
+    },
+    {
+      "epoch": 0.07984661269706007,
+      "grad_norm": 11.375,
+      "learning_rate": 2.492095852694011e-06,
+      "loss": 0.7253,
+      "step": 937
+    },
+    {
+      "epoch": 0.07993182786536003,
+      "grad_norm": 18.875,
+      "learning_rate": 2.492076974549531e-06,
+      "loss": 0.893,
+      "step": 938
+    },
+    {
+      "epoch": 0.08001704303365999,
+      "grad_norm": 18.75,
+      "learning_rate": 2.492058073959472e-06,
+      "loss": 0.863,
+      "step": 939
+    },
+    {
+      "epoch": 0.08010225820195994,
+      "grad_norm": 21.25,
+      "learning_rate": 2.492039150924174e-06,
+      "loss": 1.0865,
+      "step": 940
+    },
+    {
+      "epoch": 0.0801874733702599,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.492020205443979e-06,
+      "loss": 1.0112,
+      "step": 941
+    },
+    {
+      "epoch": 0.08027268853855986,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.4920012375192305e-06,
+      "loss": 0.53,
+      "step": 942
+    },
+    {
+      "epoch": 0.08035790370685982,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.491982247150271e-06,
+      "loss": 0.679,
+      "step": 943
+    },
+    {
+      "epoch": 0.08044311887515977,
+      "grad_norm": 16.375,
+      "learning_rate": 2.491963234337443e-06,
+      "loss": 0.9403,
+      "step": 944
+    },
+    {
+      "epoch": 0.08052833404345973,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4919441990810904e-06,
+      "loss": 0.8987,
+      "step": 945
+    },
+    {
+      "epoch": 0.0806135492117597,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4919251413815577e-06,
+      "loss": 0.9483,
+      "step": 946
+    },
+    {
+      "epoch": 0.08069876438005966,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4919060612391882e-06,
+      "loss": 1.0219,
+      "step": 947
+    },
+    {
+      "epoch": 0.08078397954835961,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4918869586543276e-06,
+      "loss": 0.8508,
+      "step": 948
+    },
+    {
+      "epoch": 0.08086919471665957,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.491867833627321e-06,
+      "loss": 0.8654,
+      "step": 949
+    },
+    {
+      "epoch": 0.08095440988495953,
+      "grad_norm": 11.375,
+      "learning_rate": 2.491848686158514e-06,
+      "loss": 0.5309,
+      "step": 950
+    },
+    {
+      "epoch": 0.08103962505325948,
+      "grad_norm": 17.875,
+      "learning_rate": 2.491829516248252e-06,
+      "loss": 0.9581,
+      "step": 951
+    },
+    {
+      "epoch": 0.08112484022155944,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.491810323896882e-06,
+      "loss": 0.9936,
+      "step": 952
+    },
+    {
+      "epoch": 0.0812100553898594,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4917911091047507e-06,
+      "loss": 0.9531,
+      "step": 953
+    },
+    {
+      "epoch": 0.08129527055815935,
+      "grad_norm": 9.875,
+      "learning_rate": 2.4917718718722056e-06,
+      "loss": 0.3583,
+      "step": 954
+    },
+    {
+      "epoch": 0.08138048572645931,
+      "grad_norm": 11.75,
+      "learning_rate": 2.491752612199594e-06,
+      "loss": 0.6067,
+      "step": 955
+    },
+    {
+      "epoch": 0.08146570089475927,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.491733330087264e-06,
+      "loss": 0.9601,
+      "step": 956
+    },
+    {
+      "epoch": 0.08155091606305923,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.491714025535564e-06,
+      "loss": 0.8267,
+      "step": 957
+    },
+    {
+      "epoch": 0.08163613123135918,
+      "grad_norm": 11.0,
+      "learning_rate": 2.491694698544843e-06,
+      "loss": 0.5244,
+      "step": 958
+    },
+    {
+      "epoch": 0.08172134639965914,
+      "grad_norm": 12.375,
+      "learning_rate": 2.49167534911545e-06,
+      "loss": 0.5638,
+      "step": 959
+    },
+    {
+      "epoch": 0.0818065615679591,
+      "grad_norm": 16.625,
+      "learning_rate": 2.491655977247735e-06,
+      "loss": 0.8481,
+      "step": 960
+    },
+    {
+      "epoch": 0.08189177673625905,
+      "grad_norm": 21.125,
+      "learning_rate": 2.491636582942048e-06,
+      "loss": 0.7118,
+      "step": 961
+    },
+    {
+      "epoch": 0.08197699190455901,
+      "grad_norm": 24.625,
+      "learning_rate": 2.491617166198739e-06,
+      "loss": 1.2096,
+      "step": 962
+    },
+    {
+      "epoch": 0.08206220707285897,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4915977270181594e-06,
+      "loss": 1.4455,
+      "step": 963
+    },
+    {
+      "epoch": 0.08214742224115892,
+      "grad_norm": 19.75,
+      "learning_rate": 2.491578265400661e-06,
+      "loss": 1.0058,
+      "step": 964
+    },
+    {
+      "epoch": 0.08223263740945888,
+      "grad_norm": 25.5,
+      "learning_rate": 2.4915587813465943e-06,
+      "loss": 0.9515,
+      "step": 965
+    },
+    {
+      "epoch": 0.08231785257775884,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4915392748563115e-06,
+      "loss": 0.8341,
+      "step": 966
+    },
+    {
+      "epoch": 0.0824030677460588,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4915197459301664e-06,
+      "loss": 0.7962,
+      "step": 967
+    },
+    {
+      "epoch": 0.08248828291435875,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4915001945685106e-06,
+      "loss": 0.7048,
+      "step": 968
+    },
+    {
+      "epoch": 0.08257349808265871,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.491480620771698e-06,
+      "loss": 0.5696,
+      "step": 969
+    },
+    {
+      "epoch": 0.08265871325095867,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4914610245400816e-06,
+      "loss": 0.7926,
+      "step": 970
+    },
+    {
+      "epoch": 0.08274392841925862,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.4914414058740166e-06,
+      "loss": 0.5785,
+      "step": 971
+    },
+    {
+      "epoch": 0.08282914358755858,
+      "grad_norm": 20.0,
+      "learning_rate": 2.491421764773857e-06,
+      "loss": 0.9236,
+      "step": 972
+    },
+    {
+      "epoch": 0.08291435875585854,
+      "grad_norm": 16.625,
+      "learning_rate": 2.491402101239958e-06,
+      "loss": 0.998,
+      "step": 973
+    },
+    {
+      "epoch": 0.0829995739241585,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4913824152726746e-06,
+      "loss": 0.3975,
+      "step": 974
+    },
+    {
+      "epoch": 0.08308478909245846,
+      "grad_norm": 19.75,
+      "learning_rate": 2.491362706872362e-06,
+      "loss": 0.9719,
+      "step": 975
+    },
+    {
+      "epoch": 0.08317000426075842,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.491342976039378e-06,
+      "loss": 0.7411,
+      "step": 976
+    },
+    {
+      "epoch": 0.08325521942905838,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4913232227740775e-06,
+      "loss": 0.7216,
+      "step": 977
+    },
+    {
+      "epoch": 0.08334043459735833,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4913034470768184e-06,
+      "loss": 1.3156,
+      "step": 978
+    },
+    {
+      "epoch": 0.08342564976565829,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4912836489479575e-06,
+      "loss": 0.4551,
+      "step": 979
+    },
+    {
+      "epoch": 0.08351086493395825,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4912638283878533e-06,
+      "loss": 1.2054,
+      "step": 980
+    },
+    {
+      "epoch": 0.0835960801022582,
+      "grad_norm": 16.625,
+      "learning_rate": 2.491243985396863e-06,
+      "loss": 1.133,
+      "step": 981
+    },
+    {
+      "epoch": 0.08368129527055816,
+      "grad_norm": 12.25,
+      "learning_rate": 2.491224119975346e-06,
+      "loss": 0.8804,
+      "step": 982
+    },
+    {
+      "epoch": 0.08376651043885812,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4912042321236614e-06,
+      "loss": 0.7901,
+      "step": 983
+    },
+    {
+      "epoch": 0.08385172560715808,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4911843218421673e-06,
+      "loss": 0.7225,
+      "step": 984
+    },
+    {
+      "epoch": 0.08393694077545803,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4911643891312247e-06,
+      "loss": 0.3106,
+      "step": 985
+    },
+    {
+      "epoch": 0.08402215594375799,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.491144433991194e-06,
+      "loss": 1.1128,
+      "step": 986
+    },
+    {
+      "epoch": 0.08410737111205795,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.491124456422435e-06,
+      "loss": 0.7801,
+      "step": 987
+    },
+    {
+      "epoch": 0.0841925862803579,
+      "grad_norm": 14.25,
+      "learning_rate": 2.491104456425309e-06,
+      "loss": 0.9965,
+      "step": 988
+    },
+    {
+      "epoch": 0.08427780144865786,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4910844340001773e-06,
+      "loss": 0.5957,
+      "step": 989
+    },
+    {
+      "epoch": 0.08436301661695782,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4910643891474024e-06,
+      "loss": 0.7323,
+      "step": 990
+    },
+    {
+      "epoch": 0.08444823178525777,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4910443218673453e-06,
+      "loss": 0.9596,
+      "step": 991
+    },
+    {
+      "epoch": 0.08453344695355773,
+      "grad_norm": 16.25,
+      "learning_rate": 2.49102423216037e-06,
+      "loss": 0.764,
+      "step": 992
+    },
+    {
+      "epoch": 0.08461866212185769,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4910041200268383e-06,
+      "loss": 0.7823,
+      "step": 993
+    },
+    {
+      "epoch": 0.08470387729015764,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4909839854671147e-06,
+      "loss": 0.7277,
+      "step": 994
+    },
+    {
+      "epoch": 0.0847890924584576,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4909638284815624e-06,
+      "loss": 0.5851,
+      "step": 995
+    },
+    {
+      "epoch": 0.08487430762675756,
+      "grad_norm": 19.875,
+      "learning_rate": 2.4909436490705456e-06,
+      "loss": 1.0849,
+      "step": 996
+    },
+    {
+      "epoch": 0.08495952279505752,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4909234472344297e-06,
+      "loss": 1.1363,
+      "step": 997
+    },
+    {
+      "epoch": 0.08504473796335747,
+      "grad_norm": 30.375,
+      "learning_rate": 2.4909032229735786e-06,
+      "loss": 1.4282,
+      "step": 998
+    },
+    {
+      "epoch": 0.08512995313165743,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4908829762883586e-06,
+      "loss": 0.7139,
+      "step": 999
+    },
+    {
+      "epoch": 0.08521516829995739,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4908627071791357e-06,
+      "loss": 0.5121,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08530038346825734,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4908424156462757e-06,
+      "loss": 0.9741,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08538559863655731,
+      "grad_norm": 21.25,
+      "learning_rate": 2.490822101690146e-06,
+      "loss": 1.3311,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08547081380485727,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4908017653111125e-06,
+      "loss": 0.6252,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08555602897315723,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4907814065095437e-06,
+      "loss": 1.0719,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08564124414145718,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.490761025285807e-06,
+      "loss": 0.6015,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08572645930975714,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4907406216402714e-06,
+      "loss": 0.9579,
+      "step": 1006
+    },
+    {
+      "epoch": 0.0858116744780571,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4907201955733047e-06,
+      "loss": 0.9232,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08589688964635706,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.490699747085276e-06,
+      "loss": 0.7117,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08598210481465701,
+      "grad_norm": 17.75,
+      "learning_rate": 2.490679276176556e-06,
+      "loss": 0.7037,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08606731998295697,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4906587828475136e-06,
+      "loss": 0.9069,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08615253515125693,
+      "grad_norm": 15.625,
+      "learning_rate": 2.490638267098519e-06,
+      "loss": 0.8875,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08623775031955688,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4906177289299438e-06,
+      "loss": 0.9406,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08632296548785684,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4905971683421588e-06,
+      "loss": 0.6685,
+      "step": 1013
+    },
+    {
+      "epoch": 0.0864081806561568,
+      "grad_norm": 14.375,
+      "learning_rate": 2.490576585335535e-06,
+      "loss": 0.8152,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08649339582445675,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4905559799104457e-06,
+      "loss": 0.5096,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08657861099275671,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4905353520672617e-06,
+      "loss": 0.8894,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08666382616105667,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4905147018063563e-06,
+      "loss": 0.5815,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08674904132935662,
+      "grad_norm": 22.875,
+      "learning_rate": 2.490494029128103e-06,
+      "loss": 1.2529,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08683425649765658,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.490473334032875e-06,
+      "loss": 0.8467,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08691947166595654,
+      "grad_norm": 12.125,
+      "learning_rate": 2.490452616521047e-06,
+      "loss": 0.7415,
+      "step": 1020
+    },
+    {
+      "epoch": 0.0870046868342565,
+      "grad_norm": 20.5,
+      "learning_rate": 2.4904318765929924e-06,
+      "loss": 1.1804,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08708990200255645,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4904111142490868e-06,
+      "loss": 0.4877,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08717511717085641,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4903903294897043e-06,
+      "loss": 1.1736,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08726033233915637,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4903695223152217e-06,
+      "loss": 1.029,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08734554750745632,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4903486927260145e-06,
+      "loss": 0.6916,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08743076267575628,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4903278407224597e-06,
+      "loss": 0.8215,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08751597784405624,
+      "grad_norm": 10.625,
+      "learning_rate": 2.490306966304933e-06,
+      "loss": 0.7558,
+      "step": 1027
+    },
+    {
+      "epoch": 0.0876011930123562,
+      "grad_norm": 22.0,
+      "learning_rate": 2.490286069473812e-06,
+      "loss": 1.0058,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08768640818065615,
+      "grad_norm": 10.625,
+      "learning_rate": 2.490265150229475e-06,
+      "loss": 0.4715,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08777162334895612,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4902442085722992e-06,
+      "loss": 0.4575,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08785683851725608,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4902232445026636e-06,
+      "loss": 0.5331,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08794205368555603,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4902022580209465e-06,
+      "loss": 0.9019,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08802726885385599,
+      "grad_norm": 14.625,
+      "learning_rate": 2.490181249127528e-06,
+      "loss": 0.768,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08811248402215595,
+      "grad_norm": 14.5,
+      "learning_rate": 2.490160217822787e-06,
+      "loss": 0.8467,
+      "step": 1034
+    },
+    {
+      "epoch": 0.0881976991904559,
+      "grad_norm": 12.125,
+      "learning_rate": 2.490139164107104e-06,
+      "loss": 0.5999,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08828291435875586,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.490118087980859e-06,
+      "loss": 1.0654,
+      "step": 1036
+    },
+    {
+      "epoch": 0.08836812952705582,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4900969894444328e-06,
+      "loss": 0.7843,
+      "step": 1037
+    },
+    {
+      "epoch": 0.08845334469535578,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4900758684982072e-06,
+      "loss": 0.6245,
+      "step": 1038
+    },
+    {
+      "epoch": 0.08853855986365573,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4900547251425638e-06,
+      "loss": 0.3501,
+      "step": 1039
+    },
+    {
+      "epoch": 0.08862377503195569,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4900335593778844e-06,
+      "loss": 0.9506,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08870899020025565,
+      "grad_norm": 13.625,
+      "learning_rate": 2.490012371204552e-06,
+      "loss": 1.1008,
+      "step": 1041
+    },
+    {
+      "epoch": 0.0887942053685556,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4899911606229487e-06,
+      "loss": 0.8739,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08887942053685556,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.489969927633459e-06,
+      "loss": 0.4483,
+      "step": 1043
+    },
+    {
+      "epoch": 0.08896463570515552,
+      "grad_norm": 27.125,
+      "learning_rate": 2.4899486722364645e-06,
+      "loss": 1.1257,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08904985087345547,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4899273944323516e-06,
+      "loss": 1.1577,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08913506604175543,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4899060942215037e-06,
+      "loss": 0.8145,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08922028121005539,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4898847716043055e-06,
+      "loss": 0.5171,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08930549637835535,
+      "grad_norm": 21.0,
+      "learning_rate": 2.489863426581143e-06,
+      "loss": 1.287,
+      "step": 1048
+    },
+    {
+      "epoch": 0.0893907115466553,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4898420591524016e-06,
+      "loss": 0.7502,
+      "step": 1049
+    },
+    {
+      "epoch": 0.08947592671495526,
+      "grad_norm": 24.625,
+      "learning_rate": 2.4898206693184672e-06,
+      "loss": 1.3815,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08956114188325522,
+      "grad_norm": 13.875,
+      "learning_rate": 2.489799257079727e-06,
+      "loss": 0.6419,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08964635705155517,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4897778224365672e-06,
+      "loss": 1.0908,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08973157221985513,
+      "grad_norm": 24.5,
+      "learning_rate": 2.4897563653893754e-06,
+      "loss": 1.2304,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08981678738815509,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4897348859385395e-06,
+      "loss": 1.2056,
+      "step": 1054
+    },
+    {
+      "epoch": 0.08990200255645504,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4897133840844474e-06,
+      "loss": 0.6569,
+      "step": 1055
+    },
+    {
+      "epoch": 0.089987217724755,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4896918598274884e-06,
+      "loss": 0.5882,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09007243289305496,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4896703131680503e-06,
+      "loss": 1.1163,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09015764806135493,
+      "grad_norm": 15.375,
+      "learning_rate": 2.489648744106523e-06,
+      "loss": 0.7096,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09024286322965488,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4896271526432966e-06,
+      "loss": 0.7612,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09032807839795484,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.489605538778761e-06,
+      "loss": 0.6753,
+      "step": 1060
+    },
+    {
+      "epoch": 0.0904132935662548,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4895839025133067e-06,
+      "loss": 0.5948,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09049850873455476,
+      "grad_norm": 14.75,
+      "learning_rate": 2.489562243847325e-06,
+      "loss": 0.6922,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09058372390285471,
+      "grad_norm": 18.375,
+      "learning_rate": 2.489540562781207e-06,
+      "loss": 0.8443,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09066893907115467,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4895188593153444e-06,
+      "loss": 0.6662,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09075415423945463,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4894971334501295e-06,
+      "loss": 1.2845,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09083936940775458,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4894753851859555e-06,
+      "loss": 1.016,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09092458457605454,
+      "grad_norm": 15.0,
+      "learning_rate": 2.489453614523214e-06,
+      "loss": 0.7773,
+      "step": 1067
+    },
+    {
+      "epoch": 0.0910097997443545,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4894318214623004e-06,
+      "loss": 0.7754,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09109501491265445,
+      "grad_norm": 20.5,
+      "learning_rate": 2.4894100060036066e-06,
+      "loss": 1.1329,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09118023008095441,
+      "grad_norm": 11.75,
+      "learning_rate": 2.489388168147528e-06,
+      "loss": 0.4827,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09126544524925437,
+      "grad_norm": 18.0,
+      "learning_rate": 2.4893663078944593e-06,
+      "loss": 0.8682,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09135066041755432,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4893444252447948e-06,
+      "loss": 0.9551,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09143587558585428,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4893225201989295e-06,
+      "loss": 0.7951,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09152109075415424,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.489300592757261e-06,
+      "loss": 0.995,
+      "step": 1074
+    },
+    {
+      "epoch": 0.0916063059224542,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4892786429201843e-06,
+      "loss": 1.0128,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09169152109075415,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4892566706880963e-06,
+      "loss": 0.8333,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09177673625905411,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.489234676061394e-06,
+      "loss": 0.9616,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09186195142735407,
+      "grad_norm": 21.0,
+      "learning_rate": 2.489212659040475e-06,
+      "loss": 1.1042,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09194716659565402,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4891906196257377e-06,
+      "loss": 0.821,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09203238176395398,
+      "grad_norm": 21.625,
+      "learning_rate": 2.489168557817579e-06,
+      "loss": 1.1489,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09211759693225394,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.489146473616399e-06,
+      "loss": 0.5219,
+      "step": 1081
+    },
+    {
+      "epoch": 0.0922028121005539,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4891243670225955e-06,
+      "loss": 0.7303,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09228802726885385,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4891022380365693e-06,
+      "loss": 0.7661,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09237324243715381,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4890800866587193e-06,
+      "loss": 1.1133,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09245845760545376,
+      "grad_norm": 17.875,
+      "learning_rate": 2.489057912889446e-06,
+      "loss": 0.7884,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09254367277375372,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4890357167291507e-06,
+      "loss": 0.6475,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09262888794205369,
+      "grad_norm": 18.375,
+      "learning_rate": 2.489013498178234e-06,
+      "loss": 0.9722,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09271410311035365,
+      "grad_norm": 17.375,
+      "learning_rate": 2.488991257237097e-06,
+      "loss": 0.8523,
+      "step": 1088
+    },
+    {
+      "epoch": 0.0927993182786536,
+      "grad_norm": 18.875,
+      "learning_rate": 2.488968993906142e-06,
+      "loss": 0.7299,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09288453344695356,
+      "grad_norm": 18.375,
+      "learning_rate": 2.488946708185772e-06,
+      "loss": 1.2717,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09296974861525352,
+      "grad_norm": 17.375,
+      "learning_rate": 2.488924400076389e-06,
+      "loss": 0.8368,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09305496378355348,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4889020695783962e-06,
+      "loss": 0.7039,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09314017895185343,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4888797166921967e-06,
+      "loss": 1.0184,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09322539412015339,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4888573414181956e-06,
+      "loss": 0.8016,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09331060928845335,
+      "grad_norm": 13.75,
+      "learning_rate": 2.488834943756796e-06,
+      "loss": 0.608,
+      "step": 1095
+    },
+    {
+      "epoch": 0.0933958244567533,
+      "grad_norm": 19.875,
+      "learning_rate": 2.4888125237084037e-06,
+      "loss": 1.0198,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09348103962505326,
+      "grad_norm": 15.875,
+      "learning_rate": 2.488790081273423e-06,
+      "loss": 1.0101,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09356625479335322,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.48876761645226e-06,
+      "loss": 0.9163,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09365146996165317,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4887451292453208e-06,
+      "loss": 1.0795,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09373668512995313,
+      "grad_norm": 19.125,
+      "learning_rate": 2.488722619653011e-06,
+      "loss": 0.7447,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09382190029825309,
+      "grad_norm": 10.25,
+      "learning_rate": 2.4887000876757383e-06,
+      "loss": 0.4926,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09390711546655305,
+      "grad_norm": 18.125,
+      "learning_rate": 2.488677533313909e-06,
+      "loss": 1.1527,
+      "step": 1102
+    },
+    {
+      "epoch": 0.093992330634853,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4886549565679314e-06,
+      "loss": 1.2142,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09407754580315296,
+      "grad_norm": 16.125,
+      "learning_rate": 2.488632357438213e-06,
+      "loss": 1.0449,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09416276097145292,
+      "grad_norm": 26.625,
+      "learning_rate": 2.488609735925162e-06,
+      "loss": 1.2415,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09424797613975287,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.4885870920291883e-06,
+      "loss": 0.5278,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09433319130805283,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4885644257506997e-06,
+      "loss": 0.5945,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09441840647635279,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4885417370901066e-06,
+      "loss": 1.0857,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09450362164465274,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4885190260478194e-06,
+      "loss": 1.0826,
+      "step": 1109
+    },
+    {
+      "epoch": 0.0945888368129527,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4884962926242473e-06,
+      "loss": 0.5277,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09467405198125266,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4884735368198022e-06,
+      "loss": 0.8315,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09475926714955261,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.4884507586348943e-06,
+      "loss": 0.7635,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09484448231785257,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4884279580699367e-06,
+      "loss": 0.8582,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09492969748615253,
+      "grad_norm": 21.0,
+      "learning_rate": 2.48840513512534e-06,
+      "loss": 1.1656,
+      "step": 1114
+    },
+    {
+      "epoch": 0.0950149126544525,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4883822898015174e-06,
+      "loss": 0.7417,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09510012782275246,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4883594220988815e-06,
+      "loss": 1.1427,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09518534299105241,
+      "grad_norm": 19.5,
+      "learning_rate": 2.488336532017845e-06,
+      "loss": 0.9757,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09527055815935237,
+      "grad_norm": 21.75,
+      "learning_rate": 2.488313619558823e-06,
+      "loss": 1.0315,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09535577332765233,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.488290684722228e-06,
+      "loss": 0.6196,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09544098849595228,
+      "grad_norm": 17.0,
+      "learning_rate": 2.488267727508476e-06,
+      "loss": 1.1974,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09552620366425224,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4882447479179804e-06,
+      "loss": 0.638,
+      "step": 1121
+    },
+    {
+      "epoch": 0.0956114188325522,
+      "grad_norm": 11.375,
+      "learning_rate": 2.488221745951157e-06,
+      "loss": 0.6605,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09569663400085215,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4881987216084217e-06,
+      "loss": 0.5806,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09578184916915211,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4881756748901903e-06,
+      "loss": 1.0507,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09586706433745207,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.488152605796879e-06,
+      "loss": 1.028,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09595227950575203,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.488129514328906e-06,
+      "loss": 0.5205,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09603749467405198,
+      "grad_norm": 16.0,
+      "learning_rate": 2.488106400486687e-06,
+      "loss": 0.8333,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09612270984235194,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4880832642706404e-06,
+      "loss": 0.7568,
+      "step": 1128
+    },
+    {
+      "epoch": 0.0962079250106519,
+      "grad_norm": 26.125,
+      "learning_rate": 2.4880601056811843e-06,
+      "loss": 1.2304,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09629314017895185,
+      "grad_norm": 20.5,
+      "learning_rate": 2.488036924718737e-06,
+      "loss": 1.0342,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09637835534725181,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4880137213837176e-06,
+      "loss": 1.3621,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09646357051555177,
+      "grad_norm": 27.75,
+      "learning_rate": 2.487990495676545e-06,
+      "loss": 1.5038,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09654878568385172,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.48796724759764e-06,
+      "loss": 0.9412,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09663400085215168,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.487943977147421e-06,
+      "loss": 0.467,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09671921602045164,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4879206843263105e-06,
+      "loss": 0.7452,
+      "step": 1135
+    },
+    {
+      "epoch": 0.0968044311887516,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.487897369134728e-06,
+      "loss": 0.6697,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09688964635705155,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4878740315730947e-06,
+      "loss": 1.3572,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09697486152535151,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.487850671641834e-06,
+      "loss": 0.9798,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09706007669365146,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4878272893413656e-06,
+      "loss": 0.5652,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09714529186195142,
+      "grad_norm": 14.625,
+      "learning_rate": 2.487803884672114e-06,
+      "loss": 0.912,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09723050703025138,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4877804576345017e-06,
+      "loss": 0.3384,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09731572219855134,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4877570082289518e-06,
+      "loss": 1.1014,
+      "step": 1142
+    },
+    {
+      "epoch": 0.0974009373668513,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.487733536455888e-06,
+      "loss": 0.5668,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09748615253515126,
+      "grad_norm": 11.625,
+      "learning_rate": 2.487710042315734e-06,
+      "loss": 0.6697,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09757136770345122,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4876865258089157e-06,
+      "loss": 0.7408,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09765658287175118,
+      "grad_norm": 11.25,
+      "learning_rate": 2.487662986935857e-06,
+      "loss": 0.5894,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09774179804005113,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.487639425696984e-06,
+      "loss": 0.4808,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09782701320835109,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.487615842092721e-06,
+      "loss": 0.9066,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09791222837665105,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.487592236123496e-06,
+      "loss": 0.5614,
+      "step": 1149
+    },
+    {
+      "epoch": 0.097997443544951,
+      "grad_norm": 10.75,
+      "learning_rate": 2.487568607789735e-06,
+      "loss": 0.3832,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09808265871325096,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4875449570918644e-06,
+      "loss": 0.3155,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09816787388155092,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4875212840303115e-06,
+      "loss": 0.9819,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09825308904985088,
+      "grad_norm": 24.125,
+      "learning_rate": 2.4874975886055054e-06,
+      "loss": 1.5273,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09833830421815083,
+      "grad_norm": 15.75,
+      "learning_rate": 2.487473870817873e-06,
+      "loss": 0.7491,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09842351938645079,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4874501306678436e-06,
+      "loss": 0.9202,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09850873455475075,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4874263681558456e-06,
+      "loss": 0.8345,
+      "step": 1156
+    },
+    {
+      "epoch": 0.0985939497230507,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4874025832823096e-06,
+      "loss": 0.8706,
+      "step": 1157
+    },
+    {
+      "epoch": 0.09867916489135066,
+      "grad_norm": 26.0,
+      "learning_rate": 2.487378776047664e-06,
+      "loss": 0.9641,
+      "step": 1158
+    },
+    {
+      "epoch": 0.09876438005965062,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4873549464523392e-06,
+      "loss": 0.3667,
+      "step": 1159
+    },
+    {
+      "epoch": 0.09884959522795057,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4873310944967673e-06,
+      "loss": 0.8958,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09893481039625053,
+      "grad_norm": 22.625,
+      "learning_rate": 2.487307220181378e-06,
+      "loss": 0.8326,
+      "step": 1161
+    },
+    {
+      "epoch": 0.09902002556455049,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4872833235066024e-06,
+      "loss": 0.745,
+      "step": 1162
+    },
+    {
+      "epoch": 0.09910524073285044,
+      "grad_norm": 13.125,
+      "learning_rate": 2.487259404472874e-06,
+      "loss": 0.6665,
+      "step": 1163
+    },
+    {
+      "epoch": 0.0991904559011504,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.487235463080623e-06,
+      "loss": 0.6353,
+      "step": 1164
+    },
+    {
+      "epoch": 0.09927567106945036,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4872114993302835e-06,
+      "loss": 1.1939,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09936088623775031,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4871875132222877e-06,
+      "loss": 1.0054,
+      "step": 1166
+    },
+    {
+      "epoch": 0.09944610140605027,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.48716350475707e-06,
+      "loss": 0.5284,
+      "step": 1167
+    },
+    {
+      "epoch": 0.09953131657435023,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4871394739350636e-06,
+      "loss": 0.7509,
+      "step": 1168
+    },
+    {
+      "epoch": 0.09961653174265019,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4871154207567027e-06,
+      "loss": 0.8657,
+      "step": 1169
+    },
+    {
+      "epoch": 0.09970174691095014,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.487091345222422e-06,
+      "loss": 0.6932,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09978696207925011,
+      "grad_norm": 30.75,
+      "learning_rate": 2.4870672473326567e-06,
+      "loss": 0.9452,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09987217724755007,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4870431270878427e-06,
+      "loss": 0.6746,
+      "step": 1172
+    },
+    {
+      "epoch": 0.09995739241585003,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.487018984488415e-06,
+      "loss": 0.677,
+      "step": 1173
+    },
+    {
+      "epoch": 0.10004260758414998,
+      "grad_norm": 12.25,
+      "learning_rate": 2.486994819534811e-06,
+      "loss": 0.6327,
+      "step": 1174
+    },
+    {
+      "epoch": 0.10012782275244994,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4869706322274663e-06,
+      "loss": 0.8116,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1002130379207499,
+      "grad_norm": 16.0,
+      "learning_rate": 2.486946422566818e-06,
+      "loss": 0.6631,
+      "step": 1176
+    },
+    {
+      "epoch": 0.10029825308904985,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.486922190553304e-06,
+      "loss": 0.5065,
+      "step": 1177
+    },
+    {
+      "epoch": 0.10038346825734981,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.486897936187363e-06,
+      "loss": 1.0097,
+      "step": 1178
+    },
+    {
+      "epoch": 0.10046868342564977,
+      "grad_norm": 17.75,
+      "learning_rate": 2.486873659469432e-06,
+      "loss": 0.7886,
+      "step": 1179
+    },
+    {
+      "epoch": 0.10055389859394973,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4868493603999503e-06,
+      "loss": 0.8535,
+      "step": 1180
+    },
+    {
+      "epoch": 0.10063911376224968,
+      "grad_norm": 16.25,
+      "learning_rate": 2.486825038979357e-06,
+      "loss": 0.7982,
+      "step": 1181
+    },
+    {
+      "epoch": 0.10072432893054964,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.486800695208091e-06,
+      "loss": 0.5671,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1008095440988496,
+      "grad_norm": 16.125,
+      "learning_rate": 2.486776329086593e-06,
+      "loss": 0.8319,
+      "step": 1183
+    },
+    {
+      "epoch": 0.10089475926714955,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4867519406153036e-06,
+      "loss": 1.1765,
+      "step": 1184
+    },
+    {
+      "epoch": 0.10097997443544951,
+      "grad_norm": 22.125,
+      "learning_rate": 2.4867275297946627e-06,
+      "loss": 1.0212,
+      "step": 1185
+    },
+    {
+      "epoch": 0.10106518960374947,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4867030966251112e-06,
+      "loss": 0.9917,
+      "step": 1186
+    },
+    {
+      "epoch": 0.10115040477204942,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4866786411070914e-06,
+      "loss": 0.8056,
+      "step": 1187
+    },
+    {
+      "epoch": 0.10123561994034938,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.486654163241045e-06,
+      "loss": 0.8442,
+      "step": 1188
+    },
+    {
+      "epoch": 0.10132083510864934,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4866296630274144e-06,
+      "loss": 1.195,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1014060502769493,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4866051404666424e-06,
+      "loss": 0.9384,
+      "step": 1190
+    },
+    {
+      "epoch": 0.10149126544524925,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4865805955591723e-06,
+      "loss": 0.6707,
+      "step": 1191
+    },
+    {
+      "epoch": 0.10157648061354921,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4865560283054466e-06,
+      "loss": 0.805,
+      "step": 1192
+    },
+    {
+      "epoch": 0.10166169578184917,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.48653143870591e-06,
+      "loss": 0.4752,
+      "step": 1193
+    },
+    {
+      "epoch": 0.10174691095014912,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4865068267610077e-06,
+      "loss": 0.9507,
+      "step": 1194
+    },
+    {
+      "epoch": 0.10183212611844908,
+      "grad_norm": 22.625,
+      "learning_rate": 2.486482192471183e-06,
+      "loss": 0.9932,
+      "step": 1195
+    },
+    {
+      "epoch": 0.10191734128674904,
+      "grad_norm": 23.0,
+      "learning_rate": 2.486457535836882e-06,
+      "loss": 0.9625,
+      "step": 1196
+    },
+    {
+      "epoch": 0.10200255645504899,
+      "grad_norm": 21.125,
+      "learning_rate": 2.48643285685855e-06,
+      "loss": 1.0072,
+      "step": 1197
+    },
+    {
+      "epoch": 0.10208777162334895,
+      "grad_norm": 21.5,
+      "learning_rate": 2.4864081555366327e-06,
+      "loss": 1.1831,
+      "step": 1198
+    },
+    {
+      "epoch": 0.10217298679164892,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4863834318715767e-06,
+      "loss": 0.7271,
+      "step": 1199
+    },
+    {
+      "epoch": 0.10225820195994888,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4863586858638287e-06,
+      "loss": 0.5084,
+      "step": 1200
+    },
+    {
+      "epoch": 0.10234341712824883,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4863339175138364e-06,
+      "loss": 0.9377,
+      "step": 1201
+    },
+    {
+      "epoch": 0.10242863229654879,
+      "grad_norm": 14.0,
+      "learning_rate": 2.486309126822047e-06,
+      "loss": 0.7694,
+      "step": 1202
+    },
+    {
+      "epoch": 0.10251384746484875,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.4862843137889083e-06,
+      "loss": 0.4483,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1025990626331487,
+      "grad_norm": 22.375,
+      "learning_rate": 2.4862594784148687e-06,
+      "loss": 1.1014,
+      "step": 1204
+    },
+    {
+      "epoch": 0.10268427780144866,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4862346207003775e-06,
+      "loss": 0.4566,
+      "step": 1205
+    },
+    {
+      "epoch": 0.10276949296974862,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.486209740645883e-06,
+      "loss": 0.8798,
+      "step": 1206
+    },
+    {
+      "epoch": 0.10285470813804858,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.486184838251836e-06,
+      "loss": 0.4533,
+      "step": 1207
+    },
+    {
+      "epoch": 0.10293992330634853,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4861599135186855e-06,
+      "loss": 0.4927,
+      "step": 1208
+    },
+    {
+      "epoch": 0.10302513847464849,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4861349664468825e-06,
+      "loss": 0.7832,
+      "step": 1209
+    },
+    {
+      "epoch": 0.10311035364294845,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4861099970368772e-06,
+      "loss": 0.8285,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1031955688112484,
+      "grad_norm": 13.125,
+      "learning_rate": 2.486085005289122e-06,
+      "loss": 0.5292,
+      "step": 1211
+    },
+    {
+      "epoch": 0.10328078397954836,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4860599912040672e-06,
+      "loss": 0.7756,
+      "step": 1212
+    },
+    {
+      "epoch": 0.10336599914784832,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.486034954782166e-06,
+      "loss": 0.8251,
+      "step": 1213
+    },
+    {
+      "epoch": 0.10345121431614827,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4860098960238697e-06,
+      "loss": 0.8704,
+      "step": 1214
+    },
+    {
+      "epoch": 0.10353642948444823,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4859848149296315e-06,
+      "loss": 0.94,
+      "step": 1215
+    },
+    {
+      "epoch": 0.10362164465274819,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4859597114999054e-06,
+      "loss": 0.6716,
+      "step": 1216
+    },
+    {
+      "epoch": 0.10370685982104814,
+      "grad_norm": 29.375,
+      "learning_rate": 2.4859345857351443e-06,
+      "loss": 1.5182,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1037920749893481,
+      "grad_norm": 21.875,
+      "learning_rate": 2.4859094376358024e-06,
+      "loss": 1.1244,
+      "step": 1218
+    },
+    {
+      "epoch": 0.10387729015764806,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4858842672023337e-06,
+      "loss": 0.8572,
+      "step": 1219
+    },
+    {
+      "epoch": 0.10396250532594802,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.485859074435194e-06,
+      "loss": 0.6713,
+      "step": 1220
+    },
+    {
+      "epoch": 0.10404772049424797,
+      "grad_norm": 21.375,
+      "learning_rate": 2.485833859334838e-06,
+      "loss": 0.9288,
+      "step": 1221
+    },
+    {
+      "epoch": 0.10413293566254793,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4858086219017214e-06,
+      "loss": 0.5945,
+      "step": 1222
+    },
+    {
+      "epoch": 0.10421815083084789,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4857833621363003e-06,
+      "loss": 0.7457,
+      "step": 1223
+    },
+    {
+      "epoch": 0.10430336599914784,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4857580800390312e-06,
+      "loss": 0.7915,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1043885811674478,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.485732775610371e-06,
+      "loss": 0.9525,
+      "step": 1225
+    },
+    {
+      "epoch": 0.10447379633574776,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4857074488507765e-06,
+      "loss": 0.4845,
+      "step": 1226
+    },
+    {
+      "epoch": 0.10455901150404771,
+      "grad_norm": 16.75,
+      "learning_rate": 2.485682099760706e-06,
+      "loss": 0.889,
+      "step": 1227
+    },
+    {
+      "epoch": 0.10464422667234768,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4856567283406177e-06,
+      "loss": 0.7665,
+      "step": 1228
+    },
+    {
+      "epoch": 0.10472944184064764,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4856313345909697e-06,
+      "loss": 0.5197,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1048146570089476,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4856059185122208e-06,
+      "loss": 1.014,
+      "step": 1230
+    },
+    {
+      "epoch": 0.10489987217724756,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4855804801048304e-06,
+      "loss": 0.389,
+      "step": 1231
+    },
+    {
+      "epoch": 0.10498508734554751,
+      "grad_norm": 22.75,
+      "learning_rate": 2.4855550193692585e-06,
+      "loss": 1.3122,
+      "step": 1232
+    },
+    {
+      "epoch": 0.10507030251384747,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.4855295363059646e-06,
+      "loss": 0.7156,
+      "step": 1233
+    },
+    {
+      "epoch": 0.10515551768214743,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.48550403091541e-06,
+      "loss": 0.6026,
+      "step": 1234
+    },
+    {
+      "epoch": 0.10524073285044738,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4854785031980547e-06,
+      "loss": 0.7546,
+      "step": 1235
+    },
+    {
+      "epoch": 0.10532594801874734,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.485452953154361e-06,
+      "loss": 0.4887,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1054111631870473,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4854273807847893e-06,
+      "loss": 0.7915,
+      "step": 1237
+    },
+    {
+      "epoch": 0.10549637835534725,
+      "grad_norm": 16.5,
+      "learning_rate": 2.485401786089803e-06,
+      "loss": 1.1728,
+      "step": 1238
+    },
+    {
+      "epoch": 0.10558159352364721,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4853761690698643e-06,
+      "loss": 1.2652,
+      "step": 1239
+    },
+    {
+      "epoch": 0.10566680869194717,
+      "grad_norm": 24.125,
+      "learning_rate": 2.485350529725436e-06,
+      "loss": 1.2301,
+      "step": 1240
+    },
+    {
+      "epoch": 0.10575202386024712,
+      "grad_norm": 17.375,
+      "learning_rate": 2.485324868056981e-06,
+      "loss": 0.9542,
+      "step": 1241
+    },
+    {
+      "epoch": 0.10583723902854708,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4852991840649638e-06,
+      "loss": 0.576,
+      "step": 1242
+    },
+    {
+      "epoch": 0.10592245419684704,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4852734777498478e-06,
+      "loss": 0.972,
+      "step": 1243
+    },
+    {
+      "epoch": 0.106007669365147,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.485247749112098e-06,
+      "loss": 0.6378,
+      "step": 1244
+    },
+    {
+      "epoch": 0.10609288453344695,
+      "grad_norm": 13.875,
+      "learning_rate": 2.485221998152179e-06,
+      "loss": 0.3143,
+      "step": 1245
+    },
+    {
+      "epoch": 0.10617809970174691,
+      "grad_norm": 11.375,
+      "learning_rate": 2.485196224870557e-06,
+      "loss": 0.6087,
+      "step": 1246
+    },
+    {
+      "epoch": 0.10626331487004687,
+      "grad_norm": 12.875,
+      "learning_rate": 2.485170429267697e-06,
+      "loss": 0.6459,
+      "step": 1247
+    },
+    {
+      "epoch": 0.10634853003834682,
+      "grad_norm": 24.75,
+      "learning_rate": 2.4851446113440647e-06,
+      "loss": 0.9692,
+      "step": 1248
+    },
+    {
+      "epoch": 0.10643374520664678,
+      "grad_norm": 15.125,
+      "learning_rate": 2.485118771100128e-06,
+      "loss": 0.7668,
+      "step": 1249
+    },
+    {
+      "epoch": 0.10651896037494674,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4850929085363525e-06,
+      "loss": 0.6654,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1066041755432467,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4850670236532065e-06,
+      "loss": 0.6495,
+      "step": 1251
+    },
+    {
+      "epoch": 0.10668939071154665,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4850411164511573e-06,
+      "loss": 0.7962,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1067746058798466,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4850151869306734e-06,
+      "loss": 0.7298,
+      "step": 1253
+    },
+    {
+      "epoch": 0.10685982104814656,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4849892350922228e-06,
+      "loss": 0.8274,
+      "step": 1254
+    },
+    {
+      "epoch": 0.10694503621644652,
+      "grad_norm": 22.625,
+      "learning_rate": 2.4849632609362747e-06,
+      "loss": 1.147,
+      "step": 1255
+    },
+    {
+      "epoch": 0.10703025138474649,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4849372644632993e-06,
+      "loss": 0.8296,
+      "step": 1256
+    },
+    {
+      "epoch": 0.10711546655304645,
+      "grad_norm": 14.875,
+      "learning_rate": 2.484911245673765e-06,
+      "loss": 0.73,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1072006817213464,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4848852045681434e-06,
+      "loss": 0.6973,
+      "step": 1258
+    },
+    {
+      "epoch": 0.10728589688964636,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4848591411469035e-06,
+      "loss": 1.1448,
+      "step": 1259
+    },
+    {
+      "epoch": 0.10737111205794632,
+      "grad_norm": 14.0,
+      "learning_rate": 2.484833055410518e-06,
+      "loss": 0.7084,
+      "step": 1260
+    },
+    {
+      "epoch": 0.10745632722624628,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4848069473594573e-06,
+      "loss": 0.8765,
+      "step": 1261
+    },
+    {
+      "epoch": 0.10754154239454623,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.484780816994193e-06,
+      "loss": 0.5018,
+      "step": 1262
+    },
+    {
+      "epoch": 0.10762675756284619,
+      "grad_norm": 10.0,
+      "learning_rate": 2.484754664315198e-06,
+      "loss": 0.4434,
+      "step": 1263
+    },
+    {
+      "epoch": 0.10771197273114615,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4847284893229445e-06,
+      "loss": 0.7946,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1077971878994461,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4847022920179054e-06,
+      "loss": 0.9332,
+      "step": 1265
+    },
+    {
+      "epoch": 0.10788240306774606,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4846760724005544e-06,
+      "loss": 1.1173,
+      "step": 1266
+    },
+    {
+      "epoch": 0.10796761823604602,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.484649830471365e-06,
+      "loss": 0.4956,
+      "step": 1267
+    },
+    {
+      "epoch": 0.10805283340434597,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4846235662308123e-06,
+      "loss": 0.4356,
+      "step": 1268
+    },
+    {
+      "epoch": 0.10813804857264593,
+      "grad_norm": 17.0,
+      "learning_rate": 2.48459727967937e-06,
+      "loss": 1.0364,
+      "step": 1269
+    },
+    {
+      "epoch": 0.10822326374094589,
+      "grad_norm": 16.0,
+      "learning_rate": 2.484570970817513e-06,
+      "loss": 0.8467,
+      "step": 1270
+    },
+    {
+      "epoch": 0.10830847890924584,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4845446396457173e-06,
+      "loss": 0.591,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1083936940775458,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4845182861644584e-06,
+      "loss": 0.4932,
+      "step": 1272
+    },
+    {
+      "epoch": 0.10847890924584576,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4844919103742126e-06,
+      "loss": 0.8037,
+      "step": 1273
+    },
+    {
+      "epoch": 0.10856412441414572,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4844655122754567e-06,
+      "loss": 0.5426,
+      "step": 1274
+    },
+    {
+      "epoch": 0.10864933958244567,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4844390918686678e-06,
+      "loss": 0.7429,
+      "step": 1275
+    },
+    {
+      "epoch": 0.10873455475074563,
+      "grad_norm": 33.25,
+      "learning_rate": 2.4844126491543234e-06,
+      "loss": 1.3516,
+      "step": 1276
+    },
+    {
+      "epoch": 0.10881976991904559,
+      "grad_norm": 15.375,
+      "learning_rate": 2.484386184132901e-06,
+      "loss": 0.9762,
+      "step": 1277
+    },
+    {
+      "epoch": 0.10890498508734554,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4843596968048784e-06,
+      "loss": 1.1892,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1089902002556455,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4843331871707355e-06,
+      "loss": 0.6167,
+      "step": 1279
+    },
+    {
+      "epoch": 0.10907541542394546,
+      "grad_norm": 13.5,
+      "learning_rate": 2.48430665523095e-06,
+      "loss": 0.6848,
+      "step": 1280
+    },
+    {
+      "epoch": 0.10916063059224541,
+      "grad_norm": 17.125,
+      "learning_rate": 2.484280100986003e-06,
+      "loss": 1.2219,
+      "step": 1281
+    },
+    {
+      "epoch": 0.10924584576054537,
+      "grad_norm": 12.0,
+      "learning_rate": 2.484253524436373e-06,
+      "loss": 0.5148,
+      "step": 1282
+    },
+    {
+      "epoch": 0.10933106092884533,
+      "grad_norm": 20.5,
+      "learning_rate": 2.48422692558254e-06,
+      "loss": 1.3084,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1094162760971453,
+      "grad_norm": 14.5,
+      "learning_rate": 2.484200304424986e-06,
+      "loss": 0.6643,
+      "step": 1284
+    },
+    {
+      "epoch": 0.10950149126544526,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4841736609641916e-06,
+      "loss": 1.0793,
+      "step": 1285
+    },
+    {
+      "epoch": 0.10958670643374521,
+      "grad_norm": 15.75,
+      "learning_rate": 2.484146995200638e-06,
+      "loss": 1.0071,
+      "step": 1286
+    },
+    {
+      "epoch": 0.10967192160204517,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4841203071348067e-06,
+      "loss": 0.6509,
+      "step": 1287
+    },
+    {
+      "epoch": 0.10975713677034513,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4840935967671807e-06,
+      "loss": 0.6941,
+      "step": 1288
+    },
+    {
+      "epoch": 0.10984235193864508,
+      "grad_norm": 11.375,
+      "learning_rate": 2.484066864098243e-06,
+      "loss": 0.6386,
+      "step": 1289
+    },
+    {
+      "epoch": 0.10992756710694504,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4840401091284754e-06,
+      "loss": 0.5552,
+      "step": 1290
+    },
+    {
+      "epoch": 0.110012782275245,
+      "grad_norm": 10.25,
+      "learning_rate": 2.4840133318583625e-06,
+      "loss": 0.541,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11009799744354495,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4839865322883873e-06,
+      "loss": 1.0938,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11018321261184491,
+      "grad_norm": 16.875,
+      "learning_rate": 2.483959710419035e-06,
+      "loss": 0.8905,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11026842778014487,
+      "grad_norm": 23.375,
+      "learning_rate": 2.4839328662507896e-06,
+      "loss": 0.9578,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11035364294844482,
+      "grad_norm": 18.75,
+      "learning_rate": 2.483905999784137e-06,
+      "loss": 0.7415,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11043885811674478,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4838791110195616e-06,
+      "loss": 0.6517,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11052407328504474,
+      "grad_norm": 17.375,
+      "learning_rate": 2.48385219995755e-06,
+      "loss": 0.7366,
+      "step": 1297
+    },
+    {
+      "epoch": 0.1106092884533447,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4838252665985887e-06,
+      "loss": 0.6285,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11069450362164465,
+      "grad_norm": 14.5,
+      "learning_rate": 2.483798310943164e-06,
+      "loss": 0.6958,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11077971878994461,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4837713329917634e-06,
+      "loss": 0.5696,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11086493395824457,
+      "grad_norm": 22.875,
+      "learning_rate": 2.483744332744874e-06,
+      "loss": 1.282,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11095014912654452,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4837173102029833e-06,
+      "loss": 1.2556,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11103536429484448,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.483690265366581e-06,
+      "loss": 0.9616,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11112057946314444,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4836631982361546e-06,
+      "loss": 0.8363,
+      "step": 1304
+    },
+    {
+      "epoch": 0.1112057946314444,
+      "grad_norm": 14.0,
+      "learning_rate": 2.483636108812194e-06,
+      "loss": 0.5904,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11129100979974435,
+      "grad_norm": 20.25,
+      "learning_rate": 2.483608997095188e-06,
+      "loss": 1.1782,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11137622496804431,
+      "grad_norm": 13.875,
+      "learning_rate": 2.483581863085627e-06,
+      "loss": 0.5778,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11146144013634426,
+      "grad_norm": 18.5,
+      "learning_rate": 2.4835547067840015e-06,
+      "loss": 0.9187,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11154665530464422,
+      "grad_norm": 22.375,
+      "learning_rate": 2.483527528190802e-06,
+      "loss": 1.2057,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11163187047294418,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4835003273065197e-06,
+      "loss": 0.8024,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11171708564124413,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4834731041316457e-06,
+      "loss": 0.5612,
+      "step": 1311
+    },
+    {
+      "epoch": 0.1118023008095441,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4834458586666725e-06,
+      "loss": 0.5,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11188751597784406,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.483418590912093e-06,
+      "loss": 0.5764,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11197273114614402,
+      "grad_norm": 11.0,
+      "learning_rate": 2.4833913008683985e-06,
+      "loss": 0.5075,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11205794631444398,
+      "grad_norm": 10.5,
+      "learning_rate": 2.483363988536083e-06,
+      "loss": 0.3628,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11214316148274393,
+      "grad_norm": 19.375,
+      "learning_rate": 2.48333665391564e-06,
+      "loss": 0.8608,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11222837665104389,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4833092970075638e-06,
+      "loss": 1.0525,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11231359181934385,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4832819178123477e-06,
+      "loss": 0.9775,
+      "step": 1318
+    },
+    {
+      "epoch": 0.1123988069876438,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.483254516330488e-06,
+      "loss": 0.6232,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11248402215594376,
+      "grad_norm": 24.125,
+      "learning_rate": 2.4832270925624784e-06,
+      "loss": 0.9508,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11256923732424372,
+      "grad_norm": 22.0,
+      "learning_rate": 2.4831996465088156e-06,
+      "loss": 0.8841,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11265445249254367,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.483172178169995e-06,
+      "loss": 0.6611,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11273966766084363,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4831446875465133e-06,
+      "loss": 1.064,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11282488282914359,
+      "grad_norm": 20.75,
+      "learning_rate": 2.483117174638867e-06,
+      "loss": 0.9941,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11291009799744355,
+      "grad_norm": 17.0,
+      "learning_rate": 2.483089639447553e-06,
+      "loss": 0.9082,
+      "step": 1325
+    },
+    {
+      "epoch": 0.1129953131657435,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4830620819730693e-06,
+      "loss": 0.8576,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11308052833404346,
+      "grad_norm": 18.375,
+      "learning_rate": 2.483034502215914e-06,
+      "loss": 1.2188,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11316574350234342,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.483006900176585e-06,
+      "loss": 0.4908,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11325095867064337,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4829792758555817e-06,
+      "loss": 0.8232,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11333617383894333,
+      "grad_norm": 16.125,
+      "learning_rate": 2.482951629253403e-06,
+      "loss": 1.0177,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11342138900724329,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4829239603705486e-06,
+      "loss": 0.7121,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11350660417554324,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4828962692075187e-06,
+      "loss": 0.6933,
+      "step": 1332
+    },
+    {
+      "epoch": 0.1135918193438432,
+      "grad_norm": 16.25,
+      "learning_rate": 2.482868555764813e-06,
+      "loss": 0.9299,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11367703451214316,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.482840820042933e-06,
+      "loss": 0.5928,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11376224968044311,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4828130620423795e-06,
+      "loss": 0.4289,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11384746484874307,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4827852817636545e-06,
+      "loss": 0.7182,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11393268001704303,
+      "grad_norm": 11.0,
+      "learning_rate": 2.4827574792072594e-06,
+      "loss": 0.4529,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11401789518534299,
+      "grad_norm": 21.0,
+      "learning_rate": 2.482729654373697e-06,
+      "loss": 0.8668,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11410311035364294,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4827018072634707e-06,
+      "loss": 1.015,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11418832552194291,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4826739378770833e-06,
+      "loss": 0.7146,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11427354069024287,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.4826460462150376e-06,
+      "loss": 0.3754,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11435875585854283,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4826181322778385e-06,
+      "loss": 0.8204,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11444397102684278,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4825901960659904e-06,
+      "loss": 0.947,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11452918619514274,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.482562237579998e-06,
+      "loss": 0.415,
+      "step": 1344
+    },
+    {
+      "epoch": 0.1146144013634427,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4825342568203664e-06,
+      "loss": 0.9712,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11469961653174265,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4825062537876014e-06,
+      "loss": 0.8323,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11478483170004261,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4824782284822092e-06,
+      "loss": 0.9422,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11487004686834257,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4824501809046956e-06,
+      "loss": 0.4773,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11495526203664252,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4824221110555683e-06,
+      "loss": 0.7092,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11504047720494248,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.482394018935334e-06,
+      "loss": 0.8438,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11512569237324244,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.482365904544501e-06,
+      "loss": 0.7924,
+      "step": 1351
+    },
+    {
+      "epoch": 0.1152109075415424,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4823377678835763e-06,
+      "loss": 0.9336,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11529612270984235,
+      "grad_norm": 18.0,
+      "learning_rate": 2.4823096089530692e-06,
+      "loss": 0.926,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11538133787814231,
+      "grad_norm": 12.125,
+      "learning_rate": 2.482281427753488e-06,
+      "loss": 0.6261,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11546655304644227,
+      "grad_norm": 19.75,
+      "learning_rate": 2.482253224285342e-06,
+      "loss": 0.4516,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11555176821474222,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.4822249985491416e-06,
+      "loss": 0.5531,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11563698338304218,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4821967505453965e-06,
+      "loss": 0.7452,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11572219855134214,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4821684802746167e-06,
+      "loss": 0.7824,
+      "step": 1358
+    },
+    {
+      "epoch": 0.1158074137196421,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4821401877373135e-06,
+      "loss": 0.6703,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11589262888794205,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4821118729339985e-06,
+      "loss": 0.873,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11597784405624201,
+      "grad_norm": 27.0,
+      "learning_rate": 2.4820835358651823e-06,
+      "loss": 0.9078,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11606305922454196,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4820551765313784e-06,
+      "loss": 0.6403,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11614827439284192,
+      "grad_norm": 18.0,
+      "learning_rate": 2.482026794933098e-06,
+      "loss": 1.0153,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11623348956114188,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4819983910708546e-06,
+      "loss": 0.8461,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11631870472944184,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4819699649451613e-06,
+      "loss": 0.9227,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11640391989774179,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4819415165565317e-06,
+      "loss": 0.6959,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11648913506604175,
+      "grad_norm": 13.25,
+      "learning_rate": 2.481913045905481e-06,
+      "loss": 0.6191,
+      "step": 1367
+    },
+    {
+      "epoch": 0.1165743502343417,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4818845529925216e-06,
+      "loss": 0.3106,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11665956540264168,
+      "grad_norm": 24.25,
+      "learning_rate": 2.4818560378181707e-06,
+      "loss": 1.1964,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11674478057094163,
+      "grad_norm": 19.375,
+      "learning_rate": 2.481827500382942e-06,
+      "loss": 0.9775,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11682999573924159,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4817989406873512e-06,
+      "loss": 0.7623,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11691521090754155,
+      "grad_norm": 28.375,
+      "learning_rate": 2.481770358731915e-06,
+      "loss": 0.5829,
+      "step": 1372
+    },
+    {
+      "epoch": 0.1170004260758415,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.48174175451715e-06,
+      "loss": 0.5868,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11708564124414146,
+      "grad_norm": 12.75,
+      "learning_rate": 2.481713128043573e-06,
+      "loss": 0.5655,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11717085641244142,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4816844793117014e-06,
+      "loss": 0.9376,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11725607158074138,
+      "grad_norm": 14.25,
+      "learning_rate": 2.481655808322052e-06,
+      "loss": 0.5245,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11734128674904133,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4816271150751444e-06,
+      "loss": 0.5217,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11742650191734129,
+      "grad_norm": 14.75,
+      "learning_rate": 2.481598399571496e-06,
+      "loss": 0.8422,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11751171708564125,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4815696618116257e-06,
+      "loss": 1.1816,
+      "step": 1379
+    },
+    {
+      "epoch": 0.1175969322539412,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4815409017960538e-06,
+      "loss": 0.8437,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11768214742224116,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.481512119525299e-06,
+      "loss": 0.8476,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11776736259054112,
+      "grad_norm": 13.5,
+      "learning_rate": 2.481483314999882e-06,
+      "loss": 0.8877,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11785257775884107,
+      "grad_norm": 12.875,
+      "learning_rate": 2.481454488220323e-06,
+      "loss": 0.7549,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11793779292714103,
+      "grad_norm": 19.375,
+      "learning_rate": 2.4814256391871435e-06,
+      "loss": 0.7129,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11802300809544099,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.481396767900864e-06,
+      "loss": 0.5135,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11810822326374094,
+      "grad_norm": 14.625,
+      "learning_rate": 2.481367874362006e-06,
+      "loss": 0.7443,
+      "step": 1386
+    },
+    {
+      "epoch": 0.1181934384320409,
+      "grad_norm": 9.1875,
+      "learning_rate": 2.4813389585710933e-06,
+      "loss": 0.3536,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11827865360034086,
+      "grad_norm": 14.75,
+      "learning_rate": 2.481310020528647e-06,
+      "loss": 0.6719,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11836386876864081,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.4812810602351907e-06,
+      "loss": 0.5734,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11844908393694077,
+      "grad_norm": 19.375,
+      "learning_rate": 2.481252077691247e-06,
+      "loss": 1.0354,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11853429910524073,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4812230728973407e-06,
+      "loss": 0.8676,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11861951427354069,
+      "grad_norm": 13.5,
+      "learning_rate": 2.481194045853995e-06,
+      "loss": 0.6957,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11870472944184064,
+      "grad_norm": 16.0,
+      "learning_rate": 2.481164996561735e-06,
+      "loss": 0.9095,
+      "step": 1393
+    },
+    {
+      "epoch": 0.1187899446101406,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.4811359250210852e-06,
+      "loss": 0.4006,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11887515977844056,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4811068312325713e-06,
+      "loss": 0.6876,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11896037494674051,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4810777151967196e-06,
+      "loss": 0.7951,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11904559011504048,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4810485769140554e-06,
+      "loss": 0.6443,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11913080528334044,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.481019416385105e-06,
+      "loss": 0.9024,
+      "step": 1398
+    },
+    {
+      "epoch": 0.1192160204516404,
+      "grad_norm": 13.375,
+      "learning_rate": 2.480990233610396e-06,
+      "loss": 0.6823,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11930123561994035,
+      "grad_norm": 24.125,
+      "learning_rate": 2.4809610285904558e-06,
+      "loss": 0.7423,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11938645078824031,
+      "grad_norm": 19.0,
+      "learning_rate": 2.480931801325812e-06,
+      "loss": 0.9985,
+      "step": 1401
+    },
+    {
+      "epoch": 0.11947166595654027,
+      "grad_norm": 16.25,
+      "learning_rate": 2.480902551816993e-06,
+      "loss": 0.7579,
+      "step": 1402
+    },
+    {
+      "epoch": 0.11955688112484023,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4808732800645264e-06,
+      "loss": 1.2039,
+      "step": 1403
+    },
+    {
+      "epoch": 0.11964209629314018,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4808439860689427e-06,
+      "loss": 0.6401,
+      "step": 1404
+    },
+    {
+      "epoch": 0.11972731146144014,
+      "grad_norm": 25.125,
+      "learning_rate": 2.48081466983077e-06,
+      "loss": 0.9703,
+      "step": 1405
+    },
+    {
+      "epoch": 0.1198125266297401,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4807853313505385e-06,
+      "loss": 0.8871,
+      "step": 1406
+    },
+    {
+      "epoch": 0.11989774179804005,
+      "grad_norm": 25.875,
+      "learning_rate": 2.4807559706287786e-06,
+      "loss": 0.7812,
+      "step": 1407
+    },
+    {
+      "epoch": 0.11998295696634001,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4807265876660205e-06,
+      "loss": 0.8993,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12006817213463997,
+      "grad_norm": 22.375,
+      "learning_rate": 2.4806971824627955e-06,
+      "loss": 1.2011,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12015338730293992,
+      "grad_norm": 14.375,
+      "learning_rate": 2.480667755019635e-06,
+      "loss": 0.6522,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12023860247123988,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4806383053370704e-06,
+      "loss": 0.703,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12032381763953984,
+      "grad_norm": 16.25,
+      "learning_rate": 2.480608833415634e-06,
+      "loss": 1.1941,
+      "step": 1412
+    },
+    {
+      "epoch": 0.1204090328078398,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4805793392558587e-06,
+      "loss": 0.6089,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12049424797613975,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4805498228582776e-06,
+      "loss": 0.8614,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12057946314443971,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4805202842234233e-06,
+      "loss": 0.7704,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12066467831273966,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4804907233518305e-06,
+      "loss": 0.827,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12074989348103962,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4804611402440326e-06,
+      "loss": 0.6887,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12083510864933958,
+      "grad_norm": 15.5,
+      "learning_rate": 2.480431534900565e-06,
+      "loss": 0.703,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12092032381763954,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4804019073219617e-06,
+      "loss": 0.7831,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12100553898593949,
+      "grad_norm": 16.375,
+      "learning_rate": 2.480372257508759e-06,
+      "loss": 0.946,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12109075415423945,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.480342585461492e-06,
+      "loss": 0.7256,
+      "step": 1421
+    },
+    {
+      "epoch": 0.1211759693225394,
+      "grad_norm": 31.75,
+      "learning_rate": 2.480312891180698e-06,
+      "loss": 1.0698,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12126118449083936,
+      "grad_norm": 17.625,
+      "learning_rate": 2.480283174666912e-06,
+      "loss": 1.0257,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12134639965913932,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4802534359206727e-06,
+      "loss": 0.5837,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12143161482743929,
+      "grad_norm": 19.625,
+      "learning_rate": 2.480223674942516e-06,
+      "loss": 1.1336,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12151682999573925,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4801938917329808e-06,
+      "loss": 0.6785,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1216020451640392,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4801640862926054e-06,
+      "loss": 1.2337,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12168726033233916,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4801342586219273e-06,
+      "loss": 0.8515,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12177247550063912,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4801044087214867e-06,
+      "loss": 0.6902,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12185769066893908,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4800745365918216e-06,
+      "loss": 0.6936,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12194290583723903,
+      "grad_norm": 20.5,
+      "learning_rate": 2.480044642233473e-06,
+      "loss": 1.3433,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12202812100553899,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.480014725646981e-06,
+      "loss": 0.793,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12211333617383895,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4799847868328863e-06,
+      "loss": 0.622,
+      "step": 1433
+    },
+    {
+      "epoch": 0.1221985513421389,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4799548257917292e-06,
+      "loss": 0.741,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12228376651043886,
+      "grad_norm": 14.25,
+      "learning_rate": 2.479924842524052e-06,
+      "loss": 0.8168,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12236898167873882,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4798948370303953e-06,
+      "loss": 0.8108,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12245419684703877,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4798648093113027e-06,
+      "loss": 0.8577,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12253941201533873,
+      "grad_norm": 11.125,
+      "learning_rate": 2.479834759367316e-06,
+      "loss": 0.4022,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12262462718363869,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.479804687198979e-06,
+      "loss": 1.0363,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12270984235193864,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4797745928068346e-06,
+      "loss": 0.9002,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1227950575202386,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.479744476191426e-06,
+      "loss": 0.661,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12288027268853856,
+      "grad_norm": 18.0,
+      "learning_rate": 2.479714337353299e-06,
+      "loss": 0.9553,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12296548785683852,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4796841762929967e-06,
+      "loss": 0.9295,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12305070302513847,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4796539930110652e-06,
+      "loss": 0.8901,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12313591819343843,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.479623787508049e-06,
+      "loss": 0.6673,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12322113336173839,
+      "grad_norm": 17.25,
+      "learning_rate": 2.479593559784495e-06,
+      "loss": 1.2743,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12330634853003834,
+      "grad_norm": 14.875,
+      "learning_rate": 2.479563309840949e-06,
+      "loss": 0.8796,
+      "step": 1447
+    },
+    {
+      "epoch": 0.1233915636983383,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.479533037677957e-06,
+      "loss": 0.6922,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12347677886663826,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4795027432960673e-06,
+      "loss": 0.5181,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12356199403493821,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.479472426695826e-06,
+      "loss": 0.5184,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12364720920323817,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4794420878777818e-06,
+      "loss": 1.2067,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12373242437153813,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4794117268424834e-06,
+      "loss": 0.9037,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1238176395398381,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4793813435904782e-06,
+      "loss": 1.0361,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12390285470813805,
+      "grad_norm": 13.75,
+      "learning_rate": 2.479350938122316e-06,
+      "loss": 0.8246,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12398806987643801,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.479320510438546e-06,
+      "loss": 0.7256,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12407328504473797,
+      "grad_norm": 24.5,
+      "learning_rate": 2.4792900605397185e-06,
+      "loss": 1.6031,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12415850021303793,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4792595884263833e-06,
+      "loss": 1.1003,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12424371538133788,
+      "grad_norm": 16.875,
+      "learning_rate": 2.479229094099091e-06,
+      "loss": 0.7001,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12432893054963784,
+      "grad_norm": 20.0,
+      "learning_rate": 2.479198577558393e-06,
+      "loss": 0.9321,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1244141457179378,
+      "grad_norm": 16.125,
+      "learning_rate": 2.479168038804841e-06,
+      "loss": 0.9677,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12449936088623775,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4791374778389866e-06,
+      "loss": 0.7319,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12458457605453771,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4791068946613815e-06,
+      "loss": 0.8257,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12466979122283767,
+      "grad_norm": 13.375,
+      "learning_rate": 2.479076289272579e-06,
+      "loss": 0.7955,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12475500639113762,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.479045661673132e-06,
+      "loss": 0.4965,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12484022155943758,
+      "grad_norm": 16.5,
+      "learning_rate": 2.479015011863594e-06,
+      "loss": 0.6017,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12492543672773754,
+      "grad_norm": 15.875,
+      "learning_rate": 2.478984339844519e-06,
+      "loss": 0.7101,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1250106518960375,
+      "grad_norm": 13.625,
+      "learning_rate": 2.478953645616461e-06,
+      "loss": 0.9422,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12509586706433745,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4789229291799748e-06,
+      "loss": 0.8229,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1251810822326374,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.4788921905356157e-06,
+      "loss": 0.3218,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12526629740093737,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4788614296839387e-06,
+      "loss": 0.7974,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12535151256923732,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4788306466255e-06,
+      "loss": 0.5381,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12543672773753728,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4787998413608556e-06,
+      "loss": 0.6639,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12552194290583724,
+      "grad_norm": 23.75,
+      "learning_rate": 2.478769013890563e-06,
+      "loss": 1.1722,
+      "step": 1473
+    },
+    {
+      "epoch": 0.1256071580741372,
+      "grad_norm": 14.0,
+      "learning_rate": 2.478738164215178e-06,
+      "loss": 0.7418,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12569237324243715,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4787072923352597e-06,
+      "loss": 1.033,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1257775884107371,
+      "grad_norm": 14.25,
+      "learning_rate": 2.478676398251364e-06,
+      "loss": 0.6169,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12586280357903706,
+      "grad_norm": 16.125,
+      "learning_rate": 2.478645481964051e-06,
+      "loss": 1.163,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12594801874733702,
+      "grad_norm": 25.375,
+      "learning_rate": 2.4786145434738783e-06,
+      "loss": 1.371,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12603323391563698,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4785835827814055e-06,
+      "loss": 0.672,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12611844908393693,
+      "grad_norm": 14.0,
+      "learning_rate": 2.478552599887192e-06,
+      "loss": 0.8676,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1262036642522369,
+      "grad_norm": 22.25,
+      "learning_rate": 2.4785215947917973e-06,
+      "loss": 1.2902,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12628887942053685,
+      "grad_norm": 16.125,
+      "learning_rate": 2.478490567495782e-06,
+      "loss": 0.8189,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1263740945888368,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4784595179997073e-06,
+      "loss": 0.9453,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12645930975713676,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4784284463041334e-06,
+      "loss": 0.6488,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12654452492543672,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4783973524096226e-06,
+      "loss": 0.5557,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12662974009373668,
+      "grad_norm": 14.5,
+      "learning_rate": 2.478366236316736e-06,
+      "loss": 0.7248,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12671495526203663,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4783350980260366e-06,
+      "loss": 0.633,
+      "step": 1487
+    },
+    {
+      "epoch": 0.1268001704303366,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4783039375380864e-06,
+      "loss": 0.5183,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12688538559863655,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.478272754853449e-06,
+      "loss": 0.8545,
+      "step": 1489
+    },
+    {
+      "epoch": 0.1269706007669365,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4782415499726883e-06,
+      "loss": 0.6475,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12705581593523646,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.478210322896367e-06,
+      "loss": 0.3222,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12714103110353642,
+      "grad_norm": 23.0,
+      "learning_rate": 2.4781790736250507e-06,
+      "loss": 1.0186,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12722624627183637,
+      "grad_norm": 15.75,
+      "learning_rate": 2.478147802159303e-06,
+      "loss": 0.5955,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12731146144013633,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4781165084996895e-06,
+      "loss": 0.6227,
+      "step": 1494
+    },
+    {
+      "epoch": 0.1273966766084363,
+      "grad_norm": 27.125,
+      "learning_rate": 2.478085192646776e-06,
+      "loss": 1.0827,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12748189177673627,
+      "grad_norm": 12.875,
+      "learning_rate": 2.478053854601128e-06,
+      "loss": 0.492,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12756710694503623,
+      "grad_norm": 14.625,
+      "learning_rate": 2.478022494363312e-06,
+      "loss": 0.5216,
+      "step": 1497
+    },
+    {
+      "epoch": 0.1276523221133362,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4779911119338948e-06,
+      "loss": 0.7451,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12773753728163614,
+      "grad_norm": 12.125,
+      "learning_rate": 2.477959707313443e-06,
+      "loss": 0.6604,
+      "step": 1499
+    },
+    {
+      "epoch": 0.1278227524499361,
+      "grad_norm": 14.875,
+      "learning_rate": 2.477928280502525e-06,
+      "loss": 0.8966,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12790796761823606,
+      "grad_norm": 19.5,
+      "learning_rate": 2.477896831501708e-06,
+      "loss": 1.037,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12799318278653601,
+      "grad_norm": 12.875,
+      "learning_rate": 2.47786536031156e-06,
+      "loss": 0.5206,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12807839795483597,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4778338669326507e-06,
+      "loss": 0.9273,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12816361312313593,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.477802351365549e-06,
+      "loss": 0.7273,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12824882829143588,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4777708136108235e-06,
+      "loss": 1.0315,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12833404345973584,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4777392536690455e-06,
+      "loss": 0.8453,
+      "step": 1506
+    },
+    {
+      "epoch": 0.1284192586280358,
+      "grad_norm": 15.25,
+      "learning_rate": 2.477707671540784e-06,
+      "loss": 1.0709,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12850447379633576,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4776760672266102e-06,
+      "loss": 0.4945,
+      "step": 1508
+    },
+    {
+      "epoch": 0.1285896889646357,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.477644440727096e-06,
+      "loss": 0.7331,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12867490413293567,
+      "grad_norm": 13.125,
+      "learning_rate": 2.477612792042812e-06,
+      "loss": 0.772,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12876011930123563,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.47758112117433e-06,
+      "loss": 1.0932,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12884533446953558,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4775494281222236e-06,
+      "loss": 0.5275,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12893054963783554,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4775177128870636e-06,
+      "loss": 1.2726,
+      "step": 1513
+    },
+    {
+      "epoch": 0.1290157648061355,
+      "grad_norm": 14.5,
+      "learning_rate": 2.477485975469425e-06,
+      "loss": 1.0652,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12910097997443545,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.47745421586988e-06,
+      "loss": 0.765,
+      "step": 1515
+    },
+    {
+      "epoch": 0.1291861951427354,
+      "grad_norm": 16.125,
+      "learning_rate": 2.477422434089003e-06,
+      "loss": 0.6144,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12927141031103537,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4773906301273686e-06,
+      "loss": 0.3885,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12935662547933532,
+      "grad_norm": 14.375,
+      "learning_rate": 2.477358803985551e-06,
+      "loss": 0.2456,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12944184064763528,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.477326955664126e-06,
+      "loss": 0.4207,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12952705581593524,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4772950851636686e-06,
+      "loss": 0.7109,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1296122709842352,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4772631924847547e-06,
+      "loss": 0.747,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12969748615253515,
+      "grad_norm": 18.5,
+      "learning_rate": 2.477231277627961e-06,
+      "loss": 0.8909,
+      "step": 1522
+    },
+    {
+      "epoch": 0.1297827013208351,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4771993405938636e-06,
+      "loss": 0.7773,
+      "step": 1523
+    },
+    {
+      "epoch": 0.12986791648913507,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4771673813830405e-06,
+      "loss": 0.8698,
+      "step": 1524
+    },
+    {
+      "epoch": 0.12995313165743502,
+      "grad_norm": 18.0,
+      "learning_rate": 2.477135399996069e-06,
+      "loss": 0.9722,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13003834682573498,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4771033964335263e-06,
+      "loss": 0.7828,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13012356199403494,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4770713706959914e-06,
+      "loss": 0.7726,
+      "step": 1527
+    },
+    {
+      "epoch": 0.1302087771623349,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4770393227840432e-06,
+      "loss": 0.7019,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13029399233063485,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4770072526982607e-06,
+      "loss": 0.9926,
+      "step": 1529
+    },
+    {
+      "epoch": 0.1303792074989348,
+      "grad_norm": 17.75,
+      "learning_rate": 2.476975160439223e-06,
+      "loss": 0.8013,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13046442266723476,
+      "grad_norm": 23.125,
+      "learning_rate": 2.47694304600751e-06,
+      "loss": 1.0229,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13054963783553472,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4769109094037027e-06,
+      "loss": 0.69,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13063485300383468,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4768787506283813e-06,
+      "loss": 0.808,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13072006817213463,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4768465696821275e-06,
+      "loss": 0.7263,
+      "step": 1534
+    },
+    {
+      "epoch": 0.1308052833404346,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4768143665655224e-06,
+      "loss": 0.5559,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13089049850873455,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.4767821412791482e-06,
+      "loss": 0.5647,
+      "step": 1536
+    },
+    {
+      "epoch": 0.1309757136770345,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4767498938235866e-06,
+      "loss": 1.0662,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13106092884533446,
+      "grad_norm": 19.875,
+      "learning_rate": 2.476717624199421e-06,
+      "loss": 0.9566,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13114614401363442,
+      "grad_norm": 11.125,
+      "learning_rate": 2.476685332407235e-06,
+      "loss": 0.4021,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13123135918193438,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4766530184476107e-06,
+      "loss": 1.0331,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13131657435023433,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4766206823211335e-06,
+      "loss": 1.2049,
+      "step": 1541
+    },
+    {
+      "epoch": 0.1314017895185343,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4765883240283868e-06,
+      "loss": 0.5505,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13148700468683425,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4765559435699554e-06,
+      "loss": 0.5995,
+      "step": 1543
+    },
+    {
+      "epoch": 0.1315722198551342,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.476523540946425e-06,
+      "loss": 0.8021,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13165743502343416,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4764911161583806e-06,
+      "loss": 0.5415,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13174265019173412,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4764586692064086e-06,
+      "loss": 0.9332,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13182786536003407,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4764262000910953e-06,
+      "loss": 0.5249,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13191308052833403,
+      "grad_norm": 12.0,
+      "learning_rate": 2.476393708813027e-06,
+      "loss": 0.448,
+      "step": 1548
+    },
+    {
+      "epoch": 0.131998295696634,
+      "grad_norm": 13.75,
+      "learning_rate": 2.476361195372791e-06,
+      "loss": 0.4997,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13208351086493395,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4763286597709755e-06,
+      "loss": 1.3269,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1321687260332339,
+      "grad_norm": 13.25,
+      "learning_rate": 2.476296102008168e-06,
+      "loss": 0.6287,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1322539412015339,
+      "grad_norm": 11.125,
+      "learning_rate": 2.476263522084956e-06,
+      "loss": 0.4037,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13233915636983384,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4762309200019296e-06,
+      "loss": 0.5243,
+      "step": 1553
+    },
+    {
+      "epoch": 0.1324243715381338,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4761982957596776e-06,
+      "loss": 0.793,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13250958670643376,
+      "grad_norm": 12.75,
+      "learning_rate": 2.476165649358789e-06,
+      "loss": 0.655,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13259480187473371,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4761329807998546e-06,
+      "loss": 0.6106,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13268001704303367,
+      "grad_norm": 22.5,
+      "learning_rate": 2.4761002900834635e-06,
+      "loss": 0.905,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13276523221133363,
+      "grad_norm": 22.125,
+      "learning_rate": 2.4760675772102077e-06,
+      "loss": 0.7622,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13285044737963358,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.476034842180678e-06,
+      "loss": 0.3594,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13293566254793354,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4760020849954653e-06,
+      "loss": 0.7484,
+      "step": 1560
+    },
+    {
+      "epoch": 0.1330208777162335,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4759693056551626e-06,
+      "loss": 0.5805,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13310609288453346,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4759365041603617e-06,
+      "loss": 0.5127,
+      "step": 1562
+    },
+    {
+      "epoch": 0.1331913080528334,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.475903680511655e-06,
+      "loss": 0.9681,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13327652322113337,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4758708347096363e-06,
+      "loss": 1.1556,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13336173838943333,
+      "grad_norm": 14.625,
+      "learning_rate": 2.475837966754899e-06,
+      "loss": 0.7612,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13344695355773328,
+      "grad_norm": 17.75,
+      "learning_rate": 2.475805076648037e-06,
+      "loss": 1.1628,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13353216872603324,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4757721643896444e-06,
+      "loss": 0.9244,
+      "step": 1567
+    },
+    {
+      "epoch": 0.1336173838943332,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4757392299803163e-06,
+      "loss": 0.6452,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13370259906263315,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4757062734206476e-06,
+      "loss": 0.61,
+      "step": 1569
+    },
+    {
+      "epoch": 0.1337878142309331,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4756732947112343e-06,
+      "loss": 1.1258,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13387302939923307,
+      "grad_norm": 18.125,
+      "learning_rate": 2.475640293852672e-06,
+      "loss": 0.9646,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13395824456753302,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4756072708455565e-06,
+      "loss": 0.7332,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13404345973583298,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4755742256904856e-06,
+      "loss": 1.1055,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13412867490413294,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.475541158388056e-06,
+      "loss": 0.9313,
+      "step": 1574
+    },
+    {
+      "epoch": 0.1342138900724329,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4755080689388656e-06,
+      "loss": 1.0465,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13429910524073285,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.475474957343512e-06,
+      "loss": 0.8469,
+      "step": 1576
+    },
+    {
+      "epoch": 0.1343843204090328,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4754418236025935e-06,
+      "loss": 0.6181,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13446953557733277,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4754086677167087e-06,
+      "loss": 0.7345,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13455475074563272,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4753754896864575e-06,
+      "loss": 0.9474,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13463996591393268,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4753422895124387e-06,
+      "loss": 0.7746,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13472518108223264,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4753090671952523e-06,
+      "loss": 0.6436,
+      "step": 1581
+    },
+    {
+      "epoch": 0.1348103962505326,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.475275822735499e-06,
+      "loss": 0.4949,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13489561141883255,
+      "grad_norm": 23.75,
+      "learning_rate": 2.47524255613378e-06,
+      "loss": 0.9854,
+      "step": 1583
+    },
+    {
+      "epoch": 0.1349808265871325,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4752092673906955e-06,
+      "loss": 1.0981,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13506604175543246,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4751759565068476e-06,
+      "loss": 0.5711,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13515125692373242,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.475142623482838e-06,
+      "loss": 0.7141,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13523647209203238,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4751092683192696e-06,
+      "loss": 0.6999,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13532168726033234,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.4750758910167443e-06,
+      "loss": 0.4597,
+      "step": 1588
+    },
+    {
+      "epoch": 0.1354069024286323,
+      "grad_norm": 34.25,
+      "learning_rate": 2.475042491575866e-06,
+      "loss": 1.183,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13549211759693225,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4750090699972383e-06,
+      "loss": 0.673,
+      "step": 1590
+    },
+    {
+      "epoch": 0.1355773327652322,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4749756262814643e-06,
+      "loss": 0.5467,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13566254793353216,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.474942160429149e-06,
+      "loss": 0.5855,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13574776310183212,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4749086724408975e-06,
+      "loss": 0.7791,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13583297827013208,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4748751623173143e-06,
+      "loss": 0.8921,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13591819343843203,
+      "grad_norm": 15.25,
+      "learning_rate": 2.474841630059005e-06,
+      "loss": 0.661,
+      "step": 1595
+    },
+    {
+      "epoch": 0.136003408606732,
+      "grad_norm": 13.875,
+      "learning_rate": 2.474808075666576e-06,
+      "loss": 0.8196,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13608862377503195,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4747744991406337e-06,
+      "loss": 0.9434,
+      "step": 1597
+    },
+    {
+      "epoch": 0.1361738389433319,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4747409004817842e-06,
+      "loss": 0.6281,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13625905411163186,
+      "grad_norm": 14.25,
+      "learning_rate": 2.474707279690635e-06,
+      "loss": 0.7573,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13634426927993182,
+      "grad_norm": 23.0,
+      "learning_rate": 2.474673636767794e-06,
+      "loss": 1.0137,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13642948444823177,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.474639971713869e-06,
+      "loss": 0.6528,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13651469961653173,
+      "grad_norm": 15.625,
+      "learning_rate": 2.474606284529468e-06,
+      "loss": 1.0082,
+      "step": 1602
+    },
+    {
+      "epoch": 0.1365999147848317,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4745725752152e-06,
+      "loss": 0.6146,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13668512995313165,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.474538843771674e-06,
+      "loss": 1.1543,
+      "step": 1604
+    },
+    {
+      "epoch": 0.1367703451214316,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4745050901994998e-06,
+      "loss": 0.6986,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13685556028973156,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4744713144992875e-06,
+      "loss": 0.6663,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13694077545803152,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4744375166716473e-06,
+      "loss": 0.9152,
+      "step": 1607
+    },
+    {
+      "epoch": 0.1370259906263315,
+      "grad_norm": 16.125,
+      "learning_rate": 2.47440369671719e-06,
+      "loss": 0.894,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13711120579463146,
+      "grad_norm": 13.5,
+      "learning_rate": 2.474369854636526e-06,
+      "loss": 0.9786,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13719642096293141,
+      "grad_norm": 16.375,
+      "learning_rate": 2.474335990430268e-06,
+      "loss": 0.4418,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13728163613123137,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4743021040990272e-06,
+      "loss": 0.5231,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13736685129953133,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4742681956434165e-06,
+      "loss": 0.4635,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13745206646783129,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4742342650640485e-06,
+      "loss": 0.649,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13753728163613124,
+      "grad_norm": 19.5,
+      "learning_rate": 2.474200312361536e-06,
+      "loss": 0.8201,
+      "step": 1614
+    },
+    {
+      "epoch": 0.1376224968044312,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4741663375364934e-06,
+      "loss": 0.7977,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13770771197273116,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4741323405895334e-06,
+      "loss": 0.9941,
+      "step": 1616
+    },
+    {
+      "epoch": 0.1377929271410311,
+      "grad_norm": 21.5,
+      "learning_rate": 2.474098321521271e-06,
+      "loss": 0.8637,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13787814230933107,
+      "grad_norm": 9.8125,
+      "learning_rate": 2.4740642803323217e-06,
+      "loss": 0.3601,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13796335747763103,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4740302170232995e-06,
+      "loss": 0.6797,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13804857264593098,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4739961315948204e-06,
+      "loss": 0.8079,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13813378781423094,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4739620240475006e-06,
+      "loss": 0.9553,
+      "step": 1621
+    },
+    {
+      "epoch": 0.1382190029825309,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4739278943819563e-06,
+      "loss": 1.2068,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13830421815083085,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4738937425988037e-06,
+      "loss": 0.7125,
+      "step": 1623
+    },
+    {
+      "epoch": 0.1383894333191308,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4738595686986603e-06,
+      "loss": 0.7978,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13847464848743077,
+      "grad_norm": 20.875,
+      "learning_rate": 2.473825372682145e-06,
+      "loss": 1.0319,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13855986365573073,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4737911545498733e-06,
+      "loss": 0.7193,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13864507882403068,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4737569143024654e-06,
+      "loss": 0.5906,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13873029399233064,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4737226519405393e-06,
+      "loss": 0.8338,
+      "step": 1628
+    },
+    {
+      "epoch": 0.1388155091606306,
+      "grad_norm": 18.5,
+      "learning_rate": 2.473688367464715e-06,
+      "loss": 1.0743,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13890072432893055,
+      "grad_norm": 20.375,
+      "learning_rate": 2.4736540608756105e-06,
+      "loss": 0.6711,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1389859394972305,
+      "grad_norm": 27.125,
+      "learning_rate": 2.473619732173847e-06,
+      "loss": 1.1368,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13907115466553047,
+      "grad_norm": 18.125,
+      "learning_rate": 2.473585381360044e-06,
+      "loss": 1.0869,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13915636983383042,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4735510084348236e-06,
+      "loss": 0.8473,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13924158500213038,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4735166133988057e-06,
+      "loss": 0.5834,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13932680017043034,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4734821962526122e-06,
+      "loss": 0.8453,
+      "step": 1635
+    },
+    {
+      "epoch": 0.1394120153387303,
+      "grad_norm": 22.875,
+      "learning_rate": 2.473447756996865e-06,
+      "loss": 0.7648,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13949723050703025,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4734132956321872e-06,
+      "loss": 0.4743,
+      "step": 1637
+    },
+    {
+      "epoch": 0.1395824456753302,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4733788121592004e-06,
+      "loss": 1.1727,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13966766084363016,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4733443065785283e-06,
+      "loss": 1.2217,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13975287601193012,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4733097788907947e-06,
+      "loss": 0.8836,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13983809118023008,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4732752290966227e-06,
+      "loss": 1.039,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13992330634853004,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4732406571966373e-06,
+      "loss": 0.864,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14000852151683,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4732060631914633e-06,
+      "loss": 0.8466,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14009373668512995,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4731714470817257e-06,
+      "loss": 0.9993,
+      "step": 1644
+    },
+    {
+      "epoch": 0.1401789518534299,
+      "grad_norm": 19.625,
+      "learning_rate": 2.47313680886805e-06,
+      "loss": 0.9561,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14026416702172986,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4731021485510625e-06,
+      "loss": 0.8679,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14034938219002982,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4730674661313892e-06,
+      "loss": 0.5669,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14043459735832978,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4730327616096566e-06,
+      "loss": 1.0393,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14051981252662973,
+      "grad_norm": 19.5,
+      "learning_rate": 2.472998034986493e-06,
+      "loss": 0.9216,
+      "step": 1649
+    },
+    {
+      "epoch": 0.1406050276949297,
+      "grad_norm": 16.5,
+      "learning_rate": 2.472963286262524e-06,
+      "loss": 0.9496,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14069024286322965,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4729285154383792e-06,
+      "loss": 0.8578,
+      "step": 1651
+    },
+    {
+      "epoch": 0.1407754580315296,
+      "grad_norm": 11.875,
+      "learning_rate": 2.472893722514686e-06,
+      "loss": 0.4865,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14086067319982956,
+      "grad_norm": 16.375,
+      "learning_rate": 2.472858907492074e-06,
+      "loss": 0.9434,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14094588836812952,
+      "grad_norm": 11.75,
+      "learning_rate": 2.4728240703711712e-06,
+      "loss": 0.5191,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14103110353642948,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4727892111526084e-06,
+      "loss": 0.8929,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14111631870472943,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4727543298370148e-06,
+      "loss": 0.4518,
+      "step": 1656
+    },
+    {
+      "epoch": 0.1412015338730294,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.472719426425021e-06,
+      "loss": 0.6647,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14128674904132935,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4726845009172572e-06,
+      "loss": 0.847,
+      "step": 1658
+    },
+    {
+      "epoch": 0.1413719642096293,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.472649553314355e-06,
+      "loss": 1.0141,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14145717937792926,
+      "grad_norm": 19.25,
+      "learning_rate": 2.472614583616946e-06,
+      "loss": 0.6686,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14154239454622922,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4725795918256628e-06,
+      "loss": 0.7507,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14162760971452917,
+      "grad_norm": 17.0,
+      "learning_rate": 2.472544577941136e-06,
+      "loss": 0.6057,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14171282488282913,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.472509541964e-06,
+      "loss": 0.5135,
+      "step": 1663
+    },
+    {
+      "epoch": 0.1417980400511291,
+      "grad_norm": 16.25,
+      "learning_rate": 2.472474483894887e-06,
+      "loss": 0.7836,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14188325521942907,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4724394037344302e-06,
+      "loss": 0.4222,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14196847038772903,
+      "grad_norm": 11.125,
+      "learning_rate": 2.472404301483265e-06,
+      "loss": 0.4369,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14205368555602899,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4723691771420243e-06,
+      "loss": 0.8266,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14213890072432894,
+      "grad_norm": 18.5,
+      "learning_rate": 2.472334030711344e-06,
+      "loss": 0.9728,
+      "step": 1668
+    },
+    {
+      "epoch": 0.1422241158926289,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4722988621918575e-06,
+      "loss": 0.6679,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14230933106092886,
+      "grad_norm": 42.25,
+      "learning_rate": 2.472263671584202e-06,
+      "loss": 1.1168,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1423945462292288,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4722284588890135e-06,
+      "loss": 0.603,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14247976139752877,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4721932241069267e-06,
+      "loss": 0.8431,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14256497656582873,
+      "grad_norm": 18.25,
+      "learning_rate": 2.47215796723858e-06,
+      "loss": 1.4129,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14265019173412868,
+      "grad_norm": 19.375,
+      "learning_rate": 2.4721226882846096e-06,
+      "loss": 1.0845,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14273540690242864,
+      "grad_norm": 39.0,
+      "learning_rate": 2.4720873872456532e-06,
+      "loss": 1.4384,
+      "step": 1675
+    },
+    {
+      "epoch": 0.1428206220707286,
+      "grad_norm": 16.0,
+      "learning_rate": 2.472052064122349e-06,
+      "loss": 0.4952,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14290583723902855,
+      "grad_norm": 10.5,
+      "learning_rate": 2.4720167189153352e-06,
+      "loss": 0.4848,
+      "step": 1677
+    },
+    {
+      "epoch": 0.1429910524073285,
+      "grad_norm": 16.25,
+      "learning_rate": 2.47198135162525e-06,
+      "loss": 0.5999,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14307626757562847,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4719459622527335e-06,
+      "loss": 1.2016,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14316148274392843,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4719105507984246e-06,
+      "loss": 0.6198,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14324669791222838,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4718751172629633e-06,
+      "loss": 0.6453,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14333191308052834,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4718396616469896e-06,
+      "loss": 0.7325,
+      "step": 1682
+    },
+    {
+      "epoch": 0.1434171282488283,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.471804183951145e-06,
+      "loss": 0.6485,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14350234341712825,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4717686841760703e-06,
+      "loss": 0.7609,
+      "step": 1684
+    },
+    {
+      "epoch": 0.1435875585854282,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4717331623224064e-06,
+      "loss": 0.4967,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14367277375372817,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.471697618390796e-06,
+      "loss": 0.6717,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14375798892202812,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4716620523818805e-06,
+      "loss": 0.4425,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14384320409032808,
+      "grad_norm": 16.625,
+      "learning_rate": 2.471626464296304e-06,
+      "loss": 0.8764,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14392841925862804,
+      "grad_norm": 18.0,
+      "learning_rate": 2.471590854134709e-06,
+      "loss": 1.065,
+      "step": 1689
+    },
+    {
+      "epoch": 0.144013634426928,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4715552218977383e-06,
+      "loss": 0.647,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14409884959522795,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4715195675860366e-06,
+      "loss": 0.7388,
+      "step": 1691
+    },
+    {
+      "epoch": 0.1441840647635279,
+      "grad_norm": 17.625,
+      "learning_rate": 2.471483891200248e-06,
+      "loss": 0.9272,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14426927993182787,
+      "grad_norm": 13.0,
+      "learning_rate": 2.471448192741017e-06,
+      "loss": 0.6184,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14435449510012782,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.471412472208989e-06,
+      "loss": 0.7274,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14443971026842778,
+      "grad_norm": 16.0,
+      "learning_rate": 2.47137672960481e-06,
+      "loss": 0.5611,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14452492543672774,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4713409649291243e-06,
+      "loss": 1.1467,
+      "step": 1696
+    },
+    {
+      "epoch": 0.1446101406050277,
+      "grad_norm": 15.625,
+      "learning_rate": 2.47130517818258e-06,
+      "loss": 0.7028,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14469535577332765,
+      "grad_norm": 18.0,
+      "learning_rate": 2.4712693693658222e-06,
+      "loss": 0.9602,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1447805709416276,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4712335384794995e-06,
+      "loss": 0.6669,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14486578610992756,
+      "grad_norm": 16.375,
+      "learning_rate": 2.471197685524258e-06,
+      "loss": 0.9796,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14495100127822752,
+      "grad_norm": 18.875,
+      "learning_rate": 2.471161810500747e-06,
+      "loss": 1.2473,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14503621644652748,
+      "grad_norm": 17.0,
+      "learning_rate": 2.471125913409614e-06,
+      "loss": 0.9406,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14512143161482743,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.4710899942515077e-06,
+      "loss": 0.2845,
+      "step": 1703
+    },
+    {
+      "epoch": 0.1452066467831274,
+      "grad_norm": 16.875,
+      "learning_rate": 2.471054053027077e-06,
+      "loss": 0.9779,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14529186195142735,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4710180897369725e-06,
+      "loss": 0.6043,
+      "step": 1705
+    },
+    {
+      "epoch": 0.1453770771197273,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.470982104381843e-06,
+      "loss": 1.1665,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14546229228802726,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4709460969623385e-06,
+      "loss": 0.7901,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14554750745632722,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4709100674791108e-06,
+      "loss": 1.1028,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14563272262462718,
+      "grad_norm": 11.75,
+      "learning_rate": 2.4708740159328105e-06,
+      "loss": 0.591,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14571793779292713,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4708379423240887e-06,
+      "loss": 0.8908,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1458031529612271,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4708018466535976e-06,
+      "loss": 0.9323,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14588836812952705,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4707657289219898e-06,
+      "loss": 0.6885,
+      "step": 1712
+    },
+    {
+      "epoch": 0.145973583297827,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4707295891299175e-06,
+      "loss": 0.7523,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14605879846612696,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4706934272780342e-06,
+      "loss": 0.7756,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14614401363442692,
+      "grad_norm": 13.75,
+      "learning_rate": 2.470657243366993e-06,
+      "loss": 0.759,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14622922880272687,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4706210373974483e-06,
+      "loss": 0.834,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14631444397102683,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4705848093700534e-06,
+      "loss": 0.9064,
+      "step": 1717
+    },
+    {
+      "epoch": 0.1463996591393268,
+      "grad_norm": 17.125,
+      "learning_rate": 2.470548559285464e-06,
+      "loss": 1.0722,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14648487430762674,
+      "grad_norm": 16.25,
+      "learning_rate": 2.470512287144335e-06,
+      "loss": 0.8775,
+      "step": 1719
+    },
+    {
+      "epoch": 0.1465700894759267,
+      "grad_norm": 33.0,
+      "learning_rate": 2.4704759929473213e-06,
+      "loss": 1.0421,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14665530464422669,
+      "grad_norm": 11.125,
+      "learning_rate": 2.470439676695079e-06,
+      "loss": 0.4325,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14674051981252664,
+      "grad_norm": 13.625,
+      "learning_rate": 2.470403338388265e-06,
+      "loss": 0.5844,
+      "step": 1722
+    },
+    {
+      "epoch": 0.1468257349808266,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4703669780275353e-06,
+      "loss": 0.4866,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14691095014912656,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4703305956135466e-06,
+      "loss": 0.9827,
+      "step": 1724
+    },
+    {
+      "epoch": 0.1469961653174265,
+      "grad_norm": 18.5,
+      "learning_rate": 2.4702941911469573e-06,
+      "loss": 1.1083,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14708138048572647,
+      "grad_norm": 17.625,
+      "learning_rate": 2.470257764628425e-06,
+      "loss": 1.0822,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14716659565402643,
+      "grad_norm": 16.125,
+      "learning_rate": 2.470221316058608e-06,
+      "loss": 0.3616,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14725181082232638,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4701848454381643e-06,
+      "loss": 0.4721,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14733702599062634,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4701483527677536e-06,
+      "loss": 0.7631,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1474222411589263,
+      "grad_norm": 15.0,
+      "learning_rate": 2.470111838048035e-06,
+      "loss": 0.7506,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14750745632722626,
+      "grad_norm": 21.25,
+      "learning_rate": 2.470075301279669e-06,
+      "loss": 0.8739,
+      "step": 1731
+    },
+    {
+      "epoch": 0.1475926714955262,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4700387424633154e-06,
+      "loss": 0.8766,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14767788666382617,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4700021615996343e-06,
+      "loss": 1.1099,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14776310183212613,
+      "grad_norm": 30.25,
+      "learning_rate": 2.4699655586892877e-06,
+      "loss": 1.156,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14784831700042608,
+      "grad_norm": 24.625,
+      "learning_rate": 2.4699289337329373e-06,
+      "loss": 0.9198,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14793353216872604,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4698922867312436e-06,
+      "loss": 0.342,
+      "step": 1736
+    },
+    {
+      "epoch": 0.148018747337026,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.46985561768487e-06,
+      "loss": 0.7033,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14810396250532595,
+      "grad_norm": 20.625,
+      "learning_rate": 2.469818926594478e-06,
+      "loss": 1.0989,
+      "step": 1738
+    },
+    {
+      "epoch": 0.1481891776736259,
+      "grad_norm": 16.375,
+      "learning_rate": 2.469782213460732e-06,
+      "loss": 0.9749,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14827439284192587,
+      "grad_norm": 9.75,
+      "learning_rate": 2.4697454782842946e-06,
+      "loss": 0.4032,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14835960801022582,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4697087210658297e-06,
+      "loss": 0.4214,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14844482317852578,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4696719418060023e-06,
+      "loss": 0.5647,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14853003834682574,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4696351405054763e-06,
+      "loss": 0.9841,
+      "step": 1743
+    },
+    {
+      "epoch": 0.1486152535151257,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4695983171649162e-06,
+      "loss": 0.804,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14870046868342565,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4695614717849888e-06,
+      "loss": 0.9698,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1487856838517256,
+      "grad_norm": 16.875,
+      "learning_rate": 2.469524604366359e-06,
+      "loss": 0.7278,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14887089902002557,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4694877149096934e-06,
+      "loss": 0.5082,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14895611418832552,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4694508034156583e-06,
+      "loss": 0.9462,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14904132935662548,
+      "grad_norm": 11.75,
+      "learning_rate": 2.4694138698849206e-06,
+      "loss": 0.4739,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14912654452492544,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.469376914318149e-06,
+      "loss": 0.5964,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1492117596932254,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4693399367160095e-06,
+      "loss": 0.9735,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14929697486152535,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4693029370791715e-06,
+      "loss": 1.0902,
+      "step": 1752
+    },
+    {
+      "epoch": 0.1493821900298253,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4692659154083033e-06,
+      "loss": 0.72,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14946740519812526,
+      "grad_norm": 13.0,
+      "learning_rate": 2.469228871704074e-06,
+      "loss": 0.6453,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14955262036642522,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.469191805967153e-06,
+      "loss": 0.5758,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14963783553472518,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4691547181982102e-06,
+      "loss": 0.5256,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14972305070302513,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4691176083979155e-06,
+      "loss": 0.9599,
+      "step": 1757
+    },
+    {
+      "epoch": 0.1498082658713251,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4690804765669396e-06,
+      "loss": 0.9446,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14989348103962505,
+      "grad_norm": 19.0,
+      "learning_rate": 2.469043322705954e-06,
+      "loss": 1.1701,
+      "step": 1759
+    },
+    {
+      "epoch": 0.149978696207925,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4690061468156293e-06,
+      "loss": 0.7094,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15006391137622496,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.4689689488966377e-06,
+      "loss": 0.292,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15014912654452492,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.468931728949652e-06,
+      "loss": 0.5294,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15023434171282488,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4688944869753435e-06,
+      "loss": 0.9632,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15031955688112483,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4688572229743863e-06,
+      "loss": 0.3175,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1504047720494248,
+      "grad_norm": 30.125,
+      "learning_rate": 2.4688199369474532e-06,
+      "loss": 1.283,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15048998721772475,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4687826288952186e-06,
+      "loss": 0.9624,
+      "step": 1766
+    },
+    {
+      "epoch": 0.1505752023860247,
+      "grad_norm": 16.5,
+      "learning_rate": 2.468745298818356e-06,
+      "loss": 0.8091,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15066041755432466,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4687079467175402e-06,
+      "loss": 0.6241,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15074563272262462,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4686705725934467e-06,
+      "loss": 0.6405,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15083084789092457,
+      "grad_norm": 15.0,
+      "learning_rate": 2.46863317644675e-06,
+      "loss": 0.8592,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15091606305922453,
+      "grad_norm": 19.125,
+      "learning_rate": 2.468595758278127e-06,
+      "loss": 0.775,
+      "step": 1771
+    },
+    {
+      "epoch": 0.1510012782275245,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.468558318088253e-06,
+      "loss": 0.4237,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15108649339582444,
+      "grad_norm": 26.375,
+      "learning_rate": 2.4685208558778045e-06,
+      "loss": 0.6044,
+      "step": 1773
+    },
+    {
+      "epoch": 0.1511717085641244,
+      "grad_norm": 22.125,
+      "learning_rate": 2.4684833716474594e-06,
+      "loss": 1.1508,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15125692373242436,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.468445865397894e-06,
+      "loss": 0.6571,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15134213890072432,
+      "grad_norm": 11.75,
+      "learning_rate": 2.4684083371297866e-06,
+      "loss": 0.4636,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15142735406902427,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4683707868438156e-06,
+      "loss": 0.8693,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15151256923732426,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.468333214540659e-06,
+      "loss": 0.7335,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15159778440562421,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4682956202209963e-06,
+      "loss": 0.9285,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15168299957392417,
+      "grad_norm": 16.125,
+      "learning_rate": 2.468258003885507e-06,
+      "loss": 1.2837,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15176821474222413,
+      "grad_norm": 37.0,
+      "learning_rate": 2.46822036553487e-06,
+      "loss": 1.6967,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15185342991052408,
+      "grad_norm": 14.75,
+      "learning_rate": 2.468182705169766e-06,
+      "loss": 1.1296,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15193864507882404,
+      "grad_norm": 15.0,
+      "learning_rate": 2.468145022790876e-06,
+      "loss": 0.8443,
+      "step": 1783
+    },
+    {
+      "epoch": 0.152023860247124,
+      "grad_norm": 18.375,
+      "learning_rate": 2.46810731839888e-06,
+      "loss": 1.1151,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15210907541542396,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.46806959199446e-06,
+      "loss": 0.3695,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1521942905837239,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4680318435782977e-06,
+      "loss": 0.9289,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15227950575202387,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.467994073151075e-06,
+      "loss": 0.7206,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15236472092032383,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4679562807134746e-06,
+      "loss": 0.5083,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15244993608862378,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4679184662661797e-06,
+      "loss": 1.1709,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15253515125692374,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4678806298098732e-06,
+      "loss": 1.1609,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1526203664252237,
+      "grad_norm": 11.875,
+      "learning_rate": 2.467842771345239e-06,
+      "loss": 0.7869,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15270558159352365,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4678048908729615e-06,
+      "loss": 0.7349,
+      "step": 1792
+    },
+    {
+      "epoch": 0.1527907967618236,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.467766988393725e-06,
+      "loss": 0.8078,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15287601193012357,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4677290639082143e-06,
+      "loss": 0.7204,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15296122709842352,
+      "grad_norm": 13.125,
+      "learning_rate": 2.467691117417115e-06,
+      "loss": 0.6347,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15304644226672348,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4676531489211127e-06,
+      "loss": 0.9882,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15313165743502344,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4676151584208937e-06,
+      "loss": 0.638,
+      "step": 1797
+    },
+    {
+      "epoch": 0.1532168726033234,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4675771459171443e-06,
+      "loss": 0.7384,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15330208777162335,
+      "grad_norm": 11.75,
+      "learning_rate": 2.467539111410551e-06,
+      "loss": 0.4922,
+      "step": 1799
+    },
+    {
+      "epoch": 0.1533873029399233,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4675010549018024e-06,
+      "loss": 1.1412,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15347251810822327,
+      "grad_norm": 11.375,
+      "learning_rate": 2.467462976391585e-06,
+      "loss": 0.4821,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15355773327652322,
+      "grad_norm": 22.625,
+      "learning_rate": 2.4674248758805873e-06,
+      "loss": 1.1391,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15364294844482318,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4673867533694982e-06,
+      "loss": 1.0296,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15372816361312314,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4673486088590064e-06,
+      "loss": 0.8798,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1538133787814231,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4673104423498005e-06,
+      "loss": 1.0988,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15389859394972305,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4672722538425713e-06,
+      "loss": 1.1313,
+      "step": 1806
+    },
+    {
+      "epoch": 0.153983809118023,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4672340433380083e-06,
+      "loss": 0.7441,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15406902428632296,
+      "grad_norm": 20.375,
+      "learning_rate": 2.467195810836802e-06,
+      "loss": 0.914,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15415423945462292,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4671575563396437e-06,
+      "loss": 0.3241,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15423945462292288,
+      "grad_norm": 21.875,
+      "learning_rate": 2.467119279847224e-06,
+      "loss": 1.3322,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15432466979122283,
+      "grad_norm": 9.8125,
+      "learning_rate": 2.4670809813602355e-06,
+      "loss": 0.3766,
+      "step": 1811
+    },
+    {
+      "epoch": 0.1544098849595228,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.467042660879369e-06,
+      "loss": 0.6851,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15449510012782275,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4670043184053184e-06,
+      "loss": 0.9174,
+      "step": 1813
+    },
+    {
+      "epoch": 0.1545803152961227,
+      "grad_norm": 18.875,
+      "learning_rate": 2.466965953938776e-06,
+      "loss": 1.0705,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15466553046442266,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4669275674804347e-06,
+      "loss": 1.5079,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15475074563272262,
+      "grad_norm": 19.375,
+      "learning_rate": 2.466889159030989e-06,
+      "loss": 0.9318,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15483596080102258,
+      "grad_norm": 18.625,
+      "learning_rate": 2.466850728591132e-06,
+      "loss": 0.8093,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15492117596932253,
+      "grad_norm": 12.125,
+      "learning_rate": 2.466812276161559e-06,
+      "loss": 0.4672,
+      "step": 1818
+    },
+    {
+      "epoch": 0.1550063911376225,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4667738017429648e-06,
+      "loss": 0.6757,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15509160630592245,
+      "grad_norm": 11.375,
+      "learning_rate": 2.466735305336044e-06,
+      "loss": 0.5264,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1551768214742224,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4666967869414932e-06,
+      "loss": 0.7891,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15526203664252236,
+      "grad_norm": 21.125,
+      "learning_rate": 2.4666582465600073e-06,
+      "loss": 0.8732,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15534725181082232,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.466619684192284e-06,
+      "loss": 0.5744,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15543246697912227,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4665810998390196e-06,
+      "loss": 0.5755,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15551768214742223,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4665424935009106e-06,
+      "loss": 0.746,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1556028973157222,
+      "grad_norm": 15.125,
+      "learning_rate": 2.466503865178656e-06,
+      "loss": 0.7084,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15568811248402215,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4664652148729533e-06,
+      "loss": 0.6737,
+      "step": 1827
+    },
+    {
+      "epoch": 0.1557733276523221,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.466426542584501e-06,
+      "loss": 0.6415,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15585854282062206,
+      "grad_norm": 12.125,
+      "learning_rate": 2.466387848313998e-06,
+      "loss": 0.8392,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15594375798892202,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.466349132062143e-06,
+      "loss": 0.7633,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15602897315722197,
+      "grad_norm": 22.875,
+      "learning_rate": 2.466310393829636e-06,
+      "loss": 1.0498,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15611418832552193,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4662716336171778e-06,
+      "loss": 0.4587,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1561994034938219,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4662328514254672e-06,
+      "loss": 0.6356,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15628461866212187,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4661940472552067e-06,
+      "loss": 1.142,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15636983383042183,
+      "grad_norm": 13.25,
+      "learning_rate": 2.466155221107096e-06,
+      "loss": 0.7789,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15645504899872179,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4661163729818383e-06,
+      "loss": 0.9163,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15654026416702174,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4660775028801345e-06,
+      "loss": 0.5911,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1566254793353217,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4660386108026872e-06,
+      "loss": 0.8785,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15671069450362166,
+      "grad_norm": 23.75,
+      "learning_rate": 2.4659996967501997e-06,
+      "loss": 0.7775,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1567959096719216,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4659607607233748e-06,
+      "loss": 0.4976,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15688112484022157,
+      "grad_norm": 19.5,
+      "learning_rate": 2.465921802722916e-06,
+      "loss": 0.9605,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15696634000852153,
+      "grad_norm": 16.125,
+      "learning_rate": 2.465882822749528e-06,
+      "loss": 0.8789,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15705155517682148,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.465843820803914e-06,
+      "loss": 1.0392,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15713677034512144,
+      "grad_norm": 13.75,
+      "learning_rate": 2.46580479688678e-06,
+      "loss": 0.5005,
+      "step": 1844
+    },
+    {
+      "epoch": 0.1572219855134214,
+      "grad_norm": 20.25,
+      "learning_rate": 2.465765750998831e-06,
+      "loss": 1.0982,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15730720068172135,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4657266831407717e-06,
+      "loss": 0.7649,
+      "step": 1846
+    },
+    {
+      "epoch": 0.1573924158500213,
+      "grad_norm": 30.625,
+      "learning_rate": 2.465687593313309e-06,
+      "loss": 0.9712,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15747763101832127,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.4656484815171494e-06,
+      "loss": 0.7227,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15756284618662122,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4656093477529986e-06,
+      "loss": 0.7647,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15764806135492118,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.465570192021565e-06,
+      "loss": 0.6293,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15773327652322114,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4655310143235556e-06,
+      "loss": 0.6958,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1578184916915211,
+      "grad_norm": 27.0,
+      "learning_rate": 2.4654918146596783e-06,
+      "loss": 0.9169,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15790370685982105,
+      "grad_norm": 19.125,
+      "learning_rate": 2.465452593030642e-06,
+      "loss": 0.9063,
+      "step": 1853
+    },
+    {
+      "epoch": 0.157988922028121,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4654133494371547e-06,
+      "loss": 0.5622,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15807413719642097,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.465374083879926e-06,
+      "loss": 0.3138,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15815935236472092,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.465334796359666e-06,
+      "loss": 0.5773,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15824456753302088,
+      "grad_norm": 21.125,
+      "learning_rate": 2.4652954868770834e-06,
+      "loss": 0.8814,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15832978270132084,
+      "grad_norm": 31.0,
+      "learning_rate": 2.4652561554328895e-06,
+      "loss": 0.8206,
+      "step": 1858
+    },
+    {
+      "epoch": 0.1584149978696208,
+      "grad_norm": 12.5,
+      "learning_rate": 2.465216802027795e-06,
+      "loss": 0.6012,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15850021303792075,
+      "grad_norm": 13.75,
+      "learning_rate": 2.465177426662511e-06,
+      "loss": 0.7477,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1585854282062207,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4651380293377486e-06,
+      "loss": 1.0542,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15867064337452066,
+      "grad_norm": 15.625,
+      "learning_rate": 2.46509861005422e-06,
+      "loss": 0.745,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15875585854282062,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4650591688126384e-06,
+      "loss": 0.7779,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15884107371112058,
+      "grad_norm": 19.0,
+      "learning_rate": 2.465019705613715e-06,
+      "loss": 0.9052,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15892628887942054,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4649802204581643e-06,
+      "loss": 0.7703,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1590115040477205,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4649407133466987e-06,
+      "loss": 0.639,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15909671921602045,
+      "grad_norm": 17.125,
+      "learning_rate": 2.464901184280033e-06,
+      "loss": 0.809,
+      "step": 1867
+    },
+    {
+      "epoch": 0.1591819343843204,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4648616332588807e-06,
+      "loss": 0.6356,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15926714955262036,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4648220602839577e-06,
+      "loss": 0.9886,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15935236472092032,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4647824653559786e-06,
+      "loss": 1.0655,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15943757988922028,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4647428484756584e-06,
+      "loss": 0.9297,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15952279505752023,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4647032096437133e-06,
+      "loss": 1.1566,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1596080102258202,
+      "grad_norm": 14.875,
+      "learning_rate": 2.46466354886086e-06,
+      "loss": 0.9909,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15969322539412015,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4646238661278148e-06,
+      "loss": 0.9221,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1597784405624201,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4645841614452955e-06,
+      "loss": 0.9497,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15986365573072006,
+      "grad_norm": 8.8125,
+      "learning_rate": 2.464544434814018e-06,
+      "loss": 0.3158,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15994887089902002,
+      "grad_norm": 22.625,
+      "learning_rate": 2.4645046862347024e-06,
+      "loss": 1.2811,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16003408606731997,
+      "grad_norm": 24.375,
+      "learning_rate": 2.4644649157080647e-06,
+      "loss": 0.6415,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16011930123561993,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4644251232348253e-06,
+      "loss": 1.0919,
+      "step": 1879
+    },
+    {
+      "epoch": 0.1602045164039199,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4643853088157026e-06,
+      "loss": 0.6815,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16028973157221985,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4643454724514165e-06,
+      "loss": 0.766,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1603749467405198,
+      "grad_norm": 15.375,
+      "learning_rate": 2.464305614142686e-06,
+      "loss": 0.843,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16046016190881976,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4642657338902324e-06,
+      "loss": 0.5713,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16054537707711972,
+      "grad_norm": 16.625,
+      "learning_rate": 2.464225831694776e-06,
+      "loss": 0.2904,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16063059224541967,
+      "grad_norm": 35.5,
+      "learning_rate": 2.464185907557038e-06,
+      "loss": 0.6779,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16071580741371963,
+      "grad_norm": 10.625,
+      "learning_rate": 2.4641459614777393e-06,
+      "loss": 0.4465,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1608010225820196,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4641059934576022e-06,
+      "loss": 0.3821,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16088623775031954,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4640660034973492e-06,
+      "loss": 0.8156,
+      "step": 1888
+    },
+    {
+      "epoch": 0.1609714529186195,
+      "grad_norm": 25.0,
+      "learning_rate": 2.4640259915977024e-06,
+      "loss": 0.854,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16105666808691946,
+      "grad_norm": 10.75,
+      "learning_rate": 2.463985957759385e-06,
+      "loss": 0.3041,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16114188325521944,
+      "grad_norm": 15.25,
+      "learning_rate": 2.463945901983121e-06,
+      "loss": 0.8962,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1612270984235194,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.4639058242696337e-06,
+      "loss": 0.3486,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16131231359181936,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4638657246196474e-06,
+      "loss": 0.5969,
+      "step": 1893
+    },
+    {
+      "epoch": 0.1613975287601193,
+      "grad_norm": 34.5,
+      "learning_rate": 2.463825603033887e-06,
+      "loss": 1.2165,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16148274392841927,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4637854595130774e-06,
+      "loss": 0.6792,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16156795909671923,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4637452940579434e-06,
+      "loss": 0.7919,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16165317426501918,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.463705106669212e-06,
+      "loss": 0.5261,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16173838943331914,
+      "grad_norm": 30.25,
+      "learning_rate": 2.4636648973476086e-06,
+      "loss": 1.0703,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1618236046016191,
+      "grad_norm": 16.75,
+      "learning_rate": 2.46362466609386e-06,
+      "loss": 0.9846,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16190881976991905,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4635844129086934e-06,
+      "loss": 0.5692,
+      "step": 1900
+    },
+    {
+      "epoch": 0.161994034938219,
+      "grad_norm": 17.75,
+      "learning_rate": 2.463544137792836e-06,
+      "loss": 0.9089,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16207925010651897,
+      "grad_norm": 21.5,
+      "learning_rate": 2.463503840747016e-06,
+      "loss": 1.0333,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16216446527481893,
+      "grad_norm": 10.25,
+      "learning_rate": 2.463463521771961e-06,
+      "loss": 0.3731,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16224968044311888,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4634231808684e-06,
+      "loss": 0.5566,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16233489561141884,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.463382818037062e-06,
+      "loss": 0.5679,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1624201107797188,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4633424332786767e-06,
+      "loss": 0.8186,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16250532594801875,
+      "grad_norm": 15.625,
+      "learning_rate": 2.463302026593973e-06,
+      "loss": 0.7229,
+      "step": 1907
+    },
+    {
+      "epoch": 0.1625905411163187,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.463261597983682e-06,
+      "loss": 0.8144,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16267575628461867,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4632211474485333e-06,
+      "loss": 0.9194,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16276097145291862,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.463180674989259e-06,
+      "loss": 0.5902,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16284618662121858,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4631401806065898e-06,
+      "loss": 0.4058,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16293140178951854,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4630996643012577e-06,
+      "loss": 1.0518,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1630166169578185,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4630591260739946e-06,
+      "loss": 0.8537,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16310183212611845,
+      "grad_norm": 22.875,
+      "learning_rate": 2.4630185659255333e-06,
+      "loss": 1.2322,
+      "step": 1914
+    },
+    {
+      "epoch": 0.1631870472944184,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.462977983856607e-06,
+      "loss": 0.5618,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16327226246271836,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4629373798679487e-06,
+      "loss": 0.6039,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16335747763101832,
+      "grad_norm": 13.375,
+      "learning_rate": 2.462896753960292e-06,
+      "loss": 0.7162,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16344269279931828,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4628561061343716e-06,
+      "loss": 1.0867,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16352790796761824,
+      "grad_norm": 16.25,
+      "learning_rate": 2.462815436390922e-06,
+      "loss": 0.9963,
+      "step": 1919
+    },
+    {
+      "epoch": 0.1636131231359182,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.462774744730677e-06,
+      "loss": 0.6273,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16369833830421815,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4627340311543735e-06,
+      "loss": 0.7721,
+      "step": 1921
+    },
+    {
+      "epoch": 0.1637835534725181,
+      "grad_norm": 18.5,
+      "learning_rate": 2.462693295662746e-06,
+      "loss": 0.6745,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16386876864081806,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.462652538256532e-06,
+      "loss": 0.4459,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16395398380911802,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.462611758936467e-06,
+      "loss": 0.6698,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16403919897741798,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4625709577032876e-06,
+      "loss": 0.9035,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16412441414571793,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4625301345577318e-06,
+      "loss": 0.8877,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1642096293140179,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4624892895005377e-06,
+      "loss": 0.8654,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16429484448231785,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4624484225324424e-06,
+      "loss": 0.8497,
+      "step": 1928
+    },
+    {
+      "epoch": 0.1643800596506178,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4624075336541848e-06,
+      "loss": 0.8131,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16446527481891776,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4623666228665043e-06,
+      "loss": 0.8727,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16455048998721772,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4623256901701396e-06,
+      "loss": 0.8846,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16463570515551768,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4622847355658306e-06,
+      "loss": 0.7339,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16472092032381763,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4622437590543176e-06,
+      "loss": 0.9891,
+      "step": 1933
+    },
+    {
+      "epoch": 0.1648061354921176,
+      "grad_norm": 14.125,
+      "learning_rate": 2.46220276063634e-06,
+      "loss": 0.5347,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16489135066041755,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4621617403126407e-06,
+      "loss": 1.1538,
+      "step": 1935
+    },
+    {
+      "epoch": 0.1649765658287175,
+      "grad_norm": 20.625,
+      "learning_rate": 2.462120698083959e-06,
+      "loss": 1.0269,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16506178099701746,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4620796339510382e-06,
+      "loss": 0.7374,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16514699616531742,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.462038547914619e-06,
+      "loss": 0.4763,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16523221133361737,
+      "grad_norm": 24.375,
+      "learning_rate": 2.4619974399754444e-06,
+      "loss": 1.3747,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16531742650191733,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4619563101342577e-06,
+      "loss": 1.0067,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1654026416702173,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4619151583918013e-06,
+      "loss": 1.0633,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16548785683851724,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4618739847488196e-06,
+      "loss": 0.5575,
+      "step": 1942
+    },
+    {
+      "epoch": 0.1655730720068172,
+      "grad_norm": 14.25,
+      "learning_rate": 2.461832789206056e-06,
+      "loss": 0.5893,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16565828717511716,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.461791571764256e-06,
+      "loss": 0.7454,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16574350234341712,
+      "grad_norm": 20.5,
+      "learning_rate": 2.4617503324241633e-06,
+      "loss": 0.9566,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16582871751171707,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4617090711865235e-06,
+      "loss": 0.7483,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16591393268001706,
+      "grad_norm": 22.625,
+      "learning_rate": 2.4616677880520825e-06,
+      "loss": 0.6018,
+      "step": 1947
+    },
+    {
+      "epoch": 0.165999147848317,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.461626483021586e-06,
+      "loss": 0.7312,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16608436301661697,
+      "grad_norm": 17.5,
+      "learning_rate": 2.461585156095781e-06,
+      "loss": 0.7235,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16616957818491693,
+      "grad_norm": 29.125,
+      "learning_rate": 2.4615438072754134e-06,
+      "loss": 1.2423,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16625479335321688,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.461502436561231e-06,
+      "loss": 0.7071,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16634000852151684,
+      "grad_norm": 28.875,
+      "learning_rate": 2.4614610439539813e-06,
+      "loss": 0.7138,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1664252236898168,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4614196294544122e-06,
+      "loss": 0.4549,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16651043885811675,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4613781930632723e-06,
+      "loss": 0.7831,
+      "step": 1954
+    },
+    {
+      "epoch": 0.1665956540264167,
+      "grad_norm": 22.125,
+      "learning_rate": 2.4613367347813105e-06,
+      "loss": 1.1,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16668086919471667,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4612952546092763e-06,
+      "loss": 0.8827,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16676608436301663,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.461253752547918e-06,
+      "loss": 0.6588,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16685129953131658,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4612122285979863e-06,
+      "loss": 1.0942,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16693651469961654,
+      "grad_norm": 24.75,
+      "learning_rate": 2.4611706827602324e-06,
+      "loss": 1.3617,
+      "step": 1959
+    },
+    {
+      "epoch": 0.1670217298679165,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4611291150354054e-06,
+      "loss": 0.5499,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16710694503621645,
+      "grad_norm": 19.0,
+      "learning_rate": 2.461087525424258e-06,
+      "loss": 0.4473,
+      "step": 1961
+    },
+    {
+      "epoch": 0.1671921602045164,
+      "grad_norm": 16.875,
+      "learning_rate": 2.461045913927541e-06,
+      "loss": 0.92,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16727737537281637,
+      "grad_norm": 20.75,
+      "learning_rate": 2.461004280546007e-06,
+      "loss": 0.6915,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16736259054111632,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4609626252804074e-06,
+      "loss": 0.6387,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16744780570941628,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4609209481314957e-06,
+      "loss": 0.7349,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16753302087771624,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4608792491000245e-06,
+      "loss": 0.7464,
+      "step": 1966
+    },
+    {
+      "epoch": 0.1676182360460162,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4608375281867475e-06,
+      "loss": 0.6818,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16770345121431615,
+      "grad_norm": 12.25,
+      "learning_rate": 2.460795785392419e-06,
+      "loss": 0.6621,
+      "step": 1968
+    },
+    {
+      "epoch": 0.1677886663826161,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4607540207177934e-06,
+      "loss": 0.9225,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16787388155091607,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.460712234163625e-06,
+      "loss": 0.6546,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16795909671921602,
+      "grad_norm": 16.375,
+      "learning_rate": 2.460670425730669e-06,
+      "loss": 0.8066,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16804431188751598,
+      "grad_norm": 10.75,
+      "learning_rate": 2.4606285954196805e-06,
+      "loss": 0.5965,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16812952705581594,
+      "grad_norm": 27.125,
+      "learning_rate": 2.4605867432314167e-06,
+      "loss": 0.9837,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1682147422241159,
+      "grad_norm": 16.125,
+      "learning_rate": 2.460544869166633e-06,
+      "loss": 0.9392,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16829995739241585,
+      "grad_norm": 15.125,
+      "learning_rate": 2.460502973226086e-06,
+      "loss": 0.9626,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1683851725607158,
+      "grad_norm": 25.0,
+      "learning_rate": 2.4604610554105326e-06,
+      "loss": 1.2058,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16847038772901576,
+      "grad_norm": 17.875,
+      "learning_rate": 2.460419115720731e-06,
+      "loss": 1.023,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16855560289731572,
+      "grad_norm": 14.25,
+      "learning_rate": 2.460377154157439e-06,
+      "loss": 0.6717,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16864081806561568,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4603351707214144e-06,
+      "loss": 0.6409,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16872603323391563,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4602931654134165e-06,
+      "loss": 0.959,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1688112484022156,
+      "grad_norm": 18.25,
+      "learning_rate": 2.460251138234204e-06,
+      "loss": 1.0022,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16889646357051555,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4602090891845363e-06,
+      "loss": 0.6976,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1689816787388155,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4601670182651736e-06,
+      "loss": 0.5451,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16906689390711546,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4601249254768762e-06,
+      "loss": 1.0195,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16915210907541542,
+      "grad_norm": 19.0,
+      "learning_rate": 2.460082810820404e-06,
+      "loss": 0.8803,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16923732424371538,
+      "grad_norm": 18.875,
+      "learning_rate": 2.460040674296519e-06,
+      "loss": 1.1858,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16932253941201533,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4599985159059817e-06,
+      "loss": 1.0269,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1694077545803153,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4599563356495547e-06,
+      "loss": 0.4623,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16949296974861525,
+      "grad_norm": 12.875,
+      "learning_rate": 2.459914133528e-06,
+      "loss": 0.5695,
+      "step": 1989
+    },
+    {
+      "epoch": 0.1695781849169152,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4598719095420803e-06,
+      "loss": 0.7257,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16966340008521516,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4598296636925586e-06,
+      "loss": 0.8458,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16974861525351512,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4597873959801982e-06,
+      "loss": 0.6678,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16983383042181507,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.459745106405763e-06,
+      "loss": 0.4443,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16991904559011503,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4597027949700176e-06,
+      "loss": 0.6835,
+      "step": 1994
+    },
+    {
+      "epoch": 0.170004260758415,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4596604616737257e-06,
+      "loss": 0.7451,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17008947592671494,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.459618106517653e-06,
+      "loss": 0.6533,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1701746910950149,
+      "grad_norm": 16.875,
+      "learning_rate": 2.459575729502565e-06,
+      "loss": 1.0351,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17025990626331486,
+      "grad_norm": 21.875,
+      "learning_rate": 2.4595333306292272e-06,
+      "loss": 0.3285,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17034512143161482,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4594909098984058e-06,
+      "loss": 0.6,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17043033659991477,
+      "grad_norm": 16.0,
+      "learning_rate": 2.459448467310867e-06,
+      "loss": 1.0138,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17051555176821473,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4594060028673788e-06,
+      "loss": 0.5304,
+      "step": 2001
+    },
+    {
+      "epoch": 0.1706007669365147,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4593635165687076e-06,
+      "loss": 0.7781,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17068598210481467,
+      "grad_norm": 22.625,
+      "learning_rate": 2.459321008415622e-06,
+      "loss": 1.0972,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17077119727311463,
+      "grad_norm": 12.125,
+      "learning_rate": 2.459278478408889e-06,
+      "loss": 0.4955,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17085641244141458,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4592359265492786e-06,
+      "loss": 0.472,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17094162760971454,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4591933528375586e-06,
+      "loss": 0.475,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1710268427780145,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.459150757274499e-06,
+      "loss": 0.9155,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17111205794631446,
+      "grad_norm": 19.5,
+      "learning_rate": 2.459108139860869e-06,
+      "loss": 1.1182,
+      "step": 2008
+    },
+    {
+      "epoch": 0.1711972731146144,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.459065500597439e-06,
+      "loss": 0.619,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17128248828291437,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.45902283948498e-06,
+      "loss": 0.9494,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17136770345121433,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4589801565242627e-06,
+      "loss": 0.5641,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17145291861951428,
+      "grad_norm": 12.75,
+      "learning_rate": 2.458937451716058e-06,
+      "loss": 0.6059,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17153813378781424,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4588947250611377e-06,
+      "loss": 0.8325,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1716233489561142,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4588519765602743e-06,
+      "loss": 0.6977,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17170856412441415,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4588092062142394e-06,
+      "loss": 0.6614,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1717937792927141,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4587664140238076e-06,
+      "loss": 1.1969,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17187899446101407,
+      "grad_norm": 22.125,
+      "learning_rate": 2.458723599989751e-06,
+      "loss": 0.9492,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17196420962931402,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.458680764112843e-06,
+      "loss": 0.6721,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17204942479761398,
+      "grad_norm": 18.5,
+      "learning_rate": 2.458637906393858e-06,
+      "loss": 0.8085,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17213463996591394,
+      "grad_norm": 11.0,
+      "learning_rate": 2.458595026833571e-06,
+      "loss": 0.4131,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1722198551342139,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4585521254327567e-06,
+      "loss": 0.8082,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17230507030251385,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.45850920219219e-06,
+      "loss": 0.8042,
+      "step": 2022
+    },
+    {
+      "epoch": 0.1723902854708138,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.458466257112647e-06,
+      "loss": 0.7371,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17247550063911377,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.4584232901949036e-06,
+      "loss": 0.4148,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17256071580741372,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4583803014397355e-06,
+      "loss": 0.7513,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17264593097571368,
+      "grad_norm": 8.125,
+      "learning_rate": 2.458337290847921e-06,
+      "loss": 0.2694,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17273114614401364,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4582942584202363e-06,
+      "loss": 1.0149,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1728163613123136,
+      "grad_norm": 17.5,
+      "learning_rate": 2.458251204157459e-06,
+      "loss": 0.9005,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17290157648061355,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.458208128060368e-06,
+      "loss": 0.6792,
+      "step": 2029
+    },
+    {
+      "epoch": 0.1729867916489135,
+      "grad_norm": 16.375,
+      "learning_rate": 2.458165030129741e-06,
+      "loss": 0.6161,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17307200681721346,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.458121910366357e-06,
+      "loss": 0.4292,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17315722198551342,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4580787687709947e-06,
+      "loss": 1.0835,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17324243715381338,
+      "grad_norm": 12.875,
+      "learning_rate": 2.458035605344435e-06,
+      "loss": 0.7919,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17332765232211333,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4579924200874566e-06,
+      "loss": 0.7203,
+      "step": 2034
+    },
+    {
+      "epoch": 0.1734128674904133,
+      "grad_norm": 22.0,
+      "learning_rate": 2.4579492130008406e-06,
+      "loss": 1.0555,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17349808265871325,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4579059840853677e-06,
+      "loss": 0.8128,
+      "step": 2036
+    },
+    {
+      "epoch": 0.1735832978270132,
+      "grad_norm": 31.375,
+      "learning_rate": 2.457862733341819e-06,
+      "loss": 0.9135,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17366851299531316,
+      "grad_norm": 14.125,
+      "learning_rate": 2.457819460770976e-06,
+      "loss": 0.713,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17375372816361312,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.457776166373621e-06,
+      "loss": 0.6362,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17383894333191308,
+      "grad_norm": 16.5,
+      "learning_rate": 2.457732850150536e-06,
+      "loss": 1.0217,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17392415850021303,
+      "grad_norm": 17.625,
+      "learning_rate": 2.457689512102504e-06,
+      "loss": 0.6053,
+      "step": 2041
+    },
+    {
+      "epoch": 0.174009373668513,
+      "grad_norm": 14.75,
+      "learning_rate": 2.457646152230308e-06,
+      "loss": 0.8687,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17409458883681295,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4576027705347316e-06,
+      "loss": 0.5309,
+      "step": 2043
+    },
+    {
+      "epoch": 0.1741798040051129,
+      "grad_norm": 21.75,
+      "learning_rate": 2.457559367016559e-06,
+      "loss": 0.903,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17426501917341286,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4575159416765742e-06,
+      "loss": 0.6945,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17435023434171282,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4574724945155624e-06,
+      "loss": 0.9111,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17443544951001277,
+      "grad_norm": 14.875,
+      "learning_rate": 2.457429025534308e-06,
+      "loss": 0.8245,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17452066467831273,
+      "grad_norm": 25.125,
+      "learning_rate": 2.4573855347335972e-06,
+      "loss": 1.5115,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1746058798466127,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4573420221142155e-06,
+      "loss": 0.8505,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17469109501491265,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4572984876769495e-06,
+      "loss": 0.8712,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1747763101832126,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.457254931422586e-06,
+      "loss": 0.6953,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17486152535151256,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4572113533519116e-06,
+      "loss": 0.5377,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17494674051981252,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.457167753465714e-06,
+      "loss": 0.7744,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17503195568811247,
+      "grad_norm": 21.5,
+      "learning_rate": 2.457124131764781e-06,
+      "loss": 0.846,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17511717085641243,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4570804882499015e-06,
+      "loss": 0.9177,
+      "step": 2055
+    },
+    {
+      "epoch": 0.1752023860247124,
+      "grad_norm": 12.125,
+      "learning_rate": 2.457036822921864e-06,
+      "loss": 0.802,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17528760119301234,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.456993135781457e-06,
+      "loss": 0.5264,
+      "step": 2057
+    },
+    {
+      "epoch": 0.1753728163613123,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4569494268294702e-06,
+      "loss": 0.5958,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17545803152961226,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4569056960666937e-06,
+      "loss": 1.366,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17554324669791224,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.456861943493918e-06,
+      "loss": 0.7018,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1756284618662122,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.456818169111933e-06,
+      "loss": 0.7075,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17571367703451216,
+      "grad_norm": 16.25,
+      "learning_rate": 2.45677437292153e-06,
+      "loss": 0.8157,
+      "step": 2062
+    },
+    {
+      "epoch": 0.1757988922028121,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4567305549235008e-06,
+      "loss": 1.0475,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17588410737111207,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.456686715118637e-06,
+      "loss": 0.327,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17596932253941203,
+      "grad_norm": 19.0,
+      "learning_rate": 2.456642853507731e-06,
+      "loss": 1.1008,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17605453770771198,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4565989700915746e-06,
+      "loss": 1.0194,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17613975287601194,
+      "grad_norm": 26.0,
+      "learning_rate": 2.456555064870962e-06,
+      "loss": 0.8765,
+      "step": 2067
+    },
+    {
+      "epoch": 0.1762249680443119,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.456511137846686e-06,
+      "loss": 0.9368,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17631018321261185,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4564671890195406e-06,
+      "loss": 0.9802,
+      "step": 2069
+    },
+    {
+      "epoch": 0.1763953983809118,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4564232183903196e-06,
+      "loss": 1.0024,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17648061354921177,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4563792259598186e-06,
+      "loss": 0.9557,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17656582871751172,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4563352117288312e-06,
+      "loss": 0.5583,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17665104388581168,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4562911756981537e-06,
+      "loss": 1.0684,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17673625905411164,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4562471178685815e-06,
+      "loss": 0.9078,
+      "step": 2074
+    },
+    {
+      "epoch": 0.1768214742224116,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.456203038240911e-06,
+      "loss": 0.9359,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17690668939071155,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4561589368159388e-06,
+      "loss": 1.1046,
+      "step": 2076
+    },
+    {
+      "epoch": 0.1769919045590115,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4561148135944616e-06,
+      "loss": 0.8891,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17707711972731147,
+      "grad_norm": 28.25,
+      "learning_rate": 2.4560706685772767e-06,
+      "loss": 0.8514,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17716233489561142,
+      "grad_norm": 28.0,
+      "learning_rate": 2.4560265017651826e-06,
+      "loss": 0.5024,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17724755006391138,
+      "grad_norm": 18.5,
+      "learning_rate": 2.455982313158976e-06,
+      "loss": 1.2425,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17733276523221134,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.455938102759457e-06,
+      "loss": 0.7255,
+      "step": 2081
+    },
+    {
+      "epoch": 0.1774179804005113,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4558938705674236e-06,
+      "loss": 1.0185,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17750319556881125,
+      "grad_norm": 16.25,
+      "learning_rate": 2.455849616583675e-06,
+      "loss": 0.8518,
+      "step": 2083
+    },
+    {
+      "epoch": 0.1775884107371112,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4558053408090115e-06,
+      "loss": 0.9188,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17767362590541116,
+      "grad_norm": 12.75,
+      "learning_rate": 2.455761043244233e-06,
+      "loss": 0.8994,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17775884107371112,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4557167238901404e-06,
+      "loss": 0.6106,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17784405624201108,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4556723827475333e-06,
+      "loss": 0.9989,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17792927141031104,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4556280198172144e-06,
+      "loss": 0.6338,
+      "step": 2088
+    },
+    {
+      "epoch": 0.178014486578611,
+      "grad_norm": 15.75,
+      "learning_rate": 2.455583635099985e-06,
+      "loss": 0.8638,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17809970174691095,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4555392285966467e-06,
+      "loss": 0.8543,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1781849169152109,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.455494800308002e-06,
+      "loss": 0.3567,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17827013208351086,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4554503502348544e-06,
+      "loss": 0.5069,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17835534725181082,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4554058783780064e-06,
+      "loss": 0.6473,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17844056242011078,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.455361384738263e-06,
+      "loss": 0.735,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17852577758841073,
+      "grad_norm": 23.625,
+      "learning_rate": 2.4553168693164262e-06,
+      "loss": 1.4259,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1786109927567107,
+      "grad_norm": 12.625,
+      "learning_rate": 2.455272332113302e-06,
+      "loss": 0.6685,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17869620792501065,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4552277731296952e-06,
+      "loss": 0.7064,
+      "step": 2097
+    },
+    {
+      "epoch": 0.1787814230933106,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4551831923664098e-06,
+      "loss": 0.8756,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17886663826161056,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4551385898242525e-06,
+      "loss": 1.028,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17895185342991052,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.455093965504029e-06,
+      "loss": 0.7538,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17903706859821047,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4550493194065462e-06,
+      "loss": 0.9061,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17912228376651043,
+      "grad_norm": 19.375,
+      "learning_rate": 2.45500465153261e-06,
+      "loss": 0.9957,
+      "step": 2102
+    },
+    {
+      "epoch": 0.1792074989348104,
+      "grad_norm": 16.375,
+      "learning_rate": 2.454959961883028e-06,
+      "loss": 0.7387,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17929271410311035,
+      "grad_norm": 12.875,
+      "learning_rate": 2.454915250458608e-06,
+      "loss": 0.6402,
+      "step": 2104
+    },
+    {
+      "epoch": 0.1793779292714103,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4548705172601574e-06,
+      "loss": 0.4893,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17946314443971026,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4548257622884853e-06,
+      "loss": 0.6715,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17954835960801022,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4547809855444e-06,
+      "loss": 0.9421,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17963357477631017,
+      "grad_norm": 13.875,
+      "learning_rate": 2.454736187028711e-06,
+      "loss": 0.6871,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17971878994461013,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4546913667422272e-06,
+      "loss": 0.6145,
+      "step": 2109
+    },
+    {
+      "epoch": 0.1798040051129101,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4546465246857595e-06,
+      "loss": 0.6275,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17988922028121004,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4546016608601173e-06,
+      "loss": 0.9303,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17997443544951,
+      "grad_norm": 13.125,
+      "learning_rate": 2.454556775266112e-06,
+      "loss": 0.6837,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18005965061780996,
+      "grad_norm": 17.25,
+      "learning_rate": 2.454511867904555e-06,
+      "loss": 0.9435,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18014486578610991,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4544669387762567e-06,
+      "loss": 0.7194,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18023008095440987,
+      "grad_norm": 16.75,
+      "learning_rate": 2.45442198788203e-06,
+      "loss": 1.0713,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18031529612270986,
+      "grad_norm": 24.375,
+      "learning_rate": 2.4543770152226857e-06,
+      "loss": 1.5671,
+      "step": 2116
+    },
+    {
+      "epoch": 0.1804005112910098,
+      "grad_norm": 21.625,
+      "learning_rate": 2.454332020799039e-06,
+      "loss": 0.9992,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18048572645930977,
+      "grad_norm": 13.5,
+      "learning_rate": 2.454287004611901e-06,
+      "loss": 0.7161,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18057094162760973,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.454241966662086e-06,
+      "loss": 0.756,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18065615679590968,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4541969069504075e-06,
+      "loss": 0.7193,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18074137196420964,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.45415182547768e-06,
+      "loss": 0.5944,
+      "step": 2121
+    },
+    {
+      "epoch": 0.1808265871325096,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4541067222447184e-06,
+      "loss": 1.2357,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18091180230080955,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4540615972523373e-06,
+      "loss": 0.3164,
+      "step": 2123
+    },
+    {
+      "epoch": 0.1809970174691095,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4540164505013523e-06,
+      "loss": 0.6552,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18108223263740947,
+      "grad_norm": 19.75,
+      "learning_rate": 2.45397128199258e-06,
+      "loss": 1.3354,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18116744780570943,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4539260917268354e-06,
+      "loss": 0.708,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18125266297400938,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4538808797049357e-06,
+      "loss": 0.7667,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18133787814230934,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4538356459276977e-06,
+      "loss": 0.9075,
+      "step": 2128
+    },
+    {
+      "epoch": 0.1814230933106093,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4537903903959394e-06,
+      "loss": 0.7951,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18150830847890925,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4537451131104776e-06,
+      "loss": 0.8273,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1815935236472092,
+      "grad_norm": 14.75,
+      "learning_rate": 2.453699814072132e-06,
+      "loss": 0.7433,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18167873881550917,
+      "grad_norm": 63.5,
+      "learning_rate": 2.45365449328172e-06,
+      "loss": 1.5492,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18176395398380912,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4536091507400608e-06,
+      "loss": 0.9812,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18184916915210908,
+      "grad_norm": 14.125,
+      "learning_rate": 2.453563786447974e-06,
+      "loss": 0.7726,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18193438432040904,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4535184004062795e-06,
+      "loss": 1.0586,
+      "step": 2135
+    },
+    {
+      "epoch": 0.182019599488709,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.453472992615797e-06,
+      "loss": 0.5624,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18210481465700895,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4534275630773475e-06,
+      "loss": 0.5086,
+      "step": 2137
+    },
+    {
+      "epoch": 0.1821900298253089,
+      "grad_norm": 12.75,
+      "learning_rate": 2.453382111791752e-06,
+      "loss": 0.5329,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18227524499360886,
+      "grad_norm": 13.875,
+      "learning_rate": 2.453336638759831e-06,
+      "loss": 0.6857,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18236046016190882,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.453291143982407e-06,
+      "loss": 0.8337,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18244567533020878,
+      "grad_norm": 21.5,
+      "learning_rate": 2.4532456274603024e-06,
+      "loss": 1.211,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18253089049850874,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4532000891943393e-06,
+      "loss": 0.8509,
+      "step": 2142
+    },
+    {
+      "epoch": 0.1826161056668087,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4531545291853407e-06,
+      "loss": 0.831,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18270132083510865,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4531089474341296e-06,
+      "loss": 0.7519,
+      "step": 2144
+    },
+    {
+      "epoch": 0.1827865360034086,
+      "grad_norm": 9.5,
+      "learning_rate": 2.45306334394153e-06,
+      "loss": 0.2772,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18287175117170856,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4530177187083666e-06,
+      "loss": 0.5885,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18295696634000852,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4529720717354627e-06,
+      "loss": 1.0621,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18304218150830848,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4529264030236445e-06,
+      "loss": 0.9031,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18312739667660843,
+      "grad_norm": 16.25,
+      "learning_rate": 2.452880712573736e-06,
+      "loss": 0.8009,
+      "step": 2149
+    },
+    {
+      "epoch": 0.1832126118449084,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.452835000386563e-06,
+      "loss": 0.7724,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18329782701320835,
+      "grad_norm": 21.75,
+      "learning_rate": 2.4527892664629525e-06,
+      "loss": 0.8147,
+      "step": 2151
+    },
+    {
+      "epoch": 0.1833830421815083,
+      "grad_norm": 24.75,
+      "learning_rate": 2.452743510803731e-06,
+      "loss": 0.7664,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18346825734980826,
+      "grad_norm": 20.75,
+      "learning_rate": 2.452697733409724e-06,
+      "loss": 1.2781,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18355347251810822,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4526519342817593e-06,
+      "loss": 1.1239,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18363868768640818,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.452606113420666e-06,
+      "loss": 0.7286,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18372390285470813,
+      "grad_norm": 19.75,
+      "learning_rate": 2.45256027082727e-06,
+      "loss": 0.8668,
+      "step": 2156
+    },
+    {
+      "epoch": 0.1838091180230081,
+      "grad_norm": 17.5,
+      "learning_rate": 2.452514406502401e-06,
+      "loss": 0.8513,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18389433319130805,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.452468520446887e-06,
+      "loss": 0.9288,
+      "step": 2158
+    },
+    {
+      "epoch": 0.183979548359608,
+      "grad_norm": 28.125,
+      "learning_rate": 2.452422612661558e-06,
+      "loss": 1.1164,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18406476352790796,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.452376683147243e-06,
+      "loss": 0.846,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18414997869620792,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4523307319047726e-06,
+      "loss": 0.562,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18423519386450787,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.4522847589349768e-06,
+      "loss": 0.561,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18432040903280783,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4522387642386863e-06,
+      "loss": 0.7294,
+      "step": 2163
+    },
+    {
+      "epoch": 0.1844056242011078,
+      "grad_norm": 22.375,
+      "learning_rate": 2.4521927478167325e-06,
+      "loss": 1.2182,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18449083936940774,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4521467096699464e-06,
+      "loss": 0.5937,
+      "step": 2165
+    },
+    {
+      "epoch": 0.1845760545377077,
+      "grad_norm": 13.625,
+      "learning_rate": 2.452100649799161e-06,
+      "loss": 0.7398,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18466126970600766,
+      "grad_norm": 30.75,
+      "learning_rate": 2.452054568205208e-06,
+      "loss": 0.9536,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18474648487430761,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4520084648889198e-06,
+      "loss": 0.4099,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18483170004260757,
+      "grad_norm": 13.0,
+      "learning_rate": 2.45196233985113e-06,
+      "loss": 0.5028,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18491691521090753,
+      "grad_norm": 14.75,
+      "learning_rate": 2.451916193092672e-06,
+      "loss": 0.9825,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18500213037920749,
+      "grad_norm": 11.625,
+      "learning_rate": 2.45187002461438e-06,
+      "loss": 0.7238,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18508734554750744,
+      "grad_norm": 13.0,
+      "learning_rate": 2.451823834417088e-06,
+      "loss": 0.6578,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18517256071580743,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.451777622501631e-06,
+      "loss": 0.5017,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18525777588410738,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4517313888688437e-06,
+      "loss": 0.5247,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18534299105240734,
+      "grad_norm": 20.5,
+      "learning_rate": 2.451685133519562e-06,
+      "loss": 1.0488,
+      "step": 2175
+    },
+    {
+      "epoch": 0.1854282062207073,
+      "grad_norm": 15.625,
+      "learning_rate": 2.451638856454621e-06,
+      "loss": 0.972,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18551342138900725,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4515925576748578e-06,
+      "loss": 0.6616,
+      "step": 2177
+    },
+    {
+      "epoch": 0.1855986365573072,
+      "grad_norm": 13.25,
+      "learning_rate": 2.451546237181109e-06,
+      "loss": 0.6429,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18568385172560717,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4514998949742115e-06,
+      "loss": 1.1132,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18576906689390713,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4514535310550023e-06,
+      "loss": 1.0648,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18585428206220708,
+      "grad_norm": 13.5,
+      "learning_rate": 2.45140714542432e-06,
+      "loss": 1.0407,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18593949723050704,
+      "grad_norm": 20.5,
+      "learning_rate": 2.451360738083002e-06,
+      "loss": 0.9394,
+      "step": 2182
+    },
+    {
+      "epoch": 0.186024712398807,
+      "grad_norm": 19.875,
+      "learning_rate": 2.4513143090318877e-06,
+      "loss": 1.2023,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18610992756710695,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4512678582718156e-06,
+      "loss": 0.638,
+      "step": 2184
+    },
+    {
+      "epoch": 0.1861951427354069,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4512213858036258e-06,
+      "loss": 0.9932,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18628035790370687,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4511748916281575e-06,
+      "loss": 0.5861,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18636557307200682,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4511283757462507e-06,
+      "loss": 0.57,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18645078824030678,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4510818381587464e-06,
+      "loss": 0.8754,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18653600340860674,
+      "grad_norm": 16.5,
+      "learning_rate": 2.451035278866486e-06,
+      "loss": 0.8999,
+      "step": 2189
+    },
+    {
+      "epoch": 0.1866212185769067,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4509886978703097e-06,
+      "loss": 0.9735,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18670643374520665,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4509420951710605e-06,
+      "loss": 0.6872,
+      "step": 2191
+    },
+    {
+      "epoch": 0.1867916489135066,
+      "grad_norm": 9.625,
+      "learning_rate": 2.45089547076958e-06,
+      "loss": 0.3071,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18687686408180657,
+      "grad_norm": 14.125,
+      "learning_rate": 2.45084882466671e-06,
+      "loss": 0.5016,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18696207925010652,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4508021568632952e-06,
+      "loss": 0.5695,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18704729441840648,
+      "grad_norm": 17.0,
+      "learning_rate": 2.450755467360177e-06,
+      "loss": 0.7118,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18713250958670644,
+      "grad_norm": 12.625,
+      "learning_rate": 2.4507087561582004e-06,
+      "loss": 0.5187,
+      "step": 2196
+    },
+    {
+      "epoch": 0.1872177247550064,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4506620232582096e-06,
+      "loss": 0.855,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18730293992330635,
+      "grad_norm": 21.875,
+      "learning_rate": 2.450615268661048e-06,
+      "loss": 0.8528,
+      "step": 2198
+    },
+    {
+      "epoch": 0.1873881550916063,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4505684923675614e-06,
+      "loss": 1.024,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18747337025990626,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.450521694378595e-06,
+      "loss": 0.6702,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18755858542820622,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4504748746949946e-06,
+      "loss": 0.6394,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18764380059650618,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4504280333176054e-06,
+      "loss": 0.8527,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18772901576480613,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.450381170247275e-06,
+      "loss": 0.6188,
+      "step": 2203
+    },
+    {
+      "epoch": 0.1878142309331061,
+      "grad_norm": 18.125,
+      "learning_rate": 2.45033428548485e-06,
+      "loss": 0.4409,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18789944610140605,
+      "grad_norm": 17.375,
+      "learning_rate": 2.450287379031177e-06,
+      "loss": 1.0834,
+      "step": 2205
+    },
+    {
+      "epoch": 0.187984661269706,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4502404508871042e-06,
+      "loss": 0.7946,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18806987643800596,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4501935010534794e-06,
+      "loss": 0.8715,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18815509160630592,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4501465295311512e-06,
+      "loss": 0.5447,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18824030677460588,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4500995363209684e-06,
+      "loss": 1.1739,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18832552194290583,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4500525214237804e-06,
+      "loss": 0.6727,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1884107371112058,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4500054848404365e-06,
+      "loss": 0.6016,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18849595227950575,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4499584265717866e-06,
+      "loss": 0.8378,
+      "step": 2212
+    },
+    {
+      "epoch": 0.1885811674478057,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4499113466186812e-06,
+      "loss": 0.6287,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18866638261610566,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4498642449819716e-06,
+      "loss": 0.9848,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18875159778440562,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4498171216625085e-06,
+      "loss": 1.0045,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18883681295270557,
+      "grad_norm": 16.25,
+      "learning_rate": 2.449769976661143e-06,
+      "loss": 0.9238,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18892202812100553,
+      "grad_norm": 15.75,
+      "learning_rate": 2.449722809978728e-06,
+      "loss": 1.0048,
+      "step": 2217
+    },
+    {
+      "epoch": 0.1890072432893055,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.449675621616115e-06,
+      "loss": 0.5783,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18909245845760544,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4496284115741575e-06,
+      "loss": 0.3754,
+      "step": 2219
+    },
+    {
+      "epoch": 0.1891776736259054,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4495811798537082e-06,
+      "loss": 0.916,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18926288879420536,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4495339264556208e-06,
+      "loss": 0.7974,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18934810396250532,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.449486651380749e-06,
+      "loss": 0.761,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18943331913080527,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4494393546299473e-06,
+      "loss": 1.0153,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18951853429910523,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4493920362040703e-06,
+      "loss": 0.6706,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18960374946740519,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.449344696103973e-06,
+      "loss": 0.7151,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18968896463570514,
+      "grad_norm": 12.75,
+      "learning_rate": 2.449297334330511e-06,
+      "loss": 0.4276,
+      "step": 2226
+    },
+    {
+      "epoch": 0.1897741798040051,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4492499508845403e-06,
+      "loss": 0.7543,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18985939497230506,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.449202545766917e-06,
+      "loss": 0.6175,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18994461014060504,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4491551189784978e-06,
+      "loss": 0.7408,
+      "step": 2229
+    },
+    {
+      "epoch": 0.190029825308905,
+      "grad_norm": 13.875,
+      "learning_rate": 2.44910767052014e-06,
+      "loss": 0.7936,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19011504047720496,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4490602003927006e-06,
+      "loss": 0.5238,
+      "step": 2231
+    },
+    {
+      "epoch": 0.1902002556455049,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4490127085970376e-06,
+      "loss": 0.8359,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19028547081380487,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4489651951340093e-06,
+      "loss": 0.5499,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19037068598210483,
+      "grad_norm": 24.875,
+      "learning_rate": 2.448917660004474e-06,
+      "loss": 1.2219,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19045590115040478,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4488701032092917e-06,
+      "loss": 0.8399,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19054111631870474,
+      "grad_norm": 19.125,
+      "learning_rate": 2.44882252474932e-06,
+      "loss": 0.8813,
+      "step": 2236
+    },
+    {
+      "epoch": 0.1906263314870047,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4487749246254206e-06,
+      "loss": 0.4664,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19071154665530465,
+      "grad_norm": 14.375,
+      "learning_rate": 2.448727302838453e-06,
+      "loss": 0.8084,
+      "step": 2238
+    },
+    {
+      "epoch": 0.1907967618236046,
+      "grad_norm": 15.5,
+      "learning_rate": 2.448679659389277e-06,
+      "loss": 0.3714,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19088197699190457,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4486319942787547e-06,
+      "loss": 0.6104,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19096719216020452,
+      "grad_norm": 21.375,
+      "learning_rate": 2.4485843075077464e-06,
+      "loss": 0.9878,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19105240732850448,
+      "grad_norm": 18.5,
+      "learning_rate": 2.4485365990771145e-06,
+      "loss": 1.0321,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19113762249680444,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.448488868987721e-06,
+      "loss": 0.8145,
+      "step": 2243
+    },
+    {
+      "epoch": 0.1912228376651044,
+      "grad_norm": 18.5,
+      "learning_rate": 2.448441117240429e-06,
+      "loss": 1.2219,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19130805283340435,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4483933438361005e-06,
+      "loss": 0.6343,
+      "step": 2245
+    },
+    {
+      "epoch": 0.1913932680017043,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4483455487755992e-06,
+      "loss": 0.7822,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19147848317000427,
+      "grad_norm": 15.875,
+      "learning_rate": 2.448297732059789e-06,
+      "loss": 0.8574,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19156369833830422,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4482498936895337e-06,
+      "loss": 1.0431,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19164891350660418,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4482020336656978e-06,
+      "loss": 0.7653,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19173412867490414,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4481541519891464e-06,
+      "loss": 0.869,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1918193438432041,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4481062486607447e-06,
+      "loss": 0.7833,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19190455901150405,
+      "grad_norm": 16.875,
+      "learning_rate": 2.448058323681358e-06,
+      "loss": 0.9384,
+      "step": 2252
+    },
+    {
+      "epoch": 0.191989774179804,
+      "grad_norm": 25.0,
+      "learning_rate": 2.4480103770518528e-06,
+      "loss": 0.7465,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19207498934810396,
+      "grad_norm": 16.0,
+      "learning_rate": 2.447962408773095e-06,
+      "loss": 0.449,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19216020451640392,
+      "grad_norm": 9.125,
+      "learning_rate": 2.4479144188459524e-06,
+      "loss": 0.3205,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19224541968470388,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4478664072712916e-06,
+      "loss": 0.8656,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19233063485300383,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4478183740499805e-06,
+      "loss": 0.9468,
+      "step": 2257
+    },
+    {
+      "epoch": 0.1924158500213038,
+      "grad_norm": 12.375,
+      "learning_rate": 2.447770319182886e-06,
+      "loss": 0.5751,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19250106518960375,
+      "grad_norm": 26.125,
+      "learning_rate": 2.4477222426708783e-06,
+      "loss": 1.1603,
+      "step": 2259
+    },
+    {
+      "epoch": 0.1925862803579037,
+      "grad_norm": 11.125,
+      "learning_rate": 2.447674144514825e-06,
+      "loss": 0.4086,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19267149552620366,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4476260247155955e-06,
+      "loss": 0.9462,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19275671069450362,
+      "grad_norm": 23.625,
+      "learning_rate": 2.447577883274059e-06,
+      "loss": 0.9794,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19284192586280358,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4475297201910865e-06,
+      "loss": 0.6012,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19292714103110353,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4474815354675473e-06,
+      "loss": 0.7466,
+      "step": 2264
+    },
+    {
+      "epoch": 0.1930123561994035,
+      "grad_norm": 30.375,
+      "learning_rate": 2.4474333291043127e-06,
+      "loss": 1.0961,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19309757136770345,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4473851011022544e-06,
+      "loss": 0.5913,
+      "step": 2266
+    },
+    {
+      "epoch": 0.1931827865360034,
+      "grad_norm": 32.25,
+      "learning_rate": 2.4473368514622426e-06,
+      "loss": 1.0623,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19326800170430336,
+      "grad_norm": 16.375,
+      "learning_rate": 2.44728858018515e-06,
+      "loss": 0.8422,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19335321687260332,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4472402872718486e-06,
+      "loss": 0.5212,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19343843204090327,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4471919727232113e-06,
+      "loss": 0.7781,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19352364720920323,
+      "grad_norm": 22.625,
+      "learning_rate": 2.447143636540111e-06,
+      "loss": 0.7535,
+      "step": 2271
+    },
+    {
+      "epoch": 0.1936088623775032,
+      "grad_norm": 24.25,
+      "learning_rate": 2.447095278723422e-06,
+      "loss": 1.0257,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19369407754580314,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4470468992740173e-06,
+      "loss": 0.6115,
+      "step": 2273
+    },
+    {
+      "epoch": 0.1937792927141031,
+      "grad_norm": 15.875,
+      "learning_rate": 2.4469984981927714e-06,
+      "loss": 0.7066,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19386450788240306,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.446950075480559e-06,
+      "loss": 0.5466,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19394972305070302,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4469016311382547e-06,
+      "loss": 0.7272,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19403493821900297,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4468531651667347e-06,
+      "loss": 1.1441,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19412015338730293,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4468046775668746e-06,
+      "loss": 0.8277,
+      "step": 2278
+    },
+    {
+      "epoch": 0.1942053685556029,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4467561683395504e-06,
+      "loss": 0.9888,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19429058372390284,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.446707637485639e-06,
+      "loss": 0.6971,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1943757988922028,
+      "grad_norm": 14.625,
+      "learning_rate": 2.446659085006017e-06,
+      "loss": 0.6163,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19446101406050276,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.4466105109015618e-06,
+      "loss": 0.754,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19454622922880271,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4465619151731513e-06,
+      "loss": 0.5766,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19463144439710267,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.446513297821664e-06,
+      "loss": 0.8218,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19471665956540266,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4464646588479784e-06,
+      "loss": 0.8447,
+      "step": 2285
+    },
+    {
+      "epoch": 0.1948018747337026,
+      "grad_norm": 15.5,
+      "learning_rate": 2.446415998252973e-06,
+      "loss": 0.9637,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19488708990200257,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4463673160375274e-06,
+      "loss": 0.4817,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19497230507030253,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4463186122025217e-06,
+      "loss": 0.6415,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19505752023860248,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4462698867488356e-06,
+      "loss": 0.6263,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19514273540690244,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.4462211396773495e-06,
+      "loss": 0.9363,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1952279505752024,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4461723709889448e-06,
+      "loss": 0.7639,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19531316574350235,
+      "grad_norm": 12.0,
+      "learning_rate": 2.446123580684502e-06,
+      "loss": 0.6631,
+      "step": 2292
+    },
+    {
+      "epoch": 0.1953983809118023,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4460747687649035e-06,
+      "loss": 0.9672,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19548359608010227,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4460259352310313e-06,
+      "loss": 0.8038,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19556881124840222,
+      "grad_norm": 13.0,
+      "learning_rate": 2.445977080083768e-06,
+      "loss": 0.4675,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19565402641670218,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.4459282033239957e-06,
+      "loss": 0.3959,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19573924158500214,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4458793049525987e-06,
+      "loss": 0.9084,
+      "step": 2297
+    },
+    {
+      "epoch": 0.1958244567533021,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4458303849704597e-06,
+      "loss": 0.7232,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19590967192160205,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4457814433784633e-06,
+      "loss": 0.722,
+      "step": 2299
+    },
+    {
+      "epoch": 0.195994887089902,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4457324801774934e-06,
+      "loss": 1.0233,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19608010225820197,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4456834953684357e-06,
+      "loss": 0.5372,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19616531742650192,
+      "grad_norm": 13.375,
+      "learning_rate": 2.445634488952175e-06,
+      "loss": 0.7053,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19625053259480188,
+      "grad_norm": 22.875,
+      "learning_rate": 2.4455854609295963e-06,
+      "loss": 1.2031,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19633574776310184,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4455364113015863e-06,
+      "loss": 0.6793,
+      "step": 2304
+    },
+    {
+      "epoch": 0.1964209629314018,
+      "grad_norm": 16.5,
+      "learning_rate": 2.445487340069031e-06,
+      "loss": 1.0749,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19650617809970175,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4454382472328174e-06,
+      "loss": 0.8275,
+      "step": 2306
+    },
+    {
+      "epoch": 0.1965913932680017,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4453891327938324e-06,
+      "loss": 0.8679,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19667660843630166,
+      "grad_norm": 21.625,
+      "learning_rate": 2.445339996752964e-06,
+      "loss": 1.1166,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19676182360460162,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4452908391111e-06,
+      "loss": 0.4847,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19684703877290158,
+      "grad_norm": 20.375,
+      "learning_rate": 2.445241659869128e-06,
+      "loss": 1.0299,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19693225394120153,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4451924590279374e-06,
+      "loss": 0.5539,
+      "step": 2311
+    },
+    {
+      "epoch": 0.1970174691095015,
+      "grad_norm": 20.5,
+      "learning_rate": 2.4451432365884175e-06,
+      "loss": 0.8347,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19710268427780145,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4450939925514577e-06,
+      "loss": 0.7147,
+      "step": 2313
+    },
+    {
+      "epoch": 0.1971878994461014,
+      "grad_norm": 14.25,
+      "learning_rate": 2.445044726917947e-06,
+      "loss": 0.9938,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19727311461440136,
+      "grad_norm": 16.375,
+      "learning_rate": 2.444995439688777e-06,
+      "loss": 0.9131,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19735832978270132,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4449461308648375e-06,
+      "loss": 0.9354,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19744354495100128,
+      "grad_norm": 12.875,
+      "learning_rate": 2.44489680044702e-06,
+      "loss": 0.6274,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19752876011930123,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4448474484362157e-06,
+      "loss": 0.4424,
+      "step": 2318
+    },
+    {
+      "epoch": 0.1976139752876012,
+      "grad_norm": 17.0,
+      "learning_rate": 2.444798074833316e-06,
+      "loss": 0.7284,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19769919045590115,
+      "grad_norm": 20.75,
+      "learning_rate": 2.444748679639214e-06,
+      "loss": 1.13,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1977844056242011,
+      "grad_norm": 15.375,
+      "learning_rate": 2.444699262854802e-06,
+      "loss": 1.0087,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19786962079250106,
+      "grad_norm": 13.5,
+      "learning_rate": 2.444649824480973e-06,
+      "loss": 0.7771,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19795483596080102,
+      "grad_norm": 22.875,
+      "learning_rate": 2.4446003645186204e-06,
+      "loss": 1.2427,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19804005112910097,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4445508829686374e-06,
+      "loss": 0.9601,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19812526629740093,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4445013798319194e-06,
+      "loss": 0.8291,
+      "step": 2325
+    },
+    {
+      "epoch": 0.1982104814657009,
+      "grad_norm": 19.0,
+      "learning_rate": 2.44445185510936e-06,
+      "loss": 1.0551,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19829569663400085,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.444402308801855e-06,
+      "loss": 0.5958,
+      "step": 2327
+    },
+    {
+      "epoch": 0.1983809118023008,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4443527409102986e-06,
+      "loss": 1.0035,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19846612697060076,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4443031514355873e-06,
+      "loss": 0.6038,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19855134213890072,
+      "grad_norm": 15.625,
+      "learning_rate": 2.444253540378617e-06,
+      "loss": 1.0203,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19863655730720067,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4442039077402845e-06,
+      "loss": 1.0133,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19872177247550063,
+      "grad_norm": 17.125,
+      "learning_rate": 2.444154253521486e-06,
+      "loss": 0.9023,
+      "step": 2332
+    },
+    {
+      "epoch": 0.1988069876438006,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4441045777231197e-06,
+      "loss": 1.0655,
+      "step": 2333
+    },
+    {
+      "epoch": 0.19889220281210054,
+      "grad_norm": 21.75,
+      "learning_rate": 2.4440548803460835e-06,
+      "loss": 1.1842,
+      "step": 2334
+    },
+    {
+      "epoch": 0.1989774179804005,
+      "grad_norm": 18.5,
+      "learning_rate": 2.444005161391274e-06,
+      "loss": 1.0103,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19906263314870046,
+      "grad_norm": 21.75,
+      "learning_rate": 2.443955420859591e-06,
+      "loss": 0.8935,
+      "step": 2336
+    },
+    {
+      "epoch": 0.19914784831700041,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4439056587519327e-06,
+      "loss": 1.0824,
+      "step": 2337
+    },
+    {
+      "epoch": 0.19923306348530037,
+      "grad_norm": 14.0,
+      "learning_rate": 2.443855875069199e-06,
+      "loss": 0.8669,
+      "step": 2338
+    },
+    {
+      "epoch": 0.19931827865360033,
+      "grad_norm": 14.375,
+      "learning_rate": 2.443806069812289e-06,
+      "loss": 0.7385,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19940349382190029,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4437562429821026e-06,
+      "loss": 0.7603,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19948870899020024,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4437063945795405e-06,
+      "loss": 0.4115,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19957392415850023,
+      "grad_norm": 21.875,
+      "learning_rate": 2.4436565246055035e-06,
+      "loss": 1.0797,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19965913932680018,
+      "grad_norm": 13.75,
+      "learning_rate": 2.443606633060893e-06,
+      "loss": 0.9298,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19974435449510014,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.44355671994661e-06,
+      "loss": 0.4778,
+      "step": 2344
+    },
+    {
+      "epoch": 0.1998295696634001,
+      "grad_norm": 18.625,
+      "learning_rate": 2.443506785263557e-06,
+      "loss": 0.786,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19991478483170005,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4434568290126364e-06,
+      "loss": 0.7459,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 20.5,
+      "learning_rate": 2.443406851194751e-06,
+      "loss": 0.77,
+      "step": 2347
+    },
+    {
+      "epoch": 0.20008521516829997,
+      "grad_norm": 19.125,
+      "learning_rate": 2.443356851810803e-06,
+      "loss": 0.9424,
+      "step": 2348
+    },
+    {
+      "epoch": 0.20017043033659992,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4433068308616976e-06,
+      "loss": 0.5636,
+      "step": 2349
+    },
+    {
+      "epoch": 0.20025564550489988,
+      "grad_norm": 15.25,
+      "learning_rate": 2.443256788348337e-06,
+      "loss": 0.6556,
+      "step": 2350
+    },
+    {
+      "epoch": 0.20034086067319984,
+      "grad_norm": 16.375,
+      "learning_rate": 2.443206724271627e-06,
+      "loss": 1.2988,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2004260758414998,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.443156638632471e-06,
+      "loss": 0.6152,
+      "step": 2352
+    },
+    {
+      "epoch": 0.20051129100979975,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4431065314317747e-06,
+      "loss": 0.6758,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2005965061780997,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4430564026704444e-06,
+      "loss": 1.0294,
+      "step": 2354
+    },
+    {
+      "epoch": 0.20068172134639967,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4430062523493845e-06,
+      "loss": 0.7881,
+      "step": 2355
+    },
+    {
+      "epoch": 0.20076693651469962,
+      "grad_norm": 12.5,
+      "learning_rate": 2.442956080469502e-06,
+      "loss": 0.5759,
+      "step": 2356
+    },
+    {
+      "epoch": 0.20085215168299958,
+      "grad_norm": 13.875,
+      "learning_rate": 2.442905887031704e-06,
+      "loss": 0.7606,
+      "step": 2357
+    },
+    {
+      "epoch": 0.20093736685129954,
+      "grad_norm": 12.75,
+      "learning_rate": 2.442855672036897e-06,
+      "loss": 0.6191,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2010225820195995,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4428054354859878e-06,
+      "loss": 0.7308,
+      "step": 2359
+    },
+    {
+      "epoch": 0.20110779718789945,
+      "grad_norm": 17.25,
+      "learning_rate": 2.442755177379885e-06,
+      "loss": 0.999,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2011930123561994,
+      "grad_norm": 25.875,
+      "learning_rate": 2.4427048977194975e-06,
+      "loss": 1.0313,
+      "step": 2361
+    },
+    {
+      "epoch": 0.20127822752449936,
+      "grad_norm": 17.125,
+      "learning_rate": 2.442654596505733e-06,
+      "loss": 1.0347,
+      "step": 2362
+    },
+    {
+      "epoch": 0.20136344269279932,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4426042737395e-06,
+      "loss": 0.6473,
+      "step": 2363
+    },
+    {
+      "epoch": 0.20144865786109928,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4425539294217087e-06,
+      "loss": 0.6536,
+      "step": 2364
+    },
+    {
+      "epoch": 0.20153387302939924,
+      "grad_norm": 22.5,
+      "learning_rate": 2.442503563553269e-06,
+      "loss": 1.0198,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2016190881976992,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4424531761350906e-06,
+      "loss": 0.5529,
+      "step": 2366
+    },
+    {
+      "epoch": 0.20170430336599915,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4424027671680845e-06,
+      "loss": 0.4424,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2017895185342991,
+      "grad_norm": 21.625,
+      "learning_rate": 2.442352336653161e-06,
+      "loss": 1.1686,
+      "step": 2368
+    },
+    {
+      "epoch": 0.20187473370259906,
+      "grad_norm": 9.75,
+      "learning_rate": 2.4423018845912315e-06,
+      "loss": 0.3541,
+      "step": 2369
+    },
+    {
+      "epoch": 0.20195994887089902,
+      "grad_norm": 36.0,
+      "learning_rate": 2.4422514109832083e-06,
+      "loss": 0.951,
+      "step": 2370
+    },
+    {
+      "epoch": 0.20204516403919898,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4422009158300035e-06,
+      "loss": 0.4967,
+      "step": 2371
+    },
+    {
+      "epoch": 0.20213037920749893,
+      "grad_norm": 14.25,
+      "learning_rate": 2.442150399132529e-06,
+      "loss": 0.7606,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2022155943757989,
+      "grad_norm": 13.25,
+      "learning_rate": 2.442099860891698e-06,
+      "loss": 0.504,
+      "step": 2373
+    },
+    {
+      "epoch": 0.20230080954409885,
+      "grad_norm": 10.25,
+      "learning_rate": 2.442049301108424e-06,
+      "loss": 0.3663,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2023860247123988,
+      "grad_norm": 9.875,
+      "learning_rate": 2.4419987197836204e-06,
+      "loss": 0.4461,
+      "step": 2375
+    },
+    {
+      "epoch": 0.20247123988069876,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4419481169182006e-06,
+      "loss": 0.7311,
+      "step": 2376
+    },
+    {
+      "epoch": 0.20255645504899872,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4418974925130805e-06,
+      "loss": 1.0781,
+      "step": 2377
+    },
+    {
+      "epoch": 0.20264167021729868,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4418468465691743e-06,
+      "loss": 0.7007,
+      "step": 2378
+    },
+    {
+      "epoch": 0.20272688538559863,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4417961790873963e-06,
+      "loss": 0.8812,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2028121005538986,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4417454900686634e-06,
+      "loss": 0.5282,
+      "step": 2380
+    },
+    {
+      "epoch": 0.20289731572219855,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.441694779513891e-06,
+      "loss": 0.6174,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2029825308904985,
+      "grad_norm": 30.0,
+      "learning_rate": 2.4416440474239953e-06,
+      "loss": 1.0134,
+      "step": 2382
+    },
+    {
+      "epoch": 0.20306774605879846,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4415932937998936e-06,
+      "loss": 0.6321,
+      "step": 2383
+    },
+    {
+      "epoch": 0.20315296122709842,
+      "grad_norm": 10.625,
+      "learning_rate": 2.441542518642503e-06,
+      "loss": 0.4032,
+      "step": 2384
+    },
+    {
+      "epoch": 0.20323817639539837,
+      "grad_norm": 15.25,
+      "learning_rate": 2.441491721952741e-06,
+      "loss": 0.8512,
+      "step": 2385
+    },
+    {
+      "epoch": 0.20332339156369833,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4414409037315255e-06,
+      "loss": 0.8377,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2034086067319983,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4413900639797745e-06,
+      "loss": 0.7815,
+      "step": 2387
+    },
+    {
+      "epoch": 0.20349382190029824,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4413392026984066e-06,
+      "loss": 0.7139,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2035790370685982,
+      "grad_norm": 14.75,
+      "learning_rate": 2.441288319888342e-06,
+      "loss": 0.5836,
+      "step": 2389
+    },
+    {
+      "epoch": 0.20366425223689816,
+      "grad_norm": 10.0,
+      "learning_rate": 2.4412374155504993e-06,
+      "loss": 0.3388,
+      "step": 2390
+    },
+    {
+      "epoch": 0.20374946740519811,
+      "grad_norm": 19.875,
+      "learning_rate": 2.4411864896857986e-06,
+      "loss": 1.0113,
+      "step": 2391
+    },
+    {
+      "epoch": 0.20383468257349807,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4411355422951607e-06,
+      "loss": 0.9502,
+      "step": 2392
+    },
+    {
+      "epoch": 0.20391989774179803,
+      "grad_norm": 34.5,
+      "learning_rate": 2.4410845733795047e-06,
+      "loss": 0.6113,
+      "step": 2393
+    },
+    {
+      "epoch": 0.20400511291009799,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4410335829397535e-06,
+      "loss": 0.8717,
+      "step": 2394
+    },
+    {
+      "epoch": 0.20409032807839794,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4409825709768277e-06,
+      "loss": 0.9236,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2041755432466979,
+      "grad_norm": 23.75,
+      "learning_rate": 2.440931537491649e-06,
+      "loss": 0.8897,
+      "step": 2396
+    },
+    {
+      "epoch": 0.20426075841499786,
+      "grad_norm": 11.875,
+      "learning_rate": 2.44088048248514e-06,
+      "loss": 0.5197,
+      "step": 2397
+    },
+    {
+      "epoch": 0.20434597358329784,
+      "grad_norm": 17.25,
+      "learning_rate": 2.440829405958223e-06,
+      "loss": 1.1689,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2044311887515978,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4407783079118214e-06,
+      "loss": 1.0093,
+      "step": 2399
+    },
+    {
+      "epoch": 0.20451640391989775,
+      "grad_norm": 12.75,
+      "learning_rate": 2.440727188346858e-06,
+      "loss": 0.4181,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2046016190881977,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4406760472642576e-06,
+      "loss": 0.6745,
+      "step": 2401
+    },
+    {
+      "epoch": 0.20468683425649767,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.440624884664943e-06,
+      "loss": 0.7777,
+      "step": 2402
+    },
+    {
+      "epoch": 0.20477204942479763,
+      "grad_norm": 12.875,
+      "learning_rate": 2.44057370054984e-06,
+      "loss": 0.6037,
+      "step": 2403
+    },
+    {
+      "epoch": 0.20485726459309758,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4405224949198725e-06,
+      "loss": 0.7889,
+      "step": 2404
+    },
+    {
+      "epoch": 0.20494247976139754,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4404712677759667e-06,
+      "loss": 0.6122,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2050276949296975,
+      "grad_norm": 21.5,
+      "learning_rate": 2.440420019119048e-06,
+      "loss": 0.9379,
+      "step": 2406
+    },
+    {
+      "epoch": 0.20511291009799745,
+      "grad_norm": 19.875,
+      "learning_rate": 2.440368748950042e-06,
+      "loss": 1.0805,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2051981252662974,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4403174572698764e-06,
+      "loss": 0.7804,
+      "step": 2408
+    },
+    {
+      "epoch": 0.20528334043459737,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.440266144079477e-06,
+      "loss": 0.8601,
+      "step": 2409
+    },
+    {
+      "epoch": 0.20536855560289732,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4402148093797717e-06,
+      "loss": 0.5067,
+      "step": 2410
+    },
+    {
+      "epoch": 0.20545377077119728,
+      "grad_norm": 19.125,
+      "learning_rate": 2.440163453171688e-06,
+      "loss": 1.0207,
+      "step": 2411
+    },
+    {
+      "epoch": 0.20553898593949724,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4401120754561535e-06,
+      "loss": 0.9645,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2056242011077972,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4400606762340973e-06,
+      "loss": 0.3906,
+      "step": 2413
+    },
+    {
+      "epoch": 0.20570941627609715,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4400092555064475e-06,
+      "loss": 0.7098,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2057946314443971,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4399578132741343e-06,
+      "loss": 0.6405,
+      "step": 2415
+    },
+    {
+      "epoch": 0.20587984661269706,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.439906349538087e-06,
+      "loss": 0.6483,
+      "step": 2416
+    },
+    {
+      "epoch": 0.20596506178099702,
+      "grad_norm": 16.625,
+      "learning_rate": 2.439854864299235e-06,
+      "loss": 0.6714,
+      "step": 2417
+    },
+    {
+      "epoch": 0.20605027694929698,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4398033575585095e-06,
+      "loss": 0.7971,
+      "step": 2418
+    },
+    {
+      "epoch": 0.20613549211759694,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4397518293168403e-06,
+      "loss": 0.5174,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2062207072858969,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4397002795751595e-06,
+      "loss": 0.9742,
+      "step": 2420
+    },
+    {
+      "epoch": 0.20630592245419685,
+      "grad_norm": 20.375,
+      "learning_rate": 2.4396487083343984e-06,
+      "loss": 0.95,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2063911376224968,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.439597115595489e-06,
+      "loss": 0.4189,
+      "step": 2422
+    },
+    {
+      "epoch": 0.20647635279079676,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.4395455013593634e-06,
+      "loss": 0.4058,
+      "step": 2423
+    },
+    {
+      "epoch": 0.20656156795909672,
+      "grad_norm": 10.25,
+      "learning_rate": 2.439493865626954e-06,
+      "loss": 0.3387,
+      "step": 2424
+    },
+    {
+      "epoch": 0.20664678312739668,
+      "grad_norm": 37.25,
+      "learning_rate": 2.439442208399195e-06,
+      "loss": 0.9265,
+      "step": 2425
+    },
+    {
+      "epoch": 0.20673199829569663,
+      "grad_norm": 23.625,
+      "learning_rate": 2.4393905296770183e-06,
+      "loss": 0.9737,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2068172134639966,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4393388294613596e-06,
+      "loss": 0.8829,
+      "step": 2427
+    },
+    {
+      "epoch": 0.20690242863229655,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4392871077531516e-06,
+      "loss": 0.6078,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2069876438005965,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.43923536455333e-06,
+      "loss": 0.4589,
+      "step": 2429
+    },
+    {
+      "epoch": 0.20707285896889646,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4391835998628295e-06,
+      "loss": 1.1341,
+      "step": 2430
+    },
+    {
+      "epoch": 0.20715807413719642,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4391318136825855e-06,
+      "loss": 0.7221,
+      "step": 2431
+    },
+    {
+      "epoch": 0.20724328930549638,
+      "grad_norm": 17.25,
+      "learning_rate": 2.439080006013534e-06,
+      "loss": 0.7102,
+      "step": 2432
+    },
+    {
+      "epoch": 0.20732850447379633,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4390281768566108e-06,
+      "loss": 0.7828,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2074137196420963,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4389763262127532e-06,
+      "loss": 0.7476,
+      "step": 2434
+    },
+    {
+      "epoch": 0.20749893481039625,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.438924454082897e-06,
+      "loss": 0.5536,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2075841499786962,
+      "grad_norm": 14.125,
+      "learning_rate": 2.4388725604679807e-06,
+      "loss": 0.7197,
+      "step": 2436
+    },
+    {
+      "epoch": 0.20766936514699616,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4388206453689422e-06,
+      "loss": 0.7193,
+      "step": 2437
+    },
+    {
+      "epoch": 0.20775458031529612,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.438768708786719e-06,
+      "loss": 0.9624,
+      "step": 2438
+    },
+    {
+      "epoch": 0.20783979548359607,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.4387167507222494e-06,
+      "loss": 0.419,
+      "step": 2439
+    },
+    {
+      "epoch": 0.20792501065189603,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4386647711764732e-06,
+      "loss": 0.9562,
+      "step": 2440
+    },
+    {
+      "epoch": 0.208010225820196,
+      "grad_norm": 15.375,
+      "learning_rate": 2.438612770150329e-06,
+      "loss": 0.829,
+      "step": 2441
+    },
+    {
+      "epoch": 0.20809544098849594,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.438560747644757e-06,
+      "loss": 0.5211,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2081806561567959,
+      "grad_norm": 19.375,
+      "learning_rate": 2.438508703660697e-06,
+      "loss": 1.0678,
+      "step": 2443
+    },
+    {
+      "epoch": 0.20826587132509586,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4384566381990897e-06,
+      "loss": 0.4248,
+      "step": 2444
+    },
+    {
+      "epoch": 0.20835108649339582,
+      "grad_norm": 17.75,
+      "learning_rate": 2.438404551260876e-06,
+      "loss": 0.7505,
+      "step": 2445
+    },
+    {
+      "epoch": 0.20843630166169577,
+      "grad_norm": 17.5,
+      "learning_rate": 2.438352442846997e-06,
+      "loss": 1.0375,
+      "step": 2446
+    },
+    {
+      "epoch": 0.20852151682999573,
+      "grad_norm": 16.5,
+      "learning_rate": 2.438300312958394e-06,
+      "loss": 0.8947,
+      "step": 2447
+    },
+    {
+      "epoch": 0.20860673199829569,
+      "grad_norm": 11.0,
+      "learning_rate": 2.4382481615960096e-06,
+      "loss": 0.498,
+      "step": 2448
+    },
+    {
+      "epoch": 0.20869194716659564,
+      "grad_norm": 18.125,
+      "learning_rate": 2.438195988760786e-06,
+      "loss": 0.7602,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2087771623348956,
+      "grad_norm": 12.5,
+      "learning_rate": 2.438143794453666e-06,
+      "loss": 0.5734,
+      "step": 2450
+    },
+    {
+      "epoch": 0.20886237750319556,
+      "grad_norm": 17.75,
+      "learning_rate": 2.438091578675593e-06,
+      "loss": 1.0439,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2089475926714955,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4380393414275107e-06,
+      "loss": 1.0062,
+      "step": 2452
+    },
+    {
+      "epoch": 0.20903280783979547,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.4379870827103626e-06,
+      "loss": 0.8711,
+      "step": 2453
+    },
+    {
+      "epoch": 0.20911802300809543,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4379348025250933e-06,
+      "loss": 0.5635,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2092032381763954,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4378825008726473e-06,
+      "loss": 0.8082,
+      "step": 2455
+    },
+    {
+      "epoch": 0.20928845334469537,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.43783017775397e-06,
+      "loss": 0.7393,
+      "step": 2456
+    },
+    {
+      "epoch": 0.20937366851299533,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4377778331700077e-06,
+      "loss": 0.7298,
+      "step": 2457
+    },
+    {
+      "epoch": 0.20945888368129528,
+      "grad_norm": 24.375,
+      "learning_rate": 2.437725467121705e-06,
+      "loss": 1.121,
+      "step": 2458
+    },
+    {
+      "epoch": 0.20954409884959524,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4376730796100084e-06,
+      "loss": 0.6564,
+      "step": 2459
+    },
+    {
+      "epoch": 0.2096293140178952,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.437620670635865e-06,
+      "loss": 0.6601,
+      "step": 2460
+    },
+    {
+      "epoch": 0.20971452918619515,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4375682402002227e-06,
+      "loss": 0.8472,
+      "step": 2461
+    },
+    {
+      "epoch": 0.2097997443544951,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4375157883040273e-06,
+      "loss": 0.5964,
+      "step": 2462
+    },
+    {
+      "epoch": 0.20988495952279507,
+      "grad_norm": 20.625,
+      "learning_rate": 2.437463314948228e-06,
+      "loss": 1.0367,
+      "step": 2463
+    },
+    {
+      "epoch": 0.20997017469109502,
+      "grad_norm": 20.25,
+      "learning_rate": 2.437410820133772e-06,
+      "loss": 1.0741,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21005538985939498,
+      "grad_norm": 21.75,
+      "learning_rate": 2.437358303861609e-06,
+      "loss": 1.0369,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21014060502769494,
+      "grad_norm": 17.875,
+      "learning_rate": 2.437305766132687e-06,
+      "loss": 0.7807,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2102258201959949,
+      "grad_norm": 12.875,
+      "learning_rate": 2.437253206947956e-06,
+      "loss": 0.7198,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21031103536429485,
+      "grad_norm": 17.75,
+      "learning_rate": 2.437200626308366e-06,
+      "loss": 0.6326,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2103962505325948,
+      "grad_norm": 11.5,
+      "learning_rate": 2.437148024214866e-06,
+      "loss": 0.5773,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21048146570089477,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4370954006684083e-06,
+      "loss": 0.8714,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21056668086919472,
+      "grad_norm": 16.375,
+      "learning_rate": 2.437042755669943e-06,
+      "loss": 0.79,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21065189603749468,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4369900892204205e-06,
+      "loss": 0.7128,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21073711120579464,
+      "grad_norm": 25.75,
+      "learning_rate": 2.4369374013207946e-06,
+      "loss": 0.9695,
+      "step": 2473
+    },
+    {
+      "epoch": 0.2108223263740946,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4368846919720156e-06,
+      "loss": 0.6582,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21090754154239455,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4368319611750367e-06,
+      "loss": 0.5139,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2109927567106945,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4367792089308114e-06,
+      "loss": 0.8152,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21107797187899446,
+      "grad_norm": 16.0,
+      "learning_rate": 2.436726435240292e-06,
+      "loss": 0.6635,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21116318704729442,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.436673640104433e-06,
+      "loss": 0.5012,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21124840221559438,
+      "grad_norm": 18.375,
+      "learning_rate": 2.436620823524187e-06,
+      "loss": 0.9702,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21133361738389433,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4365679855005104e-06,
+      "loss": 0.5865,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2114188325521943,
+      "grad_norm": 14.125,
+      "learning_rate": 2.436515126034357e-06,
+      "loss": 0.8313,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21150404772049425,
+      "grad_norm": 14.75,
+      "learning_rate": 2.436462245126682e-06,
+      "loss": 0.4818,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2115892628887942,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.436409342778441e-06,
+      "loss": 0.8752,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21167447805709416,
+      "grad_norm": 11.25,
+      "learning_rate": 2.4363564189905904e-06,
+      "loss": 0.3124,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21175969322539412,
+      "grad_norm": 17.75,
+      "learning_rate": 2.436303473764086e-06,
+      "loss": 0.8358,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21184490839369408,
+      "grad_norm": 9.75,
+      "learning_rate": 2.4362505070998847e-06,
+      "loss": 0.2835,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21193012356199403,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4361975189989447e-06,
+      "loss": 0.9199,
+      "step": 2487
+    },
+    {
+      "epoch": 0.212015338730294,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4361445094622223e-06,
+      "loss": 0.7091,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21210055389859395,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4360914784906757e-06,
+      "loss": 0.8237,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2121857690668939,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.436038426085264e-06,
+      "loss": 0.9451,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21227098423519386,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4359853522469447e-06,
+      "loss": 0.7429,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21235619940349382,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4359322569766776e-06,
+      "loss": 0.9901,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21244141457179377,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4358791402754216e-06,
+      "loss": 0.5445,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21252662974009373,
+      "grad_norm": 11.75,
+      "learning_rate": 2.4358260021441376e-06,
+      "loss": 0.662,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2126118449083937,
+      "grad_norm": 13.75,
+      "learning_rate": 2.4357728425837853e-06,
+      "loss": 0.4285,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21269706007669364,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4357196615953256e-06,
+      "loss": 0.8422,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2127822752449936,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4356664591797186e-06,
+      "loss": 0.5245,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21286749041329356,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.435613235337927e-06,
+      "loss": 1.0852,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21295270558159352,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4355599900709114e-06,
+      "loss": 0.3149,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21303792074989347,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4355067233796352e-06,
+      "loss": 1.0901,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21312313591819343,
+      "grad_norm": 14.5,
+      "learning_rate": 2.43545343526506e-06,
+      "loss": 0.8986,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2132083510864934,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4354001257281497e-06,
+      "loss": 0.8688,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21329356625479334,
+      "grad_norm": 19.375,
+      "learning_rate": 2.4353467947698665e-06,
+      "loss": 1.086,
+      "step": 2503
+    },
+    {
+      "epoch": 0.2133787814230933,
+      "grad_norm": 24.375,
+      "learning_rate": 2.435293442391175e-06,
+      "loss": 1.2636,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21346399659139326,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4352400685930388e-06,
+      "loss": 0.5842,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2135492117596932,
+      "grad_norm": 11.125,
+      "learning_rate": 2.435186673376423e-06,
+      "loss": 0.5385,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21363442692799317,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4351332567422924e-06,
+      "loss": 0.6024,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21371964209629313,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4350798186916118e-06,
+      "loss": 0.6682,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21380485726459308,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4350263592253476e-06,
+      "loss": 1.0078,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21389007243289304,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4349728783444646e-06,
+      "loss": 0.6571,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21397528760119303,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.434919376049931e-06,
+      "loss": 0.5937,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21406050276949298,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4348658523427125e-06,
+      "loss": 0.7072,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21414571793779294,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4348123072237766e-06,
+      "loss": 0.8007,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2142309331060929,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4347587406940908e-06,
+      "loss": 0.8709,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21431614827439285,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4347051527546227e-06,
+      "loss": 0.9103,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2144013634426928,
+      "grad_norm": 11.625,
+      "learning_rate": 2.434651543406342e-06,
+      "loss": 0.6602,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21448657861099277,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4345979126502164e-06,
+      "loss": 0.7888,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21457179377929272,
+      "grad_norm": 13.125,
+      "learning_rate": 2.434544260487215e-06,
+      "loss": 0.6301,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21465700894759268,
+      "grad_norm": 17.25,
+      "learning_rate": 2.434490586918308e-06,
+      "loss": 1.1467,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21474222411589264,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4344368919444646e-06,
+      "loss": 0.903,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2148274392841926,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4343831755666557e-06,
+      "loss": 0.7378,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21491265445249255,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.434329437785852e-06,
+      "loss": 0.7339,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2149978696207925,
+      "grad_norm": 20.875,
+      "learning_rate": 2.4342756786030246e-06,
+      "loss": 1.2273,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21508308478909247,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.434221898019144e-06,
+      "loss": 0.4503,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21516829995739242,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.434168096035184e-06,
+      "loss": 0.5745,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21525351512569238,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.434114272652115e-06,
+      "loss": 1.0181,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21533873029399234,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.43406042787091e-06,
+      "loss": 0.8084,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2154239454622923,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4340065616925433e-06,
+      "loss": 0.8192,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21550916063059225,
+      "grad_norm": 12.875,
+      "learning_rate": 2.433952674117987e-06,
+      "loss": 0.7902,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2155943757988922,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4338987651482153e-06,
+      "loss": 0.904,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21567959096719216,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4338448347842027e-06,
+      "loss": 0.8997,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21576480613549212,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.433790883026923e-06,
+      "loss": 0.7253,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21585002130379208,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.433736909877352e-06,
+      "loss": 0.9365,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21593523647209203,
+      "grad_norm": 16.75,
+      "learning_rate": 2.433682915336465e-06,
+      "loss": 0.9074,
+      "step": 2534
+    },
+    {
+      "epoch": 0.216020451640392,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4336288994052364e-06,
+      "loss": 0.359,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21610566680869195,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4335748620846444e-06,
+      "loss": 0.6362,
+      "step": 2536
+    },
+    {
+      "epoch": 0.2161908819769919,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.433520803375664e-06,
+      "loss": 0.418,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21627609714529186,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4334667232792724e-06,
+      "loss": 0.7132,
+      "step": 2538
+    },
+    {
+      "epoch": 0.21636131231359182,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4334126217964475e-06,
+      "loss": 0.5964,
+      "step": 2539
+    },
+    {
+      "epoch": 0.21644652748189178,
+      "grad_norm": 26.75,
+      "learning_rate": 2.4333584989281663e-06,
+      "loss": 0.6838,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21653174265019173,
+      "grad_norm": 18.875,
+      "learning_rate": 2.433304354675407e-06,
+      "loss": 0.9192,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2166169578184917,
+      "grad_norm": 19.375,
+      "learning_rate": 2.433250189039148e-06,
+      "loss": 0.9965,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21670217298679165,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4331960020203686e-06,
+      "loss": 0.4586,
+      "step": 2543
+    },
+    {
+      "epoch": 0.2167873881550916,
+      "grad_norm": 21.625,
+      "learning_rate": 2.4331417936200475e-06,
+      "loss": 0.6214,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21687260332339156,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4330875638391644e-06,
+      "loss": 0.5138,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21695781849169152,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.433033312678699e-06,
+      "loss": 0.6062,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21704303365999147,
+      "grad_norm": 28.125,
+      "learning_rate": 2.4329790401396326e-06,
+      "loss": 0.8695,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21712824882829143,
+      "grad_norm": 38.75,
+      "learning_rate": 2.432924746222945e-06,
+      "loss": 1.2791,
+      "step": 2548
+    },
+    {
+      "epoch": 0.2172134639965914,
+      "grad_norm": 16.25,
+      "learning_rate": 2.432870430929618e-06,
+      "loss": 0.8814,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21729867916489135,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4328160942606324e-06,
+      "loss": 0.7691,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2173838943331913,
+      "grad_norm": 28.0,
+      "learning_rate": 2.432761736216971e-06,
+      "loss": 0.7908,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21746910950149126,
+      "grad_norm": 21.375,
+      "learning_rate": 2.4327073567996152e-06,
+      "loss": 0.9392,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21755432466979122,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.4326529560095486e-06,
+      "loss": 0.6047,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21763953983809117,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4325985338477537e-06,
+      "loss": 0.4832,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21772475500639113,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.432544090315214e-06,
+      "loss": 0.4691,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2178099701746911,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4324896254129137e-06,
+      "loss": 0.727,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21789518534299104,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4324351391418364e-06,
+      "loss": 0.7773,
+      "step": 2557
+    },
+    {
+      "epoch": 0.217980400511291,
+      "grad_norm": 18.5,
+      "learning_rate": 2.4323806315029673e-06,
+      "loss": 1.0957,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21806561567959096,
+      "grad_norm": 19.125,
+      "learning_rate": 2.432326102497291e-06,
+      "loss": 1.3076,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21815083084789091,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4322715521257934e-06,
+      "loss": 0.6208,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21823604601619087,
+      "grad_norm": 16.0,
+      "learning_rate": 2.43221698038946e-06,
+      "loss": 0.8164,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21832126118449083,
+      "grad_norm": 11.125,
+      "learning_rate": 2.432162387289277e-06,
+      "loss": 0.4064,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21840647635279078,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.432107772826231e-06,
+      "loss": 0.3819,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21849169152109074,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4320531370013084e-06,
+      "loss": 0.904,
+      "step": 2564
+    },
+    {
+      "epoch": 0.2185769066893907,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4319984798154973e-06,
+      "loss": 0.8427,
+      "step": 2565
+    },
+    {
+      "epoch": 0.21866212185769066,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.431943801269785e-06,
+      "loss": 0.6886,
+      "step": 2566
+    },
+    {
+      "epoch": 0.2187473370259906,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4318891013651594e-06,
+      "loss": 0.8962,
+      "step": 2567
+    },
+    {
+      "epoch": 0.2188325521942906,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.4318343801026095e-06,
+      "loss": 0.3265,
+      "step": 2568
+    },
+    {
+      "epoch": 0.21891776736259055,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4317796374831238e-06,
+      "loss": 0.957,
+      "step": 2569
+    },
+    {
+      "epoch": 0.2190029825308905,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.431724873507692e-06,
+      "loss": 0.8619,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21908819769919047,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.431670088177303e-06,
+      "loss": 0.507,
+      "step": 2571
+    },
+    {
+      "epoch": 0.21917341286749042,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4316152814929478e-06,
+      "loss": 0.7112,
+      "step": 2572
+    },
+    {
+      "epoch": 0.21925862803579038,
+      "grad_norm": 13.125,
+      "learning_rate": 2.431560453455616e-06,
+      "loss": 0.4879,
+      "step": 2573
+    },
+    {
+      "epoch": 0.21934384320409034,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.431505604066298e-06,
+      "loss": 0.8353,
+      "step": 2574
+    },
+    {
+      "epoch": 0.2194290583723903,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4314507333259864e-06,
+      "loss": 0.9581,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21951427354069025,
+      "grad_norm": 17.5,
+      "learning_rate": 2.431395841235672e-06,
+      "loss": 0.9309,
+      "step": 2576
+    },
+    {
+      "epoch": 0.2195994887089902,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4313409277963463e-06,
+      "loss": 0.7396,
+      "step": 2577
+    },
+    {
+      "epoch": 0.21968470387729017,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4312859930090025e-06,
+      "loss": 0.8397,
+      "step": 2578
+    },
+    {
+      "epoch": 0.21976991904559012,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4312310368746327e-06,
+      "loss": 0.9028,
+      "step": 2579
+    },
+    {
+      "epoch": 0.21985513421389008,
+      "grad_norm": 13.625,
+      "learning_rate": 2.431176059394231e-06,
+      "loss": 0.8091,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21994034938219004,
+      "grad_norm": 18.0,
+      "learning_rate": 2.431121060568789e-06,
+      "loss": 0.96,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22002556455049,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4310660403993023e-06,
+      "loss": 0.7458,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22011077971878995,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4310109988867647e-06,
+      "loss": 0.5909,
+      "step": 2583
+    },
+    {
+      "epoch": 0.2201959948870899,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4309559360321707e-06,
+      "loss": 0.7709,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22028121005538986,
+      "grad_norm": 17.0,
+      "learning_rate": 2.430900851836515e-06,
+      "loss": 0.7266,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22036642522368982,
+      "grad_norm": 14.125,
+      "learning_rate": 2.430845746300794e-06,
+      "loss": 0.3861,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22045164039198978,
+      "grad_norm": 17.375,
+      "learning_rate": 2.430790619426002e-06,
+      "loss": 1.3531,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22053685556028974,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.430735471213137e-06,
+      "loss": 0.7736,
+      "step": 2588
+    },
+    {
+      "epoch": 0.2206220707285897,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4306803016631945e-06,
+      "loss": 0.5715,
+      "step": 2589
+    },
+    {
+      "epoch": 0.22070728589688965,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4306251107771717e-06,
+      "loss": 0.9396,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2207925010651896,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.430569898556066e-06,
+      "loss": 0.821,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22087771623348956,
+      "grad_norm": 16.25,
+      "learning_rate": 2.430514665000875e-06,
+      "loss": 0.9442,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22096293140178952,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4304594101125967e-06,
+      "loss": 1.0966,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22104814657008948,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4304041338922303e-06,
+      "loss": 0.6438,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22113336173838943,
+      "grad_norm": 21.25,
+      "learning_rate": 2.4303488363407736e-06,
+      "loss": 1.0892,
+      "step": 2595
+    },
+    {
+      "epoch": 0.2212185769066894,
+      "grad_norm": 22.5,
+      "learning_rate": 2.430293517459227e-06,
+      "loss": 0.9584,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22130379207498935,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4302381772485893e-06,
+      "loss": 1.055,
+      "step": 2597
+    },
+    {
+      "epoch": 0.2213890072432893,
+      "grad_norm": 15.25,
+      "learning_rate": 2.430182815709861e-06,
+      "loss": 1.1028,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22147422241158926,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4301274328440427e-06,
+      "loss": 0.7913,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22155943757988922,
+      "grad_norm": 14.25,
+      "learning_rate": 2.430072028652135e-06,
+      "loss": 0.5136,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22164465274818917,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4300166031351384e-06,
+      "loss": 0.7976,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22172986791648913,
+      "grad_norm": 15.0,
+      "learning_rate": 2.429961156294056e-06,
+      "loss": 0.7175,
+      "step": 2602
+    },
+    {
+      "epoch": 0.2218150830847891,
+      "grad_norm": 23.125,
+      "learning_rate": 2.429905688129889e-06,
+      "loss": 1.0837,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22190029825308905,
+      "grad_norm": 23.25,
+      "learning_rate": 2.429850198643639e-06,
+      "loss": 0.8739,
+      "step": 2604
+    },
+    {
+      "epoch": 0.221985513421389,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.42979468783631e-06,
+      "loss": 0.6373,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22207072858968896,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4297391557089044e-06,
+      "loss": 0.7459,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22215594375798892,
+      "grad_norm": 16.125,
+      "learning_rate": 2.429683602262426e-06,
+      "loss": 1.026,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22224115892628887,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.429628027497879e-06,
+      "loss": 0.6947,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22232637409458883,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4295724314162668e-06,
+      "loss": 0.637,
+      "step": 2609
+    },
+    {
+      "epoch": 0.2224115892628888,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4295168140185945e-06,
+      "loss": 1.2364,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22249680443118874,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.429461175305868e-06,
+      "loss": 0.5258,
+      "step": 2611
+    },
+    {
+      "epoch": 0.2225820195994887,
+      "grad_norm": 17.75,
+      "learning_rate": 2.429405515279091e-06,
+      "loss": 0.9123,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22266723476778866,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4293498339392715e-06,
+      "loss": 0.5855,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22275244993608861,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4292941312874137e-06,
+      "loss": 0.913,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22283766510438857,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4292384073245255e-06,
+      "loss": 0.5384,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22292288027268853,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4291826620516135e-06,
+      "loss": 0.9442,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22300809544098849,
+      "grad_norm": 19.75,
+      "learning_rate": 2.429126895469685e-06,
+      "loss": 0.6352,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22309331060928844,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4290711075797475e-06,
+      "loss": 0.6309,
+      "step": 2618
+    },
+    {
+      "epoch": 0.2231785257775884,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4290152983828095e-06,
+      "loss": 0.4835,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22326374094588836,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4289594678798796e-06,
+      "loss": 0.6368,
+      "step": 2620
+    },
+    {
+      "epoch": 0.2233489561141883,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4289036160719666e-06,
+      "loss": 0.9495,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22343417128248827,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4288477429600798e-06,
+      "loss": 0.4465,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22351938645078823,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.4287918485452288e-06,
+      "loss": 0.4655,
+      "step": 2623
+    },
+    {
+      "epoch": 0.2236046016190882,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4287359328284237e-06,
+      "loss": 0.9579,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22368981678738817,
+      "grad_norm": 19.0,
+      "learning_rate": 2.428679995810675e-06,
+      "loss": 0.717,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22377503195568813,
+      "grad_norm": 23.125,
+      "learning_rate": 2.4286240374929937e-06,
+      "loss": 1.1671,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22386024712398808,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4285680578763907e-06,
+      "loss": 0.914,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22394546229228804,
+      "grad_norm": 25.25,
+      "learning_rate": 2.4285120569618777e-06,
+      "loss": 1.0151,
+      "step": 2628
+    },
+    {
+      "epoch": 0.224030677460588,
+      "grad_norm": 64.5,
+      "learning_rate": 2.4284560347504666e-06,
+      "loss": 0.6493,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22411589262888795,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4283999912431704e-06,
+      "loss": 0.8151,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2242011077971879,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.428343926441001e-06,
+      "loss": 0.7265,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22428632296548787,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4282878403449724e-06,
+      "loss": 0.7539,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22437153813378782,
+      "grad_norm": 15.75,
+      "learning_rate": 2.428231732956097e-06,
+      "loss": 0.8051,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22445675330208778,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4281756042753894e-06,
+      "loss": 1.016,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22454196847038774,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4281194543038643e-06,
+      "loss": 0.4299,
+      "step": 2635
+    },
+    {
+      "epoch": 0.2246271836386877,
+      "grad_norm": 14.25,
+      "learning_rate": 2.428063283042536e-06,
+      "loss": 0.762,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22471239880698765,
+      "grad_norm": 17.125,
+      "learning_rate": 2.428007090492419e-06,
+      "loss": 0.9848,
+      "step": 2637
+    },
+    {
+      "epoch": 0.2247976139752876,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4279508766545303e-06,
+      "loss": 0.5421,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22488282914358756,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.427894641529884e-06,
+      "loss": 0.6195,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22496804431188752,
+      "grad_norm": 12.75,
+      "learning_rate": 2.427838385119497e-06,
+      "loss": 0.6019,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22505325948018748,
+      "grad_norm": 13.0,
+      "learning_rate": 2.4277821074243862e-06,
+      "loss": 0.4765,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22513847464848744,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4277258084455682e-06,
+      "loss": 0.8688,
+      "step": 2642
+    },
+    {
+      "epoch": 0.2252236898167874,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4276694881840606e-06,
+      "loss": 0.6365,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22530890498508735,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4276131466408815e-06,
+      "loss": 1.0313,
+      "step": 2644
+    },
+    {
+      "epoch": 0.2253941201533873,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4275567838170483e-06,
+      "loss": 0.6653,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22547933532168726,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.42750039971358e-06,
+      "loss": 0.5904,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22556455048998722,
+      "grad_norm": 14.0,
+      "learning_rate": 2.427443994331495e-06,
+      "loss": 0.6693,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22564976565828718,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4273875676718133e-06,
+      "loss": 0.5469,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22573498082658713,
+      "grad_norm": 16.375,
+      "learning_rate": 2.427331119735554e-06,
+      "loss": 1.003,
+      "step": 2649
+    },
+    {
+      "epoch": 0.2258201959948871,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.427274650523738e-06,
+      "loss": 0.4572,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22590541116318705,
+      "grad_norm": 19.625,
+      "learning_rate": 2.427218160037385e-06,
+      "loss": 0.9576,
+      "step": 2651
+    },
+    {
+      "epoch": 0.225990626331487,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4271616482775164e-06,
+      "loss": 0.6773,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22607584149978696,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4271051152451526e-06,
+      "loss": 0.5406,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22616105666808692,
+      "grad_norm": 14.25,
+      "learning_rate": 2.427048560941316e-06,
+      "loss": 0.622,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22624627183638688,
+      "grad_norm": 17.875,
+      "learning_rate": 2.426991985367028e-06,
+      "loss": 1.1171,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22633148700468683,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4269353885233114e-06,
+      "loss": 1.0306,
+      "step": 2656
+    },
+    {
+      "epoch": 0.2264167021729868,
+      "grad_norm": 10.375,
+      "learning_rate": 2.426878770411189e-06,
+      "loss": 0.4994,
+      "step": 2657
+    },
+    {
+      "epoch": 0.22650191734128675,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.426822131031684e-06,
+      "loss": 0.4089,
+      "step": 2658
+    },
+    {
+      "epoch": 0.2265871325095867,
+      "grad_norm": 16.625,
+      "learning_rate": 2.426765470385819e-06,
+      "loss": 0.9012,
+      "step": 2659
+    },
+    {
+      "epoch": 0.22667234767788666,
+      "grad_norm": 21.375,
+      "learning_rate": 2.426708788474619e-06,
+      "loss": 1.65,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22675756284618662,
+      "grad_norm": 19.5,
+      "learning_rate": 2.426652085299108e-06,
+      "loss": 1.0165,
+      "step": 2661
+    },
+    {
+      "epoch": 0.22684277801448657,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4265953608603105e-06,
+      "loss": 0.6429,
+      "step": 2662
+    },
+    {
+      "epoch": 0.22692799318278653,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.426538615159252e-06,
+      "loss": 0.5751,
+      "step": 2663
+    },
+    {
+      "epoch": 0.2270132083510865,
+      "grad_norm": 16.0,
+      "learning_rate": 2.426481848196957e-06,
+      "loss": 1.1199,
+      "step": 2664
+    },
+    {
+      "epoch": 0.22709842351938644,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4264250599744522e-06,
+      "loss": 0.5531,
+      "step": 2665
+    },
+    {
+      "epoch": 0.2271836386876864,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4263682504927635e-06,
+      "loss": 0.8152,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22726885385598636,
+      "grad_norm": 43.75,
+      "learning_rate": 2.426311419752918e-06,
+      "loss": 1.0729,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22735406902428631,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4262545677559418e-06,
+      "loss": 0.9786,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22743928419258627,
+      "grad_norm": 14.875,
+      "learning_rate": 2.426197694502863e-06,
+      "loss": 0.7922,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22752449936088623,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.426140799994709e-06,
+      "loss": 0.4565,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22760971452918619,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.426083884232508e-06,
+      "loss": 0.7747,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22769492969748614,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4260269472172887e-06,
+      "loss": 0.5734,
+      "step": 2672
+    },
+    {
+      "epoch": 0.2277801448657861,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4259699889500797e-06,
+      "loss": 0.9756,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22786536003408606,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4259130094319106e-06,
+      "loss": 0.6701,
+      "step": 2674
+    },
+    {
+      "epoch": 0.227950575202386,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4258560086638113e-06,
+      "loss": 0.7146,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22803579037068597,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4257989866468105e-06,
+      "loss": 1.1415,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22812100553898593,
+      "grad_norm": 16.625,
+      "learning_rate": 2.425741943381941e-06,
+      "loss": 0.8519,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22820622070728588,
+      "grad_norm": 11.375,
+      "learning_rate": 2.4256848788702314e-06,
+      "loss": 0.7111,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22829143587558584,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.425627793112714e-06,
+      "loss": 0.7178,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22837665104388583,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.42557068611042e-06,
+      "loss": 0.8471,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22846186621218578,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4255135578643817e-06,
+      "loss": 0.7265,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22854708138048574,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4254564083756318e-06,
+      "loss": 0.5264,
+      "step": 2682
+    },
+    {
+      "epoch": 0.2286322965487857,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4253992376452017e-06,
+      "loss": 0.9509,
+      "step": 2683
+    },
+    {
+      "epoch": 0.22871751171708565,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4253420456741263e-06,
+      "loss": 0.9853,
+      "step": 2684
+    },
+    {
+      "epoch": 0.2288027268853856,
+      "grad_norm": 23.375,
+      "learning_rate": 2.4252848324634375e-06,
+      "loss": 1.2609,
+      "step": 2685
+    },
+    {
+      "epoch": 0.22888794205368557,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4252275980141705e-06,
+      "loss": 0.7184,
+      "step": 2686
+    },
+    {
+      "epoch": 0.22897315722198552,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4251703423273586e-06,
+      "loss": 0.8876,
+      "step": 2687
+    },
+    {
+      "epoch": 0.22905837239028548,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4251130654040374e-06,
+      "loss": 0.7087,
+      "step": 2688
+    },
+    {
+      "epoch": 0.22914358755858544,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4250557672452406e-06,
+      "loss": 0.8459,
+      "step": 2689
+    },
+    {
+      "epoch": 0.2292288027268854,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.424998447852005e-06,
+      "loss": 0.9062,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22931401789518535,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.424941107225366e-06,
+      "loss": 0.6738,
+      "step": 2691
+    },
+    {
+      "epoch": 0.2293992330634853,
+      "grad_norm": 16.125,
+      "learning_rate": 2.42488374536636e-06,
+      "loss": 0.8291,
+      "step": 2692
+    },
+    {
+      "epoch": 0.22948444823178527,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.424826362276023e-06,
+      "loss": 0.5306,
+      "step": 2693
+    },
+    {
+      "epoch": 0.22956966340008522,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4247689579553918e-06,
+      "loss": 0.8042,
+      "step": 2694
+    },
+    {
+      "epoch": 0.22965487856838518,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4247115324055045e-06,
+      "loss": 1.0936,
+      "step": 2695
+    },
+    {
+      "epoch": 0.22974009373668514,
+      "grad_norm": 11.5,
+      "learning_rate": 2.424654085627398e-06,
+      "loss": 0.5308,
+      "step": 2696
+    },
+    {
+      "epoch": 0.2298253089049851,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4245966176221118e-06,
+      "loss": 0.4608,
+      "step": 2697
+    },
+    {
+      "epoch": 0.22991052407328505,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4245391283906837e-06,
+      "loss": 0.6074,
+      "step": 2698
+    },
+    {
+      "epoch": 0.229995739241585,
+      "grad_norm": 15.875,
+      "learning_rate": 2.424481617934152e-06,
+      "loss": 0.9357,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23008095440988496,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4244240862535563e-06,
+      "loss": 0.9879,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23016616957818492,
+      "grad_norm": 19.125,
+      "learning_rate": 2.4243665333499365e-06,
+      "loss": 1.1863,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23025138474648488,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.424308959224333e-06,
+      "loss": 0.8961,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23033659991478483,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4242513638777854e-06,
+      "loss": 0.5673,
+      "step": 2703
+    },
+    {
+      "epoch": 0.2304218150830848,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.424193747311335e-06,
+      "loss": 0.4202,
+      "step": 2704
+    },
+    {
+      "epoch": 0.23050703025138475,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4241361095260225e-06,
+      "loss": 0.6607,
+      "step": 2705
+    },
+    {
+      "epoch": 0.2305922454196847,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4240784505228897e-06,
+      "loss": 0.6322,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23067746058798466,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4240207703029788e-06,
+      "loss": 0.6797,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23076267575628462,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.4239630688673324e-06,
+      "loss": 0.4156,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23084789092458458,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4239053462169924e-06,
+      "loss": 1.1465,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23093310609288453,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4238476023530028e-06,
+      "loss": 0.4007,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2310183212611845,
+      "grad_norm": 10.625,
+      "learning_rate": 2.423789837276406e-06,
+      "loss": 0.3558,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23110353642948445,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.423732050988247e-06,
+      "loss": 0.5028,
+      "step": 2712
+    },
+    {
+      "epoch": 0.2311887515977844,
+      "grad_norm": 13.25,
+      "learning_rate": 2.42367424348957e-06,
+      "loss": 0.7019,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23127396676608436,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.4236164147814184e-06,
+      "loss": 0.47,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23135918193438432,
+      "grad_norm": 20.0,
+      "learning_rate": 2.423558564864838e-06,
+      "loss": 0.9292,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23144439710268427,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4235006937408743e-06,
+      "loss": 1.0831,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23152961227098423,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.4234428014105735e-06,
+      "loss": 0.7353,
+      "step": 2717
+    },
+    {
+      "epoch": 0.2316148274392842,
+      "grad_norm": 21.25,
+      "learning_rate": 2.423384887874981e-06,
+      "loss": 1.2695,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23170004260758414,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4233269531351437e-06,
+      "loss": 0.7943,
+      "step": 2719
+    },
+    {
+      "epoch": 0.2317852577758841,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.423268997192108e-06,
+      "loss": 0.6412,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23187047294418406,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4232110200469223e-06,
+      "loss": 0.6674,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23195568811248402,
+      "grad_norm": 16.0,
+      "learning_rate": 2.423153021700634e-06,
+      "loss": 0.7757,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23204090328078397,
+      "grad_norm": 13.875,
+      "learning_rate": 2.42309500215429e-06,
+      "loss": 1.1354,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23212611844908393,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4230369614089404e-06,
+      "loss": 0.9389,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23221133361738389,
+      "grad_norm": 15.375,
+      "learning_rate": 2.422978899465633e-06,
+      "loss": 0.7042,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23229654878568384,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4229208163254176e-06,
+      "loss": 0.5903,
+      "step": 2726
+    },
+    {
+      "epoch": 0.2323817639539838,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4228627119893434e-06,
+      "loss": 0.7102,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23246697912228376,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.42280458645846e-06,
+      "loss": 0.9055,
+      "step": 2728
+    },
+    {
+      "epoch": 0.2325521942905837,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4227464397338195e-06,
+      "loss": 0.8032,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23263740945888367,
+      "grad_norm": 22.125,
+      "learning_rate": 2.4226882718164708e-06,
+      "loss": 1.1022,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23272262462718363,
+      "grad_norm": 19.0,
+      "learning_rate": 2.422630082707466e-06,
+      "loss": 0.8255,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23280783979548358,
+      "grad_norm": 21.0,
+      "learning_rate": 2.4225718724078565e-06,
+      "loss": 0.7442,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23289305496378354,
+      "grad_norm": 18.375,
+      "learning_rate": 2.422513640918694e-06,
+      "loss": 0.688,
+      "step": 2733
+    },
+    {
+      "epoch": 0.2329782701320835,
+      "grad_norm": 24.5,
+      "learning_rate": 2.4224553882410313e-06,
+      "loss": 0.693,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23306348530038346,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.42239711437592e-06,
+      "loss": 1.0697,
+      "step": 2735
+    },
+    {
+      "epoch": 0.2331487004686834,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4223388193244144e-06,
+      "loss": 0.9238,
+      "step": 2736
+    },
+    {
+      "epoch": 0.2332339156369834,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4222805030875677e-06,
+      "loss": 0.892,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23331913080528335,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4222221656664334e-06,
+      "loss": 0.6879,
+      "step": 2738
+    },
+    {
+      "epoch": 0.2334043459735833,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4221638070620655e-06,
+      "loss": 0.6441,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23348956114188327,
+      "grad_norm": 16.375,
+      "learning_rate": 2.422105427275519e-06,
+      "loss": 0.6626,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23357477631018322,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.422047026307849e-06,
+      "loss": 0.4695,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23365999147848318,
+      "grad_norm": 11.0,
+      "learning_rate": 2.4219886041601104e-06,
+      "loss": 0.3842,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23374520664678314,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.4219301608333595e-06,
+      "loss": 0.4441,
+      "step": 2743
+    },
+    {
+      "epoch": 0.2338304218150831,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4218716963286516e-06,
+      "loss": 0.5696,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23391563698338305,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4218132106470443e-06,
+      "loss": 0.6914,
+      "step": 2745
+    },
+    {
+      "epoch": 0.234000852151683,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4217547037895935e-06,
+      "loss": 0.5383,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23408606731998297,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.421696175757357e-06,
+      "loss": 0.6258,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23417128248828292,
+      "grad_norm": 22.5,
+      "learning_rate": 2.4216376265513928e-06,
+      "loss": 1.0195,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23425649765658288,
+      "grad_norm": 17.0,
+      "learning_rate": 2.421579056172758e-06,
+      "loss": 0.9471,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23434171282488284,
+      "grad_norm": 35.75,
+      "learning_rate": 2.4215204646225118e-06,
+      "loss": 0.9019,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2344269279931828,
+      "grad_norm": 13.625,
+      "learning_rate": 2.4214618519017125e-06,
+      "loss": 0.5979,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23451214316148275,
+      "grad_norm": 20.5,
+      "learning_rate": 2.42140321801142e-06,
+      "loss": 1.1552,
+      "step": 2752
+    },
+    {
+      "epoch": 0.2345973583297827,
+      "grad_norm": 14.0,
+      "learning_rate": 2.421344562952693e-06,
+      "loss": 0.5761,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23468257349808266,
+      "grad_norm": 13.5,
+      "learning_rate": 2.421285886726592e-06,
+      "loss": 0.7395,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23476778866638262,
+      "grad_norm": 19.125,
+      "learning_rate": 2.421227189334177e-06,
+      "loss": 0.9516,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23485300383468258,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4211684707765094e-06,
+      "loss": 1.0189,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23493821900298253,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4211097310546495e-06,
+      "loss": 1.0226,
+      "step": 2757
+    },
+    {
+      "epoch": 0.2350234341712825,
+      "grad_norm": 26.625,
+      "learning_rate": 2.4210509701696592e-06,
+      "loss": 0.9777,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23510864933958245,
+      "grad_norm": 29.125,
+      "learning_rate": 2.4209921881226e-06,
+      "loss": 1.006,
+      "step": 2759
+    },
+    {
+      "epoch": 0.2351938645078824,
+      "grad_norm": 14.125,
+      "learning_rate": 2.420933384914535e-06,
+      "loss": 0.833,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23527907967618236,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.420874560546526e-06,
+      "loss": 0.9195,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23536429484448232,
+      "grad_norm": 13.375,
+      "learning_rate": 2.420815715019636e-06,
+      "loss": 0.84,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23544951001278228,
+      "grad_norm": 18.0,
+      "learning_rate": 2.420756848334929e-06,
+      "loss": 0.5436,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23553472518108223,
+      "grad_norm": 10.625,
+      "learning_rate": 2.420697960493469e-06,
+      "loss": 0.5073,
+      "step": 2764
+    },
+    {
+      "epoch": 0.2356199403493822,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4206390514963186e-06,
+      "loss": 0.5162,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23570515551768215,
+      "grad_norm": 20.75,
+      "learning_rate": 2.420580121344544e-06,
+      "loss": 0.9209,
+      "step": 2766
+    },
+    {
+      "epoch": 0.2357903706859821,
+      "grad_norm": 12.25,
+      "learning_rate": 2.420521170039209e-06,
+      "loss": 0.5864,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23587558585428206,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.42046219758138e-06,
+      "loss": 0.9031,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23596080102258202,
+      "grad_norm": 13.0,
+      "learning_rate": 2.420403203972122e-06,
+      "loss": 0.6023,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23604601619088197,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4203441892125007e-06,
+      "loss": 0.7756,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23613123135918193,
+      "grad_norm": 16.375,
+      "learning_rate": 2.420285153303583e-06,
+      "loss": 0.83,
+      "step": 2771
+    },
+    {
+      "epoch": 0.2362164465274819,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4202260962464365e-06,
+      "loss": 0.6134,
+      "step": 2772
+    },
+    {
+      "epoch": 0.23630166169578184,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4201670180421273e-06,
+      "loss": 0.5596,
+      "step": 2773
+    },
+    {
+      "epoch": 0.2363868768640818,
+      "grad_norm": 16.0,
+      "learning_rate": 2.420107918691723e-06,
+      "loss": 1.1389,
+      "step": 2774
+    },
+    {
+      "epoch": 0.23647209203238176,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4200487981962926e-06,
+      "loss": 1.1048,
+      "step": 2775
+    },
+    {
+      "epoch": 0.23655730720068172,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4199896565569037e-06,
+      "loss": 0.7548,
+      "step": 2776
+    },
+    {
+      "epoch": 0.23664252236898167,
+      "grad_norm": 14.625,
+      "learning_rate": 2.419930493774625e-06,
+      "loss": 0.6183,
+      "step": 2777
+    },
+    {
+      "epoch": 0.23672773753728163,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.419871309850526e-06,
+      "loss": 1.0604,
+      "step": 2778
+    },
+    {
+      "epoch": 0.2368129527055816,
+      "grad_norm": 19.625,
+      "learning_rate": 2.419812104785676e-06,
+      "loss": 1.0181,
+      "step": 2779
+    },
+    {
+      "epoch": 0.23689816787388154,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4197528785811447e-06,
+      "loss": 0.9889,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2369833830421815,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4196936312380025e-06,
+      "loss": 0.946,
+      "step": 2781
+    },
+    {
+      "epoch": 0.23706859821048146,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4196343627573203e-06,
+      "loss": 0.7725,
+      "step": 2782
+    },
+    {
+      "epoch": 0.23715381337878141,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4195750731401698e-06,
+      "loss": 0.9169,
+      "step": 2783
+    },
+    {
+      "epoch": 0.23723902854708137,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4195157623876206e-06,
+      "loss": 0.891,
+      "step": 2784
+    },
+    {
+      "epoch": 0.23732424371538133,
+      "grad_norm": 17.0,
+      "learning_rate": 2.419456430500746e-06,
+      "loss": 1.0242,
+      "step": 2785
+    },
+    {
+      "epoch": 0.23740945888368128,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4193970774806175e-06,
+      "loss": 0.7015,
+      "step": 2786
+    },
+    {
+      "epoch": 0.23749467405198124,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4193377033283082e-06,
+      "loss": 0.8451,
+      "step": 2787
+    },
+    {
+      "epoch": 0.2375798892202812,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4192783080448906e-06,
+      "loss": 1.0183,
+      "step": 2788
+    },
+    {
+      "epoch": 0.23766510438858116,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.419218891631438e-06,
+      "loss": 0.5921,
+      "step": 2789
+    },
+    {
+      "epoch": 0.2377503195568811,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4191594540890246e-06,
+      "loss": 0.7494,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23783553472518107,
+      "grad_norm": 21.625,
+      "learning_rate": 2.419099995418724e-06,
+      "loss": 0.9608,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23792074989348103,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4190405156216105e-06,
+      "loss": 0.5553,
+      "step": 2792
+    },
+    {
+      "epoch": 0.238005965061781,
+      "grad_norm": 24.0,
+      "learning_rate": 2.41898101469876e-06,
+      "loss": 0.9755,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23809118023008097,
+      "grad_norm": 17.75,
+      "learning_rate": 2.4189214926512467e-06,
+      "loss": 0.8864,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23817639539838092,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4188619494801463e-06,
+      "loss": 0.8475,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23826161056668088,
+      "grad_norm": 19.5,
+      "learning_rate": 2.4188023851865354e-06,
+      "loss": 0.9995,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23834682573498084,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.41874279977149e-06,
+      "loss": 0.612,
+      "step": 2797
+    },
+    {
+      "epoch": 0.2384320409032808,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4186831932360873e-06,
+      "loss": 0.7524,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23851725607158075,
+      "grad_norm": 19.75,
+      "learning_rate": 2.4186235655814037e-06,
+      "loss": 1.0977,
+      "step": 2799
+    },
+    {
+      "epoch": 0.2386024712398807,
+      "grad_norm": 23.25,
+      "learning_rate": 2.4185639168085175e-06,
+      "loss": 1.1972,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23868768640818067,
+      "grad_norm": 18.625,
+      "learning_rate": 2.418504246918506e-06,
+      "loss": 0.735,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23877290157648062,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4184445559124476e-06,
+      "loss": 1.0113,
+      "step": 2802
+    },
+    {
+      "epoch": 0.23885811674478058,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4183848437914216e-06,
+      "loss": 0.6332,
+      "step": 2803
+    },
+    {
+      "epoch": 0.23894333191308054,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4183251105565058e-06,
+      "loss": 1.0582,
+      "step": 2804
+    },
+    {
+      "epoch": 0.2390285470813805,
+      "grad_norm": 12.75,
+      "learning_rate": 2.418265356208781e-06,
+      "loss": 0.8828,
+      "step": 2805
+    },
+    {
+      "epoch": 0.23911376224968045,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4182055807493263e-06,
+      "loss": 0.4269,
+      "step": 2806
+    },
+    {
+      "epoch": 0.2391989774179804,
+      "grad_norm": 20.875,
+      "learning_rate": 2.418145784179222e-06,
+      "loss": 1.118,
+      "step": 2807
+    },
+    {
+      "epoch": 0.23928419258628036,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4180859664995486e-06,
+      "loss": 0.8421,
+      "step": 2808
+    },
+    {
+      "epoch": 0.23936940775458032,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4180261277113874e-06,
+      "loss": 0.633,
+      "step": 2809
+    },
+    {
+      "epoch": 0.23945462292288028,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4179662678158195e-06,
+      "loss": 0.7742,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23953983809118023,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4179063868139265e-06,
+      "loss": 1.0349,
+      "step": 2811
+    },
+    {
+      "epoch": 0.2396250532594802,
+      "grad_norm": 11.0,
+      "learning_rate": 2.417846484706791e-06,
+      "loss": 0.435,
+      "step": 2812
+    },
+    {
+      "epoch": 0.23971026842778015,
+      "grad_norm": 27.5,
+      "learning_rate": 2.417786561495495e-06,
+      "loss": 1.2232,
+      "step": 2813
+    },
+    {
+      "epoch": 0.2397954835960801,
+      "grad_norm": 30.875,
+      "learning_rate": 2.417726617181121e-06,
+      "loss": 0.952,
+      "step": 2814
+    },
+    {
+      "epoch": 0.23988069876438006,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4176666517647534e-06,
+      "loss": 0.7361,
+      "step": 2815
+    },
+    {
+      "epoch": 0.23996591393268002,
+      "grad_norm": 14.375,
+      "learning_rate": 2.417606665247475e-06,
+      "loss": 0.7363,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24005112910097998,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4175466576303703e-06,
+      "loss": 0.7376,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24013634426927993,
+      "grad_norm": 18.875,
+      "learning_rate": 2.417486628914523e-06,
+      "loss": 1.02,
+      "step": 2818
+    },
+    {
+      "epoch": 0.2402215594375799,
+      "grad_norm": 21.625,
+      "learning_rate": 2.4174265791010187e-06,
+      "loss": 0.9254,
+      "step": 2819
+    },
+    {
+      "epoch": 0.24030677460587985,
+      "grad_norm": 19.5,
+      "learning_rate": 2.417366508190942e-06,
+      "loss": 0.6651,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2403919897741798,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4173064161853788e-06,
+      "loss": 0.7512,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24047720494247976,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4172463030854146e-06,
+      "loss": 0.5555,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24056242011077972,
+      "grad_norm": 13.75,
+      "learning_rate": 2.417186168892136e-06,
+      "loss": 0.715,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24064763527907967,
+      "grad_norm": 20.25,
+      "learning_rate": 2.41712601360663e-06,
+      "loss": 1.1555,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24073285044737963,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4170658372299827e-06,
+      "loss": 0.3854,
+      "step": 2825
+    },
+    {
+      "epoch": 0.2408180656156796,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.4170056397632823e-06,
+      "loss": 0.484,
+      "step": 2826
+    },
+    {
+      "epoch": 0.24090328078397955,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4169454212076167e-06,
+      "loss": 1.2465,
+      "step": 2827
+    },
+    {
+      "epoch": 0.2409884959522795,
+      "grad_norm": 14.5,
+      "learning_rate": 2.416885181564074e-06,
+      "loss": 0.839,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24107371112057946,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4168249208337422e-06,
+      "loss": 1.0157,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24115892628887942,
+      "grad_norm": 21.875,
+      "learning_rate": 2.4167646390177115e-06,
+      "loss": 1.0009,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24124414145717937,
+      "grad_norm": 23.125,
+      "learning_rate": 2.41670433611707e-06,
+      "loss": 0.9316,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24132935662547933,
+      "grad_norm": 12.25,
+      "learning_rate": 2.416644012132908e-06,
+      "loss": 0.5364,
+      "step": 2832
+    },
+    {
+      "epoch": 0.2414145717937793,
+      "grad_norm": 13.0,
+      "learning_rate": 2.416583667066315e-06,
+      "loss": 0.6709,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24149978696207924,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4165233009183827e-06,
+      "loss": 0.9483,
+      "step": 2834
+    },
+    {
+      "epoch": 0.2415850021303792,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.416462913690201e-06,
+      "loss": 0.9666,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24167021729867916,
+      "grad_norm": 12.0,
+      "learning_rate": 2.416402505382862e-06,
+      "loss": 0.5579,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24175543246697911,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.416342075997456e-06,
+      "loss": 0.734,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24184064763527907,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4162816255350764e-06,
+      "loss": 0.6405,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24192586280357903,
+      "grad_norm": 16.125,
+      "learning_rate": 2.416221153996815e-06,
+      "loss": 0.6583,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24201107797187899,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4161606613837645e-06,
+      "loss": 0.8384,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24209629314017894,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4161001476970186e-06,
+      "loss": 0.7441,
+      "step": 2841
+    },
+    {
+      "epoch": 0.2421815083084789,
+      "grad_norm": 18.25,
+      "learning_rate": 2.41603961293767e-06,
+      "loss": 1.0302,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24226672347677886,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.415979057106813e-06,
+      "loss": 0.6088,
+      "step": 2843
+    },
+    {
+      "epoch": 0.2423519386450788,
+      "grad_norm": 18.625,
+      "learning_rate": 2.415918480205542e-06,
+      "loss": 0.98,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24243715381337877,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4158578822349516e-06,
+      "loss": 1.1366,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24252236898167873,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4157972631961374e-06,
+      "loss": 0.7558,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24260758414997868,
+      "grad_norm": 26.875,
+      "learning_rate": 2.415736623090194e-06,
+      "loss": 0.5989,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24269279931827864,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4156759619182175e-06,
+      "loss": 0.7427,
+      "step": 2848
+    },
+    {
+      "epoch": 0.2427780144865786,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.4156152796813044e-06,
+      "loss": 0.5771,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24286322965487858,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.415554576380551e-06,
+      "loss": 0.6352,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24294844482317854,
+      "grad_norm": 11.0,
+      "learning_rate": 2.415493852017054e-06,
+      "loss": 0.5072,
+      "step": 2851
+    },
+    {
+      "epoch": 0.2430336599914785,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.415433106591911e-06,
+      "loss": 0.7335,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24311887515977845,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4153723401062207e-06,
+      "loss": 0.4162,
+      "step": 2853
+    },
+    {
+      "epoch": 0.2432040903280784,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4153115525610797e-06,
+      "loss": 0.6565,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24328930549637837,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4152507439575874e-06,
+      "loss": 0.5961,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24337452066467832,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.415189914296842e-06,
+      "loss": 0.6748,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24345973583297828,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4151290635799436e-06,
+      "loss": 0.6063,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24354495100127824,
+      "grad_norm": 15.375,
+      "learning_rate": 2.4150681918079914e-06,
+      "loss": 0.9039,
+      "step": 2858
+    },
+    {
+      "epoch": 0.2436301661695782,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4150072989820853e-06,
+      "loss": 0.815,
+      "step": 2859
+    },
+    {
+      "epoch": 0.24371538133787815,
+      "grad_norm": 79.0,
+      "learning_rate": 2.4149463851033257e-06,
+      "loss": 0.7938,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2438005965061781,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4148854501728137e-06,
+      "loss": 0.8861,
+      "step": 2861
+    },
+    {
+      "epoch": 0.24388581167447806,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.41482449419165e-06,
+      "loss": 0.7299,
+      "step": 2862
+    },
+    {
+      "epoch": 0.24397102684277802,
+      "grad_norm": 14.875,
+      "learning_rate": 2.414763517160936e-06,
+      "loss": 0.8017,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24405624201107798,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4147025190817745e-06,
+      "loss": 0.4447,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24414145717937794,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4146414999552677e-06,
+      "loss": 0.7857,
+      "step": 2865
+    },
+    {
+      "epoch": 0.2442266723476779,
+      "grad_norm": 18.625,
+      "learning_rate": 2.414580459782517e-06,
+      "loss": 0.5316,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24431188751597785,
+      "grad_norm": 45.0,
+      "learning_rate": 2.414519398564627e-06,
+      "loss": 0.7077,
+      "step": 2867
+    },
+    {
+      "epoch": 0.2443971026842778,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4144583163027e-06,
+      "loss": 0.5152,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24448231785257776,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4143972129978406e-06,
+      "loss": 0.6902,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24456753302087772,
+      "grad_norm": 13.25,
+      "learning_rate": 2.4143360886511523e-06,
+      "loss": 0.8894,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24465274818917768,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4142749432637404e-06,
+      "loss": 1.2033,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24473796335747763,
+      "grad_norm": 19.75,
+      "learning_rate": 2.414213776836709e-06,
+      "loss": 0.9076,
+      "step": 2872
+    },
+    {
+      "epoch": 0.2448231785257776,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4141525893711643e-06,
+      "loss": 0.72,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24490839369407755,
+      "grad_norm": 34.75,
+      "learning_rate": 2.4140913808682117e-06,
+      "loss": 1.1777,
+      "step": 2874
+    },
+    {
+      "epoch": 0.2449936088623775,
+      "grad_norm": 12.125,
+      "learning_rate": 2.414030151328957e-06,
+      "loss": 0.623,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24507882403067746,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4139689007545072e-06,
+      "loss": 0.9947,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24516403919897742,
+      "grad_norm": 19.125,
+      "learning_rate": 2.413907629145969e-06,
+      "loss": 1.0784,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24524925436727738,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.4138463365044492e-06,
+      "loss": 0.9475,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24533446953557733,
+      "grad_norm": 16.5,
+      "learning_rate": 2.413785022831056e-06,
+      "loss": 0.9571,
+      "step": 2879
+    },
+    {
+      "epoch": 0.2454196847038773,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.413723688126897e-06,
+      "loss": 0.6369,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24550489987217725,
+      "grad_norm": 13.25,
+      "learning_rate": 2.413662332393081e-06,
+      "loss": 0.6627,
+      "step": 2881
+    },
+    {
+      "epoch": 0.2455901150404772,
+      "grad_norm": 12.875,
+      "learning_rate": 2.4136009556307162e-06,
+      "loss": 0.6959,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24567533020877716,
+      "grad_norm": 13.5,
+      "learning_rate": 2.413539557840912e-06,
+      "loss": 0.6722,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24576054537707712,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4134781390247785e-06,
+      "loss": 0.5039,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24584576054537707,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4134166991834247e-06,
+      "loss": 0.5201,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24593097571367703,
+      "grad_norm": 16.75,
+      "learning_rate": 2.413355238317961e-06,
+      "loss": 0.7207,
+      "step": 2886
+    },
+    {
+      "epoch": 0.246016190881977,
+      "grad_norm": 15.75,
+      "learning_rate": 2.413293756429499e-06,
+      "loss": 1.0481,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24610140605027694,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4132322535191482e-06,
+      "loss": 0.7666,
+      "step": 2888
+    },
+    {
+      "epoch": 0.2461866212185769,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4131707295880213e-06,
+      "loss": 0.8501,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24627183638687686,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4131091846372293e-06,
+      "loss": 0.8717,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24635705155517681,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4130476186678852e-06,
+      "loss": 0.6441,
+      "step": 2891
+    },
+    {
+      "epoch": 0.24644226672347677,
+      "grad_norm": 15.5,
+      "learning_rate": 2.412986031681101e-06,
+      "loss": 0.8361,
+      "step": 2892
+    },
+    {
+      "epoch": 0.24652748189177673,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4129244236779897e-06,
+      "loss": 0.6939,
+      "step": 2893
+    },
+    {
+      "epoch": 0.24661269706007669,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4128627946596648e-06,
+      "loss": 0.8974,
+      "step": 2894
+    },
+    {
+      "epoch": 0.24669791222837664,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4128011446272397e-06,
+      "loss": 0.571,
+      "step": 2895
+    },
+    {
+      "epoch": 0.2467831273966766,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4127394735818283e-06,
+      "loss": 1.073,
+      "step": 2896
+    },
+    {
+      "epoch": 0.24686834256497656,
+      "grad_norm": 18.625,
+      "learning_rate": 2.412677781524546e-06,
+      "loss": 0.8276,
+      "step": 2897
+    },
+    {
+      "epoch": 0.2469535577332765,
+      "grad_norm": 13.875,
+      "learning_rate": 2.4126160684565065e-06,
+      "loss": 0.9196,
+      "step": 2898
+    },
+    {
+      "epoch": 0.24703877290157647,
+      "grad_norm": 21.375,
+      "learning_rate": 2.412554334378826e-06,
+      "loss": 0.8342,
+      "step": 2899
+    },
+    {
+      "epoch": 0.24712398806987643,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4124925792926195e-06,
+      "loss": 0.4633,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24720920323817638,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4124308031990028e-06,
+      "loss": 0.8989,
+      "step": 2901
+    },
+    {
+      "epoch": 0.24729441840647634,
+      "grad_norm": 11.0,
+      "learning_rate": 2.4123690060990927e-06,
+      "loss": 0.3603,
+      "step": 2902
+    },
+    {
+      "epoch": 0.2473796335747763,
+      "grad_norm": 13.0,
+      "learning_rate": 2.412307187994006e-06,
+      "loss": 0.5763,
+      "step": 2903
+    },
+    {
+      "epoch": 0.24746484874307625,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4122453488848595e-06,
+      "loss": 0.6921,
+      "step": 2904
+    },
+    {
+      "epoch": 0.2475500639113762,
+      "grad_norm": 33.25,
+      "learning_rate": 2.412183488772771e-06,
+      "loss": 0.9493,
+      "step": 2905
+    },
+    {
+      "epoch": 0.2476352790796762,
+      "grad_norm": 14.0,
+      "learning_rate": 2.4121216076588577e-06,
+      "loss": 0.7555,
+      "step": 2906
+    },
+    {
+      "epoch": 0.24772049424797615,
+      "grad_norm": 17.5,
+      "learning_rate": 2.412059705544239e-06,
+      "loss": 0.5751,
+      "step": 2907
+    },
+    {
+      "epoch": 0.2478057094162761,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.4119977824300324e-06,
+      "loss": 0.5633,
+      "step": 2908
+    },
+    {
+      "epoch": 0.24789092458457607,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4119358383173573e-06,
+      "loss": 0.7717,
+      "step": 2909
+    },
+    {
+      "epoch": 0.24797613975287602,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.411873873207334e-06,
+      "loss": 0.8224,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24806135492117598,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.41181188710108e-06,
+      "loss": 0.7632,
+      "step": 2911
+    },
+    {
+      "epoch": 0.24814657008947594,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.411749879999718e-06,
+      "loss": 0.6182,
+      "step": 2912
+    },
+    {
+      "epoch": 0.2482317852577759,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.4116878519043676e-06,
+      "loss": 0.529,
+      "step": 2913
+    },
+    {
+      "epoch": 0.24831700042607585,
+      "grad_norm": 14.0,
+      "learning_rate": 2.411625802816149e-06,
+      "loss": 0.6263,
+      "step": 2914
+    },
+    {
+      "epoch": 0.2484022155943758,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4115637327361844e-06,
+      "loss": 0.9153,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24848743076267577,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4115016416655947e-06,
+      "loss": 0.8225,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24857264593097572,
+      "grad_norm": 13.5,
+      "learning_rate": 2.411439529605503e-06,
+      "loss": 0.8688,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24865786109927568,
+      "grad_norm": 12.125,
+      "learning_rate": 2.41137739655703e-06,
+      "loss": 0.5164,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24874307626757564,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4113152425213006e-06,
+      "loss": 0.6873,
+      "step": 2919
+    },
+    {
+      "epoch": 0.2488282914358756,
+      "grad_norm": 12.375,
+      "learning_rate": 2.4112530674994367e-06,
+      "loss": 0.4927,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24891350660417555,
+      "grad_norm": 14.75,
+      "learning_rate": 2.411190871492562e-06,
+      "loss": 0.4649,
+      "step": 2921
+    },
+    {
+      "epoch": 0.2489987217724755,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4111286545018006e-06,
+      "loss": 0.8986,
+      "step": 2922
+    },
+    {
+      "epoch": 0.24908393694077546,
+      "grad_norm": 9.8125,
+      "learning_rate": 2.411066416528277e-06,
+      "loss": 0.321,
+      "step": 2923
+    },
+    {
+      "epoch": 0.24916915210907542,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4110041575731156e-06,
+      "loss": 1.0307,
+      "step": 2924
+    },
+    {
+      "epoch": 0.24925436727737538,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4109418776374417e-06,
+      "loss": 0.7543,
+      "step": 2925
+    },
+    {
+      "epoch": 0.24933958244567533,
+      "grad_norm": 31.0,
+      "learning_rate": 2.4108795767223803e-06,
+      "loss": 1.2791,
+      "step": 2926
+    },
+    {
+      "epoch": 0.2494247976139753,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4108172548290576e-06,
+      "loss": 0.8779,
+      "step": 2927
+    },
+    {
+      "epoch": 0.24951001278227525,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4107549119586e-06,
+      "loss": 1.0068,
+      "step": 2928
+    },
+    {
+      "epoch": 0.2495952279505752,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.410692548112134e-06,
+      "loss": 0.7455,
+      "step": 2929
+    },
+    {
+      "epoch": 0.24968044311887516,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4106301632907865e-06,
+      "loss": 0.8551,
+      "step": 2930
+    },
+    {
+      "epoch": 0.24976565828717512,
+      "grad_norm": 22.125,
+      "learning_rate": 2.410567757495685e-06,
+      "loss": 1.1662,
+      "step": 2931
+    },
+    {
+      "epoch": 0.24985087345547508,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4105053307279567e-06,
+      "loss": 1.0824,
+      "step": 2932
+    },
+    {
+      "epoch": 0.24993608862377503,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.41044288298873e-06,
+      "loss": 0.367,
+      "step": 2933
+    },
+    {
+      "epoch": 0.250021303792075,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.410380414279134e-06,
+      "loss": 0.5967,
+      "step": 2934
+    },
+    {
+      "epoch": 0.25010651896037495,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4103179246002967e-06,
+      "loss": 0.6324,
+      "step": 2935
+    },
+    {
+      "epoch": 0.2501917341286749,
+      "grad_norm": 15.375,
+      "learning_rate": 2.410255413953348e-06,
+      "loss": 0.8187,
+      "step": 2936
+    },
+    {
+      "epoch": 0.25027694929697486,
+      "grad_norm": 21.0,
+      "learning_rate": 2.4101928823394167e-06,
+      "loss": 0.6911,
+      "step": 2937
+    },
+    {
+      "epoch": 0.2503621644652748,
+      "grad_norm": 19.625,
+      "learning_rate": 2.410130329759634e-06,
+      "loss": 1.0849,
+      "step": 2938
+    },
+    {
+      "epoch": 0.2504473796335748,
+      "grad_norm": 21.875,
+      "learning_rate": 2.4100677562151294e-06,
+      "loss": 1.3756,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25053259480187473,
+      "grad_norm": 17.25,
+      "learning_rate": 2.4100051617070334e-06,
+      "loss": 1.3955,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2506178099701747,
+      "grad_norm": 38.0,
+      "learning_rate": 2.4099425462364784e-06,
+      "loss": 1.5249,
+      "step": 2941
+    },
+    {
+      "epoch": 0.25070302513847464,
+      "grad_norm": 33.25,
+      "learning_rate": 2.4098799098045948e-06,
+      "loss": 1.2549,
+      "step": 2942
+    },
+    {
+      "epoch": 0.2507882403067746,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4098172524125153e-06,
+      "loss": 0.566,
+      "step": 2943
+    },
+    {
+      "epoch": 0.25087345547507456,
+      "grad_norm": 13.625,
+      "learning_rate": 2.409754574061371e-06,
+      "loss": 0.7195,
+      "step": 2944
+    },
+    {
+      "epoch": 0.2509586706433745,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4096918747522956e-06,
+      "loss": 0.5083,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25104388581167447,
+      "grad_norm": 11.125,
+      "learning_rate": 2.409629154486422e-06,
+      "loss": 0.4639,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25112910097997443,
+      "grad_norm": 13.75,
+      "learning_rate": 2.409566413264884e-06,
+      "loss": 0.6159,
+      "step": 2947
+    },
+    {
+      "epoch": 0.2512143161482744,
+      "grad_norm": 18.125,
+      "learning_rate": 2.409503651088814e-06,
+      "loss": 0.7635,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25129953131657434,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4094408679593477e-06,
+      "loss": 0.8745,
+      "step": 2949
+    },
+    {
+      "epoch": 0.2513847464848743,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4093780638776188e-06,
+      "loss": 1.0396,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25146996165317426,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.409315238844762e-06,
+      "loss": 0.5989,
+      "step": 2951
+    },
+    {
+      "epoch": 0.2515551768214742,
+      "grad_norm": 14.75,
+      "learning_rate": 2.4092523928619137e-06,
+      "loss": 1.1529,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25164039198977417,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4091895259302084e-06,
+      "loss": 1.2555,
+      "step": 2953
+    },
+    {
+      "epoch": 0.2517256071580741,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.409126638050783e-06,
+      "loss": 0.6195,
+      "step": 2954
+    },
+    {
+      "epoch": 0.2518108223263741,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4090637292247737e-06,
+      "loss": 0.9051,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25189603749467404,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.409000799453317e-06,
+      "loss": 0.8554,
+      "step": 2956
+    },
+    {
+      "epoch": 0.251981252662974,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.408937848737551e-06,
+      "loss": 0.9069,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25206646783127395,
+      "grad_norm": 16.375,
+      "learning_rate": 2.408874877078612e-06,
+      "loss": 1.0019,
+      "step": 2958
+    },
+    {
+      "epoch": 0.2521516829995739,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.408811884477639e-06,
+      "loss": 0.7156,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25223689816787387,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4087488709357698e-06,
+      "loss": 0.95,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2523221133361738,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4086858364541437e-06,
+      "loss": 1.1111,
+      "step": 2961
+    },
+    {
+      "epoch": 0.2524073285044738,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.4086227810338987e-06,
+      "loss": 0.7465,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25249254367277374,
+      "grad_norm": 12.25,
+      "learning_rate": 2.4085597046761754e-06,
+      "loss": 0.5839,
+      "step": 2963
+    },
+    {
+      "epoch": 0.2525777588410737,
+      "grad_norm": 17.5,
+      "learning_rate": 2.408496607382113e-06,
+      "loss": 0.939,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25266297400937365,
+      "grad_norm": 17.0,
+      "learning_rate": 2.408433489152852e-06,
+      "loss": 1.0018,
+      "step": 2965
+    },
+    {
+      "epoch": 0.2527481891776736,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4083703499895335e-06,
+      "loss": 1.0089,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25283340434597357,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4083071898932975e-06,
+      "loss": 1.0389,
+      "step": 2967
+    },
+    {
+      "epoch": 0.2529186195142735,
+      "grad_norm": 15.625,
+      "learning_rate": 2.4082440088652857e-06,
+      "loss": 0.7878,
+      "step": 2968
+    },
+    {
+      "epoch": 0.2530038346825735,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4081808069066396e-06,
+      "loss": 1.1512,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25308904985087344,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4081175840185023e-06,
+      "loss": 0.9583,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2531742650191734,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4080543402020155e-06,
+      "loss": 0.4247,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25325948018747335,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4079910754583223e-06,
+      "loss": 0.4451,
+      "step": 2972
+    },
+    {
+      "epoch": 0.2533446953557733,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4079277897885662e-06,
+      "loss": 0.6546,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25342991052407327,
+      "grad_norm": 12.625,
+      "learning_rate": 2.40786448319389e-06,
+      "loss": 0.5652,
+      "step": 2974
+    },
+    {
+      "epoch": 0.2535151256923732,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4078011556754386e-06,
+      "loss": 0.7326,
+      "step": 2975
+    },
+    {
+      "epoch": 0.2536003408606732,
+      "grad_norm": 13.75,
+      "learning_rate": 2.407737807234356e-06,
+      "loss": 0.5198,
+      "step": 2976
+    },
+    {
+      "epoch": 0.25368555602897314,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.407674437871787e-06,
+      "loss": 1.0169,
+      "step": 2977
+    },
+    {
+      "epoch": 0.2537707711972731,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.4076110475888763e-06,
+      "loss": 0.3128,
+      "step": 2978
+    },
+    {
+      "epoch": 0.25385598636557305,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4075476363867707e-06,
+      "loss": 0.8734,
+      "step": 2979
+    },
+    {
+      "epoch": 0.253941201533873,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4074842042666147e-06,
+      "loss": 0.8517,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25402641670217296,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4074207512295557e-06,
+      "loss": 0.8789,
+      "step": 2981
+    },
+    {
+      "epoch": 0.2541116318704729,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4073572772767396e-06,
+      "loss": 0.604,
+      "step": 2982
+    },
+    {
+      "epoch": 0.2541968470387729,
+      "grad_norm": 33.75,
+      "learning_rate": 2.4072937824093138e-06,
+      "loss": 0.9227,
+      "step": 2983
+    },
+    {
+      "epoch": 0.25428206220707283,
+      "grad_norm": 18.625,
+      "learning_rate": 2.4072302666284255e-06,
+      "loss": 0.8889,
+      "step": 2984
+    },
+    {
+      "epoch": 0.2543672773753728,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4071667299352225e-06,
+      "loss": 0.7892,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25445249254367275,
+      "grad_norm": 16.125,
+      "learning_rate": 2.407103172330853e-06,
+      "loss": 0.647,
+      "step": 2986
+    },
+    {
+      "epoch": 0.2545377077119727,
+      "grad_norm": 19.625,
+      "learning_rate": 2.407039593816466e-06,
+      "loss": 1.0276,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25462292288027266,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4069759943932096e-06,
+      "loss": 0.5737,
+      "step": 2988
+    },
+    {
+      "epoch": 0.2547081380485726,
+      "grad_norm": 11.5,
+      "learning_rate": 2.406912374062234e-06,
+      "loss": 0.8007,
+      "step": 2989
+    },
+    {
+      "epoch": 0.2547933532168726,
+      "grad_norm": 11.625,
+      "learning_rate": 2.406848732824688e-06,
+      "loss": 0.4117,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2548785683851726,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4067850706817224e-06,
+      "loss": 0.6301,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25496378355347254,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4067213876344873e-06,
+      "loss": 0.7766,
+      "step": 2992
+    },
+    {
+      "epoch": 0.2550489987217725,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4066576836841334e-06,
+      "loss": 0.9498,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25513421389007246,
+      "grad_norm": 17.75,
+      "learning_rate": 2.406593958831812e-06,
+      "loss": 0.9111,
+      "step": 2994
+    },
+    {
+      "epoch": 0.2552194290583724,
+      "grad_norm": 20.625,
+      "learning_rate": 2.406530213078675e-06,
+      "loss": 0.7957,
+      "step": 2995
+    },
+    {
+      "epoch": 0.2553046442266724,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.406466446425874e-06,
+      "loss": 0.618,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25538985939497233,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4064026588745616e-06,
+      "loss": 0.6432,
+      "step": 2997
+    },
+    {
+      "epoch": 0.2554750745632723,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4063388504258897e-06,
+      "loss": 0.7925,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25556028973157224,
+      "grad_norm": 18.125,
+      "learning_rate": 2.4062750210810124e-06,
+      "loss": 1.2936,
+      "step": 2999
+    },
+    {
+      "epoch": 0.2556455048998722,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4062111708410825e-06,
+      "loss": 0.8411,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25573072006817216,
+      "grad_norm": 10.875,
+      "learning_rate": 2.406147299707254e-06,
+      "loss": 0.4682,
+      "step": 3001
+    },
+    {
+      "epoch": 0.2558159352364721,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.406083407680682e-06,
+      "loss": 0.6219,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25590115040477207,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4060194947625195e-06,
+      "loss": 0.9872,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25598636557307203,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4059555609539226e-06,
+      "loss": 0.6533,
+      "step": 3004
+    },
+    {
+      "epoch": 0.256071580741372,
+      "grad_norm": 17.875,
+      "learning_rate": 2.405891606256046e-06,
+      "loss": 1.0268,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25615679590967194,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4058276306700462e-06,
+      "loss": 0.8031,
+      "step": 3006
+    },
+    {
+      "epoch": 0.2562420110779719,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4057636341970786e-06,
+      "loss": 0.8053,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25632722624627186,
+      "grad_norm": 71.5,
+      "learning_rate": 2.4056996168382996e-06,
+      "loss": 0.8514,
+      "step": 3008
+    },
+    {
+      "epoch": 0.2564124414145718,
+      "grad_norm": 17.75,
+      "learning_rate": 2.405635578594867e-06,
+      "loss": 1.1079,
+      "step": 3009
+    },
+    {
+      "epoch": 0.25649765658287177,
+      "grad_norm": 15.0,
+      "learning_rate": 2.405571519467937e-06,
+      "loss": 0.8001,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2565828717511717,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4055074394586676e-06,
+      "loss": 0.8068,
+      "step": 3011
+    },
+    {
+      "epoch": 0.2566680869194717,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.405443338568217e-06,
+      "loss": 0.5894,
+      "step": 3012
+    },
+    {
+      "epoch": 0.25675330208777164,
+      "grad_norm": 16.5,
+      "learning_rate": 2.405379216797743e-06,
+      "loss": 0.7341,
+      "step": 3013
+    },
+    {
+      "epoch": 0.2568385172560716,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4053150741484056e-06,
+      "loss": 0.6545,
+      "step": 3014
+    },
+    {
+      "epoch": 0.25692373242437155,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4052509106213623e-06,
+      "loss": 0.7748,
+      "step": 3015
+    },
+    {
+      "epoch": 0.2570089475926715,
+      "grad_norm": 28.125,
+      "learning_rate": 2.4051867262177736e-06,
+      "loss": 1.1048,
+      "step": 3016
+    },
+    {
+      "epoch": 0.25709416276097147,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.4051225209387992e-06,
+      "loss": 0.6033,
+      "step": 3017
+    },
+    {
+      "epoch": 0.2571793779292714,
+      "grad_norm": 23.375,
+      "learning_rate": 2.4050582947855987e-06,
+      "loss": 1.2305,
+      "step": 3018
+    },
+    {
+      "epoch": 0.2572645930975714,
+      "grad_norm": 22.75,
+      "learning_rate": 2.404994047759334e-06,
+      "loss": 0.9808,
+      "step": 3019
+    },
+    {
+      "epoch": 0.25734980826587134,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4049297798611654e-06,
+      "loss": 0.8482,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2574350234341713,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4048654910922537e-06,
+      "loss": 0.5807,
+      "step": 3021
+    },
+    {
+      "epoch": 0.25752023860247125,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.404801181453762e-06,
+      "loss": 0.7013,
+      "step": 3022
+    },
+    {
+      "epoch": 0.2576054537707712,
+      "grad_norm": 17.375,
+      "learning_rate": 2.404736850946851e-06,
+      "loss": 0.7474,
+      "step": 3023
+    },
+    {
+      "epoch": 0.25769066893907117,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4046724995726844e-06,
+      "loss": 0.6789,
+      "step": 3024
+    },
+    {
+      "epoch": 0.2577758841073711,
+      "grad_norm": 10.875,
+      "learning_rate": 2.4046081273324246e-06,
+      "loss": 0.394,
+      "step": 3025
+    },
+    {
+      "epoch": 0.2578610992756711,
+      "grad_norm": 13.0,
+      "learning_rate": 2.404543734227235e-06,
+      "loss": 0.8614,
+      "step": 3026
+    },
+    {
+      "epoch": 0.25794631444397104,
+      "grad_norm": 13.125,
+      "learning_rate": 2.404479320258279e-06,
+      "loss": 0.4764,
+      "step": 3027
+    },
+    {
+      "epoch": 0.258031529612271,
+      "grad_norm": 13.375,
+      "learning_rate": 2.4044148854267203e-06,
+      "loss": 0.7893,
+      "step": 3028
+    },
+    {
+      "epoch": 0.25811674478057095,
+      "grad_norm": 14.625,
+      "learning_rate": 2.4043504297337244e-06,
+      "loss": 0.4692,
+      "step": 3029
+    },
+    {
+      "epoch": 0.2582019599488709,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.404285953180455e-06,
+      "loss": 0.8938,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25828717511717086,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4042214557680783e-06,
+      "loss": 0.6956,
+      "step": 3031
+    },
+    {
+      "epoch": 0.2583723902854708,
+      "grad_norm": 21.5,
+      "learning_rate": 2.4041569374977587e-06,
+      "loss": 1.0409,
+      "step": 3032
+    },
+    {
+      "epoch": 0.2584576054537708,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4040923983706626e-06,
+      "loss": 1.004,
+      "step": 3033
+    },
+    {
+      "epoch": 0.25854282062207073,
+      "grad_norm": 19.375,
+      "learning_rate": 2.404027838387957e-06,
+      "loss": 1.1083,
+      "step": 3034
+    },
+    {
+      "epoch": 0.2586280357903707,
+      "grad_norm": 17.125,
+      "learning_rate": 2.403963257550807e-06,
+      "loss": 0.7026,
+      "step": 3035
+    },
+    {
+      "epoch": 0.25871325095867065,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.403898655860381e-06,
+      "loss": 0.4245,
+      "step": 3036
+    },
+    {
+      "epoch": 0.2587984661269706,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.403834033317846e-06,
+      "loss": 0.6608,
+      "step": 3037
+    },
+    {
+      "epoch": 0.25888368129527056,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.4037693899243696e-06,
+      "loss": 0.6097,
+      "step": 3038
+    },
+    {
+      "epoch": 0.2589688964635705,
+      "grad_norm": 21.0,
+      "learning_rate": 2.4037047256811204e-06,
+      "loss": 1.1294,
+      "step": 3039
+    },
+    {
+      "epoch": 0.2590541116318705,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.403640040589266e-06,
+      "loss": 0.6969,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25913932680017043,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.403575334649977e-06,
+      "loss": 0.7207,
+      "step": 3041
+    },
+    {
+      "epoch": 0.2592245419684704,
+      "grad_norm": 14.625,
+      "learning_rate": 2.403510607864421e-06,
+      "loss": 0.8406,
+      "step": 3042
+    },
+    {
+      "epoch": 0.25930975713677035,
+      "grad_norm": 18.75,
+      "learning_rate": 2.4034458602337684e-06,
+      "loss": 1.0368,
+      "step": 3043
+    },
+    {
+      "epoch": 0.2593949723050703,
+      "grad_norm": 18.75,
+      "learning_rate": 2.403381091759189e-06,
+      "loss": 1.1469,
+      "step": 3044
+    },
+    {
+      "epoch": 0.25948018747337026,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4033163024418543e-06,
+      "loss": 1.0348,
+      "step": 3045
+    },
+    {
+      "epoch": 0.2595654026416702,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4032514922829337e-06,
+      "loss": 0.8978,
+      "step": 3046
+    },
+    {
+      "epoch": 0.2596506178099702,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4031866612835993e-06,
+      "loss": 0.1999,
+      "step": 3047
+    },
+    {
+      "epoch": 0.25973583297827013,
+      "grad_norm": 11.125,
+      "learning_rate": 2.4031218094450215e-06,
+      "loss": 0.4699,
+      "step": 3048
+    },
+    {
+      "epoch": 0.2598210481465701,
+      "grad_norm": 20.125,
+      "learning_rate": 2.4030569367683736e-06,
+      "loss": 0.6024,
+      "step": 3049
+    },
+    {
+      "epoch": 0.25990626331487005,
+      "grad_norm": 33.75,
+      "learning_rate": 2.4029920432548275e-06,
+      "loss": 1.1232,
+      "step": 3050
+    },
+    {
+      "epoch": 0.25999147848317,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4029271289055558e-06,
+      "loss": 0.6519,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26007669365146996,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4028621937217312e-06,
+      "loss": 0.5323,
+      "step": 3052
+    },
+    {
+      "epoch": 0.2601619088197699,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.4027972377045274e-06,
+      "loss": 0.5999,
+      "step": 3053
+    },
+    {
+      "epoch": 0.2602471239880699,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4027322608551183e-06,
+      "loss": 0.6853,
+      "step": 3054
+    },
+    {
+      "epoch": 0.26033233915636983,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.4026672631746783e-06,
+      "loss": 0.8008,
+      "step": 3055
+    },
+    {
+      "epoch": 0.2604175543246698,
+      "grad_norm": 17.375,
+      "learning_rate": 2.402602244664382e-06,
+      "loss": 0.6959,
+      "step": 3056
+    },
+    {
+      "epoch": 0.26050276949296974,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.402537205325404e-06,
+      "loss": 0.8843,
+      "step": 3057
+    },
+    {
+      "epoch": 0.2605879846612697,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.4024721451589194e-06,
+      "loss": 0.4014,
+      "step": 3058
+    },
+    {
+      "epoch": 0.26067319982956966,
+      "grad_norm": 15.25,
+      "learning_rate": 2.4024070641661043e-06,
+      "loss": 0.9268,
+      "step": 3059
+    },
+    {
+      "epoch": 0.2607584149978696,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4023419623481346e-06,
+      "loss": 0.3544,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26084363016616957,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4022768397061873e-06,
+      "loss": 0.7147,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26092884533446953,
+      "grad_norm": 17.875,
+      "learning_rate": 2.4022116962414384e-06,
+      "loss": 0.7709,
+      "step": 3062
+    },
+    {
+      "epoch": 0.2610140605027695,
+      "grad_norm": 13.375,
+      "learning_rate": 2.402146531955066e-06,
+      "loss": 0.3639,
+      "step": 3063
+    },
+    {
+      "epoch": 0.26109927567106944,
+      "grad_norm": 17.5,
+      "learning_rate": 2.402081346848247e-06,
+      "loss": 0.7288,
+      "step": 3064
+    },
+    {
+      "epoch": 0.2611844908393694,
+      "grad_norm": 26.5,
+      "learning_rate": 2.4020161409221594e-06,
+      "loss": 1.3285,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26126970600766936,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4019509141779816e-06,
+      "loss": 1.0347,
+      "step": 3066
+    },
+    {
+      "epoch": 0.2613549211759693,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4018856666168925e-06,
+      "loss": 0.8393,
+      "step": 3067
+    },
+    {
+      "epoch": 0.26144013634426927,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4018203982400713e-06,
+      "loss": 0.7065,
+      "step": 3068
+    },
+    {
+      "epoch": 0.2615253515125692,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.401755109048697e-06,
+      "loss": 0.6089,
+      "step": 3069
+    },
+    {
+      "epoch": 0.2616105666808692,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4016897990439498e-06,
+      "loss": 0.587,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26169578184916914,
+      "grad_norm": 16.25,
+      "learning_rate": 2.4016244682270097e-06,
+      "loss": 0.8635,
+      "step": 3071
+    },
+    {
+      "epoch": 0.2617809970174691,
+      "grad_norm": 30.0,
+      "learning_rate": 2.4015591165990576e-06,
+      "loss": 0.9933,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26186621218576905,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.401493744161274e-06,
+      "loss": 0.4795,
+      "step": 3073
+    },
+    {
+      "epoch": 0.261951427354069,
+      "grad_norm": 13.5,
+      "learning_rate": 2.4014283509148406e-06,
+      "loss": 0.5747,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26203664252236897,
+      "grad_norm": 15.75,
+      "learning_rate": 2.4013629368609393e-06,
+      "loss": 0.5161,
+      "step": 3075
+    },
+    {
+      "epoch": 0.2621218576906689,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4012975020007516e-06,
+      "loss": 0.6793,
+      "step": 3076
+    },
+    {
+      "epoch": 0.2622070728589689,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4012320463354606e-06,
+      "loss": 0.4181,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26229228802726884,
+      "grad_norm": 18.0,
+      "learning_rate": 2.401166569866249e-06,
+      "loss": 0.6657,
+      "step": 3078
+    },
+    {
+      "epoch": 0.2623775031955688,
+      "grad_norm": 26.25,
+      "learning_rate": 2.4011010725942998e-06,
+      "loss": 1.0912,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26246271836386875,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4010355545207965e-06,
+      "loss": 0.8171,
+      "step": 3080
+    },
+    {
+      "epoch": 0.2625479335321687,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.4009700156469233e-06,
+      "loss": 0.6955,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26263314870046867,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4009044559738646e-06,
+      "loss": 0.688,
+      "step": 3082
+    },
+    {
+      "epoch": 0.2627183638687686,
+      "grad_norm": 13.0,
+      "learning_rate": 2.400838875502805e-06,
+      "loss": 0.6327,
+      "step": 3083
+    },
+    {
+      "epoch": 0.2628035790370686,
+      "grad_norm": 16.375,
+      "learning_rate": 2.40077327423493e-06,
+      "loss": 1.2231,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26288879420536854,
+      "grad_norm": 16.75,
+      "learning_rate": 2.400707652171424e-06,
+      "loss": 1.0987,
+      "step": 3085
+    },
+    {
+      "epoch": 0.2629740093736685,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.4006420093134744e-06,
+      "loss": 0.5817,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26305922454196845,
+      "grad_norm": 17.625,
+      "learning_rate": 2.4005763456622663e-06,
+      "loss": 0.8865,
+      "step": 3087
+    },
+    {
+      "epoch": 0.2631444397102684,
+      "grad_norm": 12.75,
+      "learning_rate": 2.4005106612189866e-06,
+      "loss": 0.5989,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26322965487856836,
+      "grad_norm": 16.5,
+      "learning_rate": 2.4004449559848224e-06,
+      "loss": 1.0143,
+      "step": 3089
+    },
+    {
+      "epoch": 0.2633148700468683,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.400379229960961e-06,
+      "loss": 0.4729,
+      "step": 3090
+    },
+    {
+      "epoch": 0.2634000852151683,
+      "grad_norm": 17.0,
+      "learning_rate": 2.40031348314859e-06,
+      "loss": 0.7569,
+      "step": 3091
+    },
+    {
+      "epoch": 0.26348530038346824,
+      "grad_norm": 24.75,
+      "learning_rate": 2.4002477155488976e-06,
+      "loss": 0.9389,
+      "step": 3092
+    },
+    {
+      "epoch": 0.2635705155517682,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4001819271630726e-06,
+      "loss": 0.7724,
+      "step": 3093
+    },
+    {
+      "epoch": 0.26365573072006815,
+      "grad_norm": 12.125,
+      "learning_rate": 2.4001161179923034e-06,
+      "loss": 0.6507,
+      "step": 3094
+    },
+    {
+      "epoch": 0.2637409458883681,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.400050288037779e-06,
+      "loss": 0.506,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26382616105666806,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3999844373006904e-06,
+      "loss": 0.5859,
+      "step": 3096
+    },
+    {
+      "epoch": 0.263911376224968,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.3999185657822256e-06,
+      "loss": 0.6689,
+      "step": 3097
+    },
+    {
+      "epoch": 0.263996591393268,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3998526734835765e-06,
+      "loss": 0.7123,
+      "step": 3098
+    },
+    {
+      "epoch": 0.26408180656156793,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3997867604059335e-06,
+      "loss": 0.7335,
+      "step": 3099
+    },
+    {
+      "epoch": 0.2641670217298679,
+      "grad_norm": 17.25,
+      "learning_rate": 2.399720826550487e-06,
+      "loss": 0.9156,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26425223689816785,
+      "grad_norm": 17.75,
+      "learning_rate": 2.3996548719184293e-06,
+      "loss": 0.7847,
+      "step": 3101
+    },
+    {
+      "epoch": 0.2643374520664678,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.399588896510952e-06,
+      "loss": 0.7428,
+      "step": 3102
+    },
+    {
+      "epoch": 0.26442266723476776,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3995229003292473e-06,
+      "loss": 0.6125,
+      "step": 3103
+    },
+    {
+      "epoch": 0.2645078824030678,
+      "grad_norm": 12.75,
+      "learning_rate": 2.3994568833745085e-06,
+      "loss": 0.5944,
+      "step": 3104
+    },
+    {
+      "epoch": 0.26459309757136773,
+      "grad_norm": 15.25,
+      "learning_rate": 2.399390845647927e-06,
+      "loss": 0.7549,
+      "step": 3105
+    },
+    {
+      "epoch": 0.2646783127396677,
+      "grad_norm": 36.25,
+      "learning_rate": 2.3993247871506976e-06,
+      "loss": 1.4105,
+      "step": 3106
+    },
+    {
+      "epoch": 0.26476352790796764,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3992587078840134e-06,
+      "loss": 0.7736,
+      "step": 3107
+    },
+    {
+      "epoch": 0.2648487430762676,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3991926078490686e-06,
+      "loss": 0.9964,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26493395824456756,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.3991264870470583e-06,
+      "loss": 0.5037,
+      "step": 3109
+    },
+    {
+      "epoch": 0.2650191734128675,
+      "grad_norm": 15.375,
+      "learning_rate": 2.399060345479176e-06,
+      "loss": 0.4565,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26510438858116747,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.3989941831466184e-06,
+      "loss": 0.4318,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26518960374946743,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.3989280000505805e-06,
+      "loss": 0.3038,
+      "step": 3112
+    },
+    {
+      "epoch": 0.2652748189177674,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.3988617961922578e-06,
+      "loss": 0.5742,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26536003408606734,
+      "grad_norm": 14.0,
+      "learning_rate": 2.398795571572848e-06,
+      "loss": 0.8746,
+      "step": 3114
+    },
+    {
+      "epoch": 0.2654452492543673,
+      "grad_norm": 17.125,
+      "learning_rate": 2.398729326193546e-06,
+      "loss": 1.365,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26553046442266726,
+      "grad_norm": 22.375,
+      "learning_rate": 2.398663060055551e-06,
+      "loss": 1.0579,
+      "step": 3116
+    },
+    {
+      "epoch": 0.2656156795909672,
+      "grad_norm": 32.0,
+      "learning_rate": 2.3985967731600586e-06,
+      "loss": 0.9658,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26570089475926717,
+      "grad_norm": 21.5,
+      "learning_rate": 2.398530465508268e-06,
+      "loss": 0.747,
+      "step": 3118
+    },
+    {
+      "epoch": 0.2657861099275671,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3984641371013765e-06,
+      "loss": 0.9848,
+      "step": 3119
+    },
+    {
+      "epoch": 0.2658713250958671,
+      "grad_norm": 20.875,
+      "learning_rate": 2.3983977879405834e-06,
+      "loss": 0.879,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26595654026416704,
+      "grad_norm": 13.0,
+      "learning_rate": 2.398331418027088e-06,
+      "loss": 0.375,
+      "step": 3121
+    },
+    {
+      "epoch": 0.266041755432467,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.3982650273620882e-06,
+      "loss": 0.5694,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26612697060076695,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.398198615946785e-06,
+      "loss": 0.6548,
+      "step": 3123
+    },
+    {
+      "epoch": 0.2662121857690669,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3981321837823784e-06,
+      "loss": 0.7613,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26629740093736687,
+      "grad_norm": 22.625,
+      "learning_rate": 2.398065730870069e-06,
+      "loss": 0.8212,
+      "step": 3125
+    },
+    {
+      "epoch": 0.2663826161056668,
+      "grad_norm": 21.75,
+      "learning_rate": 2.397999257211057e-06,
+      "loss": 0.922,
+      "step": 3126
+    },
+    {
+      "epoch": 0.2664678312739668,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.397932762806544e-06,
+      "loss": 0.5572,
+      "step": 3127
+    },
+    {
+      "epoch": 0.26655304644226674,
+      "grad_norm": 24.75,
+      "learning_rate": 2.3978662476577313e-06,
+      "loss": 0.9428,
+      "step": 3128
+    },
+    {
+      "epoch": 0.2666382616105667,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.397799711765822e-06,
+      "loss": 0.9027,
+      "step": 3129
+    },
+    {
+      "epoch": 0.26672347677886665,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.397733155132017e-06,
+      "loss": 0.6059,
+      "step": 3130
+    },
+    {
+      "epoch": 0.2668086919471666,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3976665777575203e-06,
+      "loss": 0.9804,
+      "step": 3131
+    },
+    {
+      "epoch": 0.26689390711546657,
+      "grad_norm": 10.75,
+      "learning_rate": 2.3975999796435343e-06,
+      "loss": 0.3274,
+      "step": 3132
+    },
+    {
+      "epoch": 0.2669791222837665,
+      "grad_norm": 12.625,
+      "learning_rate": 2.397533360791263e-06,
+      "loss": 0.5653,
+      "step": 3133
+    },
+    {
+      "epoch": 0.2670643374520665,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.397466721201909e-06,
+      "loss": 0.5945,
+      "step": 3134
+    },
+    {
+      "epoch": 0.26714955262036644,
+      "grad_norm": 10.5,
+      "learning_rate": 2.3974000608766777e-06,
+      "loss": 0.5088,
+      "step": 3135
+    },
+    {
+      "epoch": 0.2672347677886664,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.3973333798167744e-06,
+      "loss": 0.5068,
+      "step": 3136
+    },
+    {
+      "epoch": 0.26731998295696635,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3972666780234026e-06,
+      "loss": 0.905,
+      "step": 3137
+    },
+    {
+      "epoch": 0.2674051981252663,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3971999554977683e-06,
+      "loss": 0.713,
+      "step": 3138
+    },
+    {
+      "epoch": 0.26749041329356626,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.397133212241077e-06,
+      "loss": 0.3716,
+      "step": 3139
+    },
+    {
+      "epoch": 0.2675756284618662,
+      "grad_norm": 21.375,
+      "learning_rate": 2.3970664482545352e-06,
+      "loss": 0.9597,
+      "step": 3140
+    },
+    {
+      "epoch": 0.2676608436301662,
+      "grad_norm": 11.25,
+      "learning_rate": 2.3969996635393493e-06,
+      "loss": 0.4993,
+      "step": 3141
+    },
+    {
+      "epoch": 0.26774605879846614,
+      "grad_norm": 13.125,
+      "learning_rate": 2.396932858096726e-06,
+      "loss": 0.7593,
+      "step": 3142
+    },
+    {
+      "epoch": 0.2678312739667661,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3968660319278725e-06,
+      "loss": 0.7841,
+      "step": 3143
+    },
+    {
+      "epoch": 0.26791648913506605,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3967991850339965e-06,
+      "loss": 0.7095,
+      "step": 3144
+    },
+    {
+      "epoch": 0.268001704303366,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3967323174163066e-06,
+      "loss": 1.0058,
+      "step": 3145
+    },
+    {
+      "epoch": 0.26808691947166596,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.39666542907601e-06,
+      "loss": 0.3677,
+      "step": 3146
+    },
+    {
+      "epoch": 0.2681721346399659,
+      "grad_norm": 13.5,
+      "learning_rate": 2.3965985200143163e-06,
+      "loss": 0.4982,
+      "step": 3147
+    },
+    {
+      "epoch": 0.2682573498082659,
+      "grad_norm": 22.75,
+      "learning_rate": 2.396531590232434e-06,
+      "loss": 0.8559,
+      "step": 3148
+    },
+    {
+      "epoch": 0.26834256497656583,
+      "grad_norm": 17.375,
+      "learning_rate": 2.3964646397315735e-06,
+      "loss": 1.0388,
+      "step": 3149
+    },
+    {
+      "epoch": 0.2684277801448658,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3963976685129437e-06,
+      "loss": 0.6354,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26851299531316575,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3963306765777554e-06,
+      "loss": 0.4669,
+      "step": 3151
+    },
+    {
+      "epoch": 0.2685982104814657,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3962636639272193e-06,
+      "loss": 0.9461,
+      "step": 3152
+    },
+    {
+      "epoch": 0.26868342564976566,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.396196630562546e-06,
+      "loss": 0.6262,
+      "step": 3153
+    },
+    {
+      "epoch": 0.2687686408180656,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.3961295764849462e-06,
+      "loss": 0.374,
+      "step": 3154
+    },
+    {
+      "epoch": 0.2688538559863656,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3960625016956335e-06,
+      "loss": 0.9857,
+      "step": 3155
+    },
+    {
+      "epoch": 0.26893907115466553,
+      "grad_norm": 11.375,
+      "learning_rate": 2.3959954061958184e-06,
+      "loss": 0.5476,
+      "step": 3156
+    },
+    {
+      "epoch": 0.2690242863229655,
+      "grad_norm": 11.75,
+      "learning_rate": 2.395928289986714e-06,
+      "loss": 0.741,
+      "step": 3157
+    },
+    {
+      "epoch": 0.26910950149126545,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.395861153069533e-06,
+      "loss": 0.7711,
+      "step": 3158
+    },
+    {
+      "epoch": 0.2691947166595654,
+      "grad_norm": 11.375,
+      "learning_rate": 2.395793995445489e-06,
+      "loss": 0.6454,
+      "step": 3159
+    },
+    {
+      "epoch": 0.26927993182786536,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.395726817115795e-06,
+      "loss": 0.6697,
+      "step": 3160
+    },
+    {
+      "epoch": 0.2693651469961653,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3956596180816655e-06,
+      "loss": 0.7545,
+      "step": 3161
+    },
+    {
+      "epoch": 0.2694503621644653,
+      "grad_norm": 12.625,
+      "learning_rate": 2.395592398344315e-06,
+      "loss": 0.6582,
+      "step": 3162
+    },
+    {
+      "epoch": 0.26953557733276523,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.395525157904957e-06,
+      "loss": 0.6127,
+      "step": 3163
+    },
+    {
+      "epoch": 0.2696207925010652,
+      "grad_norm": 16.75,
+      "learning_rate": 2.395457896764808e-06,
+      "loss": 0.9324,
+      "step": 3164
+    },
+    {
+      "epoch": 0.26970600766936514,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.395390614925083e-06,
+      "loss": 0.546,
+      "step": 3165
+    },
+    {
+      "epoch": 0.2697912228376651,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3953233123869975e-06,
+      "loss": 0.648,
+      "step": 3166
+    },
+    {
+      "epoch": 0.26987643800596506,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.3952559891517685e-06,
+      "loss": 0.4571,
+      "step": 3167
+    },
+    {
+      "epoch": 0.269961653174265,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.395188645220612e-06,
+      "loss": 0.3664,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27004686834256497,
+      "grad_norm": 22.125,
+      "learning_rate": 2.395121280594745e-06,
+      "loss": 1.4256,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27013208351086493,
+      "grad_norm": 24.125,
+      "learning_rate": 2.3950538952753847e-06,
+      "loss": 1.0846,
+      "step": 3170
+    },
+    {
+      "epoch": 0.2702172986791649,
+      "grad_norm": 17.125,
+      "learning_rate": 2.394986489263749e-06,
+      "loss": 1.0623,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27030251384746484,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.394919062561057e-06,
+      "loss": 0.7516,
+      "step": 3172
+    },
+    {
+      "epoch": 0.2703877290157648,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3948516151685253e-06,
+      "loss": 0.8134,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27047294418406476,
+      "grad_norm": 16.0,
+      "learning_rate": 2.394784147087374e-06,
+      "loss": 0.8237,
+      "step": 3174
+    },
+    {
+      "epoch": 0.2705581593523647,
+      "grad_norm": 13.375,
+      "learning_rate": 2.394716658318822e-06,
+      "loss": 0.8622,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27064337452066467,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.3946491488640887e-06,
+      "loss": 0.5916,
+      "step": 3176
+    },
+    {
+      "epoch": 0.2707285896889646,
+      "grad_norm": 14.625,
+      "learning_rate": 2.394581618724395e-06,
+      "loss": 0.5921,
+      "step": 3177
+    },
+    {
+      "epoch": 0.2708138048572646,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3945140679009593e-06,
+      "loss": 0.7363,
+      "step": 3178
+    },
+    {
+      "epoch": 0.27089902002556454,
+      "grad_norm": 19.375,
+      "learning_rate": 2.394446496395004e-06,
+      "loss": 0.9161,
+      "step": 3179
+    },
+    {
+      "epoch": 0.2709842351938645,
+      "grad_norm": 32.0,
+      "learning_rate": 2.39437890420775e-06,
+      "loss": 1.0106,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27106945036216445,
+      "grad_norm": 21.25,
+      "learning_rate": 2.394311291340418e-06,
+      "loss": 0.6123,
+      "step": 3181
+    },
+    {
+      "epoch": 0.2711546655304644,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.3942436577942308e-06,
+      "loss": 0.6568,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27123988069876437,
+      "grad_norm": 20.375,
+      "learning_rate": 2.39417600357041e-06,
+      "loss": 0.804,
+      "step": 3183
+    },
+    {
+      "epoch": 0.2713250958670643,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3941083286701778e-06,
+      "loss": 0.4972,
+      "step": 3184
+    },
+    {
+      "epoch": 0.2714103110353643,
+      "grad_norm": 20.625,
+      "learning_rate": 2.3940406330947578e-06,
+      "loss": 0.8979,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27149552620366424,
+      "grad_norm": 8.8125,
+      "learning_rate": 2.393972916845373e-06,
+      "loss": 0.5903,
+      "step": 3186
+    },
+    {
+      "epoch": 0.2715807413719642,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.3939051799232476e-06,
+      "loss": 0.4858,
+      "step": 3187
+    },
+    {
+      "epoch": 0.27166595654026415,
+      "grad_norm": 22.625,
+      "learning_rate": 2.393837422329605e-06,
+      "loss": 0.8938,
+      "step": 3188
+    },
+    {
+      "epoch": 0.2717511717085641,
+      "grad_norm": 31.0,
+      "learning_rate": 2.3937696440656703e-06,
+      "loss": 1.5027,
+      "step": 3189
+    },
+    {
+      "epoch": 0.27183638687686407,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.3937018451326677e-06,
+      "loss": 0.4057,
+      "step": 3190
+    },
+    {
+      "epoch": 0.271921602045164,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3936340255318223e-06,
+      "loss": 0.8049,
+      "step": 3191
+    },
+    {
+      "epoch": 0.272006817213464,
+      "grad_norm": 10.75,
+      "learning_rate": 2.3935661852643606e-06,
+      "loss": 0.4454,
+      "step": 3192
+    },
+    {
+      "epoch": 0.27209203238176394,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.393498324331508e-06,
+      "loss": 0.6401,
+      "step": 3193
+    },
+    {
+      "epoch": 0.2721772475500639,
+      "grad_norm": 23.0,
+      "learning_rate": 2.3934304427344906e-06,
+      "loss": 1.0184,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27226246271836385,
+      "grad_norm": 10.875,
+      "learning_rate": 2.3933625404745357e-06,
+      "loss": 0.4099,
+      "step": 3195
+    },
+    {
+      "epoch": 0.2723476778866638,
+      "grad_norm": 20.875,
+      "learning_rate": 2.3932946175528693e-06,
+      "loss": 1.134,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27243289305496377,
+      "grad_norm": 29.125,
+      "learning_rate": 2.39322667397072e-06,
+      "loss": 0.8978,
+      "step": 3197
+    },
+    {
+      "epoch": 0.2725181082232637,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.393158709729315e-06,
+      "loss": 1.0133,
+      "step": 3198
+    },
+    {
+      "epoch": 0.2726033233915637,
+      "grad_norm": 18.75,
+      "learning_rate": 2.3930907248298825e-06,
+      "loss": 1.0383,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27268853855986364,
+      "grad_norm": 14.625,
+      "learning_rate": 2.393022719273651e-06,
+      "loss": 0.4778,
+      "step": 3200
+    },
+    {
+      "epoch": 0.2727737537281636,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3929546930618496e-06,
+      "loss": 0.9558,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27285896889646355,
+      "grad_norm": 17.75,
+      "learning_rate": 2.392886646195708e-06,
+      "loss": 0.7627,
+      "step": 3202
+    },
+    {
+      "epoch": 0.2729441840647635,
+      "grad_norm": 19.375,
+      "learning_rate": 2.392818578676455e-06,
+      "loss": 1.3218,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27302939923306346,
+      "grad_norm": 13.5,
+      "learning_rate": 2.392750490505321e-06,
+      "loss": 0.7897,
+      "step": 3204
+    },
+    {
+      "epoch": 0.2731146144013634,
+      "grad_norm": 13.625,
+      "learning_rate": 2.392682381683537e-06,
+      "loss": 0.7711,
+      "step": 3205
+    },
+    {
+      "epoch": 0.2731998295696634,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3926142522123327e-06,
+      "loss": 0.6208,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27328504473796333,
+      "grad_norm": 19.375,
+      "learning_rate": 2.3925461020929404e-06,
+      "loss": 0.5715,
+      "step": 3207
+    },
+    {
+      "epoch": 0.2733702599062633,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.3924779313265906e-06,
+      "loss": 0.5413,
+      "step": 3208
+    },
+    {
+      "epoch": 0.27345547507456325,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.3924097399145165e-06,
+      "loss": 0.5909,
+      "step": 3209
+    },
+    {
+      "epoch": 0.2735406902428632,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3923415278579486e-06,
+      "loss": 0.4819,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27362590541116316,
+      "grad_norm": 18.875,
+      "learning_rate": 2.3922732951581213e-06,
+      "loss": 1.0415,
+      "step": 3211
+    },
+    {
+      "epoch": 0.2737111205794631,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3922050418162667e-06,
+      "loss": 0.741,
+      "step": 3212
+    },
+    {
+      "epoch": 0.2737963357477631,
+      "grad_norm": 10.875,
+      "learning_rate": 2.392136767833618e-06,
+      "loss": 0.381,
+      "step": 3213
+    },
+    {
+      "epoch": 0.27388155091606303,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3920684732114103e-06,
+      "loss": 0.9179,
+      "step": 3214
+    },
+    {
+      "epoch": 0.273966766084363,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3920001579508764e-06,
+      "loss": 0.5845,
+      "step": 3215
+    },
+    {
+      "epoch": 0.274051981252663,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.391931822053251e-06,
+      "loss": 0.6268,
+      "step": 3216
+    },
+    {
+      "epoch": 0.27413719642096296,
+      "grad_norm": 21.375,
+      "learning_rate": 2.3918634655197698e-06,
+      "loss": 1.2118,
+      "step": 3217
+    },
+    {
+      "epoch": 0.2742224115892629,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.3917950883516674e-06,
+      "loss": 0.8733,
+      "step": 3218
+    },
+    {
+      "epoch": 0.2743076267575629,
+      "grad_norm": 27.5,
+      "learning_rate": 2.3917266905501796e-06,
+      "loss": 1.2663,
+      "step": 3219
+    },
+    {
+      "epoch": 0.27439284192586283,
+      "grad_norm": 13.75,
+      "learning_rate": 2.391658272116542e-06,
+      "loss": 0.7653,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2744780570941628,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.3915898330519917e-06,
+      "loss": 0.7122,
+      "step": 3221
+    },
+    {
+      "epoch": 0.27456327226246274,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3915213733577654e-06,
+      "loss": 0.9794,
+      "step": 3222
+    },
+    {
+      "epoch": 0.2746484874307627,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3914528930350998e-06,
+      "loss": 0.7808,
+      "step": 3223
+    },
+    {
+      "epoch": 0.27473370259906266,
+      "grad_norm": 19.5,
+      "learning_rate": 2.3913843920852325e-06,
+      "loss": 0.9757,
+      "step": 3224
+    },
+    {
+      "epoch": 0.2748189177673626,
+      "grad_norm": 18.125,
+      "learning_rate": 2.3913158705094014e-06,
+      "loss": 0.7532,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27490413293566257,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.391247328308845e-06,
+      "loss": 0.4005,
+      "step": 3226
+    },
+    {
+      "epoch": 0.2749893481039625,
+      "grad_norm": 22.375,
+      "learning_rate": 2.391178765484802e-06,
+      "loss": 1.2088,
+      "step": 3227
+    },
+    {
+      "epoch": 0.2750745632722625,
+      "grad_norm": 21.125,
+      "learning_rate": 2.391110182038511e-06,
+      "loss": 0.7223,
+      "step": 3228
+    },
+    {
+      "epoch": 0.27515977844056244,
+      "grad_norm": 16.375,
+      "learning_rate": 2.391041577971211e-06,
+      "loss": 1.0136,
+      "step": 3229
+    },
+    {
+      "epoch": 0.2752449936088624,
+      "grad_norm": 15.75,
+      "learning_rate": 2.390972953284143e-06,
+      "loss": 0.9312,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27533020877716236,
+      "grad_norm": 15.5,
+      "learning_rate": 2.390904307978546e-06,
+      "loss": 1.1171,
+      "step": 3231
+    },
+    {
+      "epoch": 0.2754154239454623,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.390835642055661e-06,
+      "loss": 0.7454,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27550063911376227,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3907669555167288e-06,
+      "loss": 0.8645,
+      "step": 3233
+    },
+    {
+      "epoch": 0.2755858542820622,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3906982483629903e-06,
+      "loss": 0.5586,
+      "step": 3234
+    },
+    {
+      "epoch": 0.2756710694503622,
+      "grad_norm": 13.875,
+      "learning_rate": 2.3906295205956877e-06,
+      "loss": 1.0113,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27575628461866214,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3905607722160625e-06,
+      "loss": 0.6721,
+      "step": 3236
+    },
+    {
+      "epoch": 0.2758414997869621,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3904920032253575e-06,
+      "loss": 1.0627,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27592671495526205,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.390423213624815e-06,
+      "loss": 0.7827,
+      "step": 3238
+    },
+    {
+      "epoch": 0.276011930123562,
+      "grad_norm": 10.0,
+      "learning_rate": 2.390354403415678e-06,
+      "loss": 0.3368,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27609714529186197,
+      "grad_norm": 15.125,
+      "learning_rate": 2.3902855725991907e-06,
+      "loss": 0.5098,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2761823604601619,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3902167211765958e-06,
+      "loss": 0.649,
+      "step": 3241
+    },
+    {
+      "epoch": 0.2762675756284619,
+      "grad_norm": 15.625,
+      "learning_rate": 2.390147849149139e-06,
+      "loss": 0.6001,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27635279079676184,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3900789565180632e-06,
+      "loss": 0.8388,
+      "step": 3243
+    },
+    {
+      "epoch": 0.2764380059650618,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.390010043284615e-06,
+      "loss": 0.587,
+      "step": 3244
+    },
+    {
+      "epoch": 0.27652322113336175,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3899411094500385e-06,
+      "loss": 0.5722,
+      "step": 3245
+    },
+    {
+      "epoch": 0.2766084363016617,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.38987215501558e-06,
+      "loss": 0.9183,
+      "step": 3246
+    },
+    {
+      "epoch": 0.27669365146996167,
+      "grad_norm": 13.0,
+      "learning_rate": 2.389803179982485e-06,
+      "loss": 0.6159,
+      "step": 3247
+    },
+    {
+      "epoch": 0.2767788666382616,
+      "grad_norm": 16.5,
+      "learning_rate": 2.389734184352001e-06,
+      "loss": 0.7564,
+      "step": 3248
+    },
+    {
+      "epoch": 0.2768640818065616,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.389665168125374e-06,
+      "loss": 0.7234,
+      "step": 3249
+    },
+    {
+      "epoch": 0.27694929697486154,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.3895961313038515e-06,
+      "loss": 0.4677,
+      "step": 3250
+    },
+    {
+      "epoch": 0.2770345121431615,
+      "grad_norm": 16.625,
+      "learning_rate": 2.3895270738886807e-06,
+      "loss": 1.1617,
+      "step": 3251
+    },
+    {
+      "epoch": 0.27711972731146145,
+      "grad_norm": 17.625,
+      "learning_rate": 2.38945799588111e-06,
+      "loss": 0.8164,
+      "step": 3252
+    },
+    {
+      "epoch": 0.2772049424797614,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.389388897282387e-06,
+      "loss": 0.5967,
+      "step": 3253
+    },
+    {
+      "epoch": 0.27729015764806136,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.389319778093761e-06,
+      "loss": 0.8305,
+      "step": 3254
+    },
+    {
+      "epoch": 0.2773753728163613,
+      "grad_norm": 19.375,
+      "learning_rate": 2.3892506383164816e-06,
+      "loss": 0.7634,
+      "step": 3255
+    },
+    {
+      "epoch": 0.2774605879846613,
+      "grad_norm": 23.25,
+      "learning_rate": 2.389181477951797e-06,
+      "loss": 1.2268,
+      "step": 3256
+    },
+    {
+      "epoch": 0.27754580315296123,
+      "grad_norm": 23.75,
+      "learning_rate": 2.3891122970009577e-06,
+      "loss": 1.534,
+      "step": 3257
+    },
+    {
+      "epoch": 0.2776310183212612,
+      "grad_norm": 15.75,
+      "learning_rate": 2.389043095465214e-06,
+      "loss": 0.8984,
+      "step": 3258
+    },
+    {
+      "epoch": 0.27771623348956115,
+      "grad_norm": 27.125,
+      "learning_rate": 2.3889738733458155e-06,
+      "loss": 0.8182,
+      "step": 3259
+    },
+    {
+      "epoch": 0.2778014486578611,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3889046306440146e-06,
+      "loss": 0.8357,
+      "step": 3260
+    },
+    {
+      "epoch": 0.27788666382616106,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3888353673610613e-06,
+      "loss": 0.8822,
+      "step": 3261
+    },
+    {
+      "epoch": 0.277971878994461,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3887660834982076e-06,
+      "loss": 0.6939,
+      "step": 3262
+    },
+    {
+      "epoch": 0.278057094162761,
+      "grad_norm": 11.375,
+      "learning_rate": 2.388696779056706e-06,
+      "loss": 0.6559,
+      "step": 3263
+    },
+    {
+      "epoch": 0.27814230933106093,
+      "grad_norm": 18.375,
+      "learning_rate": 2.3886274540378083e-06,
+      "loss": 0.5942,
+      "step": 3264
+    },
+    {
+      "epoch": 0.2782275244993609,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.388558108442768e-06,
+      "loss": 0.6218,
+      "step": 3265
+    },
+    {
+      "epoch": 0.27831273966766085,
+      "grad_norm": 13.125,
+      "learning_rate": 2.388488742272837e-06,
+      "loss": 0.5284,
+      "step": 3266
+    },
+    {
+      "epoch": 0.2783979548359608,
+      "grad_norm": 11.25,
+      "learning_rate": 2.38841935552927e-06,
+      "loss": 0.6733,
+      "step": 3267
+    },
+    {
+      "epoch": 0.27848317000426076,
+      "grad_norm": 15.5,
+      "learning_rate": 2.3883499482133206e-06,
+      "loss": 0.9241,
+      "step": 3268
+    },
+    {
+      "epoch": 0.2785683851725607,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3882805203262425e-06,
+      "loss": 0.7892,
+      "step": 3269
+    },
+    {
+      "epoch": 0.2786536003408607,
+      "grad_norm": 13.125,
+      "learning_rate": 2.388211071869291e-06,
+      "loss": 0.7357,
+      "step": 3270
+    },
+    {
+      "epoch": 0.27873881550916063,
+      "grad_norm": 15.5,
+      "learning_rate": 2.388141602843721e-06,
+      "loss": 0.8892,
+      "step": 3271
+    },
+    {
+      "epoch": 0.2788240306774606,
+      "grad_norm": 12.5,
+      "learning_rate": 2.388072113250788e-06,
+      "loss": 0.6221,
+      "step": 3272
+    },
+    {
+      "epoch": 0.27890924584576055,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.388002603091747e-06,
+      "loss": 0.6881,
+      "step": 3273
+    },
+    {
+      "epoch": 0.2789944610140605,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3879330723678547e-06,
+      "loss": 0.9123,
+      "step": 3274
+    },
+    {
+      "epoch": 0.27907967618236046,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.3878635210803675e-06,
+      "loss": 0.5552,
+      "step": 3275
+    },
+    {
+      "epoch": 0.2791648913506604,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3877939492305423e-06,
+      "loss": 0.7999,
+      "step": 3276
+    },
+    {
+      "epoch": 0.2792501065189604,
+      "grad_norm": 14.25,
+      "learning_rate": 2.3877243568196364e-06,
+      "loss": 0.5214,
+      "step": 3277
+    },
+    {
+      "epoch": 0.27933532168726033,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3876547438489073e-06,
+      "loss": 0.9477,
+      "step": 3278
+    },
+    {
+      "epoch": 0.2794205368555603,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.387585110319613e-06,
+      "loss": 0.5248,
+      "step": 3279
+    },
+    {
+      "epoch": 0.27950575202386024,
+      "grad_norm": 14.75,
+      "learning_rate": 2.3875154562330117e-06,
+      "loss": 0.5613,
+      "step": 3280
+    },
+    {
+      "epoch": 0.2795909671921602,
+      "grad_norm": 15.625,
+      "learning_rate": 2.3874457815903622e-06,
+      "loss": 0.664,
+      "step": 3281
+    },
+    {
+      "epoch": 0.27967618236046016,
+      "grad_norm": 10.875,
+      "learning_rate": 2.387376086392924e-06,
+      "loss": 0.3871,
+      "step": 3282
+    },
+    {
+      "epoch": 0.2797613975287601,
+      "grad_norm": 24.375,
+      "learning_rate": 2.3873063706419557e-06,
+      "loss": 1.2625,
+      "step": 3283
+    },
+    {
+      "epoch": 0.27984661269706007,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3872366343387178e-06,
+      "loss": 1.1833,
+      "step": 3284
+    },
+    {
+      "epoch": 0.27993182786536003,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3871668774844705e-06,
+      "loss": 0.769,
+      "step": 3285
+    },
+    {
+      "epoch": 0.28001704303366,
+      "grad_norm": 12.0,
+      "learning_rate": 2.3870971000804745e-06,
+      "loss": 0.5318,
+      "step": 3286
+    },
+    {
+      "epoch": 0.28010225820195994,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3870273021279896e-06,
+      "loss": 0.8197,
+      "step": 3287
+    },
+    {
+      "epoch": 0.2801874733702599,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3869574836282784e-06,
+      "loss": 0.8311,
+      "step": 3288
+    },
+    {
+      "epoch": 0.28027268853855986,
+      "grad_norm": 20.25,
+      "learning_rate": 2.3868876445826024e-06,
+      "loss": 1.1167,
+      "step": 3289
+    },
+    {
+      "epoch": 0.2803579037068598,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.386817784992223e-06,
+      "loss": 0.6433,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28044311887515977,
+      "grad_norm": 23.25,
+      "learning_rate": 2.3867479048584036e-06,
+      "loss": 1.0603,
+      "step": 3291
+    },
+    {
+      "epoch": 0.2805283340434597,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.386678004182406e-06,
+      "loss": 0.3652,
+      "step": 3292
+    },
+    {
+      "epoch": 0.2806135492117597,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3866080829654944e-06,
+      "loss": 0.5017,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28069876438005964,
+      "grad_norm": 19.125,
+      "learning_rate": 2.386538141208931e-06,
+      "loss": 0.6801,
+      "step": 3294
+    },
+    {
+      "epoch": 0.2807839795483596,
+      "grad_norm": 13.0,
+      "learning_rate": 2.386468178913981e-06,
+      "loss": 0.7981,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28086919471665955,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3863981960819085e-06,
+      "loss": 0.7691,
+      "step": 3296
+    },
+    {
+      "epoch": 0.2809544098849595,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3863281927139773e-06,
+      "loss": 0.8362,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28103962505325947,
+      "grad_norm": 18.5,
+      "learning_rate": 2.3862581688114535e-06,
+      "loss": 0.592,
+      "step": 3298
+    },
+    {
+      "epoch": 0.2811248402215594,
+      "grad_norm": 18.875,
+      "learning_rate": 2.3861881243756018e-06,
+      "loss": 0.9038,
+      "step": 3299
+    },
+    {
+      "epoch": 0.2812100553898594,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.386118059407688e-06,
+      "loss": 0.5841,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28129527055815934,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.386047973908978e-06,
+      "loss": 0.7435,
+      "step": 3301
+    },
+    {
+      "epoch": 0.2813804857264593,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3859778678807393e-06,
+      "loss": 0.7852,
+      "step": 3302
+    },
+    {
+      "epoch": 0.28146570089475925,
+      "grad_norm": 16.5,
+      "learning_rate": 2.385907741324238e-06,
+      "loss": 0.7224,
+      "step": 3303
+    },
+    {
+      "epoch": 0.2815509160630592,
+      "grad_norm": 16.25,
+      "learning_rate": 2.385837594240742e-06,
+      "loss": 0.8005,
+      "step": 3304
+    },
+    {
+      "epoch": 0.28163613123135917,
+      "grad_norm": 13.75,
+      "learning_rate": 2.385767426631518e-06,
+      "loss": 0.884,
+      "step": 3305
+    },
+    {
+      "epoch": 0.2817213463996591,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.385697238497835e-06,
+      "loss": 0.5361,
+      "step": 3306
+    },
+    {
+      "epoch": 0.2818065615679591,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3856270298409603e-06,
+      "loss": 0.6809,
+      "step": 3307
+    },
+    {
+      "epoch": 0.28189177673625904,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3855568006621634e-06,
+      "loss": 0.4543,
+      "step": 3308
+    },
+    {
+      "epoch": 0.281976991904559,
+      "grad_norm": 20.875,
+      "learning_rate": 2.3854865509627135e-06,
+      "loss": 1.0276,
+      "step": 3309
+    },
+    {
+      "epoch": 0.28206220707285895,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.38541628074388e-06,
+      "loss": 0.5687,
+      "step": 3310
+    },
+    {
+      "epoch": 0.2821474222411589,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.3853459900069317e-06,
+      "loss": 0.6423,
+      "step": 3311
+    },
+    {
+      "epoch": 0.28223263740945886,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.38527567875314e-06,
+      "loss": 0.7432,
+      "step": 3312
+    },
+    {
+      "epoch": 0.2823178525777588,
+      "grad_norm": 11.25,
+      "learning_rate": 2.3852053469837756e-06,
+      "loss": 0.6378,
+      "step": 3313
+    },
+    {
+      "epoch": 0.2824030677460588,
+      "grad_norm": 25.625,
+      "learning_rate": 2.3851349947001087e-06,
+      "loss": 1.0169,
+      "step": 3314
+    },
+    {
+      "epoch": 0.28248828291435873,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.3850646219034107e-06,
+      "loss": 0.5053,
+      "step": 3315
+    },
+    {
+      "epoch": 0.2825734980826587,
+      "grad_norm": 21.5,
+      "learning_rate": 2.384994228594954e-06,
+      "loss": 0.5648,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28265871325095865,
+      "grad_norm": 23.5,
+      "learning_rate": 2.38492381477601e-06,
+      "loss": 1.022,
+      "step": 3317
+    },
+    {
+      "epoch": 0.2827439284192586,
+      "grad_norm": 13.875,
+      "learning_rate": 2.384853380447852e-06,
+      "loss": 0.677,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28282914358755856,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.3847829256117516e-06,
+      "loss": 0.489,
+      "step": 3319
+    },
+    {
+      "epoch": 0.2829143587558585,
+      "grad_norm": 13.625,
+      "learning_rate": 2.384712450268983e-06,
+      "loss": 0.7879,
+      "step": 3320
+    },
+    {
+      "epoch": 0.2829995739241585,
+      "grad_norm": 18.25,
+      "learning_rate": 2.384641954420819e-06,
+      "loss": 1.1059,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28308478909245843,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3845714380685342e-06,
+      "loss": 0.2821,
+      "step": 3322
+    },
+    {
+      "epoch": 0.2831700042607584,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.3845009012134027e-06,
+      "loss": 0.4754,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28325521942905835,
+      "grad_norm": 32.5,
+      "learning_rate": 2.384430343856699e-06,
+      "loss": 1.4646,
+      "step": 3324
+    },
+    {
+      "epoch": 0.2833404345973583,
+      "grad_norm": 13.25,
+      "learning_rate": 2.384359765999698e-06,
+      "loss": 0.7753,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28342564976565826,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3842891676436756e-06,
+      "loss": 0.2746,
+      "step": 3326
+    },
+    {
+      "epoch": 0.2835108649339582,
+      "grad_norm": 11.625,
+      "learning_rate": 2.3842185487899073e-06,
+      "loss": 0.4905,
+      "step": 3327
+    },
+    {
+      "epoch": 0.2835960801022582,
+      "grad_norm": 15.375,
+      "learning_rate": 2.384147909439669e-06,
+      "loss": 0.6193,
+      "step": 3328
+    },
+    {
+      "epoch": 0.2836812952705582,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.3840772495942378e-06,
+      "loss": 0.7505,
+      "step": 3329
+    },
+    {
+      "epoch": 0.28376651043885814,
+      "grad_norm": 20.375,
+      "learning_rate": 2.3840065692548904e-06,
+      "loss": 0.8517,
+      "step": 3330
+    },
+    {
+      "epoch": 0.2838517256071581,
+      "grad_norm": 22.5,
+      "learning_rate": 2.3839358684229035e-06,
+      "loss": 0.904,
+      "step": 3331
+    },
+    {
+      "epoch": 0.28393694077545806,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3838651470995556e-06,
+      "loss": 0.5554,
+      "step": 3332
+    },
+    {
+      "epoch": 0.284022155943758,
+      "grad_norm": 14.625,
+      "learning_rate": 2.383794405286124e-06,
+      "loss": 0.6426,
+      "step": 3333
+    },
+    {
+      "epoch": 0.28410737111205797,
+      "grad_norm": 27.125,
+      "learning_rate": 2.3837236429838878e-06,
+      "loss": 0.8508,
+      "step": 3334
+    },
+    {
+      "epoch": 0.28419258628035793,
+      "grad_norm": 23.5,
+      "learning_rate": 2.383652860194125e-06,
+      "loss": 0.3894,
+      "step": 3335
+    },
+    {
+      "epoch": 0.2842778014486579,
+      "grad_norm": 13.0,
+      "learning_rate": 2.383582056918115e-06,
+      "loss": 0.5137,
+      "step": 3336
+    },
+    {
+      "epoch": 0.28436301661695784,
+      "grad_norm": 16.5,
+      "learning_rate": 2.3835112331571374e-06,
+      "loss": 0.7774,
+      "step": 3337
+    },
+    {
+      "epoch": 0.2844482317852578,
+      "grad_norm": 23.625,
+      "learning_rate": 2.383440388912472e-06,
+      "loss": 1.0577,
+      "step": 3338
+    },
+    {
+      "epoch": 0.28453344695355776,
+      "grad_norm": 22.0,
+      "learning_rate": 2.3833695241853993e-06,
+      "loss": 1.1389,
+      "step": 3339
+    },
+    {
+      "epoch": 0.2846186621218577,
+      "grad_norm": 14.875,
+      "learning_rate": 2.383298638977199e-06,
+      "loss": 1.037,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28470387729015767,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3832277332891534e-06,
+      "loss": 0.8874,
+      "step": 3341
+    },
+    {
+      "epoch": 0.2847890924584576,
+      "grad_norm": 13.5,
+      "learning_rate": 2.3831568071225424e-06,
+      "loss": 0.7628,
+      "step": 3342
+    },
+    {
+      "epoch": 0.2848743076267576,
+      "grad_norm": 27.25,
+      "learning_rate": 2.383085860478649e-06,
+      "loss": 0.9163,
+      "step": 3343
+    },
+    {
+      "epoch": 0.28495952279505754,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3830148933587545e-06,
+      "loss": 0.469,
+      "step": 3344
+    },
+    {
+      "epoch": 0.2850447379633575,
+      "grad_norm": 13.875,
+      "learning_rate": 2.3829439057641415e-06,
+      "loss": 0.4983,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28512995313165745,
+      "grad_norm": 16.25,
+      "learning_rate": 2.382872897696093e-06,
+      "loss": 0.5694,
+      "step": 3346
+    },
+    {
+      "epoch": 0.2852151682999574,
+      "grad_norm": 12.875,
+      "learning_rate": 2.3828018691558915e-06,
+      "loss": 0.589,
+      "step": 3347
+    },
+    {
+      "epoch": 0.28530038346825737,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3827308201448215e-06,
+      "loss": 1.1654,
+      "step": 3348
+    },
+    {
+      "epoch": 0.2853855986365573,
+      "grad_norm": 12.75,
+      "learning_rate": 2.382659750664166e-06,
+      "loss": 0.5626,
+      "step": 3349
+    },
+    {
+      "epoch": 0.2854708138048573,
+      "grad_norm": 15.0,
+      "learning_rate": 2.382588660715211e-06,
+      "loss": 0.9217,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28555602897315724,
+      "grad_norm": 20.875,
+      "learning_rate": 2.3825175502992394e-06,
+      "loss": 0.9546,
+      "step": 3351
+    },
+    {
+      "epoch": 0.2856412441414572,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3824464194175366e-06,
+      "loss": 0.3705,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28572645930975715,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3823752680713887e-06,
+      "loss": 0.4902,
+      "step": 3353
+    },
+    {
+      "epoch": 0.2858116744780571,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3823040962620805e-06,
+      "loss": 1.0502,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28589688964635707,
+      "grad_norm": 13.625,
+      "learning_rate": 2.382232903990899e-06,
+      "loss": 0.547,
+      "step": 3355
+    },
+    {
+      "epoch": 0.285982104814657,
+      "grad_norm": 19.0,
+      "learning_rate": 2.38216169125913e-06,
+      "loss": 1.1682,
+      "step": 3356
+    },
+    {
+      "epoch": 0.286067319982957,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.382090458068061e-06,
+      "loss": 0.8007,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28615253515125694,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3820192044189794e-06,
+      "loss": 0.952,
+      "step": 3358
+    },
+    {
+      "epoch": 0.2862377503195569,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.3819479303131722e-06,
+      "loss": 0.6862,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28632296548785685,
+      "grad_norm": 30.875,
+      "learning_rate": 2.3818766357519273e-06,
+      "loss": 0.8938,
+      "step": 3360
+    },
+    {
+      "epoch": 0.2864081806561568,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3818053207365337e-06,
+      "loss": 0.5579,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28649339582445676,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.38173398526828e-06,
+      "loss": 0.4603,
+      "step": 3362
+    },
+    {
+      "epoch": 0.2865786109927567,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.381662629348455e-06,
+      "loss": 1.006,
+      "step": 3363
+    },
+    {
+      "epoch": 0.2866638261610567,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.381591252978348e-06,
+      "loss": 0.8126,
+      "step": 3364
+    },
+    {
+      "epoch": 0.28674904132935664,
+      "grad_norm": 16.875,
+      "learning_rate": 2.381519856159249e-06,
+      "loss": 0.8469,
+      "step": 3365
+    },
+    {
+      "epoch": 0.2868342564976566,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3814484388924487e-06,
+      "loss": 0.4767,
+      "step": 3366
+    },
+    {
+      "epoch": 0.28691947166595655,
+      "grad_norm": 18.125,
+      "learning_rate": 2.3813770011792373e-06,
+      "loss": 0.7269,
+      "step": 3367
+    },
+    {
+      "epoch": 0.2870046868342565,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3813055430209063e-06,
+      "loss": 0.5826,
+      "step": 3368
+    },
+    {
+      "epoch": 0.28708990200255646,
+      "grad_norm": 21.875,
+      "learning_rate": 2.381234064418746e-06,
+      "loss": 1.1915,
+      "step": 3369
+    },
+    {
+      "epoch": 0.2871751171708564,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3811625653740487e-06,
+      "loss": 0.4676,
+      "step": 3370
+    },
+    {
+      "epoch": 0.2872603323391564,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3810910458881064e-06,
+      "loss": 0.6785,
+      "step": 3371
+    },
+    {
+      "epoch": 0.28734554750745633,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3810195059622117e-06,
+      "loss": 0.6019,
+      "step": 3372
+    },
+    {
+      "epoch": 0.2874307626757563,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.380947945597657e-06,
+      "loss": 0.7969,
+      "step": 3373
+    },
+    {
+      "epoch": 0.28751597784405625,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3808763647957355e-06,
+      "loss": 0.4861,
+      "step": 3374
+    },
+    {
+      "epoch": 0.2876011930123562,
+      "grad_norm": 14.125,
+      "learning_rate": 2.380804763557741e-06,
+      "loss": 0.87,
+      "step": 3375
+    },
+    {
+      "epoch": 0.28768640818065616,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3807331418849675e-06,
+      "loss": 0.4479,
+      "step": 3376
+    },
+    {
+      "epoch": 0.2877716233489561,
+      "grad_norm": 23.625,
+      "learning_rate": 2.3806614997787084e-06,
+      "loss": 0.8922,
+      "step": 3377
+    },
+    {
+      "epoch": 0.2878568385172561,
+      "grad_norm": 20.625,
+      "learning_rate": 2.3805898372402605e-06,
+      "loss": 0.7641,
+      "step": 3378
+    },
+    {
+      "epoch": 0.28794205368555603,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.380518154270916e-06,
+      "loss": 0.5588,
+      "step": 3379
+    },
+    {
+      "epoch": 0.288027268853856,
+      "grad_norm": 13.375,
+      "learning_rate": 2.380446450871972e-06,
+      "loss": 0.6506,
+      "step": 3380
+    },
+    {
+      "epoch": 0.28811248402215595,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.380374727044724e-06,
+      "loss": 0.6244,
+      "step": 3381
+    },
+    {
+      "epoch": 0.2881976991904559,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.380302982790468e-06,
+      "loss": 1.1131,
+      "step": 3382
+    },
+    {
+      "epoch": 0.28828291435875586,
+      "grad_norm": 18.0,
+      "learning_rate": 2.3802312181105e-06,
+      "loss": 1.1048,
+      "step": 3383
+    },
+    {
+      "epoch": 0.2883681295270558,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3801594330061183e-06,
+      "loss": 0.6149,
+      "step": 3384
+    },
+    {
+      "epoch": 0.2884533446953558,
+      "grad_norm": 28.375,
+      "learning_rate": 2.3800876274786185e-06,
+      "loss": 0.8047,
+      "step": 3385
+    },
+    {
+      "epoch": 0.28853855986365573,
+      "grad_norm": 8.6875,
+      "learning_rate": 2.380015801529299e-06,
+      "loss": 0.4653,
+      "step": 3386
+    },
+    {
+      "epoch": 0.2886237750319557,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.379943955159458e-06,
+      "loss": 0.7847,
+      "step": 3387
+    },
+    {
+      "epoch": 0.28870899020025564,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.379872088370393e-06,
+      "loss": 0.5854,
+      "step": 3388
+    },
+    {
+      "epoch": 0.2887942053685556,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3798002011634035e-06,
+      "loss": 0.7127,
+      "step": 3389
+    },
+    {
+      "epoch": 0.28887942053685556,
+      "grad_norm": 24.25,
+      "learning_rate": 2.379728293539788e-06,
+      "loss": 1.0345,
+      "step": 3390
+    },
+    {
+      "epoch": 0.2889646357051555,
+      "grad_norm": 11.875,
+      "learning_rate": 2.3796563655008466e-06,
+      "loss": 0.4436,
+      "step": 3391
+    },
+    {
+      "epoch": 0.28904985087345547,
+      "grad_norm": 23.75,
+      "learning_rate": 2.3795844170478783e-06,
+      "loss": 1.2227,
+      "step": 3392
+    },
+    {
+      "epoch": 0.28913506604175543,
+      "grad_norm": 17.375,
+      "learning_rate": 2.379512448182184e-06,
+      "loss": 0.9114,
+      "step": 3393
+    },
+    {
+      "epoch": 0.2892202812100554,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3794404589050636e-06,
+      "loss": 0.8593,
+      "step": 3394
+    },
+    {
+      "epoch": 0.28930549637835534,
+      "grad_norm": 14.75,
+      "learning_rate": 2.3793684492178186e-06,
+      "loss": 0.6833,
+      "step": 3395
+    },
+    {
+      "epoch": 0.2893907115466553,
+      "grad_norm": 13.875,
+      "learning_rate": 2.37929641912175e-06,
+      "loss": 0.9602,
+      "step": 3396
+    },
+    {
+      "epoch": 0.28947592671495526,
+      "grad_norm": 16.75,
+      "learning_rate": 2.37922436861816e-06,
+      "loss": 0.7795,
+      "step": 3397
+    },
+    {
+      "epoch": 0.2895611418832552,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.379152297708349e-06,
+      "loss": 0.998,
+      "step": 3398
+    },
+    {
+      "epoch": 0.28964635705155517,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3790802063936215e-06,
+      "loss": 0.496,
+      "step": 3399
+    },
+    {
+      "epoch": 0.2897315722198551,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.379008094675279e-06,
+      "loss": 0.448,
+      "step": 3400
+    },
+    {
+      "epoch": 0.2898167873881551,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.378935962554625e-06,
+      "loss": 0.5801,
+      "step": 3401
+    },
+    {
+      "epoch": 0.28990200255645504,
+      "grad_norm": 15.375,
+      "learning_rate": 2.378863810032963e-06,
+      "loss": 0.763,
+      "step": 3402
+    },
+    {
+      "epoch": 0.289987217724755,
+      "grad_norm": 16.5,
+      "learning_rate": 2.3787916371115966e-06,
+      "loss": 1.0671,
+      "step": 3403
+    },
+    {
+      "epoch": 0.29007243289305495,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3787194437918302e-06,
+      "loss": 0.6688,
+      "step": 3404
+    },
+    {
+      "epoch": 0.2901576480613549,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3786472300749688e-06,
+      "loss": 0.5871,
+      "step": 3405
+    },
+    {
+      "epoch": 0.29024286322965487,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3785749959623167e-06,
+      "loss": 0.5202,
+      "step": 3406
+    },
+    {
+      "epoch": 0.2903280783979548,
+      "grad_norm": 21.625,
+      "learning_rate": 2.3785027414551793e-06,
+      "loss": 1.0641,
+      "step": 3407
+    },
+    {
+      "epoch": 0.2904132935662548,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.378430466554863e-06,
+      "loss": 0.3786,
+      "step": 3408
+    },
+    {
+      "epoch": 0.29049850873455474,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3783581712626733e-06,
+      "loss": 0.6681,
+      "step": 3409
+    },
+    {
+      "epoch": 0.2905837239028547,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.3782858555799167e-06,
+      "loss": 0.834,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29066893907115465,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3782135195079004e-06,
+      "loss": 0.6867,
+      "step": 3411
+    },
+    {
+      "epoch": 0.2907541542394546,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3781411630479307e-06,
+      "loss": 0.8765,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29083936940775457,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3780687862013162e-06,
+      "loss": 0.7197,
+      "step": 3413
+    },
+    {
+      "epoch": 0.2909245845760545,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3779963889693644e-06,
+      "loss": 0.7068,
+      "step": 3414
+    },
+    {
+      "epoch": 0.2910097997443545,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.377923971353384e-06,
+      "loss": 0.619,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29109501491265444,
+      "grad_norm": 13.375,
+      "learning_rate": 2.377851533354682e-06,
+      "loss": 0.4641,
+      "step": 3416
+    },
+    {
+      "epoch": 0.2911802300809544,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3777790749745695e-06,
+      "loss": 0.5483,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29126544524925435,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3777065962143547e-06,
+      "loss": 0.617,
+      "step": 3418
+    },
+    {
+      "epoch": 0.2913506604175543,
+      "grad_norm": 15.125,
+      "learning_rate": 2.377634097075348e-06,
+      "loss": 0.8582,
+      "step": 3419
+    },
+    {
+      "epoch": 0.29143587558585426,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3775615775588594e-06,
+      "loss": 0.4053,
+      "step": 3420
+    },
+    {
+      "epoch": 0.2915210907541542,
+      "grad_norm": 20.25,
+      "learning_rate": 2.377489037666199e-06,
+      "loss": 1.0821,
+      "step": 3421
+    },
+    {
+      "epoch": 0.2916063059224542,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3774164773986775e-06,
+      "loss": 1.035,
+      "step": 3422
+    },
+    {
+      "epoch": 0.29169152109075414,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3773438967576075e-06,
+      "loss": 0.4144,
+      "step": 3423
+    },
+    {
+      "epoch": 0.2917767362590541,
+      "grad_norm": 13.375,
+      "learning_rate": 2.377271295744299e-06,
+      "loss": 0.6282,
+      "step": 3424
+    },
+    {
+      "epoch": 0.29186195142735405,
+      "grad_norm": 20.25,
+      "learning_rate": 2.3771986743600643e-06,
+      "loss": 1.0134,
+      "step": 3425
+    },
+    {
+      "epoch": 0.291947166595654,
+      "grad_norm": 14.5,
+      "learning_rate": 2.377126032606217e-06,
+      "loss": 0.753,
+      "step": 3426
+    },
+    {
+      "epoch": 0.29203238176395396,
+      "grad_norm": 21.875,
+      "learning_rate": 2.377053370484068e-06,
+      "loss": 1.1791,
+      "step": 3427
+    },
+    {
+      "epoch": 0.2921175969322539,
+      "grad_norm": 19.75,
+      "learning_rate": 2.376980687994932e-06,
+      "loss": 0.9957,
+      "step": 3428
+    },
+    {
+      "epoch": 0.2922028121005539,
+      "grad_norm": 16.5,
+      "learning_rate": 2.376907985140121e-06,
+      "loss": 1.0694,
+      "step": 3429
+    },
+    {
+      "epoch": 0.29228802726885383,
+      "grad_norm": 16.25,
+      "learning_rate": 2.37683526192095e-06,
+      "loss": 1.0766,
+      "step": 3430
+    },
+    {
+      "epoch": 0.2923732424371538,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3767625183387323e-06,
+      "loss": 0.79,
+      "step": 3431
+    },
+    {
+      "epoch": 0.29245845760545375,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.376689754394783e-06,
+      "loss": 0.7916,
+      "step": 3432
+    },
+    {
+      "epoch": 0.2925436727737537,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3766169700904176e-06,
+      "loss": 0.888,
+      "step": 3433
+    },
+    {
+      "epoch": 0.29262888794205366,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3765441654269495e-06,
+      "loss": 0.8489,
+      "step": 3434
+    },
+    {
+      "epoch": 0.2927141031103536,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.376471340405696e-06,
+      "loss": 0.7272,
+      "step": 3435
+    },
+    {
+      "epoch": 0.2927993182786536,
+      "grad_norm": 21.0,
+      "learning_rate": 2.376398495027973e-06,
+      "loss": 0.9728,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29288453344695353,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.376325629295096e-06,
+      "loss": 0.3709,
+      "step": 3437
+    },
+    {
+      "epoch": 0.2929697486152535,
+      "grad_norm": 16.625,
+      "learning_rate": 2.376252743208382e-06,
+      "loss": 0.8203,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29305496378355345,
+      "grad_norm": 14.0,
+      "learning_rate": 2.376179836769149e-06,
+      "loss": 0.8175,
+      "step": 3439
+    },
+    {
+      "epoch": 0.2931401789518534,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.376106909978714e-06,
+      "loss": 0.7277,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29322539412015336,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3760339628383944e-06,
+      "loss": 0.7932,
+      "step": 3441
+    },
+    {
+      "epoch": 0.29331060928845337,
+      "grad_norm": 19.375,
+      "learning_rate": 2.375960995349509e-06,
+      "loss": 0.3535,
+      "step": 3442
+    },
+    {
+      "epoch": 0.29339582445675333,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3758880075133754e-06,
+      "loss": 0.7799,
+      "step": 3443
+    },
+    {
+      "epoch": 0.2934810396250533,
+      "grad_norm": 13.375,
+      "learning_rate": 2.3758149993313145e-06,
+      "loss": 0.6711,
+      "step": 3444
+    },
+    {
+      "epoch": 0.29356625479335324,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.3757419708046436e-06,
+      "loss": 0.5204,
+      "step": 3445
+    },
+    {
+      "epoch": 0.2936514699616532,
+      "grad_norm": 18.375,
+      "learning_rate": 2.3756689219346833e-06,
+      "loss": 0.8879,
+      "step": 3446
+    },
+    {
+      "epoch": 0.29373668512995316,
+      "grad_norm": 13.5,
+      "learning_rate": 2.375595852722754e-06,
+      "loss": 0.7165,
+      "step": 3447
+    },
+    {
+      "epoch": 0.2938219002982531,
+      "grad_norm": 11.125,
+      "learning_rate": 2.375522763170176e-06,
+      "loss": 0.4395,
+      "step": 3448
+    },
+    {
+      "epoch": 0.29390711546655307,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3754496532782696e-06,
+      "loss": 0.9505,
+      "step": 3449
+    },
+    {
+      "epoch": 0.293992330634853,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.375376523048356e-06,
+      "loss": 0.635,
+      "step": 3450
+    },
+    {
+      "epoch": 0.294077545803153,
+      "grad_norm": 13.5,
+      "learning_rate": 2.375303372481757e-06,
+      "loss": 0.6056,
+      "step": 3451
+    },
+    {
+      "epoch": 0.29416276097145294,
+      "grad_norm": 15.75,
+      "learning_rate": 2.3752302015797944e-06,
+      "loss": 0.7424,
+      "step": 3452
+    },
+    {
+      "epoch": 0.2942479761397529,
+      "grad_norm": 16.375,
+      "learning_rate": 2.375157010343791e-06,
+      "loss": 1.001,
+      "step": 3453
+    },
+    {
+      "epoch": 0.29433319130805286,
+      "grad_norm": 13.375,
+      "learning_rate": 2.375083798775069e-06,
+      "loss": 0.3734,
+      "step": 3454
+    },
+    {
+      "epoch": 0.2944184064763528,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3750105668749513e-06,
+      "loss": 0.956,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29450362164465277,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.374937314644761e-06,
+      "loss": 0.4849,
+      "step": 3456
+    },
+    {
+      "epoch": 0.2945888368129527,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3748640420858228e-06,
+      "loss": 0.5476,
+      "step": 3457
+    },
+    {
+      "epoch": 0.2946740519812527,
+      "grad_norm": 18.5,
+      "learning_rate": 2.3747907491994597e-06,
+      "loss": 0.8018,
+      "step": 3458
+    },
+    {
+      "epoch": 0.29475926714955264,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.374717435986997e-06,
+      "loss": 0.6866,
+      "step": 3459
+    },
+    {
+      "epoch": 0.2948444823178526,
+      "grad_norm": 11.5,
+      "learning_rate": 2.3746441024497586e-06,
+      "loss": 0.5413,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29492969748615255,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.374570748589071e-06,
+      "loss": 0.6204,
+      "step": 3461
+    },
+    {
+      "epoch": 0.2950149126544525,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.374497374406259e-06,
+      "loss": 0.2954,
+      "step": 3462
+    },
+    {
+      "epoch": 0.29510012782275247,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3744239799026486e-06,
+      "loss": 0.8429,
+      "step": 3463
+    },
+    {
+      "epoch": 0.2951853429910524,
+      "grad_norm": 15.125,
+      "learning_rate": 2.3743505650795663e-06,
+      "loss": 0.8303,
+      "step": 3464
+    },
+    {
+      "epoch": 0.2952705581593524,
+      "grad_norm": 17.75,
+      "learning_rate": 2.3742771299383387e-06,
+      "loss": 1.0318,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29535577332765234,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3742036744802927e-06,
+      "loss": 0.5929,
+      "step": 3466
+    },
+    {
+      "epoch": 0.2954409884959523,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3741301987067557e-06,
+      "loss": 0.8577,
+      "step": 3467
+    },
+    {
+      "epoch": 0.29552620366425225,
+      "grad_norm": 17.625,
+      "learning_rate": 2.374056702619056e-06,
+      "loss": 1.2268,
+      "step": 3468
+    },
+    {
+      "epoch": 0.2956114188325522,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.373983186218521e-06,
+      "loss": 1.0293,
+      "step": 3469
+    },
+    {
+      "epoch": 0.29569663400085217,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.37390964950648e-06,
+      "loss": 0.6229,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2957818491691521,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.373836092484261e-06,
+      "loss": 0.5964,
+      "step": 3471
+    },
+    {
+      "epoch": 0.2958670643374521,
+      "grad_norm": 11.375,
+      "learning_rate": 2.3737625151531937e-06,
+      "loss": 0.4489,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29595227950575204,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.373688917514608e-06,
+      "loss": 0.6087,
+      "step": 3473
+    },
+    {
+      "epoch": 0.296037494674052,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3736152995698335e-06,
+      "loss": 0.8158,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29612270984235195,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3735416613202005e-06,
+      "loss": 0.8275,
+      "step": 3475
+    },
+    {
+      "epoch": 0.2962079250106519,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.37346800276704e-06,
+      "loss": 0.6294,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29629314017895186,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.373394323911683e-06,
+      "loss": 0.8348,
+      "step": 3477
+    },
+    {
+      "epoch": 0.2963783553472518,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3733206247554607e-06,
+      "loss": 0.8657,
+      "step": 3478
+    },
+    {
+      "epoch": 0.2964635705155518,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3732469052997055e-06,
+      "loss": 1.0255,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29654878568385173,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3731731655457492e-06,
+      "loss": 0.646,
+      "step": 3480
+    },
+    {
+      "epoch": 0.2966340008521517,
+      "grad_norm": 20.875,
+      "learning_rate": 2.373099405494924e-06,
+      "loss": 1.4531,
+      "step": 3481
+    },
+    {
+      "epoch": 0.29671921602045165,
+      "grad_norm": 23.25,
+      "learning_rate": 2.3730256251485633e-06,
+      "loss": 1.1542,
+      "step": 3482
+    },
+    {
+      "epoch": 0.2968044311887516,
+      "grad_norm": 19.0,
+      "learning_rate": 2.372951824508e-06,
+      "loss": 1.0287,
+      "step": 3483
+    },
+    {
+      "epoch": 0.29688964635705156,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3728780035745684e-06,
+      "loss": 0.9181,
+      "step": 3484
+    },
+    {
+      "epoch": 0.2969748615253515,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.3728041623496023e-06,
+      "loss": 0.7606,
+      "step": 3485
+    },
+    {
+      "epoch": 0.2970600766936515,
+      "grad_norm": 19.0,
+      "learning_rate": 2.372730300834436e-06,
+      "loss": 0.9599,
+      "step": 3486
+    },
+    {
+      "epoch": 0.29714529186195143,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3726564190304036e-06,
+      "loss": 0.6571,
+      "step": 3487
+    },
+    {
+      "epoch": 0.2972305070302514,
+      "grad_norm": 17.125,
+      "learning_rate": 2.372582516938841e-06,
+      "loss": 0.8738,
+      "step": 3488
+    },
+    {
+      "epoch": 0.29731572219855135,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3725085945610833e-06,
+      "loss": 0.5681,
+      "step": 3489
+    },
+    {
+      "epoch": 0.2974009373668513,
+      "grad_norm": 16.75,
+      "learning_rate": 2.372434651898467e-06,
+      "loss": 0.9971,
+      "step": 3490
+    },
+    {
+      "epoch": 0.29748615253515126,
+      "grad_norm": 11.0,
+      "learning_rate": 2.3723606889523277e-06,
+      "loss": 0.5215,
+      "step": 3491
+    },
+    {
+      "epoch": 0.2975713677034512,
+      "grad_norm": 11.75,
+      "learning_rate": 2.372286705724002e-06,
+      "loss": 0.428,
+      "step": 3492
+    },
+    {
+      "epoch": 0.2976565828717512,
+      "grad_norm": 16.625,
+      "learning_rate": 2.3722127022148267e-06,
+      "loss": 0.9708,
+      "step": 3493
+    },
+    {
+      "epoch": 0.29774179804005113,
+      "grad_norm": 21.0,
+      "learning_rate": 2.3721386784261398e-06,
+      "loss": 0.6979,
+      "step": 3494
+    },
+    {
+      "epoch": 0.2978270132083511,
+      "grad_norm": 11.75,
+      "learning_rate": 2.3720646343592783e-06,
+      "loss": 0.5514,
+      "step": 3495
+    },
+    {
+      "epoch": 0.29791222837665104,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.371990570015581e-06,
+      "loss": 1.0081,
+      "step": 3496
+    },
+    {
+      "epoch": 0.297997443544951,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3719164853963857e-06,
+      "loss": 0.5804,
+      "step": 3497
+    },
+    {
+      "epoch": 0.29808265871325096,
+      "grad_norm": 21.375,
+      "learning_rate": 2.371842380503031e-06,
+      "loss": 0.7738,
+      "step": 3498
+    },
+    {
+      "epoch": 0.2981678738815509,
+      "grad_norm": 15.75,
+      "learning_rate": 2.3717682553368566e-06,
+      "loss": 0.4976,
+      "step": 3499
+    },
+    {
+      "epoch": 0.2982530890498509,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.371694109899202e-06,
+      "loss": 0.9709,
+      "step": 3500
+    },
+    {
+      "epoch": 0.29833830421815083,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3716199441914064e-06,
+      "loss": 0.9336,
+      "step": 3501
+    },
+    {
+      "epoch": 0.2984235193864508,
+      "grad_norm": 15.125,
+      "learning_rate": 2.371545758214811e-06,
+      "loss": 0.7563,
+      "step": 3502
+    },
+    {
+      "epoch": 0.29850873455475074,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3714715519707556e-06,
+      "loss": 0.4922,
+      "step": 3503
+    },
+    {
+      "epoch": 0.2985939497230507,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3713973254605816e-06,
+      "loss": 0.3999,
+      "step": 3504
+    },
+    {
+      "epoch": 0.29867916489135066,
+      "grad_norm": 15.0,
+      "learning_rate": 2.3713230786856305e-06,
+      "loss": 0.845,
+      "step": 3505
+    },
+    {
+      "epoch": 0.2987643800596506,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3712488116472437e-06,
+      "loss": 0.7819,
+      "step": 3506
+    },
+    {
+      "epoch": 0.29884959522795057,
+      "grad_norm": 17.875,
+      "learning_rate": 2.371174524346763e-06,
+      "loss": 0.9408,
+      "step": 3507
+    },
+    {
+      "epoch": 0.2989348103962505,
+      "grad_norm": 17.875,
+      "learning_rate": 2.371100216785532e-06,
+      "loss": 1.0446,
+      "step": 3508
+    },
+    {
+      "epoch": 0.2990200255645505,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.371025888964892e-06,
+      "loss": 0.6994,
+      "step": 3509
+    },
+    {
+      "epoch": 0.29910524073285044,
+      "grad_norm": 20.75,
+      "learning_rate": 2.3709515408861873e-06,
+      "loss": 1.1759,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2991904559011504,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3708771725507605e-06,
+      "loss": 0.7445,
+      "step": 3511
+    },
+    {
+      "epoch": 0.29927567106945036,
+      "grad_norm": 17.75,
+      "learning_rate": 2.3708027839599567e-06,
+      "loss": 0.8531,
+      "step": 3512
+    },
+    {
+      "epoch": 0.2993608862377503,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3707283751151195e-06,
+      "loss": 0.45,
+      "step": 3513
+    },
+    {
+      "epoch": 0.29944610140605027,
+      "grad_norm": 9.75,
+      "learning_rate": 2.3706539460175933e-06,
+      "loss": 0.373,
+      "step": 3514
+    },
+    {
+      "epoch": 0.2995313165743502,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.3705794966687236e-06,
+      "loss": 0.4872,
+      "step": 3515
+    },
+    {
+      "epoch": 0.2996165317426502,
+      "grad_norm": 15.625,
+      "learning_rate": 2.3705050270698554e-06,
+      "loss": 0.8585,
+      "step": 3516
+    },
+    {
+      "epoch": 0.29970174691095014,
+      "grad_norm": 12.625,
+      "learning_rate": 2.3704305372223346e-06,
+      "loss": 0.5092,
+      "step": 3517
+    },
+    {
+      "epoch": 0.2997869620792501,
+      "grad_norm": 18.625,
+      "learning_rate": 2.3703560271275073e-06,
+      "loss": 1.1137,
+      "step": 3518
+    },
+    {
+      "epoch": 0.29987217724755005,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.37028149678672e-06,
+      "loss": 0.4338,
+      "step": 3519
+    },
+    {
+      "epoch": 0.29995739241585,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3702069462013196e-06,
+      "loss": 0.4063,
+      "step": 3520
+    },
+    {
+      "epoch": 0.30004260758414997,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3701323753726536e-06,
+      "loss": 0.8559,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3001278227524499,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3700577843020686e-06,
+      "loss": 1.0492,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3002130379207499,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.3699831729909135e-06,
+      "loss": 0.8995,
+      "step": 3523
+    },
+    {
+      "epoch": 0.30029825308904984,
+      "grad_norm": 11.125,
+      "learning_rate": 2.369908541440536e-06,
+      "loss": 0.5174,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3003834682573498,
+      "grad_norm": 11.625,
+      "learning_rate": 2.369833889652285e-06,
+      "loss": 0.4916,
+      "step": 3525
+    },
+    {
+      "epoch": 0.30046868342564975,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.36975921762751e-06,
+      "loss": 0.8715,
+      "step": 3526
+    },
+    {
+      "epoch": 0.3005538985939497,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.369684525367559e-06,
+      "loss": 0.441,
+      "step": 3527
+    },
+    {
+      "epoch": 0.30063911376224967,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3696098128737833e-06,
+      "loss": 0.4105,
+      "step": 3528
+    },
+    {
+      "epoch": 0.3007243289305496,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3695350801475325e-06,
+      "loss": 0.6402,
+      "step": 3529
+    },
+    {
+      "epoch": 0.3008095440988496,
+      "grad_norm": 14.875,
+      "learning_rate": 2.369460327190157e-06,
+      "loss": 0.8228,
+      "step": 3530
+    },
+    {
+      "epoch": 0.30089475926714954,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3693855540030074e-06,
+      "loss": 0.6175,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3009799744354495,
+      "grad_norm": 46.5,
+      "learning_rate": 2.3693107605874355e-06,
+      "loss": 0.6063,
+      "step": 3532
+    },
+    {
+      "epoch": 0.30106518960374945,
+      "grad_norm": 21.125,
+      "learning_rate": 2.3692359469447923e-06,
+      "loss": 1.1743,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3011504047720494,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3691611130764303e-06,
+      "loss": 0.9115,
+      "step": 3534
+    },
+    {
+      "epoch": 0.30123561994034936,
+      "grad_norm": 12.875,
+      "learning_rate": 2.3690862589837015e-06,
+      "loss": 0.5827,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3013208351086493,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3690113846679586e-06,
+      "loss": 0.9201,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3014060502769493,
+      "grad_norm": 21.125,
+      "learning_rate": 2.368936490130555e-06,
+      "loss": 1.3266,
+      "step": 3537
+    },
+    {
+      "epoch": 0.30149126544524923,
+      "grad_norm": 17.75,
+      "learning_rate": 2.3688615753728435e-06,
+      "loss": 0.9453,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3015764806135492,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3687866403961784e-06,
+      "loss": 0.6202,
+      "step": 3539
+    },
+    {
+      "epoch": 0.30166169578184915,
+      "grad_norm": 13.25,
+      "learning_rate": 2.368711685201914e-06,
+      "loss": 0.553,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3017469109501491,
+      "grad_norm": 16.625,
+      "learning_rate": 2.3686367097914038e-06,
+      "loss": 0.5821,
+      "step": 3541
+    },
+    {
+      "epoch": 0.30183212611844906,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.368561714166004e-06,
+      "loss": 0.5321,
+      "step": 3542
+    },
+    {
+      "epoch": 0.301917341286749,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.3684866983270687e-06,
+      "loss": 0.4729,
+      "step": 3543
+    },
+    {
+      "epoch": 0.302002556455049,
+      "grad_norm": 13.5,
+      "learning_rate": 2.3684116622759547e-06,
+      "loss": 0.6648,
+      "step": 3544
+    },
+    {
+      "epoch": 0.30208777162334893,
+      "grad_norm": 21.125,
+      "learning_rate": 2.3683366060140166e-06,
+      "loss": 0.8463,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3021729867916489,
+      "grad_norm": 14.25,
+      "learning_rate": 2.368261529542612e-06,
+      "loss": 0.7165,
+      "step": 3546
+    },
+    {
+      "epoch": 0.30225820195994885,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.368186432863097e-06,
+      "loss": 0.597,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3023434171282488,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.3681113159768284e-06,
+      "loss": 0.7239,
+      "step": 3548
+    },
+    {
+      "epoch": 0.30242863229654876,
+      "grad_norm": 27.875,
+      "learning_rate": 2.368036178885164e-06,
+      "loss": 1.1407,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3025138474648487,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.367961021589462e-06,
+      "loss": 0.4478,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3025990626331487,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3678858440910795e-06,
+      "loss": 1.024,
+      "step": 3551
+    },
+    {
+      "epoch": 0.30268427780144863,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.3678106463913765e-06,
+      "loss": 0.3483,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3027694929697486,
+      "grad_norm": 14.75,
+      "learning_rate": 2.3677354284917102e-06,
+      "loss": 0.7777,
+      "step": 3553
+    },
+    {
+      "epoch": 0.30285470813804855,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3676601903934417e-06,
+      "loss": 1.0319,
+      "step": 3554
+    },
+    {
+      "epoch": 0.30293992330634856,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.367584932097929e-06,
+      "loss": 0.8394,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3030251384746485,
+      "grad_norm": 16.625,
+      "learning_rate": 2.3675096536065327e-06,
+      "loss": 0.7785,
+      "step": 3556
+    },
+    {
+      "epoch": 0.30311035364294847,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3674343549206137e-06,
+      "loss": 0.7054,
+      "step": 3557
+    },
+    {
+      "epoch": 0.30319556881124843,
+      "grad_norm": 12.625,
+      "learning_rate": 2.3673590360415318e-06,
+      "loss": 0.5428,
+      "step": 3558
+    },
+    {
+      "epoch": 0.3032807839795484,
+      "grad_norm": 18.125,
+      "learning_rate": 2.3672836969706484e-06,
+      "loss": 1.0269,
+      "step": 3559
+    },
+    {
+      "epoch": 0.30336599914784834,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.3672083377093253e-06,
+      "loss": 0.7792,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3034512143161483,
+      "grad_norm": 16.25,
+      "learning_rate": 2.367132958258924e-06,
+      "loss": 0.8149,
+      "step": 3561
+    },
+    {
+      "epoch": 0.30353642948444826,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.367057558620807e-06,
+      "loss": 1.015,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3036216446527482,
+      "grad_norm": 14.625,
+      "learning_rate": 2.366982138796337e-06,
+      "loss": 0.9353,
+      "step": 3563
+    },
+    {
+      "epoch": 0.30370685982104817,
+      "grad_norm": 21.125,
+      "learning_rate": 2.366906698786876e-06,
+      "loss": 0.5762,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3037920749893481,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.366831238593788e-06,
+      "loss": 1.0013,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3038772901576481,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.366755758218436e-06,
+      "loss": 0.3632,
+      "step": 3566
+    },
+    {
+      "epoch": 0.30396250532594804,
+      "grad_norm": 15.875,
+      "learning_rate": 2.366680257662185e-06,
+      "loss": 0.9475,
+      "step": 3567
+    },
+    {
+      "epoch": 0.304047720494248,
+      "grad_norm": 13.0,
+      "learning_rate": 2.366604736926399e-06,
+      "loss": 0.5595,
+      "step": 3568
+    },
+    {
+      "epoch": 0.30413293566254795,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3665291960124422e-06,
+      "loss": 0.5741,
+      "step": 3569
+    },
+    {
+      "epoch": 0.3042181508308479,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3664536349216797e-06,
+      "loss": 0.4291,
+      "step": 3570
+    },
+    {
+      "epoch": 0.30430336599914787,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.366378053655478e-06,
+      "loss": 0.7145,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3043885811674478,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.3663024522152015e-06,
+      "loss": 0.8785,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3044737963357478,
+      "grad_norm": 10.125,
+      "learning_rate": 2.366226830602218e-06,
+      "loss": 0.3718,
+      "step": 3573
+    },
+    {
+      "epoch": 0.30455901150404774,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3661511888178928e-06,
+      "loss": 0.5046,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3046442266723477,
+      "grad_norm": 21.875,
+      "learning_rate": 2.3660755268635933e-06,
+      "loss": 0.8949,
+      "step": 3575
+    },
+    {
+      "epoch": 0.30472944184064765,
+      "grad_norm": 17.875,
+      "learning_rate": 2.365999844740687e-06,
+      "loss": 1.0063,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3048146570089476,
+      "grad_norm": 14.0,
+      "learning_rate": 2.365924142450541e-06,
+      "loss": 0.7833,
+      "step": 3577
+    },
+    {
+      "epoch": 0.30489987217724757,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3658484199945237e-06,
+      "loss": 0.6318,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3049850873455475,
+      "grad_norm": 16.625,
+      "learning_rate": 2.365772677374003e-06,
+      "loss": 0.8016,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3050703025138475,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3656969145903486e-06,
+      "loss": 0.9323,
+      "step": 3580
+    },
+    {
+      "epoch": 0.30515551768214744,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3656211316449287e-06,
+      "loss": 1.0339,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3052407328504474,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3655453285391133e-06,
+      "loss": 0.6346,
+      "step": 3582
+    },
+    {
+      "epoch": 0.30532594801874735,
+      "grad_norm": 21.125,
+      "learning_rate": 2.3654695052742717e-06,
+      "loss": 0.9334,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3054111631870473,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3653936618517747e-06,
+      "loss": 0.9575,
+      "step": 3584
+    },
+    {
+      "epoch": 0.30549637835534726,
+      "grad_norm": 13.625,
+      "learning_rate": 2.365317798272993e-06,
+      "loss": 0.615,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3055815935236472,
+      "grad_norm": 12.75,
+      "learning_rate": 2.365241914539296e-06,
+      "loss": 0.6607,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3056668086919472,
+      "grad_norm": 15.375,
+      "learning_rate": 2.3651660106520574e-06,
+      "loss": 0.6869,
+      "step": 3587
+    },
+    {
+      "epoch": 0.30575202386024714,
+      "grad_norm": 15.75,
+      "learning_rate": 2.365090086612647e-06,
+      "loss": 0.7421,
+      "step": 3588
+    },
+    {
+      "epoch": 0.3058372390285471,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.365014142422438e-06,
+      "loss": 0.6018,
+      "step": 3589
+    },
+    {
+      "epoch": 0.30592245419684705,
+      "grad_norm": 15.75,
+      "learning_rate": 2.364938178082801e-06,
+      "loss": 0.9748,
+      "step": 3590
+    },
+    {
+      "epoch": 0.306007669365147,
+      "grad_norm": 16.75,
+      "learning_rate": 2.364862193595111e-06,
+      "loss": 0.93,
+      "step": 3591
+    },
+    {
+      "epoch": 0.30609288453344696,
+      "grad_norm": 12.5,
+      "learning_rate": 2.36478618896074e-06,
+      "loss": 0.6869,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3061780997017469,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.364710164181061e-06,
+      "loss": 0.691,
+      "step": 3593
+    },
+    {
+      "epoch": 0.3062633148700469,
+      "grad_norm": 25.75,
+      "learning_rate": 2.364634119257449e-06,
+      "loss": 1.1043,
+      "step": 3594
+    },
+    {
+      "epoch": 0.30634853003834683,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.3645580541912773e-06,
+      "loss": 0.6804,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3064337452066468,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3644819689839205e-06,
+      "loss": 0.9131,
+      "step": 3596
+    },
+    {
+      "epoch": 0.30651896037494675,
+      "grad_norm": 11.375,
+      "learning_rate": 2.3644058636367544e-06,
+      "loss": 0.5796,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3066041755432467,
+      "grad_norm": 11.5,
+      "learning_rate": 2.364329738151153e-06,
+      "loss": 0.5598,
+      "step": 3598
+    },
+    {
+      "epoch": 0.30668939071154666,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3642535925284937e-06,
+      "loss": 0.78,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3067746058798466,
+      "grad_norm": 10.125,
+      "learning_rate": 2.3641774267701505e-06,
+      "loss": 0.4551,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3068598210481466,
+      "grad_norm": 12.375,
+      "learning_rate": 2.3641012408775018e-06,
+      "loss": 0.4729,
+      "step": 3601
+    },
+    {
+      "epoch": 0.30694503621644653,
+      "grad_norm": 15.125,
+      "learning_rate": 2.364025034851923e-06,
+      "loss": 0.8796,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3070302513847465,
+      "grad_norm": 17.625,
+      "learning_rate": 2.363948808694791e-06,
+      "loss": 0.7959,
+      "step": 3603
+    },
+    {
+      "epoch": 0.30711546655304645,
+      "grad_norm": 10.875,
+      "learning_rate": 2.3638725624074844e-06,
+      "loss": 0.9934,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3072006817213464,
+      "grad_norm": 21.875,
+      "learning_rate": 2.36379629599138e-06,
+      "loss": 0.9451,
+      "step": 3605
+    },
+    {
+      "epoch": 0.30728589688964636,
+      "grad_norm": 20.875,
+      "learning_rate": 2.363720009447857e-06,
+      "loss": 1.09,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3073711120579463,
+      "grad_norm": 12.5,
+      "learning_rate": 2.3636437027782937e-06,
+      "loss": 0.6467,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3074563272262463,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.3635673759840685e-06,
+      "loss": 0.6728,
+      "step": 3608
+    },
+    {
+      "epoch": 0.30754154239454623,
+      "grad_norm": 23.5,
+      "learning_rate": 2.3634910290665614e-06,
+      "loss": 1.1796,
+      "step": 3609
+    },
+    {
+      "epoch": 0.3076267575628462,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.363414662027151e-06,
+      "loss": 1.0791,
+      "step": 3610
+    },
+    {
+      "epoch": 0.30771197273114614,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3633382748672186e-06,
+      "loss": 0.5334,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3077971878994461,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.363261867588144e-06,
+      "loss": 0.7565,
+      "step": 3612
+    },
+    {
+      "epoch": 0.30788240306774606,
+      "grad_norm": 15.375,
+      "learning_rate": 2.3631854401913077e-06,
+      "loss": 0.8815,
+      "step": 3613
+    },
+    {
+      "epoch": 0.307967618236046,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3631089926780913e-06,
+      "loss": 1.0213,
+      "step": 3614
+    },
+    {
+      "epoch": 0.30805283340434597,
+      "grad_norm": 18.125,
+      "learning_rate": 2.363032525049876e-06,
+      "loss": 0.7114,
+      "step": 3615
+    },
+    {
+      "epoch": 0.30813804857264593,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3629560373080436e-06,
+      "loss": 0.8506,
+      "step": 3616
+    },
+    {
+      "epoch": 0.3082232637409459,
+      "grad_norm": 29.625,
+      "learning_rate": 2.3628795294539767e-06,
+      "loss": 1.0163,
+      "step": 3617
+    },
+    {
+      "epoch": 0.30830847890924584,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.3628030014890576e-06,
+      "loss": 0.3581,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3083936940775458,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.362726453414669e-06,
+      "loss": 0.5291,
+      "step": 3619
+    },
+    {
+      "epoch": 0.30847890924584576,
+      "grad_norm": 40.25,
+      "learning_rate": 2.3626498852321947e-06,
+      "loss": 0.9466,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3085641244141457,
+      "grad_norm": 12.0,
+      "learning_rate": 2.362573296943018e-06,
+      "loss": 0.5573,
+      "step": 3621
+    },
+    {
+      "epoch": 0.30864933958244567,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.362496688548523e-06,
+      "loss": 0.4074,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3087345547507456,
+      "grad_norm": 15.25,
+      "learning_rate": 2.3624200600500943e-06,
+      "loss": 1.2185,
+      "step": 3623
+    },
+    {
+      "epoch": 0.3088197699190456,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3623434114491166e-06,
+      "loss": 0.8924,
+      "step": 3624
+    },
+    {
+      "epoch": 0.30890498508734554,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.3622667427469743e-06,
+      "loss": 0.7894,
+      "step": 3625
+    },
+    {
+      "epoch": 0.3089902002556455,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3621900539450544e-06,
+      "loss": 1.0706,
+      "step": 3626
+    },
+    {
+      "epoch": 0.30907541542394545,
+      "grad_norm": 23.875,
+      "learning_rate": 2.362113345044741e-06,
+      "loss": 1.1409,
+      "step": 3627
+    },
+    {
+      "epoch": 0.3091606305922454,
+      "grad_norm": 12.125,
+      "learning_rate": 2.362036616047422e-06,
+      "loss": 0.7046,
+      "step": 3628
+    },
+    {
+      "epoch": 0.30924584576054537,
+      "grad_norm": 13.25,
+      "learning_rate": 2.3619598669544824e-06,
+      "loss": 0.6346,
+      "step": 3629
+    },
+    {
+      "epoch": 0.3093310609288453,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.3618830977673096e-06,
+      "loss": 0.5055,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3094162760971453,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3618063084872917e-06,
+      "loss": 0.9511,
+      "step": 3631
+    },
+    {
+      "epoch": 0.30950149126544524,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.361729499115816e-06,
+      "loss": 0.8389,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3095867064337452,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3616526696542698e-06,
+      "loss": 0.8227,
+      "step": 3633
+    },
+    {
+      "epoch": 0.30967192160204515,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.361575820104042e-06,
+      "loss": 0.9879,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3097571367703451,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3614989504665216e-06,
+      "loss": 0.9241,
+      "step": 3635
+    },
+    {
+      "epoch": 0.30984235193864507,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3614220607430973e-06,
+      "loss": 0.8256,
+      "step": 3636
+    },
+    {
+      "epoch": 0.309927567106945,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.361345150935159e-06,
+      "loss": 0.5973,
+      "step": 3637
+    },
+    {
+      "epoch": 0.310012782275245,
+      "grad_norm": 12.75,
+      "learning_rate": 2.361268221044096e-06,
+      "loss": 0.5119,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31009799744354494,
+      "grad_norm": 18.375,
+      "learning_rate": 2.3611912710712987e-06,
+      "loss": 0.7814,
+      "step": 3639
+    },
+    {
+      "epoch": 0.3101832126118449,
+      "grad_norm": 10.5,
+      "learning_rate": 2.361114301018158e-06,
+      "loss": 0.4291,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31026842778014485,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3610373108860646e-06,
+      "loss": 0.7523,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3103536429484448,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.36096030067641e-06,
+      "loss": 0.7863,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31043885811674476,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.3608832703905853e-06,
+      "loss": 0.7521,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3105240732850447,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.3608062200299825e-06,
+      "loss": 1.0106,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3106092884533447,
+      "grad_norm": 16.375,
+      "learning_rate": 2.3607291495959947e-06,
+      "loss": 0.7794,
+      "step": 3645
+    },
+    {
+      "epoch": 0.31069450362164464,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.360652059090014e-06,
+      "loss": 0.7421,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3107797187899446,
+      "grad_norm": 15.0,
+      "learning_rate": 2.360574948513434e-06,
+      "loss": 0.78,
+      "step": 3647
+    },
+    {
+      "epoch": 0.31086493395824455,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.360497817867648e-06,
+      "loss": 0.5243,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3109501491265445,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3604206671540495e-06,
+      "loss": 0.5473,
+      "step": 3649
+    },
+    {
+      "epoch": 0.31103536429484446,
+      "grad_norm": 14.25,
+      "learning_rate": 2.360343496374033e-06,
+      "loss": 0.9088,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3111205794631444,
+      "grad_norm": 17.25,
+      "learning_rate": 2.360266305528993e-06,
+      "loss": 0.6185,
+      "step": 3651
+    },
+    {
+      "epoch": 0.3112057946314444,
+      "grad_norm": 13.375,
+      "learning_rate": 2.3601890946203245e-06,
+      "loss": 0.641,
+      "step": 3652
+    },
+    {
+      "epoch": 0.31129100979974433,
+      "grad_norm": 18.25,
+      "learning_rate": 2.3601118636494223e-06,
+      "loss": 0.6078,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3113762249680443,
+      "grad_norm": 16.625,
+      "learning_rate": 2.3600346126176827e-06,
+      "loss": 0.7837,
+      "step": 3654
+    },
+    {
+      "epoch": 0.31146144013634425,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.3599573415265013e-06,
+      "loss": 0.4894,
+      "step": 3655
+    },
+    {
+      "epoch": 0.3115466553046442,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.3598800503772747e-06,
+      "loss": 0.8935,
+      "step": 3656
+    },
+    {
+      "epoch": 0.31163187047294416,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.3598027391713994e-06,
+      "loss": 0.5816,
+      "step": 3657
+    },
+    {
+      "epoch": 0.3117170856412441,
+      "grad_norm": 18.375,
+      "learning_rate": 2.359725407910273e-06,
+      "loss": 0.977,
+      "step": 3658
+    },
+    {
+      "epoch": 0.3118023008095441,
+      "grad_norm": 14.125,
+      "learning_rate": 2.3596480565952923e-06,
+      "loss": 0.5791,
+      "step": 3659
+    },
+    {
+      "epoch": 0.31188751597784403,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3595706852278554e-06,
+      "loss": 0.8229,
+      "step": 3660
+    },
+    {
+      "epoch": 0.311972731146144,
+      "grad_norm": 17.0,
+      "learning_rate": 2.35949329380936e-06,
+      "loss": 0.7805,
+      "step": 3661
+    },
+    {
+      "epoch": 0.31205794631444395,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.3594158823412057e-06,
+      "loss": 0.5918,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3121431614827439,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3593384508247906e-06,
+      "loss": 0.552,
+      "step": 3663
+    },
+    {
+      "epoch": 0.31222837665104386,
+      "grad_norm": 17.75,
+      "learning_rate": 2.359260999261514e-06,
+      "loss": 1.0187,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3123135918193438,
+      "grad_norm": 22.75,
+      "learning_rate": 2.359183527652776e-06,
+      "loss": 0.9452,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3123988069876438,
+      "grad_norm": 17.125,
+      "learning_rate": 2.359106035999976e-06,
+      "loss": 0.9554,
+      "step": 3666
+    },
+    {
+      "epoch": 0.31248402215594373,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.359028524304515e-06,
+      "loss": 0.4401,
+      "step": 3667
+    },
+    {
+      "epoch": 0.31256923732424374,
+      "grad_norm": 19.125,
+      "learning_rate": 2.358950992567793e-06,
+      "loss": 1.0082,
+      "step": 3668
+    },
+    {
+      "epoch": 0.3126544524925437,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3588734407912116e-06,
+      "loss": 0.6147,
+      "step": 3669
+    },
+    {
+      "epoch": 0.31273966766084366,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.358795868976172e-06,
+      "loss": 0.7906,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3128248828291436,
+      "grad_norm": 16.5,
+      "learning_rate": 2.3587182771240764e-06,
+      "loss": 0.7904,
+      "step": 3671
+    },
+    {
+      "epoch": 0.31291009799744357,
+      "grad_norm": 17.0,
+      "learning_rate": 2.358640665236326e-06,
+      "loss": 0.8578,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3129953131657435,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3585630333143244e-06,
+      "loss": 1.0,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3130805283340435,
+      "grad_norm": 12.375,
+      "learning_rate": 2.358485381359474e-06,
+      "loss": 0.4281,
+      "step": 3674
+    },
+    {
+      "epoch": 0.31316574350234344,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3584077093731782e-06,
+      "loss": 0.855,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3132509586706434,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.35833001735684e-06,
+      "loss": 0.5559,
+      "step": 3676
+    },
+    {
+      "epoch": 0.31333617383894335,
+      "grad_norm": 18.875,
+      "learning_rate": 2.3582523053118644e-06,
+      "loss": 0.8259,
+      "step": 3677
+    },
+    {
+      "epoch": 0.3134213890072433,
+      "grad_norm": 30.0,
+      "learning_rate": 2.358174573239655e-06,
+      "loss": 1.0447,
+      "step": 3678
+    },
+    {
+      "epoch": 0.31350660417554327,
+      "grad_norm": 11.5,
+      "learning_rate": 2.358096821141617e-06,
+      "loss": 0.349,
+      "step": 3679
+    },
+    {
+      "epoch": 0.3135918193438432,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3580190490191545e-06,
+      "loss": 0.4728,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3136770345121432,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.3579412568736743e-06,
+      "loss": 1.0822,
+      "step": 3681
+    },
+    {
+      "epoch": 0.31376224968044314,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3578634447065808e-06,
+      "loss": 0.6368,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3138474648487431,
+      "grad_norm": 14.5,
+      "learning_rate": 2.357785612519281e-06,
+      "loss": 0.5831,
+      "step": 3683
+    },
+    {
+      "epoch": 0.31393268001704305,
+      "grad_norm": 18.125,
+      "learning_rate": 2.3577077603131816e-06,
+      "loss": 1.168,
+      "step": 3684
+    },
+    {
+      "epoch": 0.314017895185343,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3576298880896886e-06,
+      "loss": 0.9287,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31410311035364297,
+      "grad_norm": 21.75,
+      "learning_rate": 2.3575519958502103e-06,
+      "loss": 0.7875,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3141883255219429,
+      "grad_norm": 20.625,
+      "learning_rate": 2.357474083596153e-06,
+      "loss": 0.7885,
+      "step": 3687
+    },
+    {
+      "epoch": 0.3142735406902429,
+      "grad_norm": 19.5,
+      "learning_rate": 2.357396151328926e-06,
+      "loss": 0.9576,
+      "step": 3688
+    },
+    {
+      "epoch": 0.31435875585854284,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.3573181990499367e-06,
+      "loss": 0.5863,
+      "step": 3689
+    },
+    {
+      "epoch": 0.3144439710268428,
+      "grad_norm": 21.75,
+      "learning_rate": 2.357240226760594e-06,
+      "loss": 0.7171,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31452918619514275,
+      "grad_norm": 11.625,
+      "learning_rate": 2.357162234462307e-06,
+      "loss": 0.6142,
+      "step": 3691
+    },
+    {
+      "epoch": 0.3146144013634427,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.357084222156485e-06,
+      "loss": 0.6429,
+      "step": 3692
+    },
+    {
+      "epoch": 0.31469961653174267,
+      "grad_norm": 16.875,
+      "learning_rate": 2.357006189844538e-06,
+      "loss": 1.093,
+      "step": 3693
+    },
+    {
+      "epoch": 0.3147848317000426,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.356928137527876e-06,
+      "loss": 0.5054,
+      "step": 3694
+    },
+    {
+      "epoch": 0.3148700468683426,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.3568500652079096e-06,
+      "loss": 0.6249,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31495526203664254,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.356771972886049e-06,
+      "loss": 0.7313,
+      "step": 3696
+    },
+    {
+      "epoch": 0.3150404772049425,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3566938605637063e-06,
+      "loss": 0.7174,
+      "step": 3697
+    },
+    {
+      "epoch": 0.31512569237324245,
+      "grad_norm": 12.375,
+      "learning_rate": 2.3566157282422924e-06,
+      "loss": 0.4473,
+      "step": 3698
+    },
+    {
+      "epoch": 0.3152109075415424,
+      "grad_norm": 16.375,
+      "learning_rate": 2.35653757592322e-06,
+      "loss": 1.2287,
+      "step": 3699
+    },
+    {
+      "epoch": 0.31529612270984236,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.3564594036079003e-06,
+      "loss": 0.5213,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3153813378781423,
+      "grad_norm": 15.125,
+      "learning_rate": 2.356381211297747e-06,
+      "loss": 0.6685,
+      "step": 3701
+    },
+    {
+      "epoch": 0.3154665530464423,
+      "grad_norm": 18.25,
+      "learning_rate": 2.356302998994172e-06,
+      "loss": 1.0843,
+      "step": 3702
+    },
+    {
+      "epoch": 0.31555176821474223,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.3562247666985903e-06,
+      "loss": 0.5893,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3156369833830422,
+      "grad_norm": 18.25,
+      "learning_rate": 2.356146514412414e-06,
+      "loss": 0.9412,
+      "step": 3704
+    },
+    {
+      "epoch": 0.31572219855134215,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.356068242137058e-06,
+      "loss": 0.4863,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3158074137196421,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3559899498739367e-06,
+      "loss": 1.0099,
+      "step": 3706
+    },
+    {
+      "epoch": 0.31589262888794206,
+      "grad_norm": 19.875,
+      "learning_rate": 2.355911637624465e-06,
+      "loss": 0.7066,
+      "step": 3707
+    },
+    {
+      "epoch": 0.315977844056242,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.3558333053900574e-06,
+      "loss": 0.6905,
+      "step": 3708
+    },
+    {
+      "epoch": 0.316063059224542,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3557549531721306e-06,
+      "loss": 0.6421,
+      "step": 3709
+    },
+    {
+      "epoch": 0.31614827439284193,
+      "grad_norm": 12.125,
+      "learning_rate": 2.3556765809720994e-06,
+      "loss": 0.6885,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3162334895611419,
+      "grad_norm": 15.0,
+      "learning_rate": 2.355598188791381e-06,
+      "loss": 0.7999,
+      "step": 3711
+    },
+    {
+      "epoch": 0.31631870472944185,
+      "grad_norm": 28.875,
+      "learning_rate": 2.355519776631391e-06,
+      "loss": 1.2331,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3164039198977418,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.3554413444935474e-06,
+      "loss": 0.4101,
+      "step": 3713
+    },
+    {
+      "epoch": 0.31648913506604176,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3553628923792666e-06,
+      "loss": 0.5618,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3165743502343417,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3552844202899673e-06,
+      "loss": 0.9511,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3166595654026417,
+      "grad_norm": 29.375,
+      "learning_rate": 2.355205928227067e-06,
+      "loss": 1.1118,
+      "step": 3716
+    },
+    {
+      "epoch": 0.31674478057094163,
+      "grad_norm": 14.375,
+      "learning_rate": 2.355127416191984e-06,
+      "loss": 0.8699,
+      "step": 3717
+    },
+    {
+      "epoch": 0.3168299957392416,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3550488841861374e-06,
+      "loss": 0.5536,
+      "step": 3718
+    },
+    {
+      "epoch": 0.31691521090754154,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.354970332210946e-06,
+      "loss": 0.7641,
+      "step": 3719
+    },
+    {
+      "epoch": 0.3170004260758415,
+      "grad_norm": 15.75,
+      "learning_rate": 2.35489176026783e-06,
+      "loss": 0.5738,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31708564124414146,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.3548131683582088e-06,
+      "loss": 0.497,
+      "step": 3721
+    },
+    {
+      "epoch": 0.3171708564124414,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.354734556483503e-06,
+      "loss": 1.0743,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3172560715807414,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.354655924645132e-06,
+      "loss": 0.7099,
+      "step": 3723
+    },
+    {
+      "epoch": 0.31734128674904133,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.354577272844518e-06,
+      "loss": 0.6531,
+      "step": 3724
+    },
+    {
+      "epoch": 0.3174265019173413,
+      "grad_norm": 19.25,
+      "learning_rate": 2.354498601083082e-06,
+      "loss": 0.8511,
+      "step": 3725
+    },
+    {
+      "epoch": 0.31751171708564124,
+      "grad_norm": 22.5,
+      "learning_rate": 2.3544199093622455e-06,
+      "loss": 1.1105,
+      "step": 3726
+    },
+    {
+      "epoch": 0.3175969322539412,
+      "grad_norm": 14.125,
+      "learning_rate": 2.3543411976834305e-06,
+      "loss": 0.6921,
+      "step": 3727
+    },
+    {
+      "epoch": 0.31768214742224116,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3542624660480603e-06,
+      "loss": 0.4625,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3177673625905411,
+      "grad_norm": 14.75,
+      "learning_rate": 2.3541837144575565e-06,
+      "loss": 1.1254,
+      "step": 3729
+    },
+    {
+      "epoch": 0.31785257775884107,
+      "grad_norm": 17.375,
+      "learning_rate": 2.354104942913342e-06,
+      "loss": 1.0532,
+      "step": 3730
+    },
+    {
+      "epoch": 0.317937792927141,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3540261514168418e-06,
+      "loss": 0.6351,
+      "step": 3731
+    },
+    {
+      "epoch": 0.318023008095441,
+      "grad_norm": 252.0,
+      "learning_rate": 2.3539473399694782e-06,
+      "loss": 1.0827,
+      "step": 3732
+    },
+    {
+      "epoch": 0.31810822326374094,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.3538685085726766e-06,
+      "loss": 0.4128,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3181934384320409,
+      "grad_norm": 20.125,
+      "learning_rate": 2.353789657227861e-06,
+      "loss": 0.355,
+      "step": 3734
+    },
+    {
+      "epoch": 0.31827865360034086,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.353710785936456e-06,
+      "loss": 0.444,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3183638687686408,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3536318946998877e-06,
+      "loss": 0.5497,
+      "step": 3736
+    },
+    {
+      "epoch": 0.31844908393694077,
+      "grad_norm": 13.0,
+      "learning_rate": 2.353552983519581e-06,
+      "loss": 0.6666,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3185342991052407,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.353474052396962e-06,
+      "loss": 0.4177,
+      "step": 3738
+    },
+    {
+      "epoch": 0.3186195142735407,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3533951013334575e-06,
+      "loss": 0.6591,
+      "step": 3739
+    },
+    {
+      "epoch": 0.31870472944184064,
+      "grad_norm": 14.125,
+      "learning_rate": 2.353316130330494e-06,
+      "loss": 0.8239,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3187899446101406,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.3532371393894983e-06,
+      "loss": 0.3792,
+      "step": 3741
+    },
+    {
+      "epoch": 0.31887515977844055,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3531581285118983e-06,
+      "loss": 0.7582,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3189603749467405,
+      "grad_norm": 18.75,
+      "learning_rate": 2.3530790976991214e-06,
+      "loss": 1.1234,
+      "step": 3743
+    },
+    {
+      "epoch": 0.31904559011504047,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.353000046952596e-06,
+      "loss": 0.7819,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3191308052833404,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.352920976273751e-06,
+      "loss": 0.6734,
+      "step": 3745
+    },
+    {
+      "epoch": 0.3192160204516404,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3528418856640145e-06,
+      "loss": 0.542,
+      "step": 3746
+    },
+    {
+      "epoch": 0.31930123561994034,
+      "grad_norm": 17.0,
+      "learning_rate": 2.352762775124816e-06,
+      "loss": 1.0065,
+      "step": 3747
+    },
+    {
+      "epoch": 0.3193864507882403,
+      "grad_norm": 34.5,
+      "learning_rate": 2.352683644657586e-06,
+      "loss": 0.9629,
+      "step": 3748
+    },
+    {
+      "epoch": 0.31947166595654025,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3526044942637526e-06,
+      "loss": 0.7039,
+      "step": 3749
+    },
+    {
+      "epoch": 0.3195568811248402,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.352525323944748e-06,
+      "loss": 0.7703,
+      "step": 3750
+    },
+    {
+      "epoch": 0.31964209629314017,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.352446133702002e-06,
+      "loss": 0.7529,
+      "step": 3751
+    },
+    {
+      "epoch": 0.3197273114614401,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3523669235369457e-06,
+      "loss": 0.4563,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3198125266297401,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3522876934510104e-06,
+      "loss": 0.5116,
+      "step": 3753
+    },
+    {
+      "epoch": 0.31989774179804004,
+      "grad_norm": 15.25,
+      "learning_rate": 2.352208443445628e-06,
+      "loss": 0.9719,
+      "step": 3754
+    },
+    {
+      "epoch": 0.31998295696634,
+      "grad_norm": 15.0,
+      "learning_rate": 2.352129173522231e-06,
+      "loss": 0.6953,
+      "step": 3755
+    },
+    {
+      "epoch": 0.32006817213463995,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.3520498836822515e-06,
+      "loss": 0.288,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3201533873029399,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.3519705739271224e-06,
+      "loss": 0.5002,
+      "step": 3757
+    },
+    {
+      "epoch": 0.32023860247123986,
+      "grad_norm": 14.625,
+      "learning_rate": 2.351891244258277e-06,
+      "loss": 0.7877,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3203238176395398,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.3518118946771484e-06,
+      "loss": 0.3697,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3204090328078398,
+      "grad_norm": 15.25,
+      "learning_rate": 2.3517325251851716e-06,
+      "loss": 0.6415,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32049424797613973,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3516531357837793e-06,
+      "loss": 0.9096,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3205794631444397,
+      "grad_norm": 15.375,
+      "learning_rate": 2.3515737264744077e-06,
+      "loss": 1.1126,
+      "step": 3762
+    },
+    {
+      "epoch": 0.32066467831273965,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.351494297258491e-06,
+      "loss": 0.4516,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3207498934810396,
+      "grad_norm": 17.75,
+      "learning_rate": 2.3514148481374643e-06,
+      "loss": 0.9108,
+      "step": 3764
+    },
+    {
+      "epoch": 0.32083510864933956,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.351335379112764e-06,
+      "loss": 0.8066,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3209203238176395,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.351255890185826e-06,
+      "loss": 0.6022,
+      "step": 3766
+    },
+    {
+      "epoch": 0.3210055389859395,
+      "grad_norm": 15.75,
+      "learning_rate": 2.351176381358087e-06,
+      "loss": 1.2152,
+      "step": 3767
+    },
+    {
+      "epoch": 0.32109075415423943,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.351096852630983e-06,
+      "loss": 0.8962,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3211759693225394,
+      "grad_norm": 15.125,
+      "learning_rate": 2.351017304005952e-06,
+      "loss": 0.7988,
+      "step": 3769
+    },
+    {
+      "epoch": 0.32126118449083935,
+      "grad_norm": 11.625,
+      "learning_rate": 2.350937735484431e-06,
+      "loss": 0.3469,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3213463996591393,
+      "grad_norm": 17.375,
+      "learning_rate": 2.3508581470678576e-06,
+      "loss": 1.0225,
+      "step": 3771
+    },
+    {
+      "epoch": 0.32143161482743926,
+      "grad_norm": 16.625,
+      "learning_rate": 2.350778538757671e-06,
+      "loss": 0.7168,
+      "step": 3772
+    },
+    {
+      "epoch": 0.3215168299957392,
+      "grad_norm": 17.625,
+      "learning_rate": 2.350698910555309e-06,
+      "loss": 0.7265,
+      "step": 3773
+    },
+    {
+      "epoch": 0.3216020451640392,
+      "grad_norm": 19.625,
+      "learning_rate": 2.350619262462211e-06,
+      "loss": 0.8753,
+      "step": 3774
+    },
+    {
+      "epoch": 0.32168726033233913,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.350539594479816e-06,
+      "loss": 0.6382,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3217724755006391,
+      "grad_norm": 12.25,
+      "learning_rate": 2.350459906609564e-06,
+      "loss": 0.5823,
+      "step": 3776
+    },
+    {
+      "epoch": 0.32185769066893904,
+      "grad_norm": 16.875,
+      "learning_rate": 2.350380198852895e-06,
+      "loss": 0.6452,
+      "step": 3777
+    },
+    {
+      "epoch": 0.321942905837239,
+      "grad_norm": 22.375,
+      "learning_rate": 2.3503004712112495e-06,
+      "loss": 0.9861,
+      "step": 3778
+    },
+    {
+      "epoch": 0.32202812100553896,
+      "grad_norm": 29.375,
+      "learning_rate": 2.3502207236860673e-06,
+      "loss": 1.3897,
+      "step": 3779
+    },
+    {
+      "epoch": 0.3221133361738389,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.350140956278791e-06,
+      "loss": 0.4045,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32219855134213893,
+      "grad_norm": 12.5,
+      "learning_rate": 2.3500611689908607e-06,
+      "loss": 0.6438,
+      "step": 3781
+    },
+    {
+      "epoch": 0.3222837665104389,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.349981361823719e-06,
+      "loss": 0.8372,
+      "step": 3782
+    },
+    {
+      "epoch": 0.32236898167873884,
+      "grad_norm": 11.0,
+      "learning_rate": 2.3499015347788083e-06,
+      "loss": 0.6228,
+      "step": 3783
+    },
+    {
+      "epoch": 0.3224541968470388,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.349821687857571e-06,
+      "loss": 0.4362,
+      "step": 3784
+    },
+    {
+      "epoch": 0.32253941201533876,
+      "grad_norm": 12.375,
+      "learning_rate": 2.3497418210614494e-06,
+      "loss": 0.6263,
+      "step": 3785
+    },
+    {
+      "epoch": 0.3226246271836387,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.3496619343918873e-06,
+      "loss": 0.5625,
+      "step": 3786
+    },
+    {
+      "epoch": 0.32270984235193867,
+      "grad_norm": 14.75,
+      "learning_rate": 2.3495820278503284e-06,
+      "loss": 0.7221,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3227950575202386,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3495021014382163e-06,
+      "loss": 0.8427,
+      "step": 3788
+    },
+    {
+      "epoch": 0.3228802726885386,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3494221551569958e-06,
+      "loss": 1.0603,
+      "step": 3789
+    },
+    {
+      "epoch": 0.32296548785683854,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.3493421890081113e-06,
+      "loss": 0.6749,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3230507030251385,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3492622029930085e-06,
+      "loss": 0.5493,
+      "step": 3791
+    },
+    {
+      "epoch": 0.32313591819343845,
+      "grad_norm": 12.25,
+      "learning_rate": 2.349182197113132e-06,
+      "loss": 0.6549,
+      "step": 3792
+    },
+    {
+      "epoch": 0.3232211333617384,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3491021713699273e-06,
+      "loss": 1.0181,
+      "step": 3793
+    },
+    {
+      "epoch": 0.32330634853003837,
+      "grad_norm": 19.5,
+      "learning_rate": 2.3490221257648415e-06,
+      "loss": 0.5747,
+      "step": 3794
+    },
+    {
+      "epoch": 0.3233915636983383,
+      "grad_norm": 27.375,
+      "learning_rate": 2.348942060299321e-06,
+      "loss": 0.975,
+      "step": 3795
+    },
+    {
+      "epoch": 0.3234767788666383,
+      "grad_norm": 11.5,
+      "learning_rate": 2.348861974974812e-06,
+      "loss": 0.4122,
+      "step": 3796
+    },
+    {
+      "epoch": 0.32356199403493824,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.3487818697927623e-06,
+      "loss": 1.05,
+      "step": 3797
+    },
+    {
+      "epoch": 0.3236472092032382,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.3487017447546196e-06,
+      "loss": 0.8526,
+      "step": 3798
+    },
+    {
+      "epoch": 0.32373242437153815,
+      "grad_norm": 17.625,
+      "learning_rate": 2.348621599861831e-06,
+      "loss": 1.1766,
+      "step": 3799
+    },
+    {
+      "epoch": 0.3238176395398381,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3485414351158454e-06,
+      "loss": 0.5928,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32390285470813807,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.3484612505181117e-06,
+      "loss": 0.888,
+      "step": 3801
+    },
+    {
+      "epoch": 0.323988069876438,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3483810460700783e-06,
+      "loss": 0.5663,
+      "step": 3802
+    },
+    {
+      "epoch": 0.324073285044738,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3483008217731953e-06,
+      "loss": 0.4324,
+      "step": 3803
+    },
+    {
+      "epoch": 0.32415850021303794,
+      "grad_norm": 12.25,
+      "learning_rate": 2.348220577628912e-06,
+      "loss": 0.5498,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3242437153813379,
+      "grad_norm": 28.25,
+      "learning_rate": 2.348140313638678e-06,
+      "loss": 1.1247,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32432893054963785,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3480600298039446e-06,
+      "loss": 0.9968,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3244141457179378,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3479797261261623e-06,
+      "loss": 0.5682,
+      "step": 3807
+    },
+    {
+      "epoch": 0.32449936088623776,
+      "grad_norm": 19.5,
+      "learning_rate": 2.3478994026067824e-06,
+      "loss": 0.9215,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3245845760545377,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.347819059247256e-06,
+      "loss": 0.9493,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3246697912228377,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3477386960490358e-06,
+      "loss": 0.4373,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32475500639113764,
+      "grad_norm": 20.375,
+      "learning_rate": 2.3476583130135727e-06,
+      "loss": 0.9943,
+      "step": 3811
+    },
+    {
+      "epoch": 0.3248402215594376,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3475779101423205e-06,
+      "loss": 0.6435,
+      "step": 3812
+    },
+    {
+      "epoch": 0.32492543672773755,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3474974874367317e-06,
+      "loss": 0.7465,
+      "step": 3813
+    },
+    {
+      "epoch": 0.3250106518960375,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3474170448982596e-06,
+      "loss": 0.8447,
+      "step": 3814
+    },
+    {
+      "epoch": 0.32509586706433746,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3473365825283583e-06,
+      "loss": 1.0348,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3251810822326374,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.347256100328481e-06,
+      "loss": 0.4889,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3252662974009374,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3471755983000825e-06,
+      "loss": 0.6483,
+      "step": 3817
+    },
+    {
+      "epoch": 0.32535151256923733,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.347095076444618e-06,
+      "loss": 0.5665,
+      "step": 3818
+    },
+    {
+      "epoch": 0.3254367277375373,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3470145347635427e-06,
+      "loss": 0.8475,
+      "step": 3819
+    },
+    {
+      "epoch": 0.32552194290583725,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.346933973258311e-06,
+      "loss": 0.908,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3256071580741372,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3468533919303797e-06,
+      "loss": 0.6939,
+      "step": 3821
+    },
+    {
+      "epoch": 0.32569237324243716,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3467727907812044e-06,
+      "loss": 1.059,
+      "step": 3822
+    },
+    {
+      "epoch": 0.3257775884107371,
+      "grad_norm": 12.875,
+      "learning_rate": 2.346692169812242e-06,
+      "loss": 0.7456,
+      "step": 3823
+    },
+    {
+      "epoch": 0.3258628035790371,
+      "grad_norm": 13.25,
+      "learning_rate": 2.3466115290249493e-06,
+      "loss": 0.6671,
+      "step": 3824
+    },
+    {
+      "epoch": 0.32594801874733703,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.346530868420783e-06,
+      "loss": 0.7496,
+      "step": 3825
+    },
+    {
+      "epoch": 0.326033233915637,
+      "grad_norm": 14.75,
+      "learning_rate": 2.346450188001202e-06,
+      "loss": 0.7022,
+      "step": 3826
+    },
+    {
+      "epoch": 0.32611844908393695,
+      "grad_norm": 16.75,
+      "learning_rate": 2.346369487767663e-06,
+      "loss": 0.7458,
+      "step": 3827
+    },
+    {
+      "epoch": 0.3262036642522369,
+      "grad_norm": 18.0,
+      "learning_rate": 2.3462887677216254e-06,
+      "loss": 0.9712,
+      "step": 3828
+    },
+    {
+      "epoch": 0.32628887942053686,
+      "grad_norm": 13.625,
+      "learning_rate": 2.346208027864547e-06,
+      "loss": 0.667,
+      "step": 3829
+    },
+    {
+      "epoch": 0.3263740945888368,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3461272681978873e-06,
+      "loss": 0.9095,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3264593097571368,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.346046488723106e-06,
+      "loss": 0.7896,
+      "step": 3831
+    },
+    {
+      "epoch": 0.32654452492543673,
+      "grad_norm": 12.5,
+      "learning_rate": 2.3459656894416616e-06,
+      "loss": 0.5909,
+      "step": 3832
+    },
+    {
+      "epoch": 0.3266297400937367,
+      "grad_norm": 15.75,
+      "learning_rate": 2.345884870355016e-06,
+      "loss": 0.6627,
+      "step": 3833
+    },
+    {
+      "epoch": 0.32671495526203664,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.345804031464628e-06,
+      "loss": 0.4878,
+      "step": 3834
+    },
+    {
+      "epoch": 0.3268001704303366,
+      "grad_norm": 17.75,
+      "learning_rate": 2.34572317277196e-06,
+      "loss": 1.0159,
+      "step": 3835
+    },
+    {
+      "epoch": 0.32688538559863656,
+      "grad_norm": 18.125,
+      "learning_rate": 2.3456422942784718e-06,
+      "loss": 0.7609,
+      "step": 3836
+    },
+    {
+      "epoch": 0.3269706007669365,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.345561395985626e-06,
+      "loss": 0.5056,
+      "step": 3837
+    },
+    {
+      "epoch": 0.32705581593523647,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.345480477894884e-06,
+      "loss": 0.5038,
+      "step": 3838
+    },
+    {
+      "epoch": 0.32714103110353643,
+      "grad_norm": 10.25,
+      "learning_rate": 2.3453995400077085e-06,
+      "loss": 0.4173,
+      "step": 3839
+    },
+    {
+      "epoch": 0.3272262462718364,
+      "grad_norm": 27.25,
+      "learning_rate": 2.345318582325561e-06,
+      "loss": 1.0164,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32731146144013634,
+      "grad_norm": 12.125,
+      "learning_rate": 2.345237604849906e-06,
+      "loss": 0.5116,
+      "step": 3841
+    },
+    {
+      "epoch": 0.3273966766084363,
+      "grad_norm": 11.375,
+      "learning_rate": 2.345156607582206e-06,
+      "loss": 0.3717,
+      "step": 3842
+    },
+    {
+      "epoch": 0.32748189177673626,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3450755905239246e-06,
+      "loss": 0.555,
+      "step": 3843
+    },
+    {
+      "epoch": 0.3275671069450362,
+      "grad_norm": 13.625,
+      "learning_rate": 2.344994553676526e-06,
+      "loss": 0.5815,
+      "step": 3844
+    },
+    {
+      "epoch": 0.32765232211333617,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.344913497041475e-06,
+      "loss": 0.5783,
+      "step": 3845
+    },
+    {
+      "epoch": 0.3277375372816361,
+      "grad_norm": 12.5,
+      "learning_rate": 2.344832420620236e-06,
+      "loss": 0.6687,
+      "step": 3846
+    },
+    {
+      "epoch": 0.3278227524499361,
+      "grad_norm": 19.5,
+      "learning_rate": 2.3447513244142745e-06,
+      "loss": 0.8924,
+      "step": 3847
+    },
+    {
+      "epoch": 0.32790796761823604,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.3446702084250552e-06,
+      "loss": 0.9233,
+      "step": 3848
+    },
+    {
+      "epoch": 0.327993182786536,
+      "grad_norm": 19.125,
+      "learning_rate": 2.344589072654045e-06,
+      "loss": 0.9593,
+      "step": 3849
+    },
+    {
+      "epoch": 0.32807839795483595,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.344507917102709e-06,
+      "loss": 0.5729,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3281636131231359,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3444267417725147e-06,
+      "loss": 1.1154,
+      "step": 3851
+    },
+    {
+      "epoch": 0.32824882829143587,
+      "grad_norm": 20.75,
+      "learning_rate": 2.3443455466649285e-06,
+      "loss": 0.7493,
+      "step": 3852
+    },
+    {
+      "epoch": 0.3283340434597358,
+      "grad_norm": 25.875,
+      "learning_rate": 2.344264331781418e-06,
+      "loss": 0.8772,
+      "step": 3853
+    },
+    {
+      "epoch": 0.3284192586280358,
+      "grad_norm": 29.125,
+      "learning_rate": 2.3441830971234503e-06,
+      "loss": 1.1599,
+      "step": 3854
+    },
+    {
+      "epoch": 0.32850447379633574,
+      "grad_norm": 12.875,
+      "learning_rate": 2.3441018426924937e-06,
+      "loss": 0.5125,
+      "step": 3855
+    },
+    {
+      "epoch": 0.3285896889646357,
+      "grad_norm": 19.375,
+      "learning_rate": 2.344020568490017e-06,
+      "loss": 1.3051,
+      "step": 3856
+    },
+    {
+      "epoch": 0.32867490413293565,
+      "grad_norm": 18.5,
+      "learning_rate": 2.3439392745174885e-06,
+      "loss": 0.6417,
+      "step": 3857
+    },
+    {
+      "epoch": 0.3287601193012356,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3438579607763774e-06,
+      "loss": 1.1153,
+      "step": 3858
+    },
+    {
+      "epoch": 0.32884533446953557,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3437766272681523e-06,
+      "loss": 1.2917,
+      "step": 3859
+    },
+    {
+      "epoch": 0.3289305496378355,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3436952739942843e-06,
+      "loss": 0.7366,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3290157648061355,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3436139009562425e-06,
+      "loss": 0.9502,
+      "step": 3861
+    },
+    {
+      "epoch": 0.32910097997443544,
+      "grad_norm": 19.25,
+      "learning_rate": 2.343532508155498e-06,
+      "loss": 0.8141,
+      "step": 3862
+    },
+    {
+      "epoch": 0.3291861951427354,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.3434510955935217e-06,
+      "loss": 0.5179,
+      "step": 3863
+    },
+    {
+      "epoch": 0.32927141031103535,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.343369663271784e-06,
+      "loss": 0.5175,
+      "step": 3864
+    },
+    {
+      "epoch": 0.3293566254793353,
+      "grad_norm": 21.5,
+      "learning_rate": 2.3432882111917574e-06,
+      "loss": 0.9953,
+      "step": 3865
+    },
+    {
+      "epoch": 0.32944184064763526,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3432067393549134e-06,
+      "loss": 0.5388,
+      "step": 3866
+    },
+    {
+      "epoch": 0.3295270558159352,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.3431252477627242e-06,
+      "loss": 0.7496,
+      "step": 3867
+    },
+    {
+      "epoch": 0.3296122709842352,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3430437364166624e-06,
+      "loss": 0.4234,
+      "step": 3868
+    },
+    {
+      "epoch": 0.32969748615253514,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.3429622053182018e-06,
+      "loss": 0.6817,
+      "step": 3869
+    },
+    {
+      "epoch": 0.3297827013208351,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3428806544688145e-06,
+      "loss": 0.8044,
+      "step": 3870
+    },
+    {
+      "epoch": 0.32986791648913505,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.342799083869975e-06,
+      "loss": 0.8683,
+      "step": 3871
+    },
+    {
+      "epoch": 0.329953131657435,
+      "grad_norm": 15.625,
+      "learning_rate": 2.342717493523157e-06,
+      "loss": 1.0609,
+      "step": 3872
+    },
+    {
+      "epoch": 0.33003834682573496,
+      "grad_norm": 14.375,
+      "learning_rate": 2.342635883429835e-06,
+      "loss": 0.5665,
+      "step": 3873
+    },
+    {
+      "epoch": 0.3301235619940349,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.342554253591484e-06,
+      "loss": 0.5236,
+      "step": 3874
+    },
+    {
+      "epoch": 0.3302087771623349,
+      "grad_norm": 15.75,
+      "learning_rate": 2.3424726040095793e-06,
+      "loss": 0.7493,
+      "step": 3875
+    },
+    {
+      "epoch": 0.33029399233063483,
+      "grad_norm": 24.625,
+      "learning_rate": 2.342390934685596e-06,
+      "loss": 0.9125,
+      "step": 3876
+    },
+    {
+      "epoch": 0.3303792074989348,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.34230924562101e-06,
+      "loss": 0.368,
+      "step": 3877
+    },
+    {
+      "epoch": 0.33046442266723475,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.342227536817297e-06,
+      "loss": 0.5587,
+      "step": 3878
+    },
+    {
+      "epoch": 0.3305496378355347,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.3421458082759347e-06,
+      "loss": 0.4354,
+      "step": 3879
+    },
+    {
+      "epoch": 0.33063485300383466,
+      "grad_norm": 18.125,
+      "learning_rate": 2.342064059998399e-06,
+      "loss": 0.7428,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3307200681721346,
+      "grad_norm": 13.875,
+      "learning_rate": 2.341982291986168e-06,
+      "loss": 0.6956,
+      "step": 3881
+    },
+    {
+      "epoch": 0.3308052833404346,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3419005042407184e-06,
+      "loss": 1.0907,
+      "step": 3882
+    },
+    {
+      "epoch": 0.33089049850873453,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.341818696763529e-06,
+      "loss": 0.8007,
+      "step": 3883
+    },
+    {
+      "epoch": 0.3309757136770345,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3417368695560777e-06,
+      "loss": 1.1786,
+      "step": 3884
+    },
+    {
+      "epoch": 0.33106092884533445,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3416550226198435e-06,
+      "loss": 0.6097,
+      "step": 3885
+    },
+    {
+      "epoch": 0.3311461440136344,
+      "grad_norm": 25.375,
+      "learning_rate": 2.341573155956305e-06,
+      "loss": 0.7952,
+      "step": 3886
+    },
+    {
+      "epoch": 0.33123135918193436,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.3414912695669424e-06,
+      "loss": 0.6954,
+      "step": 3887
+    },
+    {
+      "epoch": 0.3313165743502343,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.3414093634532345e-06,
+      "loss": 0.619,
+      "step": 3888
+    },
+    {
+      "epoch": 0.3314017895185343,
+      "grad_norm": 19.875,
+      "learning_rate": 2.3413274376166623e-06,
+      "loss": 0.7765,
+      "step": 3889
+    },
+    {
+      "epoch": 0.33148700468683423,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3412454920587057e-06,
+      "loss": 1.0166,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3315722198551342,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.341163526780846e-06,
+      "loss": 0.9749,
+      "step": 3891
+    },
+    {
+      "epoch": 0.33165743502343414,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.341081541784563e-06,
+      "loss": 0.8066,
+      "step": 3892
+    },
+    {
+      "epoch": 0.33174265019173416,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.3409995370713405e-06,
+      "loss": 0.5688,
+      "step": 3893
+    },
+    {
+      "epoch": 0.3318278653600341,
+      "grad_norm": 12.375,
+      "learning_rate": 2.340917512642659e-06,
+      "loss": 0.8135,
+      "step": 3894
+    },
+    {
+      "epoch": 0.33191308052833407,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.340835468500001e-06,
+      "loss": 0.6521,
+      "step": 3895
+    },
+    {
+      "epoch": 0.331998295696634,
+      "grad_norm": 17.75,
+      "learning_rate": 2.340753404644849e-06,
+      "loss": 0.8416,
+      "step": 3896
+    },
+    {
+      "epoch": 0.332083510864934,
+      "grad_norm": 10.375,
+      "learning_rate": 2.3406713210786867e-06,
+      "loss": 0.3349,
+      "step": 3897
+    },
+    {
+      "epoch": 0.33216872603323394,
+      "grad_norm": 17.375,
+      "learning_rate": 2.3405892178029964e-06,
+      "loss": 1.0132,
+      "step": 3898
+    },
+    {
+      "epoch": 0.3322539412015339,
+      "grad_norm": 18.625,
+      "learning_rate": 2.3405070948192625e-06,
+      "loss": 1.0266,
+      "step": 3899
+    },
+    {
+      "epoch": 0.33233915636983385,
+      "grad_norm": 15.875,
+      "learning_rate": 2.340424952128969e-06,
+      "loss": 1.0015,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3324243715381338,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3403427897335995e-06,
+      "loss": 0.7218,
+      "step": 3901
+    },
+    {
+      "epoch": 0.33250958670643377,
+      "grad_norm": 16.75,
+      "learning_rate": 2.34026060763464e-06,
+      "loss": 0.7724,
+      "step": 3902
+    },
+    {
+      "epoch": 0.3325948018747337,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3401784058335744e-06,
+      "loss": 0.7607,
+      "step": 3903
+    },
+    {
+      "epoch": 0.3326800170430337,
+      "grad_norm": 12.375,
+      "learning_rate": 2.340096184331889e-06,
+      "loss": 0.7528,
+      "step": 3904
+    },
+    {
+      "epoch": 0.33276523221133364,
+      "grad_norm": 21.0,
+      "learning_rate": 2.3400139431310694e-06,
+      "loss": 0.6997,
+      "step": 3905
+    },
+    {
+      "epoch": 0.3328504473796336,
+      "grad_norm": 11.25,
+      "learning_rate": 2.3399316822326018e-06,
+      "loss": 0.6647,
+      "step": 3906
+    },
+    {
+      "epoch": 0.33293566254793355,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.3398494016379725e-06,
+      "loss": 0.4979,
+      "step": 3907
+    },
+    {
+      "epoch": 0.3330208777162335,
+      "grad_norm": 18.125,
+      "learning_rate": 2.339767101348669e-06,
+      "loss": 0.8077,
+      "step": 3908
+    },
+    {
+      "epoch": 0.33310609288453347,
+      "grad_norm": 19.5,
+      "learning_rate": 2.339684781366178e-06,
+      "loss": 0.9445,
+      "step": 3909
+    },
+    {
+      "epoch": 0.3331913080528334,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.3396024416919864e-06,
+      "loss": 0.8323,
+      "step": 3910
+    },
+    {
+      "epoch": 0.3332765232211334,
+      "grad_norm": 17.625,
+      "learning_rate": 2.339520082327584e-06,
+      "loss": 0.9099,
+      "step": 3911
+    },
+    {
+      "epoch": 0.33336173838943334,
+      "grad_norm": 17.25,
+      "learning_rate": 2.339437703274458e-06,
+      "loss": 0.6971,
+      "step": 3912
+    },
+    {
+      "epoch": 0.3334469535577333,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3393553045340967e-06,
+      "loss": 0.9556,
+      "step": 3913
+    },
+    {
+      "epoch": 0.33353216872603325,
+      "grad_norm": 12.0,
+      "learning_rate": 2.33927288610799e-06,
+      "loss": 0.5052,
+      "step": 3914
+    },
+    {
+      "epoch": 0.3336173838943332,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3391904479976267e-06,
+      "loss": 0.8101,
+      "step": 3915
+    },
+    {
+      "epoch": 0.33370259906263317,
+      "grad_norm": 13.5,
+      "learning_rate": 2.3391079902044968e-06,
+      "loss": 0.6602,
+      "step": 3916
+    },
+    {
+      "epoch": 0.3337878142309331,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3390255127300905e-06,
+      "loss": 0.96,
+      "step": 3917
+    },
+    {
+      "epoch": 0.3338730293992331,
+      "grad_norm": 14.5,
+      "learning_rate": 2.338943015575898e-06,
+      "loss": 0.9262,
+      "step": 3918
+    },
+    {
+      "epoch": 0.33395824456753304,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3388604987434097e-06,
+      "loss": 0.6437,
+      "step": 3919
+    },
+    {
+      "epoch": 0.334043459735833,
+      "grad_norm": 13.0,
+      "learning_rate": 2.338777962234118e-06,
+      "loss": 0.8867,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33412867490413295,
+      "grad_norm": 12.625,
+      "learning_rate": 2.338695406049513e-06,
+      "loss": 0.4617,
+      "step": 3921
+    },
+    {
+      "epoch": 0.3342138900724329,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.338612830191088e-06,
+      "loss": 0.5451,
+      "step": 3922
+    },
+    {
+      "epoch": 0.33429910524073286,
+      "grad_norm": 13.625,
+      "learning_rate": 2.338530234660334e-06,
+      "loss": 0.2578,
+      "step": 3923
+    },
+    {
+      "epoch": 0.3343843204090328,
+      "grad_norm": 28.25,
+      "learning_rate": 2.338447619458744e-06,
+      "loss": 1.023,
+      "step": 3924
+    },
+    {
+      "epoch": 0.3344695355773328,
+      "grad_norm": 22.5,
+      "learning_rate": 2.3383649845878116e-06,
+      "loss": 0.9503,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33455475074563273,
+      "grad_norm": 25.0,
+      "learning_rate": 2.3382823300490286e-06,
+      "loss": 1.301,
+      "step": 3926
+    },
+    {
+      "epoch": 0.3346399659139327,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.33819965584389e-06,
+      "loss": 0.5245,
+      "step": 3927
+    },
+    {
+      "epoch": 0.33472518108223265,
+      "grad_norm": 24.375,
+      "learning_rate": 2.33811696197389e-06,
+      "loss": 0.4129,
+      "step": 3928
+    },
+    {
+      "epoch": 0.3348103962505326,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.338034248440521e-06,
+      "loss": 0.4974,
+      "step": 3929
+    },
+    {
+      "epoch": 0.33489561141883256,
+      "grad_norm": 15.75,
+      "learning_rate": 2.33795151524528e-06,
+      "loss": 0.7743,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3349808265871325,
+      "grad_norm": 21.25,
+      "learning_rate": 2.3378687623896606e-06,
+      "loss": 0.9439,
+      "step": 3931
+    },
+    {
+      "epoch": 0.3350660417554325,
+      "grad_norm": 11.75,
+      "learning_rate": 2.337785989875159e-06,
+      "loss": 0.4533,
+      "step": 3932
+    },
+    {
+      "epoch": 0.33515125692373243,
+      "grad_norm": 15.5,
+      "learning_rate": 2.3377031977032704e-06,
+      "loss": 0.994,
+      "step": 3933
+    },
+    {
+      "epoch": 0.3352364720920324,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3376203858754916e-06,
+      "loss": 0.507,
+      "step": 3934
+    },
+    {
+      "epoch": 0.33532168726033235,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3375375543933186e-06,
+      "loss": 0.4147,
+      "step": 3935
+    },
+    {
+      "epoch": 0.3354069024286323,
+      "grad_norm": 26.875,
+      "learning_rate": 2.3374547032582482e-06,
+      "loss": 1.3156,
+      "step": 3936
+    },
+    {
+      "epoch": 0.33549211759693226,
+      "grad_norm": 12.125,
+      "learning_rate": 2.337371832471778e-06,
+      "loss": 0.6413,
+      "step": 3937
+    },
+    {
+      "epoch": 0.3355773327652322,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3372889420354055e-06,
+      "loss": 0.6281,
+      "step": 3938
+    },
+    {
+      "epoch": 0.3356625479335322,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3372060319506283e-06,
+      "loss": 0.8928,
+      "step": 3939
+    },
+    {
+      "epoch": 0.33574776310183213,
+      "grad_norm": 12.625,
+      "learning_rate": 2.3371231022189447e-06,
+      "loss": 0.7688,
+      "step": 3940
+    },
+    {
+      "epoch": 0.3358329782701321,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3370401528418535e-06,
+      "loss": 0.7262,
+      "step": 3941
+    },
+    {
+      "epoch": 0.33591819343843204,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3369571838208537e-06,
+      "loss": 0.5853,
+      "step": 3942
+    },
+    {
+      "epoch": 0.336003408606732,
+      "grad_norm": 16.0,
+      "learning_rate": 2.336874195157445e-06,
+      "loss": 0.5609,
+      "step": 3943
+    },
+    {
+      "epoch": 0.33608862377503196,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.336791186853126e-06,
+      "loss": 0.3731,
+      "step": 3944
+    },
+    {
+      "epoch": 0.3361738389433319,
+      "grad_norm": 12.75,
+      "learning_rate": 2.3367081589093976e-06,
+      "loss": 0.8264,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33625905411163187,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.3366251113277597e-06,
+      "loss": 0.4323,
+      "step": 3946
+    },
+    {
+      "epoch": 0.33634426927993183,
+      "grad_norm": 16.0,
+      "learning_rate": 2.336542044109714e-06,
+      "loss": 0.5537,
+      "step": 3947
+    },
+    {
+      "epoch": 0.3364294844482318,
+      "grad_norm": 20.125,
+      "learning_rate": 2.3364589572567608e-06,
+      "loss": 1.0855,
+      "step": 3948
+    },
+    {
+      "epoch": 0.33651469961653174,
+      "grad_norm": 11.875,
+      "learning_rate": 2.3363758507704015e-06,
+      "loss": 0.679,
+      "step": 3949
+    },
+    {
+      "epoch": 0.3365999147848317,
+      "grad_norm": 18.0,
+      "learning_rate": 2.336292724652138e-06,
+      "loss": 0.8954,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33668512995313166,
+      "grad_norm": 15.0,
+      "learning_rate": 2.336209578903473e-06,
+      "loss": 0.9358,
+      "step": 3951
+    },
+    {
+      "epoch": 0.3367703451214316,
+      "grad_norm": 10.25,
+      "learning_rate": 2.3361264135259083e-06,
+      "loss": 0.3877,
+      "step": 3952
+    },
+    {
+      "epoch": 0.33685556028973157,
+      "grad_norm": 16.25,
+      "learning_rate": 2.336043228520947e-06,
+      "loss": 0.6617,
+      "step": 3953
+    },
+    {
+      "epoch": 0.3369407754580315,
+      "grad_norm": 17.75,
+      "learning_rate": 2.3359600238900932e-06,
+      "loss": 1.2946,
+      "step": 3954
+    },
+    {
+      "epoch": 0.3370259906263315,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.3358767996348492e-06,
+      "loss": 0.2643,
+      "step": 3955
+    },
+    {
+      "epoch": 0.33711120579463144,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3357935557567195e-06,
+      "loss": 0.8753,
+      "step": 3956
+    },
+    {
+      "epoch": 0.3371964209629314,
+      "grad_norm": 20.875,
+      "learning_rate": 2.3357102922572083e-06,
+      "loss": 0.9863,
+      "step": 3957
+    },
+    {
+      "epoch": 0.33728163613123135,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.335627009137821e-06,
+      "loss": 0.4275,
+      "step": 3958
+    },
+    {
+      "epoch": 0.3373668512995313,
+      "grad_norm": 15.125,
+      "learning_rate": 2.335543706400061e-06,
+      "loss": 0.7731,
+      "step": 3959
+    },
+    {
+      "epoch": 0.33745206646783127,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.3354603840454355e-06,
+      "loss": 0.6382,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3375372816361312,
+      "grad_norm": 20.25,
+      "learning_rate": 2.335377042075449e-06,
+      "loss": 1.0024,
+      "step": 3961
+    },
+    {
+      "epoch": 0.3376224968044312,
+      "grad_norm": 12.875,
+      "learning_rate": 2.3352936804916076e-06,
+      "loss": 0.6543,
+      "step": 3962
+    },
+    {
+      "epoch": 0.33770771197273114,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3352102992954183e-06,
+      "loss": 0.9571,
+      "step": 3963
+    },
+    {
+      "epoch": 0.3377929271410311,
+      "grad_norm": 16.125,
+      "learning_rate": 2.335126898488388e-06,
+      "loss": 1.1574,
+      "step": 3964
+    },
+    {
+      "epoch": 0.33787814230933105,
+      "grad_norm": 17.875,
+      "learning_rate": 2.335043478072023e-06,
+      "loss": 0.7375,
+      "step": 3965
+    },
+    {
+      "epoch": 0.337963357477631,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3349600380478317e-06,
+      "loss": 0.486,
+      "step": 3966
+    },
+    {
+      "epoch": 0.33804857264593097,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.334876578417321e-06,
+      "loss": 0.3549,
+      "step": 3967
+    },
+    {
+      "epoch": 0.3381337878142309,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3347930991819996e-06,
+      "loss": 0.7313,
+      "step": 3968
+    },
+    {
+      "epoch": 0.3382190029825309,
+      "grad_norm": 16.125,
+      "learning_rate": 2.334709600343376e-06,
+      "loss": 0.911,
+      "step": 3969
+    },
+    {
+      "epoch": 0.33830421815083084,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3346260819029597e-06,
+      "loss": 0.797,
+      "step": 3970
+    },
+    {
+      "epoch": 0.3383894333191308,
+      "grad_norm": 8.6875,
+      "learning_rate": 2.3345425438622593e-06,
+      "loss": 0.2527,
+      "step": 3971
+    },
+    {
+      "epoch": 0.33847464848743075,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3344589862227844e-06,
+      "loss": 0.5059,
+      "step": 3972
+    },
+    {
+      "epoch": 0.3385598636557307,
+      "grad_norm": 13.625,
+      "learning_rate": 2.334375408986045e-06,
+      "loss": 0.8461,
+      "step": 3973
+    },
+    {
+      "epoch": 0.33864507882403067,
+      "grad_norm": 25.5,
+      "learning_rate": 2.3342918121535516e-06,
+      "loss": 1.3825,
+      "step": 3974
+    },
+    {
+      "epoch": 0.3387302939923306,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3342081957268146e-06,
+      "loss": 0.9739,
+      "step": 3975
+    },
+    {
+      "epoch": 0.3388155091606306,
+      "grad_norm": 22.125,
+      "learning_rate": 2.334124559707346e-06,
+      "loss": 0.7812,
+      "step": 3976
+    },
+    {
+      "epoch": 0.33890072432893054,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3340409040966555e-06,
+      "loss": 0.8652,
+      "step": 3977
+    },
+    {
+      "epoch": 0.3389859394972305,
+      "grad_norm": 30.875,
+      "learning_rate": 2.3339572288962564e-06,
+      "loss": 1.031,
+      "step": 3978
+    },
+    {
+      "epoch": 0.33907115466553045,
+      "grad_norm": 12.375,
+      "learning_rate": 2.3338735341076597e-06,
+      "loss": 0.7422,
+      "step": 3979
+    },
+    {
+      "epoch": 0.3391563698338304,
+      "grad_norm": 18.75,
+      "learning_rate": 2.333789819732379e-06,
+      "loss": 0.711,
+      "step": 3980
+    },
+    {
+      "epoch": 0.33924158500213036,
+      "grad_norm": 24.25,
+      "learning_rate": 2.333706085771926e-06,
+      "loss": 1.0343,
+      "step": 3981
+    },
+    {
+      "epoch": 0.3393268001704303,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.333622332227814e-06,
+      "loss": 0.6232,
+      "step": 3982
+    },
+    {
+      "epoch": 0.3394120153387303,
+      "grad_norm": 18.25,
+      "learning_rate": 2.333538559101557e-06,
+      "loss": 0.7608,
+      "step": 3983
+    },
+    {
+      "epoch": 0.33949723050703023,
+      "grad_norm": 20.0,
+      "learning_rate": 2.333454766394669e-06,
+      "loss": 0.8179,
+      "step": 3984
+    },
+    {
+      "epoch": 0.3395824456753302,
+      "grad_norm": 21.875,
+      "learning_rate": 2.3333709541086636e-06,
+      "loss": 1.131,
+      "step": 3985
+    },
+    {
+      "epoch": 0.33966766084363015,
+      "grad_norm": 18.25,
+      "learning_rate": 2.3332871222450554e-06,
+      "loss": 0.9852,
+      "step": 3986
+    },
+    {
+      "epoch": 0.3397528760119301,
+      "grad_norm": 19.75,
+      "learning_rate": 2.33320327080536e-06,
+      "loss": 0.8217,
+      "step": 3987
+    },
+    {
+      "epoch": 0.33983809118023006,
+      "grad_norm": 14.375,
+      "learning_rate": 2.333119399791092e-06,
+      "loss": 0.4423,
+      "step": 3988
+    },
+    {
+      "epoch": 0.33992330634853,
+      "grad_norm": 14.125,
+      "learning_rate": 2.3330355092037672e-06,
+      "loss": 0.6775,
+      "step": 3989
+    },
+    {
+      "epoch": 0.34000852151683,
+      "grad_norm": 18.25,
+      "learning_rate": 2.332951599044902e-06,
+      "loss": 0.6429,
+      "step": 3990
+    },
+    {
+      "epoch": 0.34009373668512993,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3328676693160124e-06,
+      "loss": 0.7431,
+      "step": 3991
+    },
+    {
+      "epoch": 0.3401789518534299,
+      "grad_norm": 15.75,
+      "learning_rate": 2.332783720018615e-06,
+      "loss": 0.6359,
+      "step": 3992
+    },
+    {
+      "epoch": 0.34026416702172985,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3326997511542267e-06,
+      "loss": 1.0031,
+      "step": 3993
+    },
+    {
+      "epoch": 0.3403493821900298,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3326157627243653e-06,
+      "loss": 0.5978,
+      "step": 3994
+    },
+    {
+      "epoch": 0.34043459735832976,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.332531754730549e-06,
+      "loss": 0.5315,
+      "step": 3995
+    },
+    {
+      "epoch": 0.3405198125266297,
+      "grad_norm": 13.875,
+      "learning_rate": 2.3324477271742943e-06,
+      "loss": 0.6428,
+      "step": 3996
+    },
+    {
+      "epoch": 0.3406050276949297,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3323636800571212e-06,
+      "loss": 0.6487,
+      "step": 3997
+    },
+    {
+      "epoch": 0.34069024286322963,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.332279613380548e-06,
+      "loss": 0.8349,
+      "step": 3998
+    },
+    {
+      "epoch": 0.3407754580315296,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.332195527146094e-06,
+      "loss": 0.641,
+      "step": 3999
+    },
+    {
+      "epoch": 0.34086067319982954,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.332111421355278e-06,
+      "loss": 1.024,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3409458883681295,
+      "grad_norm": 19.625,
+      "learning_rate": 2.3320272960096212e-06,
+      "loss": 1.2106,
+      "step": 4001
+    },
+    {
+      "epoch": 0.34103110353642946,
+      "grad_norm": 16.875,
+      "learning_rate": 2.331943151110643e-06,
+      "loss": 1.1761,
+      "step": 4002
+    },
+    {
+      "epoch": 0.3411163187047294,
+      "grad_norm": 17.875,
+      "learning_rate": 2.331858986659864e-06,
+      "loss": 0.8235,
+      "step": 4003
+    },
+    {
+      "epoch": 0.3412015338730294,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.331774802658805e-06,
+      "loss": 0.4329,
+      "step": 4004
+    },
+    {
+      "epoch": 0.34128674904132933,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3316905991089877e-06,
+      "loss": 0.5224,
+      "step": 4005
+    },
+    {
+      "epoch": 0.34137196420962934,
+      "grad_norm": 21.0,
+      "learning_rate": 2.331606376011933e-06,
+      "loss": 1.1005,
+      "step": 4006
+    },
+    {
+      "epoch": 0.3414571793779293,
+      "grad_norm": 15.25,
+      "learning_rate": 2.331522133369164e-06,
+      "loss": 1.0837,
+      "step": 4007
+    },
+    {
+      "epoch": 0.34154239454622926,
+      "grad_norm": 15.125,
+      "learning_rate": 2.3314378711822024e-06,
+      "loss": 0.9544,
+      "step": 4008
+    },
+    {
+      "epoch": 0.3416276097145292,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3313535894525713e-06,
+      "loss": 0.815,
+      "step": 4009
+    },
+    {
+      "epoch": 0.34171282488282917,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.331269288181793e-06,
+      "loss": 0.4748,
+      "step": 4010
+    },
+    {
+      "epoch": 0.3417980400511291,
+      "grad_norm": 18.75,
+      "learning_rate": 2.3311849673713913e-06,
+      "loss": 1.3333,
+      "step": 4011
+    },
+    {
+      "epoch": 0.3418832552194291,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3311006270228905e-06,
+      "loss": 0.4996,
+      "step": 4012
+    },
+    {
+      "epoch": 0.34196847038772904,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.331016267137814e-06,
+      "loss": 0.5874,
+      "step": 4013
+    },
+    {
+      "epoch": 0.342053685556029,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3309318877176865e-06,
+      "loss": 0.9413,
+      "step": 4014
+    },
+    {
+      "epoch": 0.34213890072432895,
+      "grad_norm": 25.375,
+      "learning_rate": 2.3308474887640327e-06,
+      "loss": 1.0336,
+      "step": 4015
+    },
+    {
+      "epoch": 0.3422241158926289,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.330763070278378e-06,
+      "loss": 0.5361,
+      "step": 4016
+    },
+    {
+      "epoch": 0.34230933106092887,
+      "grad_norm": 17.375,
+      "learning_rate": 2.330678632262248e-06,
+      "loss": 0.3863,
+      "step": 4017
+    },
+    {
+      "epoch": 0.3423945462292288,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.3305941747171684e-06,
+      "loss": 0.7383,
+      "step": 4018
+    },
+    {
+      "epoch": 0.3424797613975288,
+      "grad_norm": 18.0,
+      "learning_rate": 2.3305096976446652e-06,
+      "loss": 0.9696,
+      "step": 4019
+    },
+    {
+      "epoch": 0.34256497656582874,
+      "grad_norm": 12.5,
+      "learning_rate": 2.3304252010462654e-06,
+      "loss": 0.5396,
+      "step": 4020
+    },
+    {
+      "epoch": 0.3426501917341287,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.3303406849234956e-06,
+      "loss": 0.6174,
+      "step": 4021
+    },
+    {
+      "epoch": 0.34273540690242865,
+      "grad_norm": 20.875,
+      "learning_rate": 2.330256149277883e-06,
+      "loss": 1.2139,
+      "step": 4022
+    },
+    {
+      "epoch": 0.3428206220707286,
+      "grad_norm": 20.25,
+      "learning_rate": 2.330171594110956e-06,
+      "loss": 0.904,
+      "step": 4023
+    },
+    {
+      "epoch": 0.34290583723902857,
+      "grad_norm": 33.25,
+      "learning_rate": 2.330087019424242e-06,
+      "loss": 0.9877,
+      "step": 4024
+    },
+    {
+      "epoch": 0.3429910524073285,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3300024252192692e-06,
+      "loss": 0.6767,
+      "step": 4025
+    },
+    {
+      "epoch": 0.3430762675756285,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.329917811497567e-06,
+      "loss": 0.6389,
+      "step": 4026
+    },
+    {
+      "epoch": 0.34316148274392844,
+      "grad_norm": 18.375,
+      "learning_rate": 2.3298331782606638e-06,
+      "loss": 0.7641,
+      "step": 4027
+    },
+    {
+      "epoch": 0.3432466979122284,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3297485255100892e-06,
+      "loss": 0.6028,
+      "step": 4028
+    },
+    {
+      "epoch": 0.34333191308052835,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.329663853247373e-06,
+      "loss": 0.4676,
+      "step": 4029
+    },
+    {
+      "epoch": 0.3434171282488283,
+      "grad_norm": 24.25,
+      "learning_rate": 2.3295791614740456e-06,
+      "loss": 0.8223,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34350234341712826,
+      "grad_norm": 11.75,
+      "learning_rate": 2.3294944501916364e-06,
+      "loss": 0.4249,
+      "step": 4031
+    },
+    {
+      "epoch": 0.3435875585854282,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3294097194016773e-06,
+      "loss": 0.5445,
+      "step": 4032
+    },
+    {
+      "epoch": 0.3436727737537282,
+      "grad_norm": 22.0,
+      "learning_rate": 2.3293249691056993e-06,
+      "loss": 1.0898,
+      "step": 4033
+    },
+    {
+      "epoch": 0.34375798892202813,
+      "grad_norm": 23.625,
+      "learning_rate": 2.3292401993052337e-06,
+      "loss": 0.9155,
+      "step": 4034
+    },
+    {
+      "epoch": 0.3438432040903281,
+      "grad_norm": 13.5,
+      "learning_rate": 2.329155410001812e-06,
+      "loss": 0.6164,
+      "step": 4035
+    },
+    {
+      "epoch": 0.34392841925862805,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3290706011969676e-06,
+      "loss": 0.6982,
+      "step": 4036
+    },
+    {
+      "epoch": 0.344013634426928,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.328985772892232e-06,
+      "loss": 0.5167,
+      "step": 4037
+    },
+    {
+      "epoch": 0.34409884959522796,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3289009250891385e-06,
+      "loss": 1.0156,
+      "step": 4038
+    },
+    {
+      "epoch": 0.3441840647635279,
+      "grad_norm": 16.0,
+      "learning_rate": 2.32881605778922e-06,
+      "loss": 0.8291,
+      "step": 4039
+    },
+    {
+      "epoch": 0.3442692799318279,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.328731170994011e-06,
+      "loss": 0.749,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34435449510012783,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.328646264705045e-06,
+      "loss": 0.8931,
+      "step": 4041
+    },
+    {
+      "epoch": 0.3444397102684278,
+      "grad_norm": 13.0,
+      "learning_rate": 2.328561338923856e-06,
+      "loss": 0.7366,
+      "step": 4042
+    },
+    {
+      "epoch": 0.34452492543672775,
+      "grad_norm": 26.25,
+      "learning_rate": 2.3284763936519795e-06,
+      "loss": 0.9489,
+      "step": 4043
+    },
+    {
+      "epoch": 0.3446101406050277,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.3283914288909498e-06,
+      "loss": 0.4795,
+      "step": 4044
+    },
+    {
+      "epoch": 0.34469535577332766,
+      "grad_norm": 14.25,
+      "learning_rate": 2.3283064446423024e-06,
+      "loss": 0.7264,
+      "step": 4045
+    },
+    {
+      "epoch": 0.3447805709416276,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.328221440907573e-06,
+      "loss": 0.7318,
+      "step": 4046
+    },
+    {
+      "epoch": 0.3448657861099276,
+      "grad_norm": 16.375,
+      "learning_rate": 2.3281364176882986e-06,
+      "loss": 0.992,
+      "step": 4047
+    },
+    {
+      "epoch": 0.34495100127822753,
+      "grad_norm": 24.25,
+      "learning_rate": 2.3280513749860147e-06,
+      "loss": 0.7895,
+      "step": 4048
+    },
+    {
+      "epoch": 0.3450362164465275,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3279663128022585e-06,
+      "loss": 0.7572,
+      "step": 4049
+    },
+    {
+      "epoch": 0.34512143161482745,
+      "grad_norm": 17.375,
+      "learning_rate": 2.327881231138567e-06,
+      "loss": 1.0086,
+      "step": 4050
+    },
+    {
+      "epoch": 0.3452066467831274,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.3277961299964773e-06,
+      "loss": 1.0919,
+      "step": 4051
+    },
+    {
+      "epoch": 0.34529186195142736,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3277110093775283e-06,
+      "loss": 0.9007,
+      "step": 4052
+    },
+    {
+      "epoch": 0.3453770771197273,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3276258692832577e-06,
+      "loss": 0.8951,
+      "step": 4053
+    },
+    {
+      "epoch": 0.3454622922880273,
+      "grad_norm": 25.75,
+      "learning_rate": 2.3275407097152036e-06,
+      "loss": 1.2109,
+      "step": 4054
+    },
+    {
+      "epoch": 0.34554750745632723,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3274555306749053e-06,
+      "loss": 0.9206,
+      "step": 4055
+    },
+    {
+      "epoch": 0.3456327226246272,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3273703321639025e-06,
+      "loss": 0.7756,
+      "step": 4056
+    },
+    {
+      "epoch": 0.34571793779292714,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3272851141837343e-06,
+      "loss": 0.8646,
+      "step": 4057
+    },
+    {
+      "epoch": 0.3458031529612271,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3271998767359407e-06,
+      "loss": 1.0957,
+      "step": 4058
+    },
+    {
+      "epoch": 0.34588836812952706,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.327114619822062e-06,
+      "loss": 0.5483,
+      "step": 4059
+    },
+    {
+      "epoch": 0.345973583297827,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3270293434436387e-06,
+      "loss": 1.0238,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34605879846612697,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.3269440476022124e-06,
+      "loss": 0.4951,
+      "step": 4061
+    },
+    {
+      "epoch": 0.34614401363442693,
+      "grad_norm": 11.875,
+      "learning_rate": 2.326858732299325e-06,
+      "loss": 0.5912,
+      "step": 4062
+    },
+    {
+      "epoch": 0.3462292288027269,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.3267733975365164e-06,
+      "loss": 0.3455,
+      "step": 4063
+    },
+    {
+      "epoch": 0.34631444397102684,
+      "grad_norm": 17.375,
+      "learning_rate": 2.32668804331533e-06,
+      "loss": 0.8845,
+      "step": 4064
+    },
+    {
+      "epoch": 0.3463996591393268,
+      "grad_norm": 11.625,
+      "learning_rate": 2.3266026696373086e-06,
+      "loss": 0.9656,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34648487430762676,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.3265172765039936e-06,
+      "loss": 0.8902,
+      "step": 4066
+    },
+    {
+      "epoch": 0.3465700894759267,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.326431863916929e-06,
+      "loss": 0.5413,
+      "step": 4067
+    },
+    {
+      "epoch": 0.34665530464422667,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.3263464318776583e-06,
+      "loss": 0.3425,
+      "step": 4068
+    },
+    {
+      "epoch": 0.3467405198125266,
+      "grad_norm": 15.375,
+      "learning_rate": 2.326260980387725e-06,
+      "loss": 1.1196,
+      "step": 4069
+    },
+    {
+      "epoch": 0.3468257349808266,
+      "grad_norm": 17.125,
+      "learning_rate": 2.326175509448674e-06,
+      "loss": 1.0365,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34691095014912654,
+      "grad_norm": 18.25,
+      "learning_rate": 2.326090019062049e-06,
+      "loss": 0.4944,
+      "step": 4071
+    },
+    {
+      "epoch": 0.3469961653174265,
+      "grad_norm": 16.0,
+      "learning_rate": 2.326004509229395e-06,
+      "loss": 0.9812,
+      "step": 4072
+    },
+    {
+      "epoch": 0.34708138048572645,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.325918979952258e-06,
+      "loss": 0.7273,
+      "step": 4073
+    },
+    {
+      "epoch": 0.3471665956540264,
+      "grad_norm": 12.75,
+      "learning_rate": 2.3258334312321832e-06,
+      "loss": 0.5361,
+      "step": 4074
+    },
+    {
+      "epoch": 0.34725181082232637,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3257478630707164e-06,
+      "loss": 0.6556,
+      "step": 4075
+    },
+    {
+      "epoch": 0.3473370259906263,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3256622754694037e-06,
+      "loss": 0.6668,
+      "step": 4076
+    },
+    {
+      "epoch": 0.3474222411589263,
+      "grad_norm": 20.75,
+      "learning_rate": 2.3255766684297924e-06,
+      "loss": 0.7155,
+      "step": 4077
+    },
+    {
+      "epoch": 0.34750745632722624,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.3254910419534288e-06,
+      "loss": 0.4792,
+      "step": 4078
+    },
+    {
+      "epoch": 0.3475926714955262,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.3254053960418603e-06,
+      "loss": 0.577,
+      "step": 4079
+    },
+    {
+      "epoch": 0.34767788666382615,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.3253197306966357e-06,
+      "loss": 0.7516,
+      "step": 4080
+    },
+    {
+      "epoch": 0.3477631018321261,
+      "grad_norm": 21.875,
+      "learning_rate": 2.3252340459193017e-06,
+      "loss": 0.8775,
+      "step": 4081
+    },
+    {
+      "epoch": 0.34784831700042607,
+      "grad_norm": 17.875,
+      "learning_rate": 2.325148341711407e-06,
+      "loss": 0.8316,
+      "step": 4082
+    },
+    {
+      "epoch": 0.347933532168726,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.3250626180745013e-06,
+      "loss": 0.3671,
+      "step": 4083
+    },
+    {
+      "epoch": 0.348018747337026,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.3249768750101325e-06,
+      "loss": 0.585,
+      "step": 4084
+    },
+    {
+      "epoch": 0.34810396250532594,
+      "grad_norm": 20.0,
+      "learning_rate": 2.3248911125198507e-06,
+      "loss": 1.0553,
+      "step": 4085
+    },
+    {
+      "epoch": 0.3481891776736259,
+      "grad_norm": 12.625,
+      "learning_rate": 2.324805330605206e-06,
+      "loss": 0.6835,
+      "step": 4086
+    },
+    {
+      "epoch": 0.34827439284192585,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.3247195292677475e-06,
+      "loss": 0.5109,
+      "step": 4087
+    },
+    {
+      "epoch": 0.3483596080102258,
+      "grad_norm": 16.0,
+      "learning_rate": 2.324633708509027e-06,
+      "loss": 0.9071,
+      "step": 4088
+    },
+    {
+      "epoch": 0.34844482317852576,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3245478683305948e-06,
+      "loss": 0.853,
+      "step": 4089
+    },
+    {
+      "epoch": 0.3485300383468257,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3244620087340017e-06,
+      "loss": 0.844,
+      "step": 4090
+    },
+    {
+      "epoch": 0.3486152535151257,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3243761297207995e-06,
+      "loss": 0.8373,
+      "step": 4091
+    },
+    {
+      "epoch": 0.34870046868342564,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3242902312925408e-06,
+      "loss": 0.7101,
+      "step": 4092
+    },
+    {
+      "epoch": 0.3487856838517256,
+      "grad_norm": 20.625,
+      "learning_rate": 2.324204313450777e-06,
+      "loss": 0.6203,
+      "step": 4093
+    },
+    {
+      "epoch": 0.34887089902002555,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.324118376197061e-06,
+      "loss": 0.6374,
+      "step": 4094
+    },
+    {
+      "epoch": 0.3489561141883255,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.324032419532946e-06,
+      "loss": 0.5917,
+      "step": 4095
+    },
+    {
+      "epoch": 0.34904132935662546,
+      "grad_norm": 13.375,
+      "learning_rate": 2.323946443459985e-06,
+      "loss": 0.7285,
+      "step": 4096
+    },
+    {
+      "epoch": 0.3491265445249254,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.323860447979732e-06,
+      "loss": 0.7153,
+      "step": 4097
+    },
+    {
+      "epoch": 0.3492117596932254,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.3237744330937407e-06,
+      "loss": 0.5982,
+      "step": 4098
+    },
+    {
+      "epoch": 0.34929697486152533,
+      "grad_norm": 9.5,
+      "learning_rate": 2.323688398803566e-06,
+      "loss": 0.5317,
+      "step": 4099
+    },
+    {
+      "epoch": 0.3493821900298253,
+      "grad_norm": 14.125,
+      "learning_rate": 2.323602345110762e-06,
+      "loss": 0.6596,
+      "step": 4100
+    },
+    {
+      "epoch": 0.34946740519812525,
+      "grad_norm": 14.0,
+      "learning_rate": 2.323516272016884e-06,
+      "loss": 0.6538,
+      "step": 4101
+    },
+    {
+      "epoch": 0.3495526203664252,
+      "grad_norm": 21.125,
+      "learning_rate": 2.3234301795234875e-06,
+      "loss": 0.5747,
+      "step": 4102
+    },
+    {
+      "epoch": 0.34963783553472516,
+      "grad_norm": 13.375,
+      "learning_rate": 2.3233440676321285e-06,
+      "loss": 0.7308,
+      "step": 4103
+    },
+    {
+      "epoch": 0.3497230507030251,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3232579363443627e-06,
+      "loss": 0.6837,
+      "step": 4104
+    },
+    {
+      "epoch": 0.3498082658713251,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3231717856617463e-06,
+      "loss": 0.4639,
+      "step": 4105
+    },
+    {
+      "epoch": 0.34989348103962503,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3230856155858366e-06,
+      "loss": 0.5849,
+      "step": 4106
+    },
+    {
+      "epoch": 0.349978696207925,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.322999426118191e-06,
+      "loss": 0.5821,
+      "step": 4107
+    },
+    {
+      "epoch": 0.35006391137622495,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.3229132172603667e-06,
+      "loss": 0.3634,
+      "step": 4108
+    },
+    {
+      "epoch": 0.3501491265445249,
+      "grad_norm": 14.875,
+      "learning_rate": 2.3228269890139222e-06,
+      "loss": 0.8087,
+      "step": 4109
+    },
+    {
+      "epoch": 0.35023434171282486,
+      "grad_norm": 16.75,
+      "learning_rate": 2.322740741380415e-06,
+      "loss": 0.8326,
+      "step": 4110
+    },
+    {
+      "epoch": 0.3503195568811248,
+      "grad_norm": 18.25,
+      "learning_rate": 2.3226544743614036e-06,
+      "loss": 0.8339,
+      "step": 4111
+    },
+    {
+      "epoch": 0.3504047720494248,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.3225681879584473e-06,
+      "loss": 0.6817,
+      "step": 4112
+    },
+    {
+      "epoch": 0.35048998721772473,
+      "grad_norm": 18.125,
+      "learning_rate": 2.3224818821731055e-06,
+      "loss": 1.0308,
+      "step": 4113
+    },
+    {
+      "epoch": 0.3505752023860247,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3223955570069375e-06,
+      "loss": 0.8212,
+      "step": 4114
+    },
+    {
+      "epoch": 0.35066041755432464,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.322309212461503e-06,
+      "loss": 0.8081,
+      "step": 4115
+    },
+    {
+      "epoch": 0.3507456327226246,
+      "grad_norm": 15.375,
+      "learning_rate": 2.322222848538363e-06,
+      "loss": 0.7543,
+      "step": 4116
+    },
+    {
+      "epoch": 0.35083084789092456,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3221364652390782e-06,
+      "loss": 0.7831,
+      "step": 4117
+    },
+    {
+      "epoch": 0.3509160630592245,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.322050062565209e-06,
+      "loss": 0.5824,
+      "step": 4118
+    },
+    {
+      "epoch": 0.3510012782275245,
+      "grad_norm": 16.75,
+      "learning_rate": 2.321963640518317e-06,
+      "loss": 1.0646,
+      "step": 4119
+    },
+    {
+      "epoch": 0.3510864933958245,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3218771990999644e-06,
+      "loss": 0.977,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35117170856412444,
+      "grad_norm": 13.25,
+      "learning_rate": 2.3217907383117132e-06,
+      "loss": 0.4908,
+      "step": 4121
+    },
+    {
+      "epoch": 0.3512569237324244,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3217042581551253e-06,
+      "loss": 0.4513,
+      "step": 4122
+    },
+    {
+      "epoch": 0.35134213890072435,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.3216177586317635e-06,
+      "loss": 0.7099,
+      "step": 4123
+    },
+    {
+      "epoch": 0.3514273540690243,
+      "grad_norm": 15.25,
+      "learning_rate": 2.3215312397431914e-06,
+      "loss": 0.7217,
+      "step": 4124
+    },
+    {
+      "epoch": 0.35151256923732427,
+      "grad_norm": 20.25,
+      "learning_rate": 2.3214447014909727e-06,
+      "loss": 0.7117,
+      "step": 4125
+    },
+    {
+      "epoch": 0.3515977844056242,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.3213581438766704e-06,
+      "loss": 0.5743,
+      "step": 4126
+    },
+    {
+      "epoch": 0.3516829995739242,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.321271566901849e-06,
+      "loss": 0.8363,
+      "step": 4127
+    },
+    {
+      "epoch": 0.35176821474222414,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3211849705680735e-06,
+      "loss": 0.7295,
+      "step": 4128
+    },
+    {
+      "epoch": 0.3518534299105241,
+      "grad_norm": 14.125,
+      "learning_rate": 2.3210983548769085e-06,
+      "loss": 0.539,
+      "step": 4129
+    },
+    {
+      "epoch": 0.35193864507882405,
+      "grad_norm": 16.125,
+      "learning_rate": 2.321011719829919e-06,
+      "loss": 0.8455,
+      "step": 4130
+    },
+    {
+      "epoch": 0.352023860247124,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3209250654286705e-06,
+      "loss": 0.8935,
+      "step": 4131
+    },
+    {
+      "epoch": 0.35210907541542397,
+      "grad_norm": 17.375,
+      "learning_rate": 2.3208383916747294e-06,
+      "loss": 0.7844,
+      "step": 4132
+    },
+    {
+      "epoch": 0.3521942905837239,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3207516985696616e-06,
+      "loss": 0.6487,
+      "step": 4133
+    },
+    {
+      "epoch": 0.3522795057520239,
+      "grad_norm": 11.375,
+      "learning_rate": 2.320664986115034e-06,
+      "loss": 0.8521,
+      "step": 4134
+    },
+    {
+      "epoch": 0.35236472092032384,
+      "grad_norm": 13.5,
+      "learning_rate": 2.3205782543124132e-06,
+      "loss": 0.641,
+      "step": 4135
+    },
+    {
+      "epoch": 0.3524499360886238,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3204915031633675e-06,
+      "loss": 0.4281,
+      "step": 4136
+    },
+    {
+      "epoch": 0.35253515125692375,
+      "grad_norm": 16.0,
+      "learning_rate": 2.320404732669463e-06,
+      "loss": 1.0232,
+      "step": 4137
+    },
+    {
+      "epoch": 0.3526203664252237,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3203179428322693e-06,
+      "loss": 0.5465,
+      "step": 4138
+    },
+    {
+      "epoch": 0.35270558159352366,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3202311336533538e-06,
+      "loss": 0.6673,
+      "step": 4139
+    },
+    {
+      "epoch": 0.3527907967618236,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3201443051342853e-06,
+      "loss": 0.8589,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3528760119301236,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.320057457276633e-06,
+      "loss": 0.7259,
+      "step": 4141
+    },
+    {
+      "epoch": 0.35296122709842354,
+      "grad_norm": 18.25,
+      "learning_rate": 2.319970590081967e-06,
+      "loss": 0.9552,
+      "step": 4142
+    },
+    {
+      "epoch": 0.3530464422667235,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3198837035518562e-06,
+      "loss": 0.4087,
+      "step": 4143
+    },
+    {
+      "epoch": 0.35313165743502345,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.319796797687871e-06,
+      "loss": 0.6732,
+      "step": 4144
+    },
+    {
+      "epoch": 0.3532168726033234,
+      "grad_norm": 20.875,
+      "learning_rate": 2.319709872491582e-06,
+      "loss": 0.8255,
+      "step": 4145
+    },
+    {
+      "epoch": 0.35330208777162336,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.31962292796456e-06,
+      "loss": 0.5269,
+      "step": 4146
+    },
+    {
+      "epoch": 0.3533873029399233,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3195359641083757e-06,
+      "loss": 0.5091,
+      "step": 4147
+    },
+    {
+      "epoch": 0.3534725181082233,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3194489809246017e-06,
+      "loss": 0.7482,
+      "step": 4148
+    },
+    {
+      "epoch": 0.35355773327652323,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3193619784148087e-06,
+      "loss": 0.908,
+      "step": 4149
+    },
+    {
+      "epoch": 0.3536429484448232,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.3192749565805693e-06,
+      "loss": 0.4892,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35372816361312315,
+      "grad_norm": 11.875,
+      "learning_rate": 2.3191879154234564e-06,
+      "loss": 0.6026,
+      "step": 4151
+    },
+    {
+      "epoch": 0.3538133787814231,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3191008549450426e-06,
+      "loss": 0.6892,
+      "step": 4152
+    },
+    {
+      "epoch": 0.35389859394972306,
+      "grad_norm": 10.75,
+      "learning_rate": 2.3190137751469016e-06,
+      "loss": 0.4586,
+      "step": 4153
+    },
+    {
+      "epoch": 0.353983809118023,
+      "grad_norm": 14.25,
+      "learning_rate": 2.318926676030607e-06,
+      "loss": 0.8373,
+      "step": 4154
+    },
+    {
+      "epoch": 0.354069024286323,
+      "grad_norm": 23.75,
+      "learning_rate": 2.3188395575977316e-06,
+      "loss": 0.9728,
+      "step": 4155
+    },
+    {
+      "epoch": 0.35415423945462293,
+      "grad_norm": 16.25,
+      "learning_rate": 2.318752419849851e-06,
+      "loss": 0.582,
+      "step": 4156
+    },
+    {
+      "epoch": 0.3542394546229229,
+      "grad_norm": 10.875,
+      "learning_rate": 2.3186652627885393e-06,
+      "loss": 0.5042,
+      "step": 4157
+    },
+    {
+      "epoch": 0.35432466979122285,
+      "grad_norm": 17.625,
+      "learning_rate": 2.318578086415372e-06,
+      "loss": 1.0334,
+      "step": 4158
+    },
+    {
+      "epoch": 0.3544098849595228,
+      "grad_norm": 15.5,
+      "learning_rate": 2.318490890731924e-06,
+      "loss": 0.5908,
+      "step": 4159
+    },
+    {
+      "epoch": 0.35449510012782276,
+      "grad_norm": 18.5,
+      "learning_rate": 2.318403675739771e-06,
+      "loss": 0.9349,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3545803152961227,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.3183164414404895e-06,
+      "loss": 0.6485,
+      "step": 4161
+    },
+    {
+      "epoch": 0.3546655304644227,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3182291878356554e-06,
+      "loss": 0.7995,
+      "step": 4162
+    },
+    {
+      "epoch": 0.35475074563272263,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.318141914926846e-06,
+      "loss": 0.8688,
+      "step": 4163
+    },
+    {
+      "epoch": 0.3548359608010226,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3180546227156376e-06,
+      "loss": 0.7732,
+      "step": 4164
+    },
+    {
+      "epoch": 0.35492117596932254,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.3179673112036083e-06,
+      "loss": 0.669,
+      "step": 4165
+    },
+    {
+      "epoch": 0.3550063911376225,
+      "grad_norm": 16.125,
+      "learning_rate": 2.317879980392336e-06,
+      "loss": 0.5297,
+      "step": 4166
+    },
+    {
+      "epoch": 0.35509160630592246,
+      "grad_norm": 16.25,
+      "learning_rate": 2.317792630283399e-06,
+      "loss": 0.7031,
+      "step": 4167
+    },
+    {
+      "epoch": 0.3551768214742224,
+      "grad_norm": 20.625,
+      "learning_rate": 2.3177052608783747e-06,
+      "loss": 0.9453,
+      "step": 4168
+    },
+    {
+      "epoch": 0.35526203664252237,
+      "grad_norm": 27.375,
+      "learning_rate": 2.3176178721788426e-06,
+      "loss": 1.2809,
+      "step": 4169
+    },
+    {
+      "epoch": 0.35534725181082233,
+      "grad_norm": 10.625,
+      "learning_rate": 2.317530464186382e-06,
+      "loss": 0.357,
+      "step": 4170
+    },
+    {
+      "epoch": 0.3554324669791223,
+      "grad_norm": 13.25,
+      "learning_rate": 2.317443036902573e-06,
+      "loss": 0.7564,
+      "step": 4171
+    },
+    {
+      "epoch": 0.35551768214742224,
+      "grad_norm": 15.375,
+      "learning_rate": 2.3173555903289945e-06,
+      "loss": 0.6833,
+      "step": 4172
+    },
+    {
+      "epoch": 0.3556028973157222,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3172681244672274e-06,
+      "loss": 0.667,
+      "step": 4173
+    },
+    {
+      "epoch": 0.35568811248402216,
+      "grad_norm": 14.625,
+      "learning_rate": 2.317180639318852e-06,
+      "loss": 0.8292,
+      "step": 4174
+    },
+    {
+      "epoch": 0.3557733276523221,
+      "grad_norm": 18.0,
+      "learning_rate": 2.3170931348854494e-06,
+      "loss": 0.966,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35585854282062207,
+      "grad_norm": 21.125,
+      "learning_rate": 2.3170056111686008e-06,
+      "loss": 1.1164,
+      "step": 4176
+    },
+    {
+      "epoch": 0.355943757988922,
+      "grad_norm": 12.0,
+      "learning_rate": 2.3169180681698874e-06,
+      "loss": 0.5023,
+      "step": 4177
+    },
+    {
+      "epoch": 0.356028973157222,
+      "grad_norm": 14.875,
+      "learning_rate": 2.316830505890892e-06,
+      "loss": 0.6285,
+      "step": 4178
+    },
+    {
+      "epoch": 0.35611418832552194,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3167429243331966e-06,
+      "loss": 0.8044,
+      "step": 4179
+    },
+    {
+      "epoch": 0.3561994034938219,
+      "grad_norm": 12.375,
+      "learning_rate": 2.3166553234983836e-06,
+      "loss": 0.6846,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35628461866212185,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3165677033880366e-06,
+      "loss": 0.5404,
+      "step": 4181
+    },
+    {
+      "epoch": 0.3563698338304218,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3164800640037387e-06,
+      "loss": 0.776,
+      "step": 4182
+    },
+    {
+      "epoch": 0.35645504899872177,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3163924053470737e-06,
+      "loss": 0.5105,
+      "step": 4183
+    },
+    {
+      "epoch": 0.3565402641670217,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3163047274196253e-06,
+      "loss": 0.7045,
+      "step": 4184
+    },
+    {
+      "epoch": 0.3566254793353217,
+      "grad_norm": 17.25,
+      "learning_rate": 2.316217030222978e-06,
+      "loss": 0.4571,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35671069450362164,
+      "grad_norm": 11.75,
+      "learning_rate": 2.3161293137587176e-06,
+      "loss": 0.5128,
+      "step": 4186
+    },
+    {
+      "epoch": 0.3567959096719216,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.316041578028428e-06,
+      "loss": 0.8485,
+      "step": 4187
+    },
+    {
+      "epoch": 0.35688112484022155,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.315953823033695e-06,
+      "loss": 0.5175,
+      "step": 4188
+    },
+    {
+      "epoch": 0.3569663400085215,
+      "grad_norm": 17.375,
+      "learning_rate": 2.3158660487761043e-06,
+      "loss": 0.9817,
+      "step": 4189
+    },
+    {
+      "epoch": 0.35705155517682147,
+      "grad_norm": 23.875,
+      "learning_rate": 2.3157782552572426e-06,
+      "loss": 1.1117,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3571367703451214,
+      "grad_norm": 16.75,
+      "learning_rate": 2.315690442478696e-06,
+      "loss": 0.8224,
+      "step": 4191
+    },
+    {
+      "epoch": 0.3572219855134214,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3156026104420515e-06,
+      "loss": 1.0135,
+      "step": 4192
+    },
+    {
+      "epoch": 0.35730720068172134,
+      "grad_norm": 20.875,
+      "learning_rate": 2.3155147591488965e-06,
+      "loss": 1.2968,
+      "step": 4193
+    },
+    {
+      "epoch": 0.3573924158500213,
+      "grad_norm": 16.875,
+      "learning_rate": 2.315426888600818e-06,
+      "loss": 0.8112,
+      "step": 4194
+    },
+    {
+      "epoch": 0.35747763101832125,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.3153389987994046e-06,
+      "loss": 0.6947,
+      "step": 4195
+    },
+    {
+      "epoch": 0.3575628461866212,
+      "grad_norm": 11.0,
+      "learning_rate": 2.3152510897462437e-06,
+      "loss": 0.5024,
+      "step": 4196
+    },
+    {
+      "epoch": 0.35764806135492117,
+      "grad_norm": 12.875,
+      "learning_rate": 2.315163161442925e-06,
+      "loss": 0.599,
+      "step": 4197
+    },
+    {
+      "epoch": 0.3577332765232211,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.3150752138910366e-06,
+      "loss": 0.8405,
+      "step": 4198
+    },
+    {
+      "epoch": 0.3578184916915211,
+      "grad_norm": 14.25,
+      "learning_rate": 2.314987247092168e-06,
+      "loss": 0.8189,
+      "step": 4199
+    },
+    {
+      "epoch": 0.35790370685982104,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.314899261047909e-06,
+      "loss": 0.6091,
+      "step": 4200
+    },
+    {
+      "epoch": 0.357988922028121,
+      "grad_norm": 11.25,
+      "learning_rate": 2.3148112557598497e-06,
+      "loss": 0.3602,
+      "step": 4201
+    },
+    {
+      "epoch": 0.35807413719642095,
+      "grad_norm": 12.625,
+      "learning_rate": 2.31472323122958e-06,
+      "loss": 0.9743,
+      "step": 4202
+    },
+    {
+      "epoch": 0.3581593523647209,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.314635187458691e-06,
+      "loss": 0.4343,
+      "step": 4203
+    },
+    {
+      "epoch": 0.35824456753302086,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.314547124448773e-06,
+      "loss": 0.4612,
+      "step": 4204
+    },
+    {
+      "epoch": 0.3583297827013208,
+      "grad_norm": 14.0,
+      "learning_rate": 2.314459042201419e-06,
+      "loss": 0.7736,
+      "step": 4205
+    },
+    {
+      "epoch": 0.3584149978696208,
+      "grad_norm": 12.0,
+      "learning_rate": 2.314370940718219e-06,
+      "loss": 0.6694,
+      "step": 4206
+    },
+    {
+      "epoch": 0.35850021303792073,
+      "grad_norm": 17.375,
+      "learning_rate": 2.314282820000766e-06,
+      "loss": 0.6235,
+      "step": 4207
+    },
+    {
+      "epoch": 0.3585854282062207,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.314194680050652e-06,
+      "loss": 0.8902,
+      "step": 4208
+    },
+    {
+      "epoch": 0.35867064337452065,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3141065208694704e-06,
+      "loss": 0.5892,
+      "step": 4209
+    },
+    {
+      "epoch": 0.3587558585428206,
+      "grad_norm": 24.0,
+      "learning_rate": 2.3140183424588136e-06,
+      "loss": 0.6459,
+      "step": 4210
+    },
+    {
+      "epoch": 0.35884107371112056,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3139301448202757e-06,
+      "loss": 0.6411,
+      "step": 4211
+    },
+    {
+      "epoch": 0.3589262888794205,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.31384192795545e-06,
+      "loss": 0.5189,
+      "step": 4212
+    },
+    {
+      "epoch": 0.3590115040477205,
+      "grad_norm": 28.125,
+      "learning_rate": 2.3137536918659307e-06,
+      "loss": 1.1267,
+      "step": 4213
+    },
+    {
+      "epoch": 0.35909671921602043,
+      "grad_norm": 13.75,
+      "learning_rate": 2.313665436553313e-06,
+      "loss": 0.6345,
+      "step": 4214
+    },
+    {
+      "epoch": 0.3591819343843204,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.313577162019191e-06,
+      "loss": 0.5901,
+      "step": 4215
+    },
+    {
+      "epoch": 0.35926714955262035,
+      "grad_norm": 15.875,
+      "learning_rate": 2.31348886826516e-06,
+      "loss": 0.8198,
+      "step": 4216
+    },
+    {
+      "epoch": 0.3593523647209203,
+      "grad_norm": 16.875,
+      "learning_rate": 2.313400555292816e-06,
+      "loss": 0.8849,
+      "step": 4217
+    },
+    {
+      "epoch": 0.35943757988922026,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.3133122231037547e-06,
+      "loss": 0.4802,
+      "step": 4218
+    },
+    {
+      "epoch": 0.3595227950575202,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.313223871699572e-06,
+      "loss": 0.592,
+      "step": 4219
+    },
+    {
+      "epoch": 0.3596080102258202,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.313135501081865e-06,
+      "loss": 0.8414,
+      "step": 4220
+    },
+    {
+      "epoch": 0.35969322539412013,
+      "grad_norm": 14.125,
+      "learning_rate": 2.3130471112522303e-06,
+      "loss": 0.7617,
+      "step": 4221
+    },
+    {
+      "epoch": 0.3597784405624201,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3129587022122652e-06,
+      "loss": 0.7365,
+      "step": 4222
+    },
+    {
+      "epoch": 0.35986365573072004,
+      "grad_norm": 10.875,
+      "learning_rate": 2.3128702739635677e-06,
+      "loss": 0.4697,
+      "step": 4223
+    },
+    {
+      "epoch": 0.35994887089902,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3127818265077357e-06,
+      "loss": 0.6869,
+      "step": 4224
+    },
+    {
+      "epoch": 0.36003408606731996,
+      "grad_norm": 24.125,
+      "learning_rate": 2.312693359846367e-06,
+      "loss": 0.863,
+      "step": 4225
+    },
+    {
+      "epoch": 0.3601193012356199,
+      "grad_norm": 17.5,
+      "learning_rate": 2.312604873981061e-06,
+      "loss": 0.8449,
+      "step": 4226
+    },
+    {
+      "epoch": 0.36020451640391987,
+      "grad_norm": 20.5,
+      "learning_rate": 2.312516368913416e-06,
+      "loss": 0.9352,
+      "step": 4227
+    },
+    {
+      "epoch": 0.36028973157221983,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.312427844645032e-06,
+      "loss": 1.0289,
+      "step": 4228
+    },
+    {
+      "epoch": 0.3603749467405198,
+      "grad_norm": 13.875,
+      "learning_rate": 2.3123393011775085e-06,
+      "loss": 0.7591,
+      "step": 4229
+    },
+    {
+      "epoch": 0.36046016190881974,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.3122507385124453e-06,
+      "loss": 0.5287,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3605453770771197,
+      "grad_norm": 11.5,
+      "learning_rate": 2.3121621566514433e-06,
+      "loss": 0.5529,
+      "step": 4231
+    },
+    {
+      "epoch": 0.3606305922454197,
+      "grad_norm": 15.25,
+      "learning_rate": 2.3120735555961034e-06,
+      "loss": 0.7931,
+      "step": 4232
+    },
+    {
+      "epoch": 0.36071580741371967,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.311984935348026e-06,
+      "loss": 1.0952,
+      "step": 4233
+    },
+    {
+      "epoch": 0.3608010225820196,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.311896295908813e-06,
+      "loss": 0.6672,
+      "step": 4234
+    },
+    {
+      "epoch": 0.3608862377503196,
+      "grad_norm": 11.625,
+      "learning_rate": 2.311807637280066e-06,
+      "loss": 0.4892,
+      "step": 4235
+    },
+    {
+      "epoch": 0.36097145291861954,
+      "grad_norm": 14.75,
+      "learning_rate": 2.3117189594633873e-06,
+      "loss": 0.9461,
+      "step": 4236
+    },
+    {
+      "epoch": 0.3610566680869195,
+      "grad_norm": 57.0,
+      "learning_rate": 2.3116302624603797e-06,
+      "loss": 1.0312,
+      "step": 4237
+    },
+    {
+      "epoch": 0.36114188325521945,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.311541546272645e-06,
+      "loss": 0.7095,
+      "step": 4238
+    },
+    {
+      "epoch": 0.3612270984235194,
+      "grad_norm": 12.5,
+      "learning_rate": 2.3114528109017876e-06,
+      "loss": 0.733,
+      "step": 4239
+    },
+    {
+      "epoch": 0.36131231359181937,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3113640563494105e-06,
+      "loss": 0.7853,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3613975287601193,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.3112752826171174e-06,
+      "loss": 0.3749,
+      "step": 4241
+    },
+    {
+      "epoch": 0.3614827439284193,
+      "grad_norm": 21.375,
+      "learning_rate": 2.3111864897065127e-06,
+      "loss": 0.9213,
+      "step": 4242
+    },
+    {
+      "epoch": 0.36156795909671924,
+      "grad_norm": 21.625,
+      "learning_rate": 2.3110976776192016e-06,
+      "loss": 0.694,
+      "step": 4243
+    },
+    {
+      "epoch": 0.3616531742650192,
+      "grad_norm": 11.25,
+      "learning_rate": 2.311008846356788e-06,
+      "loss": 0.5355,
+      "step": 4244
+    },
+    {
+      "epoch": 0.36173838943331915,
+      "grad_norm": 19.625,
+      "learning_rate": 2.3109199959208777e-06,
+      "loss": 1.1726,
+      "step": 4245
+    },
+    {
+      "epoch": 0.3618236046016191,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.310831126313076e-06,
+      "loss": 0.6114,
+      "step": 4246
+    },
+    {
+      "epoch": 0.36190881976991907,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3107422375349894e-06,
+      "loss": 0.9449,
+      "step": 4247
+    },
+    {
+      "epoch": 0.361994034938219,
+      "grad_norm": 17.25,
+      "learning_rate": 2.310653329588224e-06,
+      "loss": 0.9274,
+      "step": 4248
+    },
+    {
+      "epoch": 0.362079250106519,
+      "grad_norm": 16.625,
+      "learning_rate": 2.310564402474386e-06,
+      "loss": 0.9639,
+      "step": 4249
+    },
+    {
+      "epoch": 0.36216446527481894,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.3104754561950828e-06,
+      "loss": 0.8019,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3622496804431189,
+      "grad_norm": 15.375,
+      "learning_rate": 2.310386490751922e-06,
+      "loss": 0.8768,
+      "step": 4251
+    },
+    {
+      "epoch": 0.36233489561141885,
+      "grad_norm": 13.125,
+      "learning_rate": 2.310297506146511e-06,
+      "loss": 0.7227,
+      "step": 4252
+    },
+    {
+      "epoch": 0.3624201107797188,
+      "grad_norm": 12.375,
+      "learning_rate": 2.3102085023804576e-06,
+      "loss": 0.4608,
+      "step": 4253
+    },
+    {
+      "epoch": 0.36250532594801876,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3101194794553706e-06,
+      "loss": 0.9351,
+      "step": 4254
+    },
+    {
+      "epoch": 0.3625905411163187,
+      "grad_norm": 15.125,
+      "learning_rate": 2.3100304373728585e-06,
+      "loss": 0.6416,
+      "step": 4255
+    },
+    {
+      "epoch": 0.3626757562846187,
+      "grad_norm": 15.0,
+      "learning_rate": 2.3099413761345303e-06,
+      "loss": 0.781,
+      "step": 4256
+    },
+    {
+      "epoch": 0.36276097145291863,
+      "grad_norm": 16.625,
+      "learning_rate": 2.309852295741996e-06,
+      "loss": 0.5596,
+      "step": 4257
+    },
+    {
+      "epoch": 0.3628461866212186,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3097631961968646e-06,
+      "loss": 0.8913,
+      "step": 4258
+    },
+    {
+      "epoch": 0.36293140178951855,
+      "grad_norm": 20.375,
+      "learning_rate": 2.3096740775007472e-06,
+      "loss": 0.9099,
+      "step": 4259
+    },
+    {
+      "epoch": 0.3630166169578185,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.309584939655253e-06,
+      "loss": 0.9584,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36310183212611846,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.3094957826619933e-06,
+      "loss": 0.5139,
+      "step": 4261
+    },
+    {
+      "epoch": 0.3631870472944184,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3094066065225794e-06,
+      "loss": 1.0313,
+      "step": 4262
+    },
+    {
+      "epoch": 0.3632722624627184,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3093174112386234e-06,
+      "loss": 0.6139,
+      "step": 4263
+    },
+    {
+      "epoch": 0.36335747763101833,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3092281968117363e-06,
+      "loss": 0.6167,
+      "step": 4264
+    },
+    {
+      "epoch": 0.3634426927993183,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3091389632435305e-06,
+      "loss": 0.7595,
+      "step": 4265
+    },
+    {
+      "epoch": 0.36352790796761825,
+      "grad_norm": 15.75,
+      "learning_rate": 2.3090497105356187e-06,
+      "loss": 0.8667,
+      "step": 4266
+    },
+    {
+      "epoch": 0.3636131231359182,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.3089604386896132e-06,
+      "loss": 0.5804,
+      "step": 4267
+    },
+    {
+      "epoch": 0.36369833830421816,
+      "grad_norm": 16.25,
+      "learning_rate": 2.308871147707128e-06,
+      "loss": 0.7245,
+      "step": 4268
+    },
+    {
+      "epoch": 0.3637835534725181,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3087818375897767e-06,
+      "loss": 0.4998,
+      "step": 4269
+    },
+    {
+      "epoch": 0.3638687686408181,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.3086925083391727e-06,
+      "loss": 0.5352,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36395398380911803,
+      "grad_norm": 12.875,
+      "learning_rate": 2.3086031599569303e-06,
+      "loss": 0.6056,
+      "step": 4271
+    },
+    {
+      "epoch": 0.364039198977418,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3085137924446644e-06,
+      "loss": 0.7291,
+      "step": 4272
+    },
+    {
+      "epoch": 0.36412441414571795,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3084244058039902e-06,
+      "loss": 0.7364,
+      "step": 4273
+    },
+    {
+      "epoch": 0.3642096293140179,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3083350000365223e-06,
+      "loss": 0.6466,
+      "step": 4274
+    },
+    {
+      "epoch": 0.36429484448231786,
+      "grad_norm": 12.75,
+      "learning_rate": 2.3082455751438766e-06,
+      "loss": 0.3638,
+      "step": 4275
+    },
+    {
+      "epoch": 0.3643800596506178,
+      "grad_norm": 10.75,
+      "learning_rate": 2.308156131127669e-06,
+      "loss": 0.5419,
+      "step": 4276
+    },
+    {
+      "epoch": 0.3644652748189178,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3080666679895165e-06,
+      "loss": 0.622,
+      "step": 4277
+    },
+    {
+      "epoch": 0.36455048998721773,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.307977185731035e-06,
+      "loss": 0.6674,
+      "step": 4278
+    },
+    {
+      "epoch": 0.3646357051555177,
+      "grad_norm": 13.875,
+      "learning_rate": 2.3078876843538423e-06,
+      "loss": 0.4298,
+      "step": 4279
+    },
+    {
+      "epoch": 0.36472092032381764,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.3077981638595554e-06,
+      "loss": 0.5843,
+      "step": 4280
+    },
+    {
+      "epoch": 0.3648061354921176,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3077086242497916e-06,
+      "loss": 0.7285,
+      "step": 4281
+    },
+    {
+      "epoch": 0.36489135066041756,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3076190655261694e-06,
+      "loss": 0.5327,
+      "step": 4282
+    },
+    {
+      "epoch": 0.3649765658287175,
+      "grad_norm": 15.625,
+      "learning_rate": 2.3075294876903076e-06,
+      "loss": 1.0993,
+      "step": 4283
+    },
+    {
+      "epoch": 0.36506178099701747,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.307439890743824e-06,
+      "loss": 0.6302,
+      "step": 4284
+    },
+    {
+      "epoch": 0.36514699616531743,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3073502746883387e-06,
+      "loss": 0.6014,
+      "step": 4285
+    },
+    {
+      "epoch": 0.3652322113336174,
+      "grad_norm": 12.25,
+      "learning_rate": 2.30726063952547e-06,
+      "loss": 0.5546,
+      "step": 4286
+    },
+    {
+      "epoch": 0.36531742650191734,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.3071709852568392e-06,
+      "loss": 0.7087,
+      "step": 4287
+    },
+    {
+      "epoch": 0.3654026416702173,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.307081311884065e-06,
+      "loss": 0.9438,
+      "step": 4288
+    },
+    {
+      "epoch": 0.36548785683851726,
+      "grad_norm": 10.125,
+      "learning_rate": 2.3069916194087687e-06,
+      "loss": 0.2662,
+      "step": 4289
+    },
+    {
+      "epoch": 0.3655730720068172,
+      "grad_norm": 18.0,
+      "learning_rate": 2.3069019078325713e-06,
+      "loss": 0.9825,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36565828717511717,
+      "grad_norm": 11.0,
+      "learning_rate": 2.306812177157093e-06,
+      "loss": 0.4794,
+      "step": 4291
+    },
+    {
+      "epoch": 0.3657435023434171,
+      "grad_norm": 23.625,
+      "learning_rate": 2.3067224273839562e-06,
+      "loss": 0.8132,
+      "step": 4292
+    },
+    {
+      "epoch": 0.3658287175117171,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3066326585147827e-06,
+      "loss": 0.6164,
+      "step": 4293
+    },
+    {
+      "epoch": 0.36591393268001704,
+      "grad_norm": 13.875,
+      "learning_rate": 2.3065428705511946e-06,
+      "loss": 0.8203,
+      "step": 4294
+    },
+    {
+      "epoch": 0.365999147848317,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.306453063494814e-06,
+      "loss": 0.5603,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36608436301661695,
+      "grad_norm": 25.875,
+      "learning_rate": 2.3063632373472645e-06,
+      "loss": 0.9298,
+      "step": 4296
+    },
+    {
+      "epoch": 0.3661695781849169,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.306273392110169e-06,
+      "loss": 0.4272,
+      "step": 4297
+    },
+    {
+      "epoch": 0.36625479335321687,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.3061835277851514e-06,
+      "loss": 0.4046,
+      "step": 4298
+    },
+    {
+      "epoch": 0.3663400085215168,
+      "grad_norm": 27.0,
+      "learning_rate": 2.3060936443738347e-06,
+      "loss": 0.7893,
+      "step": 4299
+    },
+    {
+      "epoch": 0.3664252236898168,
+      "grad_norm": 9.875,
+      "learning_rate": 2.3060037418778446e-06,
+      "loss": 0.9641,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36651043885811674,
+      "grad_norm": 10.75,
+      "learning_rate": 2.3059138202988044e-06,
+      "loss": 0.5021,
+      "step": 4301
+    },
+    {
+      "epoch": 0.3665956540264167,
+      "grad_norm": 30.375,
+      "learning_rate": 2.3058238796383396e-06,
+      "loss": 0.8703,
+      "step": 4302
+    },
+    {
+      "epoch": 0.36668086919471665,
+      "grad_norm": 18.0,
+      "learning_rate": 2.305733919898076e-06,
+      "loss": 1.1952,
+      "step": 4303
+    },
+    {
+      "epoch": 0.3667660843630166,
+      "grad_norm": 18.375,
+      "learning_rate": 2.305643941079639e-06,
+      "loss": 0.6789,
+      "step": 4304
+    },
+    {
+      "epoch": 0.36685129953131657,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.305553943184654e-06,
+      "loss": 0.8883,
+      "step": 4305
+    },
+    {
+      "epoch": 0.3669365146996165,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.3054639262147483e-06,
+      "loss": 0.4696,
+      "step": 4306
+    },
+    {
+      "epoch": 0.3670217298679165,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.305373890171548e-06,
+      "loss": 0.9156,
+      "step": 4307
+    },
+    {
+      "epoch": 0.36710694503621644,
+      "grad_norm": 21.5,
+      "learning_rate": 2.3052838350566797e-06,
+      "loss": 0.8899,
+      "step": 4308
+    },
+    {
+      "epoch": 0.3671921602045164,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3051937608717716e-06,
+      "loss": 0.5692,
+      "step": 4309
+    },
+    {
+      "epoch": 0.36727737537281635,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3051036676184514e-06,
+      "loss": 0.7317,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3673625905411163,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.3050135552983467e-06,
+      "loss": 0.4207,
+      "step": 4311
+    },
+    {
+      "epoch": 0.36744780570941626,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3049234239130864e-06,
+      "loss": 0.6036,
+      "step": 4312
+    },
+    {
+      "epoch": 0.3675330208777162,
+      "grad_norm": 16.5,
+      "learning_rate": 2.304833273464299e-06,
+      "loss": 0.8124,
+      "step": 4313
+    },
+    {
+      "epoch": 0.3676182360460162,
+      "grad_norm": 19.0,
+      "learning_rate": 2.304743103953613e-06,
+      "loss": 0.7817,
+      "step": 4314
+    },
+    {
+      "epoch": 0.36770345121431613,
+      "grad_norm": 17.75,
+      "learning_rate": 2.304652915382659e-06,
+      "loss": 0.9005,
+      "step": 4315
+    },
+    {
+      "epoch": 0.3677886663826161,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3045627077530664e-06,
+      "loss": 0.859,
+      "step": 4316
+    },
+    {
+      "epoch": 0.36787388155091605,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.3044724810664647e-06,
+      "loss": 0.6488,
+      "step": 4317
+    },
+    {
+      "epoch": 0.367959096719216,
+      "grad_norm": 19.0,
+      "learning_rate": 2.3043822353244856e-06,
+      "loss": 0.6512,
+      "step": 4318
+    },
+    {
+      "epoch": 0.36804431188751596,
+      "grad_norm": 14.125,
+      "learning_rate": 2.304291970528759e-06,
+      "loss": 0.5366,
+      "step": 4319
+    },
+    {
+      "epoch": 0.3681295270558159,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.3042016866809165e-06,
+      "loss": 0.5852,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3682147422241159,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.304111383782589e-06,
+      "loss": 0.6552,
+      "step": 4321
+    },
+    {
+      "epoch": 0.36829995739241583,
+      "grad_norm": 13.5,
+      "learning_rate": 2.304021061835409e-06,
+      "loss": 0.7635,
+      "step": 4322
+    },
+    {
+      "epoch": 0.3683851725607158,
+      "grad_norm": 14.0,
+      "learning_rate": 2.303930720841008e-06,
+      "loss": 0.5679,
+      "step": 4323
+    },
+    {
+      "epoch": 0.36847038772901575,
+      "grad_norm": 12.625,
+      "learning_rate": 2.30384036080102e-06,
+      "loss": 0.7854,
+      "step": 4324
+    },
+    {
+      "epoch": 0.3685556028973157,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.3037499817170765e-06,
+      "loss": 0.7057,
+      "step": 4325
+    },
+    {
+      "epoch": 0.36864081806561566,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.303659583590811e-06,
+      "loss": 0.7431,
+      "step": 4326
+    },
+    {
+      "epoch": 0.3687260332339156,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.303569166423858e-06,
+      "loss": 0.6521,
+      "step": 4327
+    },
+    {
+      "epoch": 0.3688112484022156,
+      "grad_norm": 12.25,
+      "learning_rate": 2.3034787302178503e-06,
+      "loss": 0.559,
+      "step": 4328
+    },
+    {
+      "epoch": 0.36889646357051553,
+      "grad_norm": 14.25,
+      "learning_rate": 2.3033882749744223e-06,
+      "loss": 0.7696,
+      "step": 4329
+    },
+    {
+      "epoch": 0.3689816787388155,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.3032978006952087e-06,
+      "loss": 0.6558,
+      "step": 4330
+    },
+    {
+      "epoch": 0.36906689390711545,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3032073073818456e-06,
+      "loss": 0.8422,
+      "step": 4331
+    },
+    {
+      "epoch": 0.3691521090754154,
+      "grad_norm": 19.625,
+      "learning_rate": 2.3031167950359668e-06,
+      "loss": 1.0157,
+      "step": 4332
+    },
+    {
+      "epoch": 0.36923732424371536,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.3030262636592084e-06,
+      "loss": 0.9339,
+      "step": 4333
+    },
+    {
+      "epoch": 0.3693225394120153,
+      "grad_norm": 20.25,
+      "learning_rate": 2.3029357132532072e-06,
+      "loss": 0.6715,
+      "step": 4334
+    },
+    {
+      "epoch": 0.3694077545803153,
+      "grad_norm": 16.875,
+      "learning_rate": 2.302845143819598e-06,
+      "loss": 0.7813,
+      "step": 4335
+    },
+    {
+      "epoch": 0.36949296974861523,
+      "grad_norm": 20.75,
+      "learning_rate": 2.302754555360019e-06,
+      "loss": 1.2755,
+      "step": 4336
+    },
+    {
+      "epoch": 0.3695781849169152,
+      "grad_norm": 13.125,
+      "learning_rate": 2.302663947876106e-06,
+      "loss": 0.504,
+      "step": 4337
+    },
+    {
+      "epoch": 0.36966340008521514,
+      "grad_norm": 9.25,
+      "learning_rate": 2.3025733213694972e-06,
+      "loss": 0.5204,
+      "step": 4338
+    },
+    {
+      "epoch": 0.3697486152535151,
+      "grad_norm": 12.75,
+      "learning_rate": 2.30248267584183e-06,
+      "loss": 0.5122,
+      "step": 4339
+    },
+    {
+      "epoch": 0.36983383042181506,
+      "grad_norm": 24.625,
+      "learning_rate": 2.3023920112947426e-06,
+      "loss": 0.917,
+      "step": 4340
+    },
+    {
+      "epoch": 0.369919045590115,
+      "grad_norm": 11.75,
+      "learning_rate": 2.3023013277298733e-06,
+      "loss": 0.7611,
+      "step": 4341
+    },
+    {
+      "epoch": 0.37000426075841497,
+      "grad_norm": 15.625,
+      "learning_rate": 2.3022106251488606e-06,
+      "loss": 0.9692,
+      "step": 4342
+    },
+    {
+      "epoch": 0.37008947592671493,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3021199035533444e-06,
+      "loss": 0.5794,
+      "step": 4343
+    },
+    {
+      "epoch": 0.3701746910950149,
+      "grad_norm": 12.125,
+      "learning_rate": 2.302029162944963e-06,
+      "loss": 0.6374,
+      "step": 4344
+    },
+    {
+      "epoch": 0.3702599062633149,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.301938403325357e-06,
+      "loss": 0.6702,
+      "step": 4345
+    },
+    {
+      "epoch": 0.37034512143161485,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.301847624696166e-06,
+      "loss": 0.5704,
+      "step": 4346
+    },
+    {
+      "epoch": 0.3704303365999148,
+      "grad_norm": 26.25,
+      "learning_rate": 2.301756827059031e-06,
+      "loss": 1.0872,
+      "step": 4347
+    },
+    {
+      "epoch": 0.37051555176821477,
+      "grad_norm": 15.125,
+      "learning_rate": 2.3016660104155917e-06,
+      "loss": 0.7761,
+      "step": 4348
+    },
+    {
+      "epoch": 0.3706007669365147,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.3015751747674907e-06,
+      "loss": 0.5179,
+      "step": 4349
+    },
+    {
+      "epoch": 0.3706859821048147,
+      "grad_norm": 14.625,
+      "learning_rate": 2.301484320116369e-06,
+      "loss": 0.9032,
+      "step": 4350
+    },
+    {
+      "epoch": 0.37077119727311464,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.301393446463868e-06,
+      "loss": 1.1451,
+      "step": 4351
+    },
+    {
+      "epoch": 0.3708564124414146,
+      "grad_norm": 27.125,
+      "learning_rate": 2.3013025538116305e-06,
+      "loss": 1.2107,
+      "step": 4352
+    },
+    {
+      "epoch": 0.37094162760971455,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3012116421612985e-06,
+      "loss": 0.9511,
+      "step": 4353
+    },
+    {
+      "epoch": 0.3710268427780145,
+      "grad_norm": 14.25,
+      "learning_rate": 2.301120711514515e-06,
+      "loss": 0.7931,
+      "step": 4354
+    },
+    {
+      "epoch": 0.37111205794631447,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3010297618729233e-06,
+      "loss": 0.5906,
+      "step": 4355
+    },
+    {
+      "epoch": 0.3711972731146144,
+      "grad_norm": 14.0,
+      "learning_rate": 2.3009387932381665e-06,
+      "loss": 0.6769,
+      "step": 4356
+    },
+    {
+      "epoch": 0.3712824882829144,
+      "grad_norm": 15.875,
+      "learning_rate": 2.3008478056118896e-06,
+      "loss": 0.8732,
+      "step": 4357
+    },
+    {
+      "epoch": 0.37136770345121434,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.3007567989957358e-06,
+      "loss": 0.3986,
+      "step": 4358
+    },
+    {
+      "epoch": 0.3714529186195143,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.30066577339135e-06,
+      "loss": 0.7206,
+      "step": 4359
+    },
+    {
+      "epoch": 0.37153813378781425,
+      "grad_norm": 16.375,
+      "learning_rate": 2.3005747288003774e-06,
+      "loss": 0.8717,
+      "step": 4360
+    },
+    {
+      "epoch": 0.3716233489561142,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3004836652244626e-06,
+      "loss": 0.6998,
+      "step": 4361
+    },
+    {
+      "epoch": 0.37170856412441416,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.300392582665252e-06,
+      "loss": 0.579,
+      "step": 4362
+    },
+    {
+      "epoch": 0.3717937792927141,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.300301481124391e-06,
+      "loss": 0.8073,
+      "step": 4363
+    },
+    {
+      "epoch": 0.3718789944610141,
+      "grad_norm": 13.75,
+      "learning_rate": 2.3002103606035265e-06,
+      "loss": 0.6729,
+      "step": 4364
+    },
+    {
+      "epoch": 0.37196420962931404,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.300119221104304e-06,
+      "loss": 0.3388,
+      "step": 4365
+    },
+    {
+      "epoch": 0.372049424797614,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3000280626283718e-06,
+      "loss": 0.6479,
+      "step": 4366
+    },
+    {
+      "epoch": 0.37213463996591395,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2999368851773764e-06,
+      "loss": 0.8203,
+      "step": 4367
+    },
+    {
+      "epoch": 0.3722198551342139,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2998456887529654e-06,
+      "loss": 0.7895,
+      "step": 4368
+    },
+    {
+      "epoch": 0.37230507030251386,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.299754473356787e-06,
+      "loss": 0.8478,
+      "step": 4369
+    },
+    {
+      "epoch": 0.3723902854708138,
+      "grad_norm": 20.625,
+      "learning_rate": 2.2996632389904903e-06,
+      "loss": 1.0137,
+      "step": 4370
+    },
+    {
+      "epoch": 0.3724755006391138,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2995719856557232e-06,
+      "loss": 1.0345,
+      "step": 4371
+    },
+    {
+      "epoch": 0.37256071580741373,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.2994807133541347e-06,
+      "loss": 0.7889,
+      "step": 4372
+    },
+    {
+      "epoch": 0.3726459309757137,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2993894220873743e-06,
+      "loss": 0.9149,
+      "step": 4373
+    },
+    {
+      "epoch": 0.37273114614401365,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.299298111857092e-06,
+      "loss": 0.4575,
+      "step": 4374
+    },
+    {
+      "epoch": 0.3728163613123136,
+      "grad_norm": 28.0,
+      "learning_rate": 2.299206782664938e-06,
+      "loss": 0.4605,
+      "step": 4375
+    },
+    {
+      "epoch": 0.37290157648061356,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2991154345125617e-06,
+      "loss": 0.9573,
+      "step": 4376
+    },
+    {
+      "epoch": 0.3729867916489135,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.299024067401615e-06,
+      "loss": 0.4191,
+      "step": 4377
+    },
+    {
+      "epoch": 0.3730720068172135,
+      "grad_norm": 13.5,
+      "learning_rate": 2.2989326813337483e-06,
+      "loss": 0.6309,
+      "step": 4378
+    },
+    {
+      "epoch": 0.37315722198551343,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2988412763106135e-06,
+      "loss": 0.458,
+      "step": 4379
+    },
+    {
+      "epoch": 0.3732424371538134,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.2987498523338615e-06,
+      "loss": 0.7933,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37332765232211335,
+      "grad_norm": 12.875,
+      "learning_rate": 2.2986584094051455e-06,
+      "loss": 0.6122,
+      "step": 4381
+    },
+    {
+      "epoch": 0.3734128674904133,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2985669475261175e-06,
+      "loss": 0.6858,
+      "step": 4382
+    },
+    {
+      "epoch": 0.37349808265871326,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2984754666984306e-06,
+      "loss": 0.7758,
+      "step": 4383
+    },
+    {
+      "epoch": 0.3735832978270132,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2983839669237372e-06,
+      "loss": 0.813,
+      "step": 4384
+    },
+    {
+      "epoch": 0.3736685129953132,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2982924482036917e-06,
+      "loss": 0.5505,
+      "step": 4385
+    },
+    {
+      "epoch": 0.37375372816361313,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.298200910539947e-06,
+      "loss": 0.7563,
+      "step": 4386
+    },
+    {
+      "epoch": 0.3738389433319131,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2981093539341588e-06,
+      "loss": 0.7459,
+      "step": 4387
+    },
+    {
+      "epoch": 0.37392415850021304,
+      "grad_norm": 18.25,
+      "learning_rate": 2.29801777838798e-06,
+      "loss": 0.9212,
+      "step": 4388
+    },
+    {
+      "epoch": 0.374009373668513,
+      "grad_norm": 19.125,
+      "learning_rate": 2.2979261839030657e-06,
+      "loss": 0.8884,
+      "step": 4389
+    },
+    {
+      "epoch": 0.37409458883681296,
+      "grad_norm": 16.375,
+      "learning_rate": 2.297834570481072e-06,
+      "loss": 0.8419,
+      "step": 4390
+    },
+    {
+      "epoch": 0.3741798040051129,
+      "grad_norm": 18.75,
+      "learning_rate": 2.297742938123654e-06,
+      "loss": 0.7831,
+      "step": 4391
+    },
+    {
+      "epoch": 0.37426501917341287,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.2976512868324675e-06,
+      "loss": 0.4998,
+      "step": 4392
+    },
+    {
+      "epoch": 0.37435023434171283,
+      "grad_norm": 10.5,
+      "learning_rate": 2.2975596166091686e-06,
+      "loss": 0.3399,
+      "step": 4393
+    },
+    {
+      "epoch": 0.3744354495100128,
+      "grad_norm": 10.625,
+      "learning_rate": 2.297467927455414e-06,
+      "loss": 0.3401,
+      "step": 4394
+    },
+    {
+      "epoch": 0.37452066467831274,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2973762193728606e-06,
+      "loss": 1.1352,
+      "step": 4395
+    },
+    {
+      "epoch": 0.3746058798466127,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.2972844923631655e-06,
+      "loss": 0.8937,
+      "step": 4396
+    },
+    {
+      "epoch": 0.37469109501491266,
+      "grad_norm": 12.875,
+      "learning_rate": 2.297192746427987e-06,
+      "loss": 0.8557,
+      "step": 4397
+    },
+    {
+      "epoch": 0.3747763101832126,
+      "grad_norm": 28.25,
+      "learning_rate": 2.297100981568982e-06,
+      "loss": 0.9301,
+      "step": 4398
+    },
+    {
+      "epoch": 0.37486152535151257,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2970091977878097e-06,
+      "loss": 0.7017,
+      "step": 4399
+    },
+    {
+      "epoch": 0.3749467405198125,
+      "grad_norm": 13.375,
+      "learning_rate": 2.296917395086128e-06,
+      "loss": 0.8333,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3750319556881125,
+      "grad_norm": 17.625,
+      "learning_rate": 2.2968255734655963e-06,
+      "loss": 0.895,
+      "step": 4401
+    },
+    {
+      "epoch": 0.37511717085641244,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2967337329278746e-06,
+      "loss": 0.9576,
+      "step": 4402
+    },
+    {
+      "epoch": 0.3752023860247124,
+      "grad_norm": 18.5,
+      "learning_rate": 2.296641873474621e-06,
+      "loss": 0.9124,
+      "step": 4403
+    },
+    {
+      "epoch": 0.37528760119301235,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.2965499951074963e-06,
+      "loss": 0.7816,
+      "step": 4404
+    },
+    {
+      "epoch": 0.3753728163613123,
+      "grad_norm": 19.625,
+      "learning_rate": 2.296458097828161e-06,
+      "loss": 0.8399,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37545803152961227,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2963661816382755e-06,
+      "loss": 1.1387,
+      "step": 4406
+    },
+    {
+      "epoch": 0.3755432466979122,
+      "grad_norm": 16.75,
+      "learning_rate": 2.296274246539501e-06,
+      "loss": 1.0137,
+      "step": 4407
+    },
+    {
+      "epoch": 0.3756284618662122,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2961822925334985e-06,
+      "loss": 0.9218,
+      "step": 4408
+    },
+    {
+      "epoch": 0.37571367703451214,
+      "grad_norm": 9.4375,
+      "learning_rate": 2.29609031962193e-06,
+      "loss": 0.3351,
+      "step": 4409
+    },
+    {
+      "epoch": 0.3757988922028121,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.2959983278064576e-06,
+      "loss": 1.0254,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37588410737111205,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.2959063170887435e-06,
+      "loss": 0.7664,
+      "step": 4411
+    },
+    {
+      "epoch": 0.375969322539412,
+      "grad_norm": 14.25,
+      "learning_rate": 2.295814287470451e-06,
+      "loss": 0.4341,
+      "step": 4412
+    },
+    {
+      "epoch": 0.37605453770771197,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.2957222389532423e-06,
+      "loss": 0.8277,
+      "step": 4413
+    },
+    {
+      "epoch": 0.3761397528760119,
+      "grad_norm": 19.375,
+      "learning_rate": 2.295630171538781e-06,
+      "loss": 0.5334,
+      "step": 4414
+    },
+    {
+      "epoch": 0.3762249680443119,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.2955380852287314e-06,
+      "loss": 0.572,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37631018321261184,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.295445980024757e-06,
+      "loss": 0.5813,
+      "step": 4416
+    },
+    {
+      "epoch": 0.3763953983809118,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2953538559285223e-06,
+      "loss": 1.1529,
+      "step": 4417
+    },
+    {
+      "epoch": 0.37648061354921175,
+      "grad_norm": 14.875,
+      "learning_rate": 2.2952617129416926e-06,
+      "loss": 0.7758,
+      "step": 4418
+    },
+    {
+      "epoch": 0.3765658287175117,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.295169551065932e-06,
+      "loss": 0.3814,
+      "step": 4419
+    },
+    {
+      "epoch": 0.37665104388581166,
+      "grad_norm": 20.625,
+      "learning_rate": 2.295077370302907e-06,
+      "loss": 0.6561,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3767362590541116,
+      "grad_norm": 16.0,
+      "learning_rate": 2.2949851706542837e-06,
+      "loss": 1.204,
+      "step": 4421
+    },
+    {
+      "epoch": 0.3768214742224116,
+      "grad_norm": 13.0,
+      "learning_rate": 2.2948929521217268e-06,
+      "loss": 0.7617,
+      "step": 4422
+    },
+    {
+      "epoch": 0.37690668939071154,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.294800714706904e-06,
+      "loss": 0.4446,
+      "step": 4423
+    },
+    {
+      "epoch": 0.3769919045590115,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2947084584114808e-06,
+      "loss": 1.0536,
+      "step": 4424
+    },
+    {
+      "epoch": 0.37707711972731145,
+      "grad_norm": 11.375,
+      "learning_rate": 2.294616183237126e-06,
+      "loss": 0.3647,
+      "step": 4425
+    },
+    {
+      "epoch": 0.3771623348956114,
+      "grad_norm": 16.625,
+      "learning_rate": 2.294523889185506e-06,
+      "loss": 0.9839,
+      "step": 4426
+    },
+    {
+      "epoch": 0.37724755006391136,
+      "grad_norm": 13.375,
+      "learning_rate": 2.2944315762582893e-06,
+      "loss": 0.5052,
+      "step": 4427
+    },
+    {
+      "epoch": 0.3773327652322113,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.2943392444571435e-06,
+      "loss": 0.8716,
+      "step": 4428
+    },
+    {
+      "epoch": 0.3774179804005113,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2942468937837377e-06,
+      "loss": 0.9572,
+      "step": 4429
+    },
+    {
+      "epoch": 0.37750319556881123,
+      "grad_norm": 14.0,
+      "learning_rate": 2.29415452423974e-06,
+      "loss": 0.4185,
+      "step": 4430
+    },
+    {
+      "epoch": 0.3775884107371112,
+      "grad_norm": 17.75,
+      "learning_rate": 2.29406213582682e-06,
+      "loss": 1.0528,
+      "step": 4431
+    },
+    {
+      "epoch": 0.37767362590541115,
+      "grad_norm": 11.625,
+      "learning_rate": 2.293969728546648e-06,
+      "loss": 0.4112,
+      "step": 4432
+    },
+    {
+      "epoch": 0.3777588410737111,
+      "grad_norm": 18.25,
+      "learning_rate": 2.2938773024008924e-06,
+      "loss": 1.0375,
+      "step": 4433
+    },
+    {
+      "epoch": 0.37784405624201106,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2937848573912246e-06,
+      "loss": 0.776,
+      "step": 4434
+    },
+    {
+      "epoch": 0.377929271410311,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.293692393519315e-06,
+      "loss": 0.6619,
+      "step": 4435
+    },
+    {
+      "epoch": 0.378014486578611,
+      "grad_norm": 24.5,
+      "learning_rate": 2.293599910786834e-06,
+      "loss": 1.2235,
+      "step": 4436
+    },
+    {
+      "epoch": 0.37809970174691093,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2935074091954535e-06,
+      "loss": 0.6463,
+      "step": 4437
+    },
+    {
+      "epoch": 0.3781849169152109,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2934148887468448e-06,
+      "loss": 0.5819,
+      "step": 4438
+    },
+    {
+      "epoch": 0.37827013208351085,
+      "grad_norm": 10.0,
+      "learning_rate": 2.29332234944268e-06,
+      "loss": 0.3957,
+      "step": 4439
+    },
+    {
+      "epoch": 0.3783553472518108,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.2932297912846307e-06,
+      "loss": 0.5868,
+      "step": 4440
+    },
+    {
+      "epoch": 0.37844056242011076,
+      "grad_norm": 18.875,
+      "learning_rate": 2.2931372142743703e-06,
+      "loss": 0.9452,
+      "step": 4441
+    },
+    {
+      "epoch": 0.3785257775884107,
+      "grad_norm": 17.25,
+      "learning_rate": 2.2930446184135712e-06,
+      "loss": 1.2007,
+      "step": 4442
+    },
+    {
+      "epoch": 0.3786109927567107,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.292952003703907e-06,
+      "loss": 0.769,
+      "step": 4443
+    },
+    {
+      "epoch": 0.37869620792501063,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.2928593701470516e-06,
+      "loss": 0.5402,
+      "step": 4444
+    },
+    {
+      "epoch": 0.3787814230933106,
+      "grad_norm": 17.5,
+      "learning_rate": 2.292766717744679e-06,
+      "loss": 0.5832,
+      "step": 4445
+    },
+    {
+      "epoch": 0.37886663826161054,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.2926740464984625e-06,
+      "loss": 0.7219,
+      "step": 4446
+    },
+    {
+      "epoch": 0.3789518534299105,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.292581356410078e-06,
+      "loss": 0.5683,
+      "step": 4447
+    },
+    {
+      "epoch": 0.37903706859821046,
+      "grad_norm": 13.5,
+      "learning_rate": 2.2924886474811995e-06,
+      "loss": 0.4263,
+      "step": 4448
+    },
+    {
+      "epoch": 0.3791222837665104,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.2923959197135033e-06,
+      "loss": 0.4058,
+      "step": 4449
+    },
+    {
+      "epoch": 0.37920749893481037,
+      "grad_norm": 15.875,
+      "learning_rate": 2.2923031731086646e-06,
+      "loss": 0.7828,
+      "step": 4450
+    },
+    {
+      "epoch": 0.37929271410311033,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2922104076683594e-06,
+      "loss": 0.5502,
+      "step": 4451
+    },
+    {
+      "epoch": 0.3793779292714103,
+      "grad_norm": 14.25,
+      "learning_rate": 2.292117623394264e-06,
+      "loss": 0.7631,
+      "step": 4452
+    },
+    {
+      "epoch": 0.37946314443971024,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2920248202880552e-06,
+      "loss": 0.8136,
+      "step": 4453
+    },
+    {
+      "epoch": 0.3795483596080102,
+      "grad_norm": 20.0,
+      "learning_rate": 2.29193199835141e-06,
+      "loss": 1.064,
+      "step": 4454
+    },
+    {
+      "epoch": 0.37963357477631016,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.291839157586006e-06,
+      "loss": 0.7133,
+      "step": 4455
+    },
+    {
+      "epoch": 0.3797187899446101,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2917462979935207e-06,
+      "loss": 0.6152,
+      "step": 4456
+    },
+    {
+      "epoch": 0.37980400511291007,
+      "grad_norm": 18.75,
+      "learning_rate": 2.2916534195756323e-06,
+      "loss": 0.8759,
+      "step": 4457
+    },
+    {
+      "epoch": 0.3798892202812101,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2915605223340193e-06,
+      "loss": 0.964,
+      "step": 4458
+    },
+    {
+      "epoch": 0.37997443544951004,
+      "grad_norm": 20.5,
+      "learning_rate": 2.2914676062703602e-06,
+      "loss": 0.7727,
+      "step": 4459
+    },
+    {
+      "epoch": 0.38005965061781,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.2913746713863338e-06,
+      "loss": 0.6933,
+      "step": 4460
+    },
+    {
+      "epoch": 0.38014486578610995,
+      "grad_norm": 19.625,
+      "learning_rate": 2.2912817176836204e-06,
+      "loss": 1.1998,
+      "step": 4461
+    },
+    {
+      "epoch": 0.3802300809544099,
+      "grad_norm": 12.75,
+      "learning_rate": 2.291188745163899e-06,
+      "loss": 0.6076,
+      "step": 4462
+    },
+    {
+      "epoch": 0.38031529612270987,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.29109575382885e-06,
+      "loss": 0.5314,
+      "step": 4463
+    },
+    {
+      "epoch": 0.3804005112910098,
+      "grad_norm": 14.375,
+      "learning_rate": 2.291002743680154e-06,
+      "loss": 0.4575,
+      "step": 4464
+    },
+    {
+      "epoch": 0.3804857264593098,
+      "grad_norm": 14.0,
+      "learning_rate": 2.2909097147194912e-06,
+      "loss": 1.0536,
+      "step": 4465
+    },
+    {
+      "epoch": 0.38057094162760974,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.2908166669485433e-06,
+      "loss": 0.7613,
+      "step": 4466
+    },
+    {
+      "epoch": 0.3806561567959097,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.290723600368992e-06,
+      "loss": 0.9637,
+      "step": 4467
+    },
+    {
+      "epoch": 0.38074137196420965,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2906305149825185e-06,
+      "loss": 0.8168,
+      "step": 4468
+    },
+    {
+      "epoch": 0.3808265871325096,
+      "grad_norm": 12.75,
+      "learning_rate": 2.2905374107908047e-06,
+      "loss": 0.7389,
+      "step": 4469
+    },
+    {
+      "epoch": 0.38091180230080957,
+      "grad_norm": 20.25,
+      "learning_rate": 2.290444287795534e-06,
+      "loss": 1.0844,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3809970174691095,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.290351145998389e-06,
+      "loss": 0.4004,
+      "step": 4471
+    },
+    {
+      "epoch": 0.3810822326374095,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.290257985401052e-06,
+      "loss": 0.5302,
+      "step": 4472
+    },
+    {
+      "epoch": 0.38116744780570944,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2901648060052077e-06,
+      "loss": 0.711,
+      "step": 4473
+    },
+    {
+      "epoch": 0.3812526629740094,
+      "grad_norm": 23.625,
+      "learning_rate": 2.290071607812539e-06,
+      "loss": 0.9568,
+      "step": 4474
+    },
+    {
+      "epoch": 0.38133787814230935,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.289978390824731e-06,
+      "loss": 0.8809,
+      "step": 4475
+    },
+    {
+      "epoch": 0.3814230933106093,
+      "grad_norm": 16.625,
+      "learning_rate": 2.289885155043467e-06,
+      "loss": 0.6751,
+      "step": 4476
+    },
+    {
+      "epoch": 0.38150830847890926,
+      "grad_norm": 17.625,
+      "learning_rate": 2.289791900470433e-06,
+      "loss": 1.0595,
+      "step": 4477
+    },
+    {
+      "epoch": 0.3815935236472092,
+      "grad_norm": 15.75,
+      "learning_rate": 2.289698627107314e-06,
+      "loss": 0.8042,
+      "step": 4478
+    },
+    {
+      "epoch": 0.3816787388155092,
+      "grad_norm": 17.875,
+      "learning_rate": 2.289605334955795e-06,
+      "loss": 1.0193,
+      "step": 4479
+    },
+    {
+      "epoch": 0.38176395398380913,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2895120240175622e-06,
+      "loss": 0.6235,
+      "step": 4480
+    },
+    {
+      "epoch": 0.3818491691521091,
+      "grad_norm": 12.5,
+      "learning_rate": 2.2894186942943015e-06,
+      "loss": 0.6292,
+      "step": 4481
+    },
+    {
+      "epoch": 0.38193438432040905,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.2893253457877e-06,
+      "loss": 0.4333,
+      "step": 4482
+    },
+    {
+      "epoch": 0.382019599488709,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2892319784994447e-06,
+      "loss": 0.7367,
+      "step": 4483
+    },
+    {
+      "epoch": 0.38210481465700896,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.2891385924312224e-06,
+      "loss": 0.7588,
+      "step": 4484
+    },
+    {
+      "epoch": 0.3821900298253089,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2890451875847205e-06,
+      "loss": 0.7638,
+      "step": 4485
+    },
+    {
+      "epoch": 0.3822752449936089,
+      "grad_norm": 16.375,
+      "learning_rate": 2.2889517639616277e-06,
+      "loss": 0.916,
+      "step": 4486
+    },
+    {
+      "epoch": 0.38236046016190883,
+      "grad_norm": 14.125,
+      "learning_rate": 2.288858321563631e-06,
+      "loss": 0.5469,
+      "step": 4487
+    },
+    {
+      "epoch": 0.3824456753302088,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.2887648603924206e-06,
+      "loss": 0.4791,
+      "step": 4488
+    },
+    {
+      "epoch": 0.38253089049850875,
+      "grad_norm": 19.375,
+      "learning_rate": 2.288671380449684e-06,
+      "loss": 1.0754,
+      "step": 4489
+    },
+    {
+      "epoch": 0.3826161056668087,
+      "grad_norm": 19.625,
+      "learning_rate": 2.2885778817371115e-06,
+      "loss": 0.995,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38270132083510866,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.2884843642563924e-06,
+      "loss": 0.4355,
+      "step": 4491
+    },
+    {
+      "epoch": 0.3827865360034086,
+      "grad_norm": 17.0,
+      "learning_rate": 2.288390828009216e-06,
+      "loss": 1.1606,
+      "step": 4492
+    },
+    {
+      "epoch": 0.3828717511717086,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.2882972729972737e-06,
+      "loss": 0.7709,
+      "step": 4493
+    },
+    {
+      "epoch": 0.38295696634000853,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2882036992222554e-06,
+      "loss": 0.7014,
+      "step": 4494
+    },
+    {
+      "epoch": 0.3830421815083085,
+      "grad_norm": 13.75,
+      "learning_rate": 2.288110106685852e-06,
+      "loss": 0.687,
+      "step": 4495
+    },
+    {
+      "epoch": 0.38312739667660844,
+      "grad_norm": 19.375,
+      "learning_rate": 2.2880164953897556e-06,
+      "loss": 1.0258,
+      "step": 4496
+    },
+    {
+      "epoch": 0.3832126118449084,
+      "grad_norm": 19.0,
+      "learning_rate": 2.287922865335657e-06,
+      "loss": 1.1486,
+      "step": 4497
+    },
+    {
+      "epoch": 0.38329782701320836,
+      "grad_norm": 24.5,
+      "learning_rate": 2.2878292165252486e-06,
+      "loss": 1.1778,
+      "step": 4498
+    },
+    {
+      "epoch": 0.3833830421815083,
+      "grad_norm": 11.0,
+      "learning_rate": 2.2877355489602223e-06,
+      "loss": 0.6552,
+      "step": 4499
+    },
+    {
+      "epoch": 0.3834682573498083,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2876418626422716e-06,
+      "loss": 0.6636,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38355347251810823,
+      "grad_norm": 14.625,
+      "learning_rate": 2.2875481575730883e-06,
+      "loss": 0.6893,
+      "step": 4501
+    },
+    {
+      "epoch": 0.3836386876864082,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.2874544337543674e-06,
+      "loss": 0.8543,
+      "step": 4502
+    },
+    {
+      "epoch": 0.38372390285470814,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.287360691187801e-06,
+      "loss": 0.5006,
+      "step": 4503
+    },
+    {
+      "epoch": 0.3838091180230081,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.287266929875084e-06,
+      "loss": 0.5273,
+      "step": 4504
+    },
+    {
+      "epoch": 0.38389433319130806,
+      "grad_norm": 18.875,
+      "learning_rate": 2.2871731498179105e-06,
+      "loss": 0.9154,
+      "step": 4505
+    },
+    {
+      "epoch": 0.383979548359608,
+      "grad_norm": 16.875,
+      "learning_rate": 2.287079351017975e-06,
+      "loss": 0.9958,
+      "step": 4506
+    },
+    {
+      "epoch": 0.38406476352790797,
+      "grad_norm": 18.25,
+      "learning_rate": 2.286985533476973e-06,
+      "loss": 0.9313,
+      "step": 4507
+    },
+    {
+      "epoch": 0.3841499786962079,
+      "grad_norm": 30.875,
+      "learning_rate": 2.2868916971966e-06,
+      "loss": 0.5992,
+      "step": 4508
+    },
+    {
+      "epoch": 0.3842351938645079,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.286797842178551e-06,
+      "loss": 0.495,
+      "step": 4509
+    },
+    {
+      "epoch": 0.38432040903280784,
+      "grad_norm": 15.375,
+      "learning_rate": 2.2867039684245224e-06,
+      "loss": 0.8634,
+      "step": 4510
+    },
+    {
+      "epoch": 0.3844056242011078,
+      "grad_norm": 13.5,
+      "learning_rate": 2.2866100759362103e-06,
+      "loss": 0.61,
+      "step": 4511
+    },
+    {
+      "epoch": 0.38449083936940776,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2865161647153124e-06,
+      "loss": 1.007,
+      "step": 4512
+    },
+    {
+      "epoch": 0.3845760545377077,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.2864222347635246e-06,
+      "loss": 0.7041,
+      "step": 4513
+    },
+    {
+      "epoch": 0.38466126970600767,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2863282860825446e-06,
+      "loss": 0.6604,
+      "step": 4514
+    },
+    {
+      "epoch": 0.3847464848743076,
+      "grad_norm": 14.5,
+      "learning_rate": 2.286234318674071e-06,
+      "loss": 0.8003,
+      "step": 4515
+    },
+    {
+      "epoch": 0.3848317000426076,
+      "grad_norm": 15.75,
+      "learning_rate": 2.2861403325398008e-06,
+      "loss": 0.9918,
+      "step": 4516
+    },
+    {
+      "epoch": 0.38491691521090754,
+      "grad_norm": 17.0,
+      "learning_rate": 2.286046327681433e-06,
+      "loss": 0.8309,
+      "step": 4517
+    },
+    {
+      "epoch": 0.3850021303792075,
+      "grad_norm": 14.125,
+      "learning_rate": 2.2859523041006665e-06,
+      "loss": 0.8778,
+      "step": 4518
+    },
+    {
+      "epoch": 0.38508734554750745,
+      "grad_norm": 24.625,
+      "learning_rate": 2.2858582617992e-06,
+      "loss": 1.1862,
+      "step": 4519
+    },
+    {
+      "epoch": 0.3851725607158074,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2857642007787324e-06,
+      "loss": 0.7936,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38525777588410737,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.285670121040965e-06,
+      "loss": 1.0927,
+      "step": 4521
+    },
+    {
+      "epoch": 0.3853429910524073,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2855760225875967e-06,
+      "loss": 0.523,
+      "step": 4522
+    },
+    {
+      "epoch": 0.3854282062207073,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2854819054203285e-06,
+      "loss": 1.069,
+      "step": 4523
+    },
+    {
+      "epoch": 0.38551342138900724,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2853877695408608e-06,
+      "loss": 0.79,
+      "step": 4524
+    },
+    {
+      "epoch": 0.3855986365573072,
+      "grad_norm": 13.0,
+      "learning_rate": 2.285293614950895e-06,
+      "loss": 0.7708,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38568385172560715,
+      "grad_norm": 17.625,
+      "learning_rate": 2.285199441652133e-06,
+      "loss": 1.1711,
+      "step": 4526
+    },
+    {
+      "epoch": 0.3857690668939071,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2851052496462754e-06,
+      "loss": 0.8402,
+      "step": 4527
+    },
+    {
+      "epoch": 0.38585428206220707,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.2850110389350254e-06,
+      "loss": 0.5961,
+      "step": 4528
+    },
+    {
+      "epoch": 0.385939497230507,
+      "grad_norm": 16.875,
+      "learning_rate": 2.284916809520085e-06,
+      "loss": 0.9529,
+      "step": 4529
+    },
+    {
+      "epoch": 0.386024712398807,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.2848225614031573e-06,
+      "loss": 0.6457,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38610992756710694,
+      "grad_norm": 16.75,
+      "learning_rate": 2.284728294585945e-06,
+      "loss": 0.9751,
+      "step": 4531
+    },
+    {
+      "epoch": 0.3861951427354069,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2846340090701522e-06,
+      "loss": 0.5709,
+      "step": 4532
+    },
+    {
+      "epoch": 0.38628035790370685,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2845397048574823e-06,
+      "loss": 0.8488,
+      "step": 4533
+    },
+    {
+      "epoch": 0.3863655730720068,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.28444538194964e-06,
+      "loss": 0.5948,
+      "step": 4534
+    },
+    {
+      "epoch": 0.38645078824030676,
+      "grad_norm": 14.625,
+      "learning_rate": 2.284351040348329e-06,
+      "loss": 0.7408,
+      "step": 4535
+    },
+    {
+      "epoch": 0.3865360034086067,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.284256680055255e-06,
+      "loss": 0.8908,
+      "step": 4536
+    },
+    {
+      "epoch": 0.3866212185769067,
+      "grad_norm": 19.375,
+      "learning_rate": 2.2841623010721228e-06,
+      "loss": 1.0765,
+      "step": 4537
+    },
+    {
+      "epoch": 0.38670643374520663,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2840679034006373e-06,
+      "loss": 0.9415,
+      "step": 4538
+    },
+    {
+      "epoch": 0.3867916489135066,
+      "grad_norm": 17.75,
+      "learning_rate": 2.283973487042505e-06,
+      "loss": 0.945,
+      "step": 4539
+    },
+    {
+      "epoch": 0.38687686408180655,
+      "grad_norm": 20.25,
+      "learning_rate": 2.2838790519994326e-06,
+      "loss": 1.1383,
+      "step": 4540
+    },
+    {
+      "epoch": 0.3869620792501065,
+      "grad_norm": 23.125,
+      "learning_rate": 2.2837845982731257e-06,
+      "loss": 0.7768,
+      "step": 4541
+    },
+    {
+      "epoch": 0.38704729441840646,
+      "grad_norm": 12.875,
+      "learning_rate": 2.2836901258652915e-06,
+      "loss": 0.7491,
+      "step": 4542
+    },
+    {
+      "epoch": 0.3871325095867064,
+      "grad_norm": 17.25,
+      "learning_rate": 2.2835956347776377e-06,
+      "loss": 0.8438,
+      "step": 4543
+    },
+    {
+      "epoch": 0.3872177247550064,
+      "grad_norm": 16.625,
+      "learning_rate": 2.283501125011871e-06,
+      "loss": 0.7922,
+      "step": 4544
+    },
+    {
+      "epoch": 0.38730293992330633,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2834065965697e-06,
+      "loss": 0.4668,
+      "step": 4545
+    },
+    {
+      "epoch": 0.3873881550916063,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2833120494528325e-06,
+      "loss": 0.7096,
+      "step": 4546
+    },
+    {
+      "epoch": 0.38747337025990625,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2832174836629774e-06,
+      "loss": 0.7342,
+      "step": 4547
+    },
+    {
+      "epoch": 0.3875585854282062,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.283122899201843e-06,
+      "loss": 0.7097,
+      "step": 4548
+    },
+    {
+      "epoch": 0.38764380059650616,
+      "grad_norm": 16.125,
+      "learning_rate": 2.283028296071139e-06,
+      "loss": 1.0786,
+      "step": 4549
+    },
+    {
+      "epoch": 0.3877290157648061,
+      "grad_norm": 13.75,
+      "learning_rate": 2.282933674272575e-06,
+      "loss": 0.791,
+      "step": 4550
+    },
+    {
+      "epoch": 0.3878142309331061,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.282839033807861e-06,
+      "loss": 0.5015,
+      "step": 4551
+    },
+    {
+      "epoch": 0.38789944610140603,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2827443746787068e-06,
+      "loss": 1.1809,
+      "step": 4552
+    },
+    {
+      "epoch": 0.387984661269706,
+      "grad_norm": 12.0,
+      "learning_rate": 2.2826496968868234e-06,
+      "loss": 0.524,
+      "step": 4553
+    },
+    {
+      "epoch": 0.38806987643800595,
+      "grad_norm": 16.375,
+      "learning_rate": 2.282555000433921e-06,
+      "loss": 0.8558,
+      "step": 4554
+    },
+    {
+      "epoch": 0.3881550916063059,
+      "grad_norm": 12.625,
+      "learning_rate": 2.282460285321712e-06,
+      "loss": 0.5823,
+      "step": 4555
+    },
+    {
+      "epoch": 0.38824030677460586,
+      "grad_norm": 10.5,
+      "learning_rate": 2.2823655515519076e-06,
+      "loss": 0.5241,
+      "step": 4556
+    },
+    {
+      "epoch": 0.3883255219429058,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2822707991262196e-06,
+      "loss": 0.6566,
+      "step": 4557
+    },
+    {
+      "epoch": 0.3884107371112058,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.2821760280463596e-06,
+      "loss": 0.5527,
+      "step": 4558
+    },
+    {
+      "epoch": 0.38849595227950573,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.2820812383140413e-06,
+      "loss": 0.8569,
+      "step": 4559
+    },
+    {
+      "epoch": 0.3885811674478057,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.281986429930977e-06,
+      "loss": 0.8562,
+      "step": 4560
+    },
+    {
+      "epoch": 0.38866638261610564,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.2818916028988798e-06,
+      "loss": 0.8177,
+      "step": 4561
+    },
+    {
+      "epoch": 0.3887515977844056,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2817967572194644e-06,
+      "loss": 1.1979,
+      "step": 4562
+    },
+    {
+      "epoch": 0.38883681295270556,
+      "grad_norm": 16.0,
+      "learning_rate": 2.2817018928944433e-06,
+      "loss": 0.8594,
+      "step": 4563
+    },
+    {
+      "epoch": 0.3889220281210055,
+      "grad_norm": 14.0,
+      "learning_rate": 2.2816070099255324e-06,
+      "loss": 0.9151,
+      "step": 4564
+    },
+    {
+      "epoch": 0.38900724328930547,
+      "grad_norm": 10.75,
+      "learning_rate": 2.281512108314445e-06,
+      "loss": 0.3492,
+      "step": 4565
+    },
+    {
+      "epoch": 0.38909245845760543,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.2814171880628964e-06,
+      "loss": 0.8376,
+      "step": 4566
+    },
+    {
+      "epoch": 0.3891776736259054,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2813222491726022e-06,
+      "loss": 1.0273,
+      "step": 4567
+    },
+    {
+      "epoch": 0.38926288879420534,
+      "grad_norm": 14.5,
+      "learning_rate": 2.281227291645278e-06,
+      "loss": 0.8885,
+      "step": 4568
+    },
+    {
+      "epoch": 0.3893481039625053,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.28113231548264e-06,
+      "loss": 0.3477,
+      "step": 4569
+    },
+    {
+      "epoch": 0.3894333191308053,
+      "grad_norm": 16.0,
+      "learning_rate": 2.2810373206864034e-06,
+      "loss": 0.7715,
+      "step": 4570
+    },
+    {
+      "epoch": 0.38951853429910527,
+      "grad_norm": 16.75,
+      "learning_rate": 2.280942307258286e-06,
+      "loss": 1.1484,
+      "step": 4571
+    },
+    {
+      "epoch": 0.3896037494674052,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2808472752000043e-06,
+      "loss": 0.78,
+      "step": 4572
+    },
+    {
+      "epoch": 0.3896889646357052,
+      "grad_norm": 15.375,
+      "learning_rate": 2.280752224513276e-06,
+      "loss": 0.635,
+      "step": 4573
+    },
+    {
+      "epoch": 0.38977417980400514,
+      "grad_norm": 35.25,
+      "learning_rate": 2.280657155199818e-06,
+      "loss": 0.9081,
+      "step": 4574
+    },
+    {
+      "epoch": 0.3898593949723051,
+      "grad_norm": 11.375,
+      "learning_rate": 2.2805620672613494e-06,
+      "loss": 0.5269,
+      "step": 4575
+    },
+    {
+      "epoch": 0.38994461014060505,
+      "grad_norm": 15.5,
+      "learning_rate": 2.280466960699588e-06,
+      "loss": 0.7133,
+      "step": 4576
+    },
+    {
+      "epoch": 0.390029825308905,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.280371835516252e-06,
+      "loss": 0.7546,
+      "step": 4577
+    },
+    {
+      "epoch": 0.39011504047720497,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.280276691713061e-06,
+      "loss": 0.6967,
+      "step": 4578
+    },
+    {
+      "epoch": 0.3902002556455049,
+      "grad_norm": 17.375,
+      "learning_rate": 2.280181529291734e-06,
+      "loss": 1.1751,
+      "step": 4579
+    },
+    {
+      "epoch": 0.3902854708138049,
+      "grad_norm": 18.75,
+      "learning_rate": 2.2800863482539905e-06,
+      "loss": 0.8412,
+      "step": 4580
+    },
+    {
+      "epoch": 0.39037068598210484,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.279991148601551e-06,
+      "loss": 0.516,
+      "step": 4581
+    },
+    {
+      "epoch": 0.3904559011504048,
+      "grad_norm": 10.5,
+      "learning_rate": 2.2798959303361364e-06,
+      "loss": 0.4011,
+      "step": 4582
+    },
+    {
+      "epoch": 0.39054111631870475,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2798006934594663e-06,
+      "loss": 0.6922,
+      "step": 4583
+    },
+    {
+      "epoch": 0.3906263314870047,
+      "grad_norm": 14.5,
+      "learning_rate": 2.279705437973262e-06,
+      "loss": 0.7145,
+      "step": 4584
+    },
+    {
+      "epoch": 0.39071154665530466,
+      "grad_norm": 15.25,
+      "learning_rate": 2.279610163879245e-06,
+      "loss": 0.7449,
+      "step": 4585
+    },
+    {
+      "epoch": 0.3907967618236046,
+      "grad_norm": 16.5,
+      "learning_rate": 2.279514871179137e-06,
+      "loss": 0.9176,
+      "step": 4586
+    },
+    {
+      "epoch": 0.3908819769919046,
+      "grad_norm": 11.625,
+      "learning_rate": 2.27941955987466e-06,
+      "loss": 0.5752,
+      "step": 4587
+    },
+    {
+      "epoch": 0.39096719216020454,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.2793242299675366e-06,
+      "loss": 0.8102,
+      "step": 4588
+    },
+    {
+      "epoch": 0.3910524073285045,
+      "grad_norm": 15.375,
+      "learning_rate": 2.279228881459489e-06,
+      "loss": 0.8104,
+      "step": 4589
+    },
+    {
+      "epoch": 0.39113762249680445,
+      "grad_norm": 12.875,
+      "learning_rate": 2.279133514352241e-06,
+      "loss": 0.6299,
+      "step": 4590
+    },
+    {
+      "epoch": 0.3912228376651044,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.279038128647515e-06,
+      "loss": 0.6081,
+      "step": 4591
+    },
+    {
+      "epoch": 0.39130805283340436,
+      "grad_norm": 21.25,
+      "learning_rate": 2.2789427243470356e-06,
+      "loss": 0.8856,
+      "step": 4592
+    },
+    {
+      "epoch": 0.3913932680017043,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.278847301452526e-06,
+      "loss": 0.7382,
+      "step": 4593
+    },
+    {
+      "epoch": 0.3914784831700043,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2787518599657116e-06,
+      "loss": 0.9935,
+      "step": 4594
+    },
+    {
+      "epoch": 0.39156369833830423,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2786563998883167e-06,
+      "loss": 0.3733,
+      "step": 4595
+    },
+    {
+      "epoch": 0.3916489135066042,
+      "grad_norm": 9.25,
+      "learning_rate": 2.278560921222066e-06,
+      "loss": 0.3683,
+      "step": 4596
+    },
+    {
+      "epoch": 0.39173412867490415,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.278465423968685e-06,
+      "loss": 0.5061,
+      "step": 4597
+    },
+    {
+      "epoch": 0.3918193438432041,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2783699081298995e-06,
+      "loss": 0.7671,
+      "step": 4598
+    },
+    {
+      "epoch": 0.39190455901150406,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.278274373707436e-06,
+      "loss": 1.0247,
+      "step": 4599
+    },
+    {
+      "epoch": 0.391989774179804,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2781788207030203e-06,
+      "loss": 0.7135,
+      "step": 4600
+    },
+    {
+      "epoch": 0.392074989348104,
+      "grad_norm": 17.25,
+      "learning_rate": 2.2780832491183796e-06,
+      "loss": 0.9954,
+      "step": 4601
+    },
+    {
+      "epoch": 0.39216020451640393,
+      "grad_norm": 13.25,
+      "learning_rate": 2.277987658955241e-06,
+      "loss": 0.7754,
+      "step": 4602
+    },
+    {
+      "epoch": 0.3922454196847039,
+      "grad_norm": 15.875,
+      "learning_rate": 2.277892050215331e-06,
+      "loss": 0.8307,
+      "step": 4603
+    },
+    {
+      "epoch": 0.39233063485300385,
+      "grad_norm": 15.75,
+      "learning_rate": 2.2777964229003786e-06,
+      "loss": 0.7016,
+      "step": 4604
+    },
+    {
+      "epoch": 0.3924158500213038,
+      "grad_norm": 16.0,
+      "learning_rate": 2.277700777012111e-06,
+      "loss": 0.91,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39250106518960376,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.277605112552257e-06,
+      "loss": 0.6765,
+      "step": 4606
+    },
+    {
+      "epoch": 0.3925862803579037,
+      "grad_norm": 11.25,
+      "learning_rate": 2.2775094295225452e-06,
+      "loss": 0.4097,
+      "step": 4607
+    },
+    {
+      "epoch": 0.3926714955262037,
+      "grad_norm": 16.875,
+      "learning_rate": 2.277413727924705e-06,
+      "loss": 0.9516,
+      "step": 4608
+    },
+    {
+      "epoch": 0.39275671069450363,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.2773180077604657e-06,
+      "loss": 0.8189,
+      "step": 4609
+    },
+    {
+      "epoch": 0.3928419258628036,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2772222690315564e-06,
+      "loss": 0.5581,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39292714103110354,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.2771265117397083e-06,
+      "loss": 0.5962,
+      "step": 4611
+    },
+    {
+      "epoch": 0.3930123561994035,
+      "grad_norm": 21.125,
+      "learning_rate": 2.277030735886651e-06,
+      "loss": 0.6761,
+      "step": 4612
+    },
+    {
+      "epoch": 0.39309757136770346,
+      "grad_norm": 12.375,
+      "learning_rate": 2.276934941474116e-06,
+      "loss": 0.6512,
+      "step": 4613
+    },
+    {
+      "epoch": 0.3931827865360034,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2768391285038334e-06,
+      "loss": 0.5349,
+      "step": 4614
+    },
+    {
+      "epoch": 0.39326800170430337,
+      "grad_norm": 17.5,
+      "learning_rate": 2.2767432969775354e-06,
+      "loss": 0.9304,
+      "step": 4615
+    },
+    {
+      "epoch": 0.39335321687260333,
+      "grad_norm": 17.375,
+      "learning_rate": 2.2766474468969536e-06,
+      "loss": 0.5558,
+      "step": 4616
+    },
+    {
+      "epoch": 0.3934384320409033,
+      "grad_norm": 30.875,
+      "learning_rate": 2.2765515782638197e-06,
+      "loss": 0.9864,
+      "step": 4617
+    },
+    {
+      "epoch": 0.39352364720920324,
+      "grad_norm": 13.0,
+      "learning_rate": 2.2764556910798668e-06,
+      "loss": 0.6642,
+      "step": 4618
+    },
+    {
+      "epoch": 0.3936088623775032,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2763597853468274e-06,
+      "loss": 0.7602,
+      "step": 4619
+    },
+    {
+      "epoch": 0.39369407754580316,
+      "grad_norm": 21.875,
+      "learning_rate": 2.2762638610664343e-06,
+      "loss": 0.8022,
+      "step": 4620
+    },
+    {
+      "epoch": 0.3937792927141031,
+      "grad_norm": 14.5,
+      "learning_rate": 2.2761679182404216e-06,
+      "loss": 0.8596,
+      "step": 4621
+    },
+    {
+      "epoch": 0.39386450788240307,
+      "grad_norm": 15.625,
+      "learning_rate": 2.2760719568705225e-06,
+      "loss": 0.8398,
+      "step": 4622
+    },
+    {
+      "epoch": 0.393949723050703,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.275975976958471e-06,
+      "loss": 0.7829,
+      "step": 4623
+    },
+    {
+      "epoch": 0.394034938219003,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.2758799785060028e-06,
+      "loss": 0.4472,
+      "step": 4624
+    },
+    {
+      "epoch": 0.39412015338730294,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2757839615148512e-06,
+      "loss": 0.8803,
+      "step": 4625
+    },
+    {
+      "epoch": 0.3942053685556029,
+      "grad_norm": 15.625,
+      "learning_rate": 2.275687925986752e-06,
+      "loss": 0.7726,
+      "step": 4626
+    },
+    {
+      "epoch": 0.39429058372390285,
+      "grad_norm": 10.25,
+      "learning_rate": 2.2755918719234405e-06,
+      "loss": 0.3376,
+      "step": 4627
+    },
+    {
+      "epoch": 0.3943757988922028,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.275495799326653e-06,
+      "loss": 0.8363,
+      "step": 4628
+    },
+    {
+      "epoch": 0.39446101406050277,
+      "grad_norm": 19.625,
+      "learning_rate": 2.275399708198125e-06,
+      "loss": 1.1049,
+      "step": 4629
+    },
+    {
+      "epoch": 0.3945462292288027,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.2753035985395928e-06,
+      "loss": 0.5373,
+      "step": 4630
+    },
+    {
+      "epoch": 0.3946314443971027,
+      "grad_norm": 16.75,
+      "learning_rate": 2.275207470352794e-06,
+      "loss": 0.8961,
+      "step": 4631
+    },
+    {
+      "epoch": 0.39471665956540264,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2751113236394647e-06,
+      "loss": 0.7463,
+      "step": 4632
+    },
+    {
+      "epoch": 0.3948018747337026,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.2750151584013432e-06,
+      "loss": 0.5158,
+      "step": 4633
+    },
+    {
+      "epoch": 0.39488708990200255,
+      "grad_norm": 16.0,
+      "learning_rate": 2.274918974640167e-06,
+      "loss": 0.7595,
+      "step": 4634
+    },
+    {
+      "epoch": 0.3949723050703025,
+      "grad_norm": 17.0,
+      "learning_rate": 2.2748227723576747e-06,
+      "loss": 1.0543,
+      "step": 4635
+    },
+    {
+      "epoch": 0.39505752023860247,
+      "grad_norm": 27.0,
+      "learning_rate": 2.274726551555604e-06,
+      "loss": 1.278,
+      "step": 4636
+    },
+    {
+      "epoch": 0.3951427354069024,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2746303122356946e-06,
+      "loss": 0.6901,
+      "step": 4637
+    },
+    {
+      "epoch": 0.3952279505752024,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2745340543996844e-06,
+      "loss": 0.9716,
+      "step": 4638
+    },
+    {
+      "epoch": 0.39531316574350234,
+      "grad_norm": 13.0,
+      "learning_rate": 2.2744377780493136e-06,
+      "loss": 0.6546,
+      "step": 4639
+    },
+    {
+      "epoch": 0.3953983809118023,
+      "grad_norm": 19.5,
+      "learning_rate": 2.2743414831863225e-06,
+      "loss": 1.1621,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39548359608010225,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2742451698124507e-06,
+      "loss": 0.7983,
+      "step": 4641
+    },
+    {
+      "epoch": 0.3955688112484022,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2741488379294386e-06,
+      "loss": 0.6898,
+      "step": 4642
+    },
+    {
+      "epoch": 0.39565402641670216,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.274052487539027e-06,
+      "loss": 0.715,
+      "step": 4643
+    },
+    {
+      "epoch": 0.3957392415850021,
+      "grad_norm": 12.875,
+      "learning_rate": 2.273956118642957e-06,
+      "loss": 0.4977,
+      "step": 4644
+    },
+    {
+      "epoch": 0.3958244567533021,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2738597312429706e-06,
+      "loss": 0.599,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39590967192160204,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.273763325340809e-06,
+      "loss": 0.5835,
+      "step": 4646
+    },
+    {
+      "epoch": 0.395994887089902,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.273666900938215e-06,
+      "loss": 0.4101,
+      "step": 4647
+    },
+    {
+      "epoch": 0.39608010225820195,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2735704580369305e-06,
+      "loss": 0.5664,
+      "step": 4648
+    },
+    {
+      "epoch": 0.3961653174265019,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.273473996638699e-06,
+      "loss": 0.3924,
+      "step": 4649
+    },
+    {
+      "epoch": 0.39625053259480186,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.2733775167452627e-06,
+      "loss": 0.4271,
+      "step": 4650
+    },
+    {
+      "epoch": 0.3963357477631018,
+      "grad_norm": 20.625,
+      "learning_rate": 2.273281018358366e-06,
+      "loss": 1.0851,
+      "step": 4651
+    },
+    {
+      "epoch": 0.3964209629314018,
+      "grad_norm": 12.125,
+      "learning_rate": 2.2731845014797515e-06,
+      "loss": 0.5392,
+      "step": 4652
+    },
+    {
+      "epoch": 0.39650617809970173,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.273087966111165e-06,
+      "loss": 0.6814,
+      "step": 4653
+    },
+    {
+      "epoch": 0.3965913932680017,
+      "grad_norm": 14.75,
+      "learning_rate": 2.27299141225435e-06,
+      "loss": 0.7036,
+      "step": 4654
+    },
+    {
+      "epoch": 0.39667660843630165,
+      "grad_norm": 21.375,
+      "learning_rate": 2.272894839911051e-06,
+      "loss": 0.9787,
+      "step": 4655
+    },
+    {
+      "epoch": 0.3967618236046016,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.2727982490830143e-06,
+      "loss": 0.8725,
+      "step": 4656
+    },
+    {
+      "epoch": 0.39684703877290156,
+      "grad_norm": 16.125,
+      "learning_rate": 2.272701639771984e-06,
+      "loss": 1.0321,
+      "step": 4657
+    },
+    {
+      "epoch": 0.3969322539412015,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.2726050119797072e-06,
+      "loss": 0.6422,
+      "step": 4658
+    },
+    {
+      "epoch": 0.3970174691095015,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.2725083657079294e-06,
+      "loss": 0.5762,
+      "step": 4659
+    },
+    {
+      "epoch": 0.39710268427780143,
+      "grad_norm": 14.75,
+      "learning_rate": 2.272411700958397e-06,
+      "loss": 0.7513,
+      "step": 4660
+    },
+    {
+      "epoch": 0.3971878994461014,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2723150177328575e-06,
+      "loss": 0.9375,
+      "step": 4661
+    },
+    {
+      "epoch": 0.39727311461440135,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2722183160330576e-06,
+      "loss": 0.8393,
+      "step": 4662
+    },
+    {
+      "epoch": 0.3973583297827013,
+      "grad_norm": 17.25,
+      "learning_rate": 2.2721215958607446e-06,
+      "loss": 1.2914,
+      "step": 4663
+    },
+    {
+      "epoch": 0.39744354495100126,
+      "grad_norm": 11.75,
+      "learning_rate": 2.272024857217666e-06,
+      "loss": 0.6641,
+      "step": 4664
+    },
+    {
+      "epoch": 0.3975287601193012,
+      "grad_norm": 12.875,
+      "learning_rate": 2.2719281001055717e-06,
+      "loss": 0.6941,
+      "step": 4665
+    },
+    {
+      "epoch": 0.3976139752876012,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.271831324526208e-06,
+      "loss": 0.7783,
+      "step": 4666
+    },
+    {
+      "epoch": 0.39769919045590113,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.271734530481325e-06,
+      "loss": 0.7481,
+      "step": 4667
+    },
+    {
+      "epoch": 0.3977844056242011,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2716377179726716e-06,
+      "loss": 0.8422,
+      "step": 4668
+    },
+    {
+      "epoch": 0.39786962079250104,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2715408870019977e-06,
+      "loss": 1.0709,
+      "step": 4669
+    },
+    {
+      "epoch": 0.397954835960801,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.271444037571052e-06,
+      "loss": 0.7559,
+      "step": 4670
+    },
+    {
+      "epoch": 0.39804005112910096,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2713471696815863e-06,
+      "loss": 0.7677,
+      "step": 4671
+    },
+    {
+      "epoch": 0.3981252662974009,
+      "grad_norm": 12.0,
+      "learning_rate": 2.2712502833353496e-06,
+      "loss": 0.7987,
+      "step": 4672
+    },
+    {
+      "epoch": 0.39821048146570087,
+      "grad_norm": 9.5,
+      "learning_rate": 2.2711533785340933e-06,
+      "loss": 0.231,
+      "step": 4673
+    },
+    {
+      "epoch": 0.39829569663400083,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2710564552795686e-06,
+      "loss": 0.8606,
+      "step": 4674
+    },
+    {
+      "epoch": 0.3983809118023008,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.270959513573527e-06,
+      "loss": 0.8815,
+      "step": 4675
+    },
+    {
+      "epoch": 0.39846612697060074,
+      "grad_norm": 12.75,
+      "learning_rate": 2.270862553417721e-06,
+      "loss": 0.4789,
+      "step": 4676
+    },
+    {
+      "epoch": 0.3985513421389007,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2707655748139014e-06,
+      "loss": 0.7227,
+      "step": 4677
+    },
+    {
+      "epoch": 0.39863655730720066,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.2706685777638217e-06,
+      "loss": 0.7824,
+      "step": 4678
+    },
+    {
+      "epoch": 0.3987217724755006,
+      "grad_norm": 19.75,
+      "learning_rate": 2.2705715622692344e-06,
+      "loss": 0.8465,
+      "step": 4679
+    },
+    {
+      "epoch": 0.39880698764380057,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.2704745283318925e-06,
+      "loss": 0.7303,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3988922028121005,
+      "grad_norm": 15.125,
+      "learning_rate": 2.2703774759535495e-06,
+      "loss": 0.8414,
+      "step": 4681
+    },
+    {
+      "epoch": 0.3989774179804005,
+      "grad_norm": 17.25,
+      "learning_rate": 2.27028040513596e-06,
+      "loss": 0.7061,
+      "step": 4682
+    },
+    {
+      "epoch": 0.3990626331487005,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2701833158808774e-06,
+      "loss": 1.028,
+      "step": 4683
+    },
+    {
+      "epoch": 0.39914784831700045,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2700862081900564e-06,
+      "loss": 0.3536,
+      "step": 4684
+    },
+    {
+      "epoch": 0.3992330634853004,
+      "grad_norm": 12.25,
+      "learning_rate": 2.269989082065252e-06,
+      "loss": 0.5192,
+      "step": 4685
+    },
+    {
+      "epoch": 0.39931827865360037,
+      "grad_norm": 16.5,
+      "learning_rate": 2.269891937508219e-06,
+      "loss": 0.7274,
+      "step": 4686
+    },
+    {
+      "epoch": 0.3994034938219003,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.269794774520713e-06,
+      "loss": 0.6608,
+      "step": 4687
+    },
+    {
+      "epoch": 0.3994887089902003,
+      "grad_norm": 19.75,
+      "learning_rate": 2.26969759310449e-06,
+      "loss": 0.9892,
+      "step": 4688
+    },
+    {
+      "epoch": 0.39957392415850024,
+      "grad_norm": 18.25,
+      "learning_rate": 2.269600393261306e-06,
+      "loss": 0.7239,
+      "step": 4689
+    },
+    {
+      "epoch": 0.3996591393268002,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.269503174992918e-06,
+      "loss": 0.7861,
+      "step": 4690
+    },
+    {
+      "epoch": 0.39974435449510015,
+      "grad_norm": 8.625,
+      "learning_rate": 2.269405938301082e-06,
+      "loss": 0.3024,
+      "step": 4691
+    },
+    {
+      "epoch": 0.3998295696634001,
+      "grad_norm": 12.875,
+      "learning_rate": 2.269308683187556e-06,
+      "loss": 0.8666,
+      "step": 4692
+    },
+    {
+      "epoch": 0.39991478483170007,
+      "grad_norm": 35.5,
+      "learning_rate": 2.2692114096540966e-06,
+      "loss": 1.2893,
+      "step": 4693
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 14.75,
+      "learning_rate": 2.269114117702463e-06,
+      "loss": 0.9085,
+      "step": 4694
+    },
+    {
+      "epoch": 0.4000852151683,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.269016807334412e-06,
+      "loss": 0.967,
+      "step": 4695
+    },
+    {
+      "epoch": 0.40017043033659994,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.2689194785517023e-06,
+      "loss": 0.7117,
+      "step": 4696
+    },
+    {
+      "epoch": 0.4002556455048999,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.268822131356093e-06,
+      "loss": 0.9099,
+      "step": 4697
+    },
+    {
+      "epoch": 0.40034086067319985,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2687247657493437e-06,
+      "loss": 0.5604,
+      "step": 4698
+    },
+    {
+      "epoch": 0.4004260758414998,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2686273817332136e-06,
+      "loss": 0.3514,
+      "step": 4699
+    },
+    {
+      "epoch": 0.40051129100979976,
+      "grad_norm": 18.5,
+      "learning_rate": 2.268529979309462e-06,
+      "loss": 0.7446,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4005965061780997,
+      "grad_norm": 16.0,
+      "learning_rate": 2.2684325584798496e-06,
+      "loss": 0.9357,
+      "step": 4701
+    },
+    {
+      "epoch": 0.4006817213463997,
+      "grad_norm": 13.375,
+      "learning_rate": 2.268335119246137e-06,
+      "loss": 0.587,
+      "step": 4702
+    },
+    {
+      "epoch": 0.40076693651469963,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2682376616100842e-06,
+      "loss": 0.8557,
+      "step": 4703
+    },
+    {
+      "epoch": 0.4008521516829996,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.2681401855734535e-06,
+      "loss": 0.6055,
+      "step": 4704
+    },
+    {
+      "epoch": 0.40093736685129955,
+      "grad_norm": 11.125,
+      "learning_rate": 2.2680426911380055e-06,
+      "loss": 0.4985,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4010225820195995,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.2679451783055025e-06,
+      "loss": 0.4013,
+      "step": 4706
+    },
+    {
+      "epoch": 0.40110779718789946,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2678476470777067e-06,
+      "loss": 0.9624,
+      "step": 4707
+    },
+    {
+      "epoch": 0.4011930123561994,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.26775009745638e-06,
+      "loss": 0.7059,
+      "step": 4708
+    },
+    {
+      "epoch": 0.4012782275244994,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.2676525294432853e-06,
+      "loss": 0.8862,
+      "step": 4709
+    },
+    {
+      "epoch": 0.40136344269279933,
+      "grad_norm": 21.875,
+      "learning_rate": 2.267554943040187e-06,
+      "loss": 1.2246,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4014486578610993,
+      "grad_norm": 12.25,
+      "learning_rate": 2.267457338248847e-06,
+      "loss": 0.5507,
+      "step": 4711
+    },
+    {
+      "epoch": 0.40153387302939925,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2673597150710302e-06,
+      "loss": 0.7336,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4016190881976992,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2672620735085e-06,
+      "loss": 0.8719,
+      "step": 4713
+    },
+    {
+      "epoch": 0.40170430336599916,
+      "grad_norm": 19.125,
+      "learning_rate": 2.267164413563021e-06,
+      "loss": 0.7144,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4017895185342991,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2670667352363586e-06,
+      "loss": 0.6919,
+      "step": 4715
+    },
+    {
+      "epoch": 0.4018747337025991,
+      "grad_norm": 17.375,
+      "learning_rate": 2.2669690385302775e-06,
+      "loss": 0.9576,
+      "step": 4716
+    },
+    {
+      "epoch": 0.40195994887089903,
+      "grad_norm": 16.875,
+      "learning_rate": 2.266871323446543e-06,
+      "loss": 0.5785,
+      "step": 4717
+    },
+    {
+      "epoch": 0.402045164039199,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.2667735899869216e-06,
+      "loss": 0.6427,
+      "step": 4718
+    },
+    {
+      "epoch": 0.40213037920749894,
+      "grad_norm": 10.5,
+      "learning_rate": 2.2666758381531783e-06,
+      "loss": 0.3087,
+      "step": 4719
+    },
+    {
+      "epoch": 0.4022155943757989,
+      "grad_norm": 10.625,
+      "learning_rate": 2.266578067947081e-06,
+      "loss": 0.4566,
+      "step": 4720
+    },
+    {
+      "epoch": 0.40230080954409886,
+      "grad_norm": 11.875,
+      "learning_rate": 2.266480279370395e-06,
+      "loss": 0.5908,
+      "step": 4721
+    },
+    {
+      "epoch": 0.4023860247123988,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.266382472424889e-06,
+      "loss": 0.4141,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4024712398806988,
+      "grad_norm": 23.0,
+      "learning_rate": 2.266284647112329e-06,
+      "loss": 0.9698,
+      "step": 4723
+    },
+    {
+      "epoch": 0.40255645504899873,
+      "grad_norm": 15.625,
+      "learning_rate": 2.266186803434484e-06,
+      "loss": 0.8946,
+      "step": 4724
+    },
+    {
+      "epoch": 0.4026416702172987,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2660889413931213e-06,
+      "loss": 0.8829,
+      "step": 4725
+    },
+    {
+      "epoch": 0.40272688538559864,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2659910609900097e-06,
+      "loss": 0.8776,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4028121005538986,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2658931622269177e-06,
+      "loss": 0.8368,
+      "step": 4727
+    },
+    {
+      "epoch": 0.40289731572219856,
+      "grad_norm": 10.625,
+      "learning_rate": 2.2657952451056153e-06,
+      "loss": 0.3565,
+      "step": 4728
+    },
+    {
+      "epoch": 0.4029825308904985,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.265697309627871e-06,
+      "loss": 0.7575,
+      "step": 4729
+    },
+    {
+      "epoch": 0.40306774605879847,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.265599355795455e-06,
+      "loss": 0.8153,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4031529612270984,
+      "grad_norm": 11.5,
+      "learning_rate": 2.2655013836101375e-06,
+      "loss": 0.4993,
+      "step": 4731
+    },
+    {
+      "epoch": 0.4032381763953984,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.2654033930736883e-06,
+      "loss": 1.1303,
+      "step": 4732
+    },
+    {
+      "epoch": 0.40332339156369834,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2653053841878793e-06,
+      "loss": 0.8431,
+      "step": 4733
+    },
+    {
+      "epoch": 0.4034086067319983,
+      "grad_norm": 16.0,
+      "learning_rate": 2.265207356954481e-06,
+      "loss": 0.5506,
+      "step": 4734
+    },
+    {
+      "epoch": 0.40349382190029826,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.265109311375265e-06,
+      "loss": 0.6505,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4035790370685982,
+      "grad_norm": 15.625,
+      "learning_rate": 2.2650112474520024e-06,
+      "loss": 0.924,
+      "step": 4736
+    },
+    {
+      "epoch": 0.40366425223689817,
+      "grad_norm": 17.25,
+      "learning_rate": 2.2649131651864666e-06,
+      "loss": 0.6731,
+      "step": 4737
+    },
+    {
+      "epoch": 0.4037494674051981,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.264815064580429e-06,
+      "loss": 0.6855,
+      "step": 4738
+    },
+    {
+      "epoch": 0.4038346825734981,
+      "grad_norm": 12.125,
+      "learning_rate": 2.2647169456356623e-06,
+      "loss": 0.4179,
+      "step": 4739
+    },
+    {
+      "epoch": 0.40391989774179804,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2646188083539407e-06,
+      "loss": 0.8236,
+      "step": 4740
+    },
+    {
+      "epoch": 0.404005112910098,
+      "grad_norm": 16.75,
+      "learning_rate": 2.264520652737036e-06,
+      "loss": 0.8586,
+      "step": 4741
+    },
+    {
+      "epoch": 0.40409032807839795,
+      "grad_norm": 19.5,
+      "learning_rate": 2.264422478786724e-06,
+      "loss": 0.7602,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4041755432466979,
+      "grad_norm": 13.25,
+      "learning_rate": 2.2643242865047773e-06,
+      "loss": 0.628,
+      "step": 4743
+    },
+    {
+      "epoch": 0.40426075841499787,
+      "grad_norm": 17.625,
+      "learning_rate": 2.2642260758929706e-06,
+      "loss": 0.9205,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4043459735832978,
+      "grad_norm": 10.0,
+      "learning_rate": 2.264127846953079e-06,
+      "loss": 0.2715,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4044311887515978,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.2640295996868774e-06,
+      "loss": 0.4329,
+      "step": 4746
+    },
+    {
+      "epoch": 0.40451640391989774,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.263931334096141e-06,
+      "loss": 0.2412,
+      "step": 4747
+    },
+    {
+      "epoch": 0.4046016190881977,
+      "grad_norm": 19.5,
+      "learning_rate": 2.263833050182646e-06,
+      "loss": 1.1018,
+      "step": 4748
+    },
+    {
+      "epoch": 0.40468683425649765,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.2637347479481687e-06,
+      "loss": 0.7363,
+      "step": 4749
+    },
+    {
+      "epoch": 0.4047720494247976,
+      "grad_norm": 14.0,
+      "learning_rate": 2.2636364273944848e-06,
+      "loss": 0.9824,
+      "step": 4750
+    },
+    {
+      "epoch": 0.40485726459309757,
+      "grad_norm": 21.625,
+      "learning_rate": 2.263538088523371e-06,
+      "loss": 1.0033,
+      "step": 4751
+    },
+    {
+      "epoch": 0.4049424797613975,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2634397313366054e-06,
+      "loss": 0.6889,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4050276949296975,
+      "grad_norm": 16.625,
+      "learning_rate": 2.263341355835964e-06,
+      "loss": 0.677,
+      "step": 4753
+    },
+    {
+      "epoch": 0.40511291009799744,
+      "grad_norm": 22.375,
+      "learning_rate": 2.263242962023226e-06,
+      "loss": 1.0805,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4051981252662974,
+      "grad_norm": 20.875,
+      "learning_rate": 2.2631445499001687e-06,
+      "loss": 1.0173,
+      "step": 4755
+    },
+    {
+      "epoch": 0.40528334043459735,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2630461194685705e-06,
+      "loss": 0.6526,
+      "step": 4756
+    },
+    {
+      "epoch": 0.4053685556028973,
+      "grad_norm": 26.375,
+      "learning_rate": 2.26294767073021e-06,
+      "loss": 0.8009,
+      "step": 4757
+    },
+    {
+      "epoch": 0.40545377077119726,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2628492036868667e-06,
+      "loss": 0.876,
+      "step": 4758
+    },
+    {
+      "epoch": 0.4055389859394972,
+      "grad_norm": 18.5,
+      "learning_rate": 2.2627507183403195e-06,
+      "loss": 0.5533,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4056242011077972,
+      "grad_norm": 10.0,
+      "learning_rate": 2.2626522146923487e-06,
+      "loss": 0.3191,
+      "step": 4760
+    },
+    {
+      "epoch": 0.40570941627609713,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2625536927447337e-06,
+      "loss": 1.0784,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4057946314443971,
+      "grad_norm": 13.25,
+      "learning_rate": 2.262455152499256e-06,
+      "loss": 0.793,
+      "step": 4762
+    },
+    {
+      "epoch": 0.40587984661269705,
+      "grad_norm": 18.5,
+      "learning_rate": 2.262356593957695e-06,
+      "loss": 0.9842,
+      "step": 4763
+    },
+    {
+      "epoch": 0.405965061780997,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.2622580171218325e-06,
+      "loss": 0.7935,
+      "step": 4764
+    },
+    {
+      "epoch": 0.40605027694929696,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.26215942199345e-06,
+      "loss": 0.4932,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4061354921175969,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2620608085743286e-06,
+      "loss": 0.5764,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4062207072858969,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.261962176866251e-06,
+      "loss": 0.5877,
+      "step": 4767
+    },
+    {
+      "epoch": 0.40630592245419683,
+      "grad_norm": 17.875,
+      "learning_rate": 2.261863526870999e-06,
+      "loss": 0.8367,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4063911376224968,
+      "grad_norm": 13.25,
+      "learning_rate": 2.2617648585903556e-06,
+      "loss": 0.853,
+      "step": 4769
+    },
+    {
+      "epoch": 0.40647635279079675,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2616661720261034e-06,
+      "loss": 0.7978,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4065615679590967,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2615674671800266e-06,
+      "loss": 0.4212,
+      "step": 4771
+    },
+    {
+      "epoch": 0.40664678312739666,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2614687440539085e-06,
+      "loss": 0.6508,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4067319982956966,
+      "grad_norm": 23.0,
+      "learning_rate": 2.261370002649533e-06,
+      "loss": 0.9489,
+      "step": 4773
+    },
+    {
+      "epoch": 0.4068172134639966,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.2612712429686846e-06,
+      "loss": 0.4592,
+      "step": 4774
+    },
+    {
+      "epoch": 0.40690242863229653,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2611724650131477e-06,
+      "loss": 0.7842,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4069876438005965,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.261073668784708e-06,
+      "loss": 0.4253,
+      "step": 4776
+    },
+    {
+      "epoch": 0.40707285896889644,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.26097485428515e-06,
+      "loss": 0.7862,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4071580741371964,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2608760215162594e-06,
+      "loss": 0.8884,
+      "step": 4778
+    },
+    {
+      "epoch": 0.40724328930549636,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.2607771704798234e-06,
+      "loss": 0.4183,
+      "step": 4779
+    },
+    {
+      "epoch": 0.4073285044737963,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2606783011776272e-06,
+      "loss": 0.6357,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4074137196420963,
+      "grad_norm": 15.375,
+      "learning_rate": 2.260579413611458e-06,
+      "loss": 0.8026,
+      "step": 4781
+    },
+    {
+      "epoch": 0.40749893481039623,
+      "grad_norm": 18.0,
+      "learning_rate": 2.260480507783102e-06,
+      "loss": 0.8144,
+      "step": 4782
+    },
+    {
+      "epoch": 0.4075841499786962,
+      "grad_norm": 14.125,
+      "learning_rate": 2.2603815836943474e-06,
+      "loss": 0.4435,
+      "step": 4783
+    },
+    {
+      "epoch": 0.40766936514699614,
+      "grad_norm": 24.25,
+      "learning_rate": 2.2602826413469813e-06,
+      "loss": 0.9548,
+      "step": 4784
+    },
+    {
+      "epoch": 0.4077545803152961,
+      "grad_norm": 24.0,
+      "learning_rate": 2.2601836807427925e-06,
+      "loss": 0.9395,
+      "step": 4785
+    },
+    {
+      "epoch": 0.40783979548359606,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2600847018835685e-06,
+      "loss": 0.9131,
+      "step": 4786
+    },
+    {
+      "epoch": 0.407925010651896,
+      "grad_norm": 19.125,
+      "learning_rate": 2.2599857047710983e-06,
+      "loss": 1.0284,
+      "step": 4787
+    },
+    {
+      "epoch": 0.40801022582019597,
+      "grad_norm": 13.125,
+      "learning_rate": 2.259886689407171e-06,
+      "loss": 0.718,
+      "step": 4788
+    },
+    {
+      "epoch": 0.4080954409884959,
+      "grad_norm": 18.75,
+      "learning_rate": 2.259787655793575e-06,
+      "loss": 1.0749,
+      "step": 4789
+    },
+    {
+      "epoch": 0.4081806561567959,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2596886039321013e-06,
+      "loss": 0.8422,
+      "step": 4790
+    },
+    {
+      "epoch": 0.40826587132509584,
+      "grad_norm": 12.125,
+      "learning_rate": 2.2595895338245387e-06,
+      "loss": 0.5409,
+      "step": 4791
+    },
+    {
+      "epoch": 0.4083510864933958,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2594904454726784e-06,
+      "loss": 1.057,
+      "step": 4792
+    },
+    {
+      "epoch": 0.40843630166169576,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2593913388783105e-06,
+      "loss": 0.5485,
+      "step": 4793
+    },
+    {
+      "epoch": 0.4085215168299957,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.2592922140432262e-06,
+      "loss": 0.9442,
+      "step": 4794
+    },
+    {
+      "epoch": 0.40860673199829567,
+      "grad_norm": 19.5,
+      "learning_rate": 2.2591930709692166e-06,
+      "loss": 0.7642,
+      "step": 4795
+    },
+    {
+      "epoch": 0.4086919471665957,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2590939096580734e-06,
+      "loss": 0.6897,
+      "step": 4796
+    },
+    {
+      "epoch": 0.40877716233489564,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.258994730111588e-06,
+      "loss": 0.5593,
+      "step": 4797
+    },
+    {
+      "epoch": 0.4088623775031956,
+      "grad_norm": 12.25,
+      "learning_rate": 2.258895532331554e-06,
+      "loss": 0.506,
+      "step": 4798
+    },
+    {
+      "epoch": 0.40894759267149555,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.258796316319763e-06,
+      "loss": 0.6796,
+      "step": 4799
+    },
+    {
+      "epoch": 0.4090328078397955,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.258697082078008e-06,
+      "loss": 1.0086,
+      "step": 4800
+    },
+    {
+      "epoch": 0.40911802300809547,
+      "grad_norm": 18.25,
+      "learning_rate": 2.2585978296080823e-06,
+      "loss": 1.0874,
+      "step": 4801
+    },
+    {
+      "epoch": 0.4092032381763954,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.25849855891178e-06,
+      "loss": 0.649,
+      "step": 4802
+    },
+    {
+      "epoch": 0.4092884533446954,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2583992699908943e-06,
+      "loss": 0.7,
+      "step": 4803
+    },
+    {
+      "epoch": 0.40937366851299534,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.25829996284722e-06,
+      "loss": 0.6285,
+      "step": 4804
+    },
+    {
+      "epoch": 0.4094588836812953,
+      "grad_norm": 18.75,
+      "learning_rate": 2.258200637482551e-06,
+      "loss": 0.9545,
+      "step": 4805
+    },
+    {
+      "epoch": 0.40954409884959525,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.258101293898683e-06,
+      "loss": 0.8332,
+      "step": 4806
+    },
+    {
+      "epoch": 0.4096293140178952,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.258001932097411e-06,
+      "loss": 0.9381,
+      "step": 4807
+    },
+    {
+      "epoch": 0.40971452918619516,
+      "grad_norm": 18.5,
+      "learning_rate": 2.2579025520805303e-06,
+      "loss": 0.8526,
+      "step": 4808
+    },
+    {
+      "epoch": 0.4097997443544951,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.2578031538498367e-06,
+      "loss": 0.7659,
+      "step": 4809
+    },
+    {
+      "epoch": 0.4098849595227951,
+      "grad_norm": 14.125,
+      "learning_rate": 2.257703737407127e-06,
+      "loss": 0.7903,
+      "step": 4810
+    },
+    {
+      "epoch": 0.40997017469109504,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2576043027541972e-06,
+      "loss": 0.879,
+      "step": 4811
+    },
+    {
+      "epoch": 0.410055389859395,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.257504849892845e-06,
+      "loss": 0.3422,
+      "step": 4812
+    },
+    {
+      "epoch": 0.41014060502769495,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.257405378824866e-06,
+      "loss": 0.7472,
+      "step": 4813
+    },
+    {
+      "epoch": 0.4102258201959949,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.25730588955206e-06,
+      "loss": 0.6635,
+      "step": 4814
+    },
+    {
+      "epoch": 0.41031103536429486,
+      "grad_norm": 15.5,
+      "learning_rate": 2.257206382076223e-06,
+      "loss": 0.7449,
+      "step": 4815
+    },
+    {
+      "epoch": 0.4103962505325948,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.257106856399154e-06,
+      "loss": 0.6581,
+      "step": 4816
+    },
+    {
+      "epoch": 0.4104814657008948,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.257007312522651e-06,
+      "loss": 0.5158,
+      "step": 4817
+    },
+    {
+      "epoch": 0.41056668086919473,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2569077504485136e-06,
+      "loss": 0.7873,
+      "step": 4818
+    },
+    {
+      "epoch": 0.4106518960374947,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.2568081701785405e-06,
+      "loss": 0.4938,
+      "step": 4819
+    },
+    {
+      "epoch": 0.41073711120579465,
+      "grad_norm": 17.875,
+      "learning_rate": 2.2567085717145314e-06,
+      "loss": 0.675,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4108223263740946,
+      "grad_norm": 11.0,
+      "learning_rate": 2.256608955058286e-06,
+      "loss": 0.4564,
+      "step": 4821
+    },
+    {
+      "epoch": 0.41090754154239456,
+      "grad_norm": 20.125,
+      "learning_rate": 2.2565093202116048e-06,
+      "loss": 0.6711,
+      "step": 4822
+    },
+    {
+      "epoch": 0.4109927567106945,
+      "grad_norm": 18.625,
+      "learning_rate": 2.256409667176288e-06,
+      "loss": 0.9432,
+      "step": 4823
+    },
+    {
+      "epoch": 0.4110779718789945,
+      "grad_norm": 16.0,
+      "learning_rate": 2.2563099959541363e-06,
+      "loss": 0.6405,
+      "step": 4824
+    },
+    {
+      "epoch": 0.41116318704729443,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.256210306546951e-06,
+      "loss": 0.5645,
+      "step": 4825
+    },
+    {
+      "epoch": 0.4112484022155944,
+      "grad_norm": 18.625,
+      "learning_rate": 2.256110598956534e-06,
+      "loss": 1.231,
+      "step": 4826
+    },
+    {
+      "epoch": 0.41133361738389435,
+      "grad_norm": 25.125,
+      "learning_rate": 2.2560108731846862e-06,
+      "loss": 0.8709,
+      "step": 4827
+    },
+    {
+      "epoch": 0.4114188325521943,
+      "grad_norm": 20.0,
+      "learning_rate": 2.255911129233211e-06,
+      "loss": 0.9058,
+      "step": 4828
+    },
+    {
+      "epoch": 0.41150404772049426,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2558113671039097e-06,
+      "loss": 0.9308,
+      "step": 4829
+    },
+    {
+      "epoch": 0.4115892628887942,
+      "grad_norm": 11.25,
+      "learning_rate": 2.255711586798586e-06,
+      "loss": 0.5423,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4116744780570942,
+      "grad_norm": 16.25,
+      "learning_rate": 2.255611788319042e-06,
+      "loss": 1.1954,
+      "step": 4831
+    },
+    {
+      "epoch": 0.41175969322539413,
+      "grad_norm": 20.625,
+      "learning_rate": 2.255511971667082e-06,
+      "loss": 1.227,
+      "step": 4832
+    },
+    {
+      "epoch": 0.4118449083936941,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2554121368445097e-06,
+      "loss": 0.8236,
+      "step": 4833
+    },
+    {
+      "epoch": 0.41193012356199404,
+      "grad_norm": 17.375,
+      "learning_rate": 2.2553122838531286e-06,
+      "loss": 1.2031,
+      "step": 4834
+    },
+    {
+      "epoch": 0.412015338730294,
+      "grad_norm": 18.0,
+      "learning_rate": 2.2552124126947446e-06,
+      "loss": 1.0453,
+      "step": 4835
+    },
+    {
+      "epoch": 0.41210055389859396,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2551125233711605e-06,
+      "loss": 0.7285,
+      "step": 4836
+    },
+    {
+      "epoch": 0.4121857690668939,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2550126158841827e-06,
+      "loss": 0.8544,
+      "step": 4837
+    },
+    {
+      "epoch": 0.41227098423519387,
+      "grad_norm": 20.25,
+      "learning_rate": 2.2549126902356165e-06,
+      "loss": 0.985,
+      "step": 4838
+    },
+    {
+      "epoch": 0.41235619940349383,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.2548127464272674e-06,
+      "loss": 0.6943,
+      "step": 4839
+    },
+    {
+      "epoch": 0.4124414145717938,
+      "grad_norm": 15.25,
+      "learning_rate": 2.2547127844609413e-06,
+      "loss": 0.8087,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41252662974009374,
+      "grad_norm": 70.5,
+      "learning_rate": 2.2546128043384454e-06,
+      "loss": 0.5554,
+      "step": 4841
+    },
+    {
+      "epoch": 0.4126118449083937,
+      "grad_norm": 11.875,
+      "learning_rate": 2.2545128060615855e-06,
+      "loss": 0.6621,
+      "step": 4842
+    },
+    {
+      "epoch": 0.41269706007669366,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2544127896321687e-06,
+      "loss": 0.691,
+      "step": 4843
+    },
+    {
+      "epoch": 0.4127822752449936,
+      "grad_norm": 17.375,
+      "learning_rate": 2.254312755052003e-06,
+      "loss": 0.6695,
+      "step": 4844
+    },
+    {
+      "epoch": 0.41286749041329357,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2542127023228967e-06,
+      "loss": 0.4374,
+      "step": 4845
+    },
+    {
+      "epoch": 0.4129527055815935,
+      "grad_norm": 24.125,
+      "learning_rate": 2.2541126314466565e-06,
+      "loss": 0.8464,
+      "step": 4846
+    },
+    {
+      "epoch": 0.4130379207498935,
+      "grad_norm": 23.125,
+      "learning_rate": 2.2540125424250914e-06,
+      "loss": 0.702,
+      "step": 4847
+    },
+    {
+      "epoch": 0.41312313591819344,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.25391243526001e-06,
+      "loss": 0.7888,
+      "step": 4848
+    },
+    {
+      "epoch": 0.4132083510864934,
+      "grad_norm": 14.875,
+      "learning_rate": 2.2538123099532212e-06,
+      "loss": 0.9714,
+      "step": 4849
+    },
+    {
+      "epoch": 0.41329356625479335,
+      "grad_norm": 17.0,
+      "learning_rate": 2.2537121665065346e-06,
+      "loss": 0.2549,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4133787814230933,
+      "grad_norm": 10.0,
+      "learning_rate": 2.25361200492176e-06,
+      "loss": 0.4591,
+      "step": 4851
+    },
+    {
+      "epoch": 0.41346399659139327,
+      "grad_norm": 15.5,
+      "learning_rate": 2.253511825200707e-06,
+      "loss": 0.7933,
+      "step": 4852
+    },
+    {
+      "epoch": 0.4135492117596932,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2534116273451865e-06,
+      "loss": 0.8653,
+      "step": 4853
+    },
+    {
+      "epoch": 0.4136344269279932,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.253311411357009e-06,
+      "loss": 0.7351,
+      "step": 4854
+    },
+    {
+      "epoch": 0.41371964209629314,
+      "grad_norm": 16.75,
+      "learning_rate": 2.253211177237985e-06,
+      "loss": 0.8661,
+      "step": 4855
+    },
+    {
+      "epoch": 0.4138048572645931,
+      "grad_norm": 21.125,
+      "learning_rate": 2.2531109249899265e-06,
+      "loss": 0.9118,
+      "step": 4856
+    },
+    {
+      "epoch": 0.41389007243289305,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2530106546146446e-06,
+      "loss": 0.6502,
+      "step": 4857
+    },
+    {
+      "epoch": 0.413975287601193,
+      "grad_norm": 13.75,
+      "learning_rate": 2.252910366113952e-06,
+      "loss": 0.8105,
+      "step": 4858
+    },
+    {
+      "epoch": 0.41406050276949297,
+      "grad_norm": 16.0,
+      "learning_rate": 2.25281005948966e-06,
+      "loss": 1.1773,
+      "step": 4859
+    },
+    {
+      "epoch": 0.4141457179377929,
+      "grad_norm": 33.5,
+      "learning_rate": 2.2527097347435823e-06,
+      "loss": 1.5125,
+      "step": 4860
+    },
+    {
+      "epoch": 0.4142309331060929,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2526093918775313e-06,
+      "loss": 0.7659,
+      "step": 4861
+    },
+    {
+      "epoch": 0.41431614827439284,
+      "grad_norm": 12.5,
+      "learning_rate": 2.25250903089332e-06,
+      "loss": 0.5829,
+      "step": 4862
+    },
+    {
+      "epoch": 0.4144013634426928,
+      "grad_norm": 12.125,
+      "learning_rate": 2.252408651792763e-06,
+      "loss": 0.533,
+      "step": 4863
+    },
+    {
+      "epoch": 0.41448657861099275,
+      "grad_norm": 23.25,
+      "learning_rate": 2.252308254577673e-06,
+      "loss": 0.9993,
+      "step": 4864
+    },
+    {
+      "epoch": 0.4145717937792927,
+      "grad_norm": 12.625,
+      "learning_rate": 2.252207839249866e-06,
+      "loss": 0.3742,
+      "step": 4865
+    },
+    {
+      "epoch": 0.41465700894759266,
+      "grad_norm": 11.25,
+      "learning_rate": 2.252107405811155e-06,
+      "loss": 0.4757,
+      "step": 4866
+    },
+    {
+      "epoch": 0.4147422241158926,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.252006954263355e-06,
+      "loss": 0.7955,
+      "step": 4867
+    },
+    {
+      "epoch": 0.4148274392841926,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.251906484608282e-06,
+      "loss": 0.5482,
+      "step": 4868
+    },
+    {
+      "epoch": 0.41491265445249254,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.2518059968477516e-06,
+      "loss": 0.8476,
+      "step": 4869
+    },
+    {
+      "epoch": 0.4149978696207925,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.251705490983579e-06,
+      "loss": 0.3362,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41508308478909245,
+      "grad_norm": 18.375,
+      "learning_rate": 2.251604967017581e-06,
+      "loss": 0.7088,
+      "step": 4871
+    },
+    {
+      "epoch": 0.4151682999573924,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2515044249515745e-06,
+      "loss": 0.6508,
+      "step": 4872
+    },
+    {
+      "epoch": 0.41525351512569236,
+      "grad_norm": 15.375,
+      "learning_rate": 2.2514038647873758e-06,
+      "loss": 1.0892,
+      "step": 4873
+    },
+    {
+      "epoch": 0.4153387302939923,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2513032865268018e-06,
+      "loss": 0.8347,
+      "step": 4874
+    },
+    {
+      "epoch": 0.4154239454622923,
+      "grad_norm": 15.625,
+      "learning_rate": 2.251202690171671e-06,
+      "loss": 0.667,
+      "step": 4875
+    },
+    {
+      "epoch": 0.41550916063059223,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.251102075723801e-06,
+      "loss": 0.5272,
+      "step": 4876
+    },
+    {
+      "epoch": 0.4155943757988922,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.2510014431850093e-06,
+      "loss": 0.351,
+      "step": 4877
+    },
+    {
+      "epoch": 0.41567959096719215,
+      "grad_norm": 19.0,
+      "learning_rate": 2.2509007925571154e-06,
+      "loss": 1.1781,
+      "step": 4878
+    },
+    {
+      "epoch": 0.4157648061354921,
+      "grad_norm": 15.375,
+      "learning_rate": 2.250800123841937e-06,
+      "loss": 0.5345,
+      "step": 4879
+    },
+    {
+      "epoch": 0.41585002130379206,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2506994370412944e-06,
+      "loss": 0.5204,
+      "step": 4880
+    },
+    {
+      "epoch": 0.415935236472092,
+      "grad_norm": 17.0,
+      "learning_rate": 2.2505987321570066e-06,
+      "loss": 0.6551,
+      "step": 4881
+    },
+    {
+      "epoch": 0.416020451640392,
+      "grad_norm": 11.5,
+      "learning_rate": 2.250498009190894e-06,
+      "loss": 0.6634,
+      "step": 4882
+    },
+    {
+      "epoch": 0.41610566680869193,
+      "grad_norm": 14.0,
+      "learning_rate": 2.2503972681447756e-06,
+      "loss": 0.4377,
+      "step": 4883
+    },
+    {
+      "epoch": 0.4161908819769919,
+      "grad_norm": 15.125,
+      "learning_rate": 2.250296509020473e-06,
+      "loss": 0.8151,
+      "step": 4884
+    },
+    {
+      "epoch": 0.41627609714529185,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2501957318198063e-06,
+      "loss": 0.8733,
+      "step": 4885
+    },
+    {
+      "epoch": 0.4163613123135918,
+      "grad_norm": 15.375,
+      "learning_rate": 2.2500949365445974e-06,
+      "loss": 0.6452,
+      "step": 4886
+    },
+    {
+      "epoch": 0.41644652748189176,
+      "grad_norm": 16.375,
+      "learning_rate": 2.249994123196667e-06,
+      "loss": 0.9311,
+      "step": 4887
+    },
+    {
+      "epoch": 0.4165317426501917,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2498932917778373e-06,
+      "loss": 0.7952,
+      "step": 4888
+    },
+    {
+      "epoch": 0.4166169578184917,
+      "grad_norm": 13.5,
+      "learning_rate": 2.2497924422899304e-06,
+      "loss": 0.735,
+      "step": 4889
+    },
+    {
+      "epoch": 0.41670217298679163,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2496915747347684e-06,
+      "loss": 0.6473,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4167873881550916,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2495906891141746e-06,
+      "loss": 0.3715,
+      "step": 4891
+    },
+    {
+      "epoch": 0.41687260332339154,
+      "grad_norm": 16.5,
+      "learning_rate": 2.249489785429972e-06,
+      "loss": 0.5998,
+      "step": 4892
+    },
+    {
+      "epoch": 0.4169578184916915,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2493888636839837e-06,
+      "loss": 0.9581,
+      "step": 4893
+    },
+    {
+      "epoch": 0.41704303365999146,
+      "grad_norm": 14.625,
+      "learning_rate": 2.249287923878034e-06,
+      "loss": 0.504,
+      "step": 4894
+    },
+    {
+      "epoch": 0.4171282488282914,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.2491869660139463e-06,
+      "loss": 0.6031,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41721346399659137,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.2490859900935455e-06,
+      "loss": 0.5202,
+      "step": 4896
+    },
+    {
+      "epoch": 0.41729867916489133,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.248984996118656e-06,
+      "loss": 0.6677,
+      "step": 4897
+    },
+    {
+      "epoch": 0.4173838943331913,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.248883984091103e-06,
+      "loss": 0.5755,
+      "step": 4898
+    },
+    {
+      "epoch": 0.41746910950149124,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.2487829540127124e-06,
+      "loss": 0.5265,
+      "step": 4899
+    },
+    {
+      "epoch": 0.4175543246697912,
+      "grad_norm": 13.0,
+      "learning_rate": 2.248681905885309e-06,
+      "loss": 0.5974,
+      "step": 4900
+    },
+    {
+      "epoch": 0.41763953983809116,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.24858083971072e-06,
+      "loss": 0.9805,
+      "step": 4901
+    },
+    {
+      "epoch": 0.4177247550063911,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2484797554907706e-06,
+      "loss": 0.5058,
+      "step": 4902
+    },
+    {
+      "epoch": 0.41780997017469107,
+      "grad_norm": 17.0,
+      "learning_rate": 2.2483786532272885e-06,
+      "loss": 0.6768,
+      "step": 4903
+    },
+    {
+      "epoch": 0.417895185342991,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2482775329220995e-06,
+      "loss": 0.6575,
+      "step": 4904
+    },
+    {
+      "epoch": 0.417980400511291,
+      "grad_norm": 20.25,
+      "learning_rate": 2.248176394577032e-06,
+      "loss": 1.009,
+      "step": 4905
+    },
+    {
+      "epoch": 0.41806561567959094,
+      "grad_norm": 12.25,
+      "learning_rate": 2.2480752381939132e-06,
+      "loss": 0.3105,
+      "step": 4906
+    },
+    {
+      "epoch": 0.4181508308478909,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.247974063774571e-06,
+      "loss": 0.6516,
+      "step": 4907
+    },
+    {
+      "epoch": 0.41823604601619085,
+      "grad_norm": 19.875,
+      "learning_rate": 2.247872871320834e-06,
+      "loss": 0.9633,
+      "step": 4908
+    },
+    {
+      "epoch": 0.41832126118449087,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2477716608345307e-06,
+      "loss": 0.6915,
+      "step": 4909
+    },
+    {
+      "epoch": 0.4184064763527908,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.2476704323174904e-06,
+      "loss": 0.7064,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4184916915210908,
+      "grad_norm": 18.875,
+      "learning_rate": 2.247569185771542e-06,
+      "loss": 0.8686,
+      "step": 4911
+    },
+    {
+      "epoch": 0.41857690668939074,
+      "grad_norm": 13.125,
+      "learning_rate": 2.247467921198516e-06,
+      "loss": 0.7016,
+      "step": 4912
+    },
+    {
+      "epoch": 0.4186621218576907,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2473666386002403e-06,
+      "loss": 1.0099,
+      "step": 4913
+    },
+    {
+      "epoch": 0.41874733702599065,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2472653379785475e-06,
+      "loss": 0.6496,
+      "step": 4914
+    },
+    {
+      "epoch": 0.4188325521942906,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.2471640193352665e-06,
+      "loss": 0.3736,
+      "step": 4915
+    },
+    {
+      "epoch": 0.41891776736259057,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.24706268267223e-06,
+      "loss": 1.0165,
+      "step": 4916
+    },
+    {
+      "epoch": 0.4190029825308905,
+      "grad_norm": 15.625,
+      "learning_rate": 2.2469613279912672e-06,
+      "loss": 0.7243,
+      "step": 4917
+    },
+    {
+      "epoch": 0.4190881976991905,
+      "grad_norm": 14.75,
+      "learning_rate": 2.246859955294211e-06,
+      "loss": 0.5785,
+      "step": 4918
+    },
+    {
+      "epoch": 0.41917341286749044,
+      "grad_norm": 18.875,
+      "learning_rate": 2.246758564582893e-06,
+      "loss": 1.0706,
+      "step": 4919
+    },
+    {
+      "epoch": 0.4192586280357904,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2466571558591457e-06,
+      "loss": 0.7378,
+      "step": 4920
+    },
+    {
+      "epoch": 0.41934384320409035,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.2465557291248006e-06,
+      "loss": 0.6348,
+      "step": 4921
+    },
+    {
+      "epoch": 0.4194290583723903,
+      "grad_norm": 12.875,
+      "learning_rate": 2.2464542843816923e-06,
+      "loss": 0.5927,
+      "step": 4922
+    },
+    {
+      "epoch": 0.41951427354069026,
+      "grad_norm": 13.875,
+      "learning_rate": 2.246352821631653e-06,
+      "loss": 0.325,
+      "step": 4923
+    },
+    {
+      "epoch": 0.4195994887089902,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2462513408765153e-06,
+      "loss": 0.8192,
+      "step": 4924
+    },
+    {
+      "epoch": 0.4196847038772902,
+      "grad_norm": 12.5,
+      "learning_rate": 2.246149842118115e-06,
+      "loss": 0.9298,
+      "step": 4925
+    },
+    {
+      "epoch": 0.41976991904559013,
+      "grad_norm": 13.75,
+      "learning_rate": 2.246048325358285e-06,
+      "loss": 0.8035,
+      "step": 4926
+    },
+    {
+      "epoch": 0.4198551342138901,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2459467905988604e-06,
+      "loss": 0.7296,
+      "step": 4927
+    },
+    {
+      "epoch": 0.41994034938219005,
+      "grad_norm": 24.875,
+      "learning_rate": 2.2458452378416753e-06,
+      "loss": 0.9956,
+      "step": 4928
+    },
+    {
+      "epoch": 0.42002556455049,
+      "grad_norm": 20.375,
+      "learning_rate": 2.245743667088566e-06,
+      "loss": 0.8621,
+      "step": 4929
+    },
+    {
+      "epoch": 0.42011077971878996,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2456420783413667e-06,
+      "loss": 0.7533,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4201959948870899,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.245540471601914e-06,
+      "loss": 1.0693,
+      "step": 4931
+    },
+    {
+      "epoch": 0.4202812100553899,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2454388468720444e-06,
+      "loss": 0.9583,
+      "step": 4932
+    },
+    {
+      "epoch": 0.42036642522368983,
+      "grad_norm": 11.375,
+      "learning_rate": 2.245337204153593e-06,
+      "loss": 0.4919,
+      "step": 4933
+    },
+    {
+      "epoch": 0.4204516403919898,
+      "grad_norm": 12.875,
+      "learning_rate": 2.245235543448398e-06,
+      "loss": 0.5027,
+      "step": 4934
+    },
+    {
+      "epoch": 0.42053685556028975,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.245133864758296e-06,
+      "loss": 0.6087,
+      "step": 4935
+    },
+    {
+      "epoch": 0.4206220707285897,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.245032168085124e-06,
+      "loss": 1.1637,
+      "step": 4936
+    },
+    {
+      "epoch": 0.42070728589688966,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.2449304534307206e-06,
+      "loss": 0.4725,
+      "step": 4937
+    },
+    {
+      "epoch": 0.4207925010651896,
+      "grad_norm": 16.375,
+      "learning_rate": 2.2448287207969234e-06,
+      "loss": 1.0574,
+      "step": 4938
+    },
+    {
+      "epoch": 0.4208777162334896,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2447269701855706e-06,
+      "loss": 1.0514,
+      "step": 4939
+    },
+    {
+      "epoch": 0.42096293140178953,
+      "grad_norm": 14.0,
+      "learning_rate": 2.244625201598501e-06,
+      "loss": 0.764,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4210481465700895,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2445234150375542e-06,
+      "loss": 0.9954,
+      "step": 4941
+    },
+    {
+      "epoch": 0.42113336173838944,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.244421610504569e-06,
+      "loss": 0.5302,
+      "step": 4942
+    },
+    {
+      "epoch": 0.4212185769066894,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.244319788001386e-06,
+      "loss": 0.4155,
+      "step": 4943
+    },
+    {
+      "epoch": 0.42130379207498936,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.2442179475298433e-06,
+      "loss": 0.6276,
+      "step": 4944
+    },
+    {
+      "epoch": 0.4213890072432893,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2441160890917833e-06,
+      "loss": 0.7554,
+      "step": 4945
+    },
+    {
+      "epoch": 0.42147422241158927,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.2440142126890455e-06,
+      "loss": 0.7116,
+      "step": 4946
+    },
+    {
+      "epoch": 0.42155943757988923,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.243912318323472e-06,
+      "loss": 0.4652,
+      "step": 4947
+    },
+    {
+      "epoch": 0.4216446527481892,
+      "grad_norm": 15.5,
+      "learning_rate": 2.243810405996903e-06,
+      "loss": 1.0827,
+      "step": 4948
+    },
+    {
+      "epoch": 0.42172986791648914,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2437084757111804e-06,
+      "loss": 0.9546,
+      "step": 4949
+    },
+    {
+      "epoch": 0.4218150830847891,
+      "grad_norm": 11.0,
+      "learning_rate": 2.2436065274681467e-06,
+      "loss": 0.6164,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42190029825308906,
+      "grad_norm": 12.875,
+      "learning_rate": 2.243504561269644e-06,
+      "loss": 0.7249,
+      "step": 4951
+    },
+    {
+      "epoch": 0.421985513421389,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.2434025771175147e-06,
+      "loss": 0.6985,
+      "step": 4952
+    },
+    {
+      "epoch": 0.42207072858968897,
+      "grad_norm": 12.5,
+      "learning_rate": 2.243300575013602e-06,
+      "loss": 0.6818,
+      "step": 4953
+    },
+    {
+      "epoch": 0.4221559437579889,
+      "grad_norm": 29.25,
+      "learning_rate": 2.243198554959749e-06,
+      "loss": 1.2069,
+      "step": 4954
+    },
+    {
+      "epoch": 0.4222411589262889,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2430965169577993e-06,
+      "loss": 1.3152,
+      "step": 4955
+    },
+    {
+      "epoch": 0.42232637409458884,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.242994461009597e-06,
+      "loss": 0.7069,
+      "step": 4956
+    },
+    {
+      "epoch": 0.4224115892628888,
+      "grad_norm": 16.625,
+      "learning_rate": 2.242892387116986e-06,
+      "loss": 0.6766,
+      "step": 4957
+    },
+    {
+      "epoch": 0.42249680443118875,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2427902952818117e-06,
+      "loss": 0.7159,
+      "step": 4958
+    },
+    {
+      "epoch": 0.4225820195994887,
+      "grad_norm": 14.0,
+      "learning_rate": 2.242688185505918e-06,
+      "loss": 0.8027,
+      "step": 4959
+    },
+    {
+      "epoch": 0.42266723476778867,
+      "grad_norm": 15.25,
+      "learning_rate": 2.2425860577911504e-06,
+      "loss": 1.0403,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4227524499360886,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.242483912139355e-06,
+      "loss": 0.8111,
+      "step": 4961
+    },
+    {
+      "epoch": 0.4228376651043886,
+      "grad_norm": 15.5,
+      "learning_rate": 2.242381748552377e-06,
+      "loss": 0.7659,
+      "step": 4962
+    },
+    {
+      "epoch": 0.42292288027268854,
+      "grad_norm": 13.75,
+      "learning_rate": 2.242279567032063e-06,
+      "loss": 0.9226,
+      "step": 4963
+    },
+    {
+      "epoch": 0.4230080954409885,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2421773675802596e-06,
+      "loss": 0.5856,
+      "step": 4964
+    },
+    {
+      "epoch": 0.42309331060928845,
+      "grad_norm": 29.25,
+      "learning_rate": 2.242075150198813e-06,
+      "loss": 1.1184,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4231785257775884,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.241972914889571e-06,
+      "loss": 0.9654,
+      "step": 4966
+    },
+    {
+      "epoch": 0.42326374094588837,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.241870661654381e-06,
+      "loss": 0.5952,
+      "step": 4967
+    },
+    {
+      "epoch": 0.4233489561141883,
+      "grad_norm": 16.25,
+      "learning_rate": 2.241768390495091e-06,
+      "loss": 0.8882,
+      "step": 4968
+    },
+    {
+      "epoch": 0.4234341712824883,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2416661014135484e-06,
+      "loss": 1.246,
+      "step": 4969
+    },
+    {
+      "epoch": 0.42351938645078824,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.2415637944116025e-06,
+      "loss": 0.6923,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4236046016190882,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2414614694911016e-06,
+      "loss": 0.5906,
+      "step": 4971
+    },
+    {
+      "epoch": 0.42368981678738815,
+      "grad_norm": 16.25,
+      "learning_rate": 2.241359126653895e-06,
+      "loss": 0.5943,
+      "step": 4972
+    },
+    {
+      "epoch": 0.4237750319556881,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.241256765901832e-06,
+      "loss": 0.5606,
+      "step": 4973
+    },
+    {
+      "epoch": 0.42386024712398807,
+      "grad_norm": 23.25,
+      "learning_rate": 2.2411543872367623e-06,
+      "loss": 1.257,
+      "step": 4974
+    },
+    {
+      "epoch": 0.423945462292288,
+      "grad_norm": 20.375,
+      "learning_rate": 2.2410519906605368e-06,
+      "loss": 1.3547,
+      "step": 4975
+    },
+    {
+      "epoch": 0.424030677460588,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.240949576175005e-06,
+      "loss": 0.858,
+      "step": 4976
+    },
+    {
+      "epoch": 0.42411589262888794,
+      "grad_norm": 17.375,
+      "learning_rate": 2.2408471437820175e-06,
+      "loss": 0.689,
+      "step": 4977
+    },
+    {
+      "epoch": 0.4242011077971879,
+      "grad_norm": 35.75,
+      "learning_rate": 2.240744693483426e-06,
+      "loss": 0.5849,
+      "step": 4978
+    },
+    {
+      "epoch": 0.42428632296548785,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2406422252810815e-06,
+      "loss": 0.5204,
+      "step": 4979
+    },
+    {
+      "epoch": 0.4243715381337878,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2405397391768357e-06,
+      "loss": 0.6337,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42445675330208776,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.2404372351725413e-06,
+      "loss": 0.7091,
+      "step": 4981
+    },
+    {
+      "epoch": 0.4245419684703877,
+      "grad_norm": 11.0,
+      "learning_rate": 2.2403347132700493e-06,
+      "loss": 0.5954,
+      "step": 4982
+    },
+    {
+      "epoch": 0.4246271836386877,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.240232173471214e-06,
+      "loss": 0.7097,
+      "step": 4983
+    },
+    {
+      "epoch": 0.42471239880698763,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.240129615777887e-06,
+      "loss": 0.7659,
+      "step": 4984
+    },
+    {
+      "epoch": 0.4247976139752876,
+      "grad_norm": 25.375,
+      "learning_rate": 2.240027040191922e-06,
+      "loss": 0.9313,
+      "step": 4985
+    },
+    {
+      "epoch": 0.42488282914358755,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2399244467151735e-06,
+      "loss": 0.8854,
+      "step": 4986
+    },
+    {
+      "epoch": 0.4249680443118875,
+      "grad_norm": 12.5,
+      "learning_rate": 2.2398218353494945e-06,
+      "loss": 0.4812,
+      "step": 4987
+    },
+    {
+      "epoch": 0.42505325948018746,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.23971920609674e-06,
+      "loss": 0.629,
+      "step": 4988
+    },
+    {
+      "epoch": 0.4251384746484874,
+      "grad_norm": 15.25,
+      "learning_rate": 2.2396165589587636e-06,
+      "loss": 0.8386,
+      "step": 4989
+    },
+    {
+      "epoch": 0.4252236898167874,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.2395138939374213e-06,
+      "loss": 0.5445,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42530890498508733,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.2394112110345672e-06,
+      "loss": 0.5404,
+      "step": 4991
+    },
+    {
+      "epoch": 0.4253941201533873,
+      "grad_norm": 14.25,
+      "learning_rate": 2.239308510252058e-06,
+      "loss": 0.4705,
+      "step": 4992
+    },
+    {
+      "epoch": 0.42547933532168725,
+      "grad_norm": 16.875,
+      "learning_rate": 2.239205791591749e-06,
+      "loss": 1.0347,
+      "step": 4993
+    },
+    {
+      "epoch": 0.4255645504899872,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2391030550554972e-06,
+      "loss": 0.6768,
+      "step": 4994
+    },
+    {
+      "epoch": 0.42564976565828716,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.239000300645158e-06,
+      "loss": 0.6945,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4257349808265871,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.238897528362589e-06,
+      "loss": 0.5118,
+      "step": 4996
+    },
+    {
+      "epoch": 0.4258201959948871,
+      "grad_norm": 15.75,
+      "learning_rate": 2.2387947382096477e-06,
+      "loss": 1.0141,
+      "step": 4997
+    },
+    {
+      "epoch": 0.42590541116318703,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.238691930188191e-06,
+      "loss": 0.6464,
+      "step": 4998
+    },
+    {
+      "epoch": 0.425990626331487,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2385891043000766e-06,
+      "loss": 0.9191,
+      "step": 4999
+    },
+    {
+      "epoch": 0.42607584149978694,
+      "grad_norm": 16.5,
+      "learning_rate": 2.238486260547163e-06,
+      "loss": 0.4816,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4261610566680869,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2383833989313087e-06,
+      "loss": 0.5444,
+      "step": 5001
+    },
+    {
+      "epoch": 0.42624627183638686,
+      "grad_norm": 23.25,
+      "learning_rate": 2.2382805194543726e-06,
+      "loss": 0.8971,
+      "step": 5002
+    },
+    {
+      "epoch": 0.4263314870046868,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.2381776221182136e-06,
+      "loss": 0.9627,
+      "step": 5003
+    },
+    {
+      "epoch": 0.4264167021729868,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.2380747069246916e-06,
+      "loss": 0.6279,
+      "step": 5004
+    },
+    {
+      "epoch": 0.42650191734128673,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2379717738756656e-06,
+      "loss": 0.8512,
+      "step": 5005
+    },
+    {
+      "epoch": 0.4265871325095867,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2378688229729965e-06,
+      "loss": 0.7839,
+      "step": 5006
+    },
+    {
+      "epoch": 0.42667234767788664,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2377658542185442e-06,
+      "loss": 0.8248,
+      "step": 5007
+    },
+    {
+      "epoch": 0.4267575628461866,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.23766286761417e-06,
+      "loss": 0.7751,
+      "step": 5008
+    },
+    {
+      "epoch": 0.42684277801448656,
+      "grad_norm": 17.125,
+      "learning_rate": 2.237559863161734e-06,
+      "loss": 1.0821,
+      "step": 5009
+    },
+    {
+      "epoch": 0.4269279931827865,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2374568408630985e-06,
+      "loss": 0.7242,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42701320835108647,
+      "grad_norm": 16.25,
+      "learning_rate": 2.237353800720125e-06,
+      "loss": 0.8592,
+      "step": 5011
+    },
+    {
+      "epoch": 0.4270984235193864,
+      "grad_norm": 12.5,
+      "learning_rate": 2.2372507427346753e-06,
+      "loss": 0.5808,
+      "step": 5012
+    },
+    {
+      "epoch": 0.4271836386876864,
+      "grad_norm": 15.125,
+      "learning_rate": 2.2371476669086123e-06,
+      "loss": 1.0118,
+      "step": 5013
+    },
+    {
+      "epoch": 0.42726885385598634,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.237044573243798e-06,
+      "loss": 0.525,
+      "step": 5014
+    },
+    {
+      "epoch": 0.4273540690242863,
+      "grad_norm": 23.5,
+      "learning_rate": 2.236941461742096e-06,
+      "loss": 0.629,
+      "step": 5015
+    },
+    {
+      "epoch": 0.42743928419258626,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2368383324053684e-06,
+      "loss": 0.6978,
+      "step": 5016
+    },
+    {
+      "epoch": 0.4275244993608862,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.236735185235481e-06,
+      "loss": 0.5348,
+      "step": 5017
+    },
+    {
+      "epoch": 0.42760971452918617,
+      "grad_norm": 16.0,
+      "learning_rate": 2.2366320202342957e-06,
+      "loss": 0.9351,
+      "step": 5018
+    },
+    {
+      "epoch": 0.4276949296974861,
+      "grad_norm": 21.0,
+      "learning_rate": 2.236528837403678e-06,
+      "loss": 1.0668,
+      "step": 5019
+    },
+    {
+      "epoch": 0.4277801448657861,
+      "grad_norm": 23.75,
+      "learning_rate": 2.2364256367454925e-06,
+      "loss": 1.078,
+      "step": 5020
+    },
+    {
+      "epoch": 0.42786536003408604,
+      "grad_norm": 25.375,
+      "learning_rate": 2.2363224182616035e-06,
+      "loss": 1.1077,
+      "step": 5021
+    },
+    {
+      "epoch": 0.42795057520238605,
+      "grad_norm": 12.125,
+      "learning_rate": 2.2362191819538763e-06,
+      "loss": 0.5359,
+      "step": 5022
+    },
+    {
+      "epoch": 0.428035790370686,
+      "grad_norm": 11.5,
+      "learning_rate": 2.236115927824177e-06,
+      "loss": 0.4355,
+      "step": 5023
+    },
+    {
+      "epoch": 0.42812100553898597,
+      "grad_norm": 13.75,
+      "learning_rate": 2.236012655874371e-06,
+      "loss": 0.555,
+      "step": 5024
+    },
+    {
+      "epoch": 0.4282062207072859,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2359093661063254e-06,
+      "loss": 0.772,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4282914358755859,
+      "grad_norm": 16.5,
+      "learning_rate": 2.235806058521906e-06,
+      "loss": 0.9323,
+      "step": 5026
+    },
+    {
+      "epoch": 0.42837665104388584,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2357027331229796e-06,
+      "loss": 0.9413,
+      "step": 5027
+    },
+    {
+      "epoch": 0.4284618662121858,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.235599389911414e-06,
+      "loss": 0.5359,
+      "step": 5028
+    },
+    {
+      "epoch": 0.42854708138048575,
+      "grad_norm": 10.375,
+      "learning_rate": 2.2354960288890754e-06,
+      "loss": 0.4342,
+      "step": 5029
+    },
+    {
+      "epoch": 0.4286322965487857,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.2353926500578333e-06,
+      "loss": 0.766,
+      "step": 5030
+    },
+    {
+      "epoch": 0.42871751171708566,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2352892534195548e-06,
+      "loss": 0.5773,
+      "step": 5031
+    },
+    {
+      "epoch": 0.4288027268853856,
+      "grad_norm": 20.75,
+      "learning_rate": 2.2351858389761088e-06,
+      "loss": 0.7962,
+      "step": 5032
+    },
+    {
+      "epoch": 0.4288879420536856,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.2350824067293643e-06,
+      "loss": 0.6885,
+      "step": 5033
+    },
+    {
+      "epoch": 0.42897315722198553,
+      "grad_norm": 16.375,
+      "learning_rate": 2.23497895668119e-06,
+      "loss": 0.7861,
+      "step": 5034
+    },
+    {
+      "epoch": 0.4290583723902855,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2348754888334554e-06,
+      "loss": 0.8054,
+      "step": 5035
+    },
+    {
+      "epoch": 0.42914358755858545,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2347720031880303e-06,
+      "loss": 0.4438,
+      "step": 5036
+    },
+    {
+      "epoch": 0.4292288027268854,
+      "grad_norm": 17.375,
+      "learning_rate": 2.234668499746785e-06,
+      "loss": 0.8679,
+      "step": 5037
+    },
+    {
+      "epoch": 0.42931401789518536,
+      "grad_norm": 14.625,
+      "learning_rate": 2.2345649785115897e-06,
+      "loss": 0.6812,
+      "step": 5038
+    },
+    {
+      "epoch": 0.4293992330634853,
+      "grad_norm": 24.375,
+      "learning_rate": 2.234461439484315e-06,
+      "loss": 0.8815,
+      "step": 5039
+    },
+    {
+      "epoch": 0.4294844482317853,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.2343578826668322e-06,
+      "loss": 0.4515,
+      "step": 5040
+    },
+    {
+      "epoch": 0.42956966340008523,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.234254308061013e-06,
+      "loss": 0.5014,
+      "step": 5041
+    },
+    {
+      "epoch": 0.4296548785683852,
+      "grad_norm": 18.875,
+      "learning_rate": 2.234150715668728e-06,
+      "loss": 0.7075,
+      "step": 5042
+    },
+    {
+      "epoch": 0.42974009373668515,
+      "grad_norm": 20.625,
+      "learning_rate": 2.2340471054918506e-06,
+      "loss": 0.7983,
+      "step": 5043
+    },
+    {
+      "epoch": 0.4298253089049851,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.233943477532252e-06,
+      "loss": 0.6592,
+      "step": 5044
+    },
+    {
+      "epoch": 0.42991052407328506,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2338398317918054e-06,
+      "loss": 0.7493,
+      "step": 5045
+    },
+    {
+      "epoch": 0.429995739241585,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.233736168272384e-06,
+      "loss": 0.4503,
+      "step": 5046
+    },
+    {
+      "epoch": 0.430080954409885,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2336324869758607e-06,
+      "loss": 0.4597,
+      "step": 5047
+    },
+    {
+      "epoch": 0.43016616957818493,
+      "grad_norm": 18.25,
+      "learning_rate": 2.233528787904109e-06,
+      "loss": 0.822,
+      "step": 5048
+    },
+    {
+      "epoch": 0.4302513847464849,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.233425071059003e-06,
+      "loss": 0.6965,
+      "step": 5049
+    },
+    {
+      "epoch": 0.43033659991478485,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2333213364424178e-06,
+      "loss": 0.9644,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4304218150830848,
+      "grad_norm": 21.25,
+      "learning_rate": 2.2332175840562264e-06,
+      "loss": 1.0969,
+      "step": 5051
+    },
+    {
+      "epoch": 0.43050703025138476,
+      "grad_norm": 18.75,
+      "learning_rate": 2.233113813902305e-06,
+      "loss": 1.0114,
+      "step": 5052
+    },
+    {
+      "epoch": 0.4305922454196847,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2330100259825285e-06,
+      "loss": 0.7766,
+      "step": 5053
+    },
+    {
+      "epoch": 0.4306774605879847,
+      "grad_norm": 22.125,
+      "learning_rate": 2.2329062202987718e-06,
+      "loss": 1.1876,
+      "step": 5054
+    },
+    {
+      "epoch": 0.43076267575628463,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.232802396852912e-06,
+      "loss": 0.9081,
+      "step": 5055
+    },
+    {
+      "epoch": 0.4308478909245846,
+      "grad_norm": 17.5,
+      "learning_rate": 2.2326985556468238e-06,
+      "loss": 0.7736,
+      "step": 5056
+    },
+    {
+      "epoch": 0.43093310609288454,
+      "grad_norm": 18.0,
+      "learning_rate": 2.2325946966823853e-06,
+      "loss": 0.9413,
+      "step": 5057
+    },
+    {
+      "epoch": 0.4310183212611845,
+      "grad_norm": 19.125,
+      "learning_rate": 2.2324908199614722e-06,
+      "loss": 1.1306,
+      "step": 5058
+    },
+    {
+      "epoch": 0.43110353642948446,
+      "grad_norm": 11.75,
+      "learning_rate": 2.232386925485962e-06,
+      "loss": 0.6384,
+      "step": 5059
+    },
+    {
+      "epoch": 0.4311887515977844,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.232283013257732e-06,
+      "loss": 0.7943,
+      "step": 5060
+    },
+    {
+      "epoch": 0.43127396676608437,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2321790832786607e-06,
+      "loss": 0.9235,
+      "step": 5061
+    },
+    {
+      "epoch": 0.43135918193438433,
+      "grad_norm": 11.125,
+      "learning_rate": 2.2320751355506255e-06,
+      "loss": 0.5174,
+      "step": 5062
+    },
+    {
+      "epoch": 0.4314443971026843,
+      "grad_norm": 10.25,
+      "learning_rate": 2.231971170075505e-06,
+      "loss": 0.4515,
+      "step": 5063
+    },
+    {
+      "epoch": 0.43152961227098424,
+      "grad_norm": 12.75,
+      "learning_rate": 2.231867186855178e-06,
+      "loss": 0.6666,
+      "step": 5064
+    },
+    {
+      "epoch": 0.4316148274392842,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2317631858915235e-06,
+      "loss": 0.7227,
+      "step": 5065
+    },
+    {
+      "epoch": 0.43170004260758416,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2316591671864213e-06,
+      "loss": 0.7342,
+      "step": 5066
+    },
+    {
+      "epoch": 0.4317852577758841,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2315551307417506e-06,
+      "loss": 0.7317,
+      "step": 5067
+    },
+    {
+      "epoch": 0.43187047294418407,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2314510765593917e-06,
+      "loss": 0.5484,
+      "step": 5068
+    },
+    {
+      "epoch": 0.431955688112484,
+      "grad_norm": 17.875,
+      "learning_rate": 2.2313470046412247e-06,
+      "loss": 1.0629,
+      "step": 5069
+    },
+    {
+      "epoch": 0.432040903280784,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2312429149891307e-06,
+      "loss": 0.74,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43212611844908394,
+      "grad_norm": 28.875,
+      "learning_rate": 2.2311388076049903e-06,
+      "loss": 0.8439,
+      "step": 5071
+    },
+    {
+      "epoch": 0.4322113336173839,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.2310346824906855e-06,
+      "loss": 0.519,
+      "step": 5072
+    },
+    {
+      "epoch": 0.43229654878568385,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.2309305396480967e-06,
+      "loss": 0.4261,
+      "step": 5073
+    },
+    {
+      "epoch": 0.4323817639539838,
+      "grad_norm": 12.375,
+      "learning_rate": 2.2308263790791073e-06,
+      "loss": 0.6671,
+      "step": 5074
+    },
+    {
+      "epoch": 0.43246697912228377,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.230722200785599e-06,
+      "loss": 0.8671,
+      "step": 5075
+    },
+    {
+      "epoch": 0.4325521942905837,
+      "grad_norm": 25.625,
+      "learning_rate": 2.230618004769454e-06,
+      "loss": 1.1036,
+      "step": 5076
+    },
+    {
+      "epoch": 0.4326374094588837,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.230513791032556e-06,
+      "loss": 0.4433,
+      "step": 5077
+    },
+    {
+      "epoch": 0.43272262462718364,
+      "grad_norm": 15.875,
+      "learning_rate": 2.230409559576787e-06,
+      "loss": 0.6872,
+      "step": 5078
+    },
+    {
+      "epoch": 0.4328078397954836,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2303053104040322e-06,
+      "loss": 0.8941,
+      "step": 5079
+    },
+    {
+      "epoch": 0.43289305496378355,
+      "grad_norm": 13.5,
+      "learning_rate": 2.230201043516174e-06,
+      "loss": 0.6345,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4329782701320835,
+      "grad_norm": 12.125,
+      "learning_rate": 2.230096758915098e-06,
+      "loss": 0.3736,
+      "step": 5081
+    },
+    {
+      "epoch": 0.43306348530038347,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2299924566026876e-06,
+      "loss": 0.869,
+      "step": 5082
+    },
+    {
+      "epoch": 0.4331487004686834,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.2298881365808276e-06,
+      "loss": 0.9045,
+      "step": 5083
+    },
+    {
+      "epoch": 0.4332339156369834,
+      "grad_norm": 13.125,
+      "learning_rate": 2.229783798851404e-06,
+      "loss": 0.8778,
+      "step": 5084
+    },
+    {
+      "epoch": 0.43331913080528334,
+      "grad_norm": 14.25,
+      "learning_rate": 2.229679443416302e-06,
+      "loss": 0.7869,
+      "step": 5085
+    },
+    {
+      "epoch": 0.4334043459735833,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.2295750702774077e-06,
+      "loss": 0.6352,
+      "step": 5086
+    },
+    {
+      "epoch": 0.43348956114188325,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2294706794366066e-06,
+      "loss": 0.8027,
+      "step": 5087
+    },
+    {
+      "epoch": 0.4335747763101832,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.229366270895785e-06,
+      "loss": 1.0057,
+      "step": 5088
+    },
+    {
+      "epoch": 0.43365999147848316,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2292618446568305e-06,
+      "loss": 0.3933,
+      "step": 5089
+    },
+    {
+      "epoch": 0.4337452066467831,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.2291574007216295e-06,
+      "loss": 0.8552,
+      "step": 5090
+    },
+    {
+      "epoch": 0.4338304218150831,
+      "grad_norm": 16.375,
+      "learning_rate": 2.2290529390920696e-06,
+      "loss": 0.8523,
+      "step": 5091
+    },
+    {
+      "epoch": 0.43391563698338304,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.2289484597700386e-06,
+      "loss": 0.7322,
+      "step": 5092
+    },
+    {
+      "epoch": 0.434000852151683,
+      "grad_norm": 11.0,
+      "learning_rate": 2.2288439627574247e-06,
+      "loss": 0.4172,
+      "step": 5093
+    },
+    {
+      "epoch": 0.43408606731998295,
+      "grad_norm": 13.375,
+      "learning_rate": 2.228739448056116e-06,
+      "loss": 0.5626,
+      "step": 5094
+    },
+    {
+      "epoch": 0.4341712824882829,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.2286349156680014e-06,
+      "loss": 0.6058,
+      "step": 5095
+    },
+    {
+      "epoch": 0.43425649765658286,
+      "grad_norm": 15.25,
+      "learning_rate": 2.2285303655949693e-06,
+      "loss": 0.9247,
+      "step": 5096
+    },
+    {
+      "epoch": 0.4343417128248828,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.22842579783891e-06,
+      "loss": 0.6181,
+      "step": 5097
+    },
+    {
+      "epoch": 0.4344269279931828,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2283212124017124e-06,
+      "loss": 0.6834,
+      "step": 5098
+    },
+    {
+      "epoch": 0.43451214316148273,
+      "grad_norm": 17.5,
+      "learning_rate": 2.228216609285267e-06,
+      "loss": 0.6878,
+      "step": 5099
+    },
+    {
+      "epoch": 0.4345973583297827,
+      "grad_norm": 12.75,
+      "learning_rate": 2.2281119884914635e-06,
+      "loss": 0.6651,
+      "step": 5100
+    },
+    {
+      "epoch": 0.43468257349808265,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2280073500221926e-06,
+      "loss": 1.0826,
+      "step": 5101
+    },
+    {
+      "epoch": 0.4347677886663826,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.2279026938793458e-06,
+      "loss": 0.863,
+      "step": 5102
+    },
+    {
+      "epoch": 0.43485300383468256,
+      "grad_norm": 15.625,
+      "learning_rate": 2.2277980200648142e-06,
+      "loss": 0.8633,
+      "step": 5103
+    },
+    {
+      "epoch": 0.4349382190029825,
+      "grad_norm": 14.875,
+      "learning_rate": 2.2276933285804884e-06,
+      "loss": 0.6028,
+      "step": 5104
+    },
+    {
+      "epoch": 0.4350234341712825,
+      "grad_norm": 19.5,
+      "learning_rate": 2.2275886194282615e-06,
+      "loss": 0.7762,
+      "step": 5105
+    },
+    {
+      "epoch": 0.43510864933958243,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.227483892610025e-06,
+      "loss": 0.5226,
+      "step": 5106
+    },
+    {
+      "epoch": 0.4351938645078824,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.227379148127672e-06,
+      "loss": 0.8069,
+      "step": 5107
+    },
+    {
+      "epoch": 0.43527907967618235,
+      "grad_norm": 16.5,
+      "learning_rate": 2.227274385983095e-06,
+      "loss": 0.7514,
+      "step": 5108
+    },
+    {
+      "epoch": 0.4353642948444823,
+      "grad_norm": 36.25,
+      "learning_rate": 2.227169606178186e-06,
+      "loss": 0.8766,
+      "step": 5109
+    },
+    {
+      "epoch": 0.43544951001278226,
+      "grad_norm": 10.75,
+      "learning_rate": 2.2270648087148412e-06,
+      "loss": 0.6408,
+      "step": 5110
+    },
+    {
+      "epoch": 0.4355347251810822,
+      "grad_norm": 17.625,
+      "learning_rate": 2.2269599935949513e-06,
+      "loss": 1.1137,
+      "step": 5111
+    },
+    {
+      "epoch": 0.4356199403493822,
+      "grad_norm": 15.375,
+      "learning_rate": 2.226855160820413e-06,
+      "loss": 0.893,
+      "step": 5112
+    },
+    {
+      "epoch": 0.43570515551768213,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.2267503103931194e-06,
+      "loss": 0.4471,
+      "step": 5113
+    },
+    {
+      "epoch": 0.4357903706859821,
+      "grad_norm": 18.5,
+      "learning_rate": 2.2266454423149653e-06,
+      "loss": 0.8746,
+      "step": 5114
+    },
+    {
+      "epoch": 0.43587558585428204,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.2265405565878463e-06,
+      "loss": 0.6705,
+      "step": 5115
+    },
+    {
+      "epoch": 0.435960801022582,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.226435653213657e-06,
+      "loss": 0.7691,
+      "step": 5116
+    },
+    {
+      "epoch": 0.43604601619088196,
+      "grad_norm": 17.5,
+      "learning_rate": 2.2263307321942935e-06,
+      "loss": 0.9185,
+      "step": 5117
+    },
+    {
+      "epoch": 0.4361312313591819,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2262257935316523e-06,
+      "loss": 0.6799,
+      "step": 5118
+    },
+    {
+      "epoch": 0.43621644652748187,
+      "grad_norm": 12.625,
+      "learning_rate": 2.226120837227629e-06,
+      "loss": 0.6413,
+      "step": 5119
+    },
+    {
+      "epoch": 0.43630166169578183,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2260158632841207e-06,
+      "loss": 0.5855,
+      "step": 5120
+    },
+    {
+      "epoch": 0.4363868768640818,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2259108717030245e-06,
+      "loss": 0.6167,
+      "step": 5121
+    },
+    {
+      "epoch": 0.43647209203238174,
+      "grad_norm": 15.125,
+      "learning_rate": 2.2258058624862374e-06,
+      "loss": 0.6559,
+      "step": 5122
+    },
+    {
+      "epoch": 0.4365573072006817,
+      "grad_norm": 18.75,
+      "learning_rate": 2.225700835635657e-06,
+      "loss": 0.9811,
+      "step": 5123
+    },
+    {
+      "epoch": 0.43664252236898166,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2255957911531817e-06,
+      "loss": 0.7894,
+      "step": 5124
+    },
+    {
+      "epoch": 0.4367277375372816,
+      "grad_norm": 17.125,
+      "learning_rate": 2.225490729040709e-06,
+      "loss": 1.1962,
+      "step": 5125
+    },
+    {
+      "epoch": 0.43681295270558157,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.225385649300138e-06,
+      "loss": 0.5451,
+      "step": 5126
+    },
+    {
+      "epoch": 0.4368981678738815,
+      "grad_norm": 12.75,
+      "learning_rate": 2.2252805519333676e-06,
+      "loss": 0.6483,
+      "step": 5127
+    },
+    {
+      "epoch": 0.4369833830421815,
+      "grad_norm": 15.125,
+      "learning_rate": 2.2251754369422968e-06,
+      "loss": 0.8555,
+      "step": 5128
+    },
+    {
+      "epoch": 0.43706859821048144,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2250703043288253e-06,
+      "loss": 0.7367,
+      "step": 5129
+    },
+    {
+      "epoch": 0.4371538133787814,
+      "grad_norm": 12.75,
+      "learning_rate": 2.224965154094853e-06,
+      "loss": 0.524,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43723902854708135,
+      "grad_norm": 15.875,
+      "learning_rate": 2.2248599862422798e-06,
+      "loss": 0.7958,
+      "step": 5131
+    },
+    {
+      "epoch": 0.4373242437153813,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2247548007730063e-06,
+      "loss": 0.7912,
+      "step": 5132
+    },
+    {
+      "epoch": 0.43740945888368127,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2246495976889335e-06,
+      "loss": 0.8527,
+      "step": 5133
+    },
+    {
+      "epoch": 0.4374946740519812,
+      "grad_norm": 18.875,
+      "learning_rate": 2.224544376991962e-06,
+      "loss": 0.9801,
+      "step": 5134
+    },
+    {
+      "epoch": 0.43757988922028124,
+      "grad_norm": 11.375,
+      "learning_rate": 2.224439138683994e-06,
+      "loss": 0.651,
+      "step": 5135
+    },
+    {
+      "epoch": 0.4376651043885812,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.224333882766931e-06,
+      "loss": 0.7407,
+      "step": 5136
+    },
+    {
+      "epoch": 0.43775031955688115,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.2242286092426747e-06,
+      "loss": 1.0061,
+      "step": 5137
+    },
+    {
+      "epoch": 0.4378355347251811,
+      "grad_norm": 14.25,
+      "learning_rate": 2.224123318113128e-06,
+      "loss": 0.9623,
+      "step": 5138
+    },
+    {
+      "epoch": 0.43792074989348106,
+      "grad_norm": 16.75,
+      "learning_rate": 2.2240180093801935e-06,
+      "loss": 0.4397,
+      "step": 5139
+    },
+    {
+      "epoch": 0.438005965061781,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2239126830457736e-06,
+      "loss": 1.0909,
+      "step": 5140
+    },
+    {
+      "epoch": 0.438091180230081,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2238073391117726e-06,
+      "loss": 1.1586,
+      "step": 5141
+    },
+    {
+      "epoch": 0.43817639539838094,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2237019775800938e-06,
+      "loss": 0.9064,
+      "step": 5142
+    },
+    {
+      "epoch": 0.4382616105666809,
+      "grad_norm": 15.375,
+      "learning_rate": 2.2235965984526408e-06,
+      "loss": 0.7616,
+      "step": 5143
+    },
+    {
+      "epoch": 0.43834682573498085,
+      "grad_norm": 25.75,
+      "learning_rate": 2.2234912017313182e-06,
+      "loss": 1.1972,
+      "step": 5144
+    },
+    {
+      "epoch": 0.4384320409032808,
+      "grad_norm": 12.75,
+      "learning_rate": 2.223385787418031e-06,
+      "loss": 0.7455,
+      "step": 5145
+    },
+    {
+      "epoch": 0.43851725607158076,
+      "grad_norm": 13.5,
+      "learning_rate": 2.223280355514684e-06,
+      "loss": 0.7262,
+      "step": 5146
+    },
+    {
+      "epoch": 0.4386024712398807,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.2231749060231816e-06,
+      "loss": 0.4949,
+      "step": 5147
+    },
+    {
+      "epoch": 0.4386876864081807,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2230694389454305e-06,
+      "loss": 0.9748,
+      "step": 5148
+    },
+    {
+      "epoch": 0.43877290157648063,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.222963954283336e-06,
+      "loss": 0.9741,
+      "step": 5149
+    },
+    {
+      "epoch": 0.4388581167447806,
+      "grad_norm": 11.5,
+      "learning_rate": 2.2228584520388043e-06,
+      "loss": 0.5132,
+      "step": 5150
+    },
+    {
+      "epoch": 0.43894333191308055,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.2227529322137425e-06,
+      "loss": 0.6055,
+      "step": 5151
+    },
+    {
+      "epoch": 0.4390285470813805,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.2226473948100564e-06,
+      "loss": 0.9458,
+      "step": 5152
+    },
+    {
+      "epoch": 0.43911376224968046,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.2225418398296545e-06,
+      "loss": 0.5338,
+      "step": 5153
+    },
+    {
+      "epoch": 0.4391989774179804,
+      "grad_norm": 19.375,
+      "learning_rate": 2.222436267274443e-06,
+      "loss": 0.3762,
+      "step": 5154
+    },
+    {
+      "epoch": 0.4392841925862804,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2223306771463305e-06,
+      "loss": 0.6674,
+      "step": 5155
+    },
+    {
+      "epoch": 0.43936940775458033,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.222225069447225e-06,
+      "loss": 0.6346,
+      "step": 5156
+    },
+    {
+      "epoch": 0.4394546229228803,
+      "grad_norm": 18.0,
+      "learning_rate": 2.222119444179035e-06,
+      "loss": 1.0271,
+      "step": 5157
+    },
+    {
+      "epoch": 0.43953983809118025,
+      "grad_norm": 23.75,
+      "learning_rate": 2.222013801343669e-06,
+      "loss": 1.0399,
+      "step": 5158
+    },
+    {
+      "epoch": 0.4396250532594802,
+      "grad_norm": 15.625,
+      "learning_rate": 2.221908140943036e-06,
+      "loss": 0.9165,
+      "step": 5159
+    },
+    {
+      "epoch": 0.43971026842778016,
+      "grad_norm": 11.0,
+      "learning_rate": 2.221802462979046e-06,
+      "loss": 0.401,
+      "step": 5160
+    },
+    {
+      "epoch": 0.4397954835960801,
+      "grad_norm": 20.25,
+      "learning_rate": 2.2216967674536074e-06,
+      "loss": 0.7724,
+      "step": 5161
+    },
+    {
+      "epoch": 0.4398806987643801,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.2215910543686317e-06,
+      "loss": 0.3656,
+      "step": 5162
+    },
+    {
+      "epoch": 0.43996591393268003,
+      "grad_norm": 11.375,
+      "learning_rate": 2.221485323726029e-06,
+      "loss": 0.3966,
+      "step": 5163
+    },
+    {
+      "epoch": 0.44005112910098,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.221379575527709e-06,
+      "loss": 0.3691,
+      "step": 5164
+    },
+    {
+      "epoch": 0.44013634426927994,
+      "grad_norm": 12.0,
+      "learning_rate": 2.221273809775583e-06,
+      "loss": 0.4409,
+      "step": 5165
+    },
+    {
+      "epoch": 0.4402215594375799,
+      "grad_norm": 15.375,
+      "learning_rate": 2.221168026471563e-06,
+      "loss": 1.0218,
+      "step": 5166
+    },
+    {
+      "epoch": 0.44030677460587986,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2210622256175606e-06,
+      "loss": 0.7113,
+      "step": 5167
+    },
+    {
+      "epoch": 0.4403919897741798,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2209564072154864e-06,
+      "loss": 0.7158,
+      "step": 5168
+    },
+    {
+      "epoch": 0.44047720494247977,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2208505712672543e-06,
+      "loss": 0.739,
+      "step": 5169
+    },
+    {
+      "epoch": 0.44056242011077973,
+      "grad_norm": 22.5,
+      "learning_rate": 2.220744717774776e-06,
+      "loss": 1.0637,
+      "step": 5170
+    },
+    {
+      "epoch": 0.4406476352790797,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2206388467399642e-06,
+      "loss": 0.7977,
+      "step": 5171
+    },
+    {
+      "epoch": 0.44073285044737964,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2205329581647323e-06,
+      "loss": 0.649,
+      "step": 5172
+    },
+    {
+      "epoch": 0.4408180656156796,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.220427052050994e-06,
+      "loss": 0.4986,
+      "step": 5173
+    },
+    {
+      "epoch": 0.44090328078397956,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2203211284006632e-06,
+      "loss": 0.9302,
+      "step": 5174
+    },
+    {
+      "epoch": 0.4409884959522795,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.2202151872156542e-06,
+      "loss": 0.4803,
+      "step": 5175
+    },
+    {
+      "epoch": 0.44107371112057947,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.22010922849788e-06,
+      "loss": 0.5799,
+      "step": 5176
+    },
+    {
+      "epoch": 0.4411589262888794,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.220003252249258e-06,
+      "loss": 0.5021,
+      "step": 5177
+    },
+    {
+      "epoch": 0.4412441414571794,
+      "grad_norm": 16.25,
+      "learning_rate": 2.219897258471701e-06,
+      "loss": 0.8595,
+      "step": 5178
+    },
+    {
+      "epoch": 0.44132935662547934,
+      "grad_norm": 17.25,
+      "learning_rate": 2.219791247167125e-06,
+      "loss": 0.7109,
+      "step": 5179
+    },
+    {
+      "epoch": 0.4414145717937793,
+      "grad_norm": 12.125,
+      "learning_rate": 2.2196852183374466e-06,
+      "loss": 0.6619,
+      "step": 5180
+    },
+    {
+      "epoch": 0.44149978696207925,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.219579171984581e-06,
+      "loss": 0.5405,
+      "step": 5181
+    },
+    {
+      "epoch": 0.4415850021303792,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.2194731081104446e-06,
+      "loss": 0.5325,
+      "step": 5182
+    },
+    {
+      "epoch": 0.44167021729867917,
+      "grad_norm": 12.25,
+      "learning_rate": 2.2193670267169544e-06,
+      "loss": 0.7233,
+      "step": 5183
+    },
+    {
+      "epoch": 0.4417554324669791,
+      "grad_norm": 18.75,
+      "learning_rate": 2.219260927806027e-06,
+      "loss": 0.7228,
+      "step": 5184
+    },
+    {
+      "epoch": 0.4418406476352791,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.2191548113795804e-06,
+      "loss": 0.4754,
+      "step": 5185
+    },
+    {
+      "epoch": 0.44192586280357904,
+      "grad_norm": 12.0,
+      "learning_rate": 2.2190486774395315e-06,
+      "loss": 0.677,
+      "step": 5186
+    },
+    {
+      "epoch": 0.442011077971879,
+      "grad_norm": 26.375,
+      "learning_rate": 2.2189425259877984e-06,
+      "loss": 0.9605,
+      "step": 5187
+    },
+    {
+      "epoch": 0.44209629314017895,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2188363570263e-06,
+      "loss": 0.7215,
+      "step": 5188
+    },
+    {
+      "epoch": 0.4421815083084789,
+      "grad_norm": 12.5,
+      "learning_rate": 2.218730170556954e-06,
+      "loss": 0.5369,
+      "step": 5189
+    },
+    {
+      "epoch": 0.44226672347677887,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.2186239665816793e-06,
+      "loss": 0.8766,
+      "step": 5190
+    },
+    {
+      "epoch": 0.4423519386450788,
+      "grad_norm": 24.0,
+      "learning_rate": 2.2185177451023965e-06,
+      "loss": 0.8766,
+      "step": 5191
+    },
+    {
+      "epoch": 0.4424371538133788,
+      "grad_norm": 8.75,
+      "learning_rate": 2.2184115061210235e-06,
+      "loss": 0.2994,
+      "step": 5192
+    },
+    {
+      "epoch": 0.44252236898167874,
+      "grad_norm": 12.875,
+      "learning_rate": 2.218305249639481e-06,
+      "loss": 0.5088,
+      "step": 5193
+    },
+    {
+      "epoch": 0.4426075841499787,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2181989756596887e-06,
+      "loss": 0.5516,
+      "step": 5194
+    },
+    {
+      "epoch": 0.44269279931827865,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2180926841835674e-06,
+      "loss": 0.6515,
+      "step": 5195
+    },
+    {
+      "epoch": 0.4427780144865786,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.2179863752130377e-06,
+      "loss": 0.6983,
+      "step": 5196
+    },
+    {
+      "epoch": 0.44286322965487857,
+      "grad_norm": 17.25,
+      "learning_rate": 2.217880048750021e-06,
+      "loss": 0.9351,
+      "step": 5197
+    },
+    {
+      "epoch": 0.4429484448231785,
+      "grad_norm": 15.625,
+      "learning_rate": 2.2177737047964385e-06,
+      "loss": 1.06,
+      "step": 5198
+    },
+    {
+      "epoch": 0.4430336599914785,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.217667343354212e-06,
+      "loss": 0.6594,
+      "step": 5199
+    },
+    {
+      "epoch": 0.44311887515977844,
+      "grad_norm": 14.875,
+      "learning_rate": 2.2175609644252635e-06,
+      "loss": 0.5199,
+      "step": 5200
+    },
+    {
+      "epoch": 0.4432040903280784,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.2174545680115157e-06,
+      "loss": 0.4343,
+      "step": 5201
+    },
+    {
+      "epoch": 0.44328930549637835,
+      "grad_norm": 11.25,
+      "learning_rate": 2.2173481541148906e-06,
+      "loss": 0.5872,
+      "step": 5202
+    },
+    {
+      "epoch": 0.4433745206646783,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.2172417227373118e-06,
+      "loss": 0.6242,
+      "step": 5203
+    },
+    {
+      "epoch": 0.44345973583297826,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.2171352738807024e-06,
+      "loss": 0.5877,
+      "step": 5204
+    },
+    {
+      "epoch": 0.4435449510012782,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.217028807546986e-06,
+      "loss": 0.8408,
+      "step": 5205
+    },
+    {
+      "epoch": 0.4436301661695782,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.2169223237380864e-06,
+      "loss": 0.829,
+      "step": 5206
+    },
+    {
+      "epoch": 0.44371538133787813,
+      "grad_norm": 11.5,
+      "learning_rate": 2.2168158224559283e-06,
+      "loss": 0.4527,
+      "step": 5207
+    },
+    {
+      "epoch": 0.4438005965061781,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.2167093037024362e-06,
+      "loss": 0.3851,
+      "step": 5208
+    },
+    {
+      "epoch": 0.44388581167447805,
+      "grad_norm": 22.5,
+      "learning_rate": 2.2166027674795346e-06,
+      "loss": 1.1393,
+      "step": 5209
+    },
+    {
+      "epoch": 0.443971026842778,
+      "grad_norm": 13.375,
+      "learning_rate": 2.2164962137891493e-06,
+      "loss": 0.5187,
+      "step": 5210
+    },
+    {
+      "epoch": 0.44405624201107796,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2163896426332054e-06,
+      "loss": 1.0089,
+      "step": 5211
+    },
+    {
+      "epoch": 0.4441414571793779,
+      "grad_norm": 19.75,
+      "learning_rate": 2.216283054013629e-06,
+      "loss": 0.9306,
+      "step": 5212
+    },
+    {
+      "epoch": 0.4442266723476779,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.216176447932346e-06,
+      "loss": 0.394,
+      "step": 5213
+    },
+    {
+      "epoch": 0.44431188751597783,
+      "grad_norm": 27.0,
+      "learning_rate": 2.216069824391283e-06,
+      "loss": 1.0792,
+      "step": 5214
+    },
+    {
+      "epoch": 0.4443971026842778,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2159631833923673e-06,
+      "loss": 0.7715,
+      "step": 5215
+    },
+    {
+      "epoch": 0.44448231785257775,
+      "grad_norm": 13.125,
+      "learning_rate": 2.215856524937525e-06,
+      "loss": 0.6128,
+      "step": 5216
+    },
+    {
+      "epoch": 0.4445675330208777,
+      "grad_norm": 12.125,
+      "learning_rate": 2.2157498490286845e-06,
+      "loss": 0.6043,
+      "step": 5217
+    },
+    {
+      "epoch": 0.44465274818917766,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.2156431556677724e-06,
+      "loss": 0.6091,
+      "step": 5218
+    },
+    {
+      "epoch": 0.4447379633574776,
+      "grad_norm": 18.875,
+      "learning_rate": 2.2155364448567178e-06,
+      "loss": 0.6945,
+      "step": 5219
+    },
+    {
+      "epoch": 0.4448231785257776,
+      "grad_norm": 20.0,
+      "learning_rate": 2.2154297165974486e-06,
+      "loss": 0.7325,
+      "step": 5220
+    },
+    {
+      "epoch": 0.44490839369407753,
+      "grad_norm": 12.375,
+      "learning_rate": 2.215322970891894e-06,
+      "loss": 0.7625,
+      "step": 5221
+    },
+    {
+      "epoch": 0.4449936088623775,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2152162077419822e-06,
+      "loss": 1.0698,
+      "step": 5222
+    },
+    {
+      "epoch": 0.44507882403067744,
+      "grad_norm": 18.375,
+      "learning_rate": 2.215109427149643e-06,
+      "loss": 0.8082,
+      "step": 5223
+    },
+    {
+      "epoch": 0.4451640391989774,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.215002629116806e-06,
+      "loss": 0.5166,
+      "step": 5224
+    },
+    {
+      "epoch": 0.44524925436727736,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.214895813645401e-06,
+      "loss": 0.8749,
+      "step": 5225
+    },
+    {
+      "epoch": 0.4453344695355773,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.214788980737359e-06,
+      "loss": 0.966,
+      "step": 5226
+    },
+    {
+      "epoch": 0.44541968470387727,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.2146821303946093e-06,
+      "loss": 0.7744,
+      "step": 5227
+    },
+    {
+      "epoch": 0.44550489987217723,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2145752626190835e-06,
+      "loss": 0.807,
+      "step": 5228
+    },
+    {
+      "epoch": 0.4455901150404772,
+      "grad_norm": 15.25,
+      "learning_rate": 2.214468377412713e-06,
+      "loss": 0.8426,
+      "step": 5229
+    },
+    {
+      "epoch": 0.44567533020877714,
+      "grad_norm": 19.125,
+      "learning_rate": 2.214361474777429e-06,
+      "loss": 0.8497,
+      "step": 5230
+    },
+    {
+      "epoch": 0.4457605453770771,
+      "grad_norm": 33.0,
+      "learning_rate": 2.2142545547151633e-06,
+      "loss": 1.0836,
+      "step": 5231
+    },
+    {
+      "epoch": 0.44584576054537706,
+      "grad_norm": 19.625,
+      "learning_rate": 2.214147617227848e-06,
+      "loss": 0.6466,
+      "step": 5232
+    },
+    {
+      "epoch": 0.445930975713677,
+      "grad_norm": 9.625,
+      "learning_rate": 2.214040662317416e-06,
+      "loss": 0.5071,
+      "step": 5233
+    },
+    {
+      "epoch": 0.44601619088197697,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.2139336899858e-06,
+      "loss": 0.7577,
+      "step": 5234
+    },
+    {
+      "epoch": 0.4461014060502769,
+      "grad_norm": 15.75,
+      "learning_rate": 2.2138267002349324e-06,
+      "loss": 0.992,
+      "step": 5235
+    },
+    {
+      "epoch": 0.4461866212185769,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.2137196930667477e-06,
+      "loss": 0.5531,
+      "step": 5236
+    },
+    {
+      "epoch": 0.44627183638687684,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.213612668483178e-06,
+      "loss": 0.4828,
+      "step": 5237
+    },
+    {
+      "epoch": 0.4463570515551768,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2135056264861596e-06,
+      "loss": 0.72,
+      "step": 5238
+    },
+    {
+      "epoch": 0.44644226672347675,
+      "grad_norm": 17.125,
+      "learning_rate": 2.213398567077625e-06,
+      "loss": 0.6722,
+      "step": 5239
+    },
+    {
+      "epoch": 0.4465274818917767,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.21329149025951e-06,
+      "loss": 1.135,
+      "step": 5240
+    },
+    {
+      "epoch": 0.44661269706007667,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2131843960337483e-06,
+      "loss": 0.6865,
+      "step": 5241
+    },
+    {
+      "epoch": 0.4466979122283766,
+      "grad_norm": 11.125,
+      "learning_rate": 2.213077284402277e-06,
+      "loss": 0.511,
+      "step": 5242
+    },
+    {
+      "epoch": 0.4467831273966766,
+      "grad_norm": 25.0,
+      "learning_rate": 2.21297015536703e-06,
+      "loss": 0.7436,
+      "step": 5243
+    },
+    {
+      "epoch": 0.44686834256497654,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.212863008929944e-06,
+      "loss": 0.5247,
+      "step": 5244
+    },
+    {
+      "epoch": 0.4469535577332765,
+      "grad_norm": 24.0,
+      "learning_rate": 2.2127558450929556e-06,
+      "loss": 1.2036,
+      "step": 5245
+    },
+    {
+      "epoch": 0.44703877290157645,
+      "grad_norm": 13.875,
+      "learning_rate": 2.212648663858001e-06,
+      "loss": 0.7338,
+      "step": 5246
+    },
+    {
+      "epoch": 0.44712398806987647,
+      "grad_norm": 20.375,
+      "learning_rate": 2.2125414652270162e-06,
+      "loss": 0.6491,
+      "step": 5247
+    },
+    {
+      "epoch": 0.4472092032381764,
+      "grad_norm": 16.25,
+      "learning_rate": 2.21243424920194e-06,
+      "loss": 0.9395,
+      "step": 5248
+    },
+    {
+      "epoch": 0.4472944184064764,
+      "grad_norm": 18.75,
+      "learning_rate": 2.212327015784709e-06,
+      "loss": 0.6624,
+      "step": 5249
+    },
+    {
+      "epoch": 0.44737963357477634,
+      "grad_norm": 19.5,
+      "learning_rate": 2.2122197649772608e-06,
+      "loss": 0.9869,
+      "step": 5250
+    },
+    {
+      "epoch": 0.4474648487430763,
+      "grad_norm": 19.0,
+      "learning_rate": 2.212112496781534e-06,
+      "loss": 0.717,
+      "step": 5251
+    },
+    {
+      "epoch": 0.44755006391137625,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.212005211199467e-06,
+      "loss": 0.7496,
+      "step": 5252
+    },
+    {
+      "epoch": 0.4476352790796762,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2118979082329984e-06,
+      "loss": 0.7765,
+      "step": 5253
+    },
+    {
+      "epoch": 0.44772049424797616,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2117905878840678e-06,
+      "loss": 0.9994,
+      "step": 5254
+    },
+    {
+      "epoch": 0.4478057094162761,
+      "grad_norm": 18.0,
+      "learning_rate": 2.2116832501546135e-06,
+      "loss": 0.9808,
+      "step": 5255
+    },
+    {
+      "epoch": 0.4478909245845761,
+      "grad_norm": 20.875,
+      "learning_rate": 2.211575895046576e-06,
+      "loss": 0.8259,
+      "step": 5256
+    },
+    {
+      "epoch": 0.44797613975287603,
+      "grad_norm": 22.0,
+      "learning_rate": 2.2114685225618953e-06,
+      "loss": 1.1549,
+      "step": 5257
+    },
+    {
+      "epoch": 0.448061354921176,
+      "grad_norm": 12.875,
+      "learning_rate": 2.211361132702511e-06,
+      "loss": 0.6495,
+      "step": 5258
+    },
+    {
+      "epoch": 0.44814657008947595,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2112537254703647e-06,
+      "loss": 0.8431,
+      "step": 5259
+    },
+    {
+      "epoch": 0.4482317852577759,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2111463008673974e-06,
+      "loss": 0.6806,
+      "step": 5260
+    },
+    {
+      "epoch": 0.44831700042607586,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.2110388588955495e-06,
+      "loss": 0.5808,
+      "step": 5261
+    },
+    {
+      "epoch": 0.4484022155943758,
+      "grad_norm": 10.75,
+      "learning_rate": 2.2109313995567628e-06,
+      "loss": 0.4957,
+      "step": 5262
+    },
+    {
+      "epoch": 0.4484874307626758,
+      "grad_norm": 11.25,
+      "learning_rate": 2.2108239228529798e-06,
+      "loss": 0.3631,
+      "step": 5263
+    },
+    {
+      "epoch": 0.44857264593097573,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.210716428786142e-06,
+      "loss": 0.7375,
+      "step": 5264
+    },
+    {
+      "epoch": 0.4486578610992757,
+      "grad_norm": 12.875,
+      "learning_rate": 2.2106089173581926e-06,
+      "loss": 0.4065,
+      "step": 5265
+    },
+    {
+      "epoch": 0.44874307626757565,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.210501388571074e-06,
+      "loss": 0.7736,
+      "step": 5266
+    },
+    {
+      "epoch": 0.4488282914358756,
+      "grad_norm": 11.625,
+      "learning_rate": 2.2103938424267292e-06,
+      "loss": 0.5118,
+      "step": 5267
+    },
+    {
+      "epoch": 0.44891350660417556,
+      "grad_norm": 25.125,
+      "learning_rate": 2.2102862789271024e-06,
+      "loss": 0.8588,
+      "step": 5268
+    },
+    {
+      "epoch": 0.4489987217724755,
+      "grad_norm": 37.0,
+      "learning_rate": 2.210178698074136e-06,
+      "loss": 1.5926,
+      "step": 5269
+    },
+    {
+      "epoch": 0.4490839369407755,
+      "grad_norm": 19.125,
+      "learning_rate": 2.210071099869776e-06,
+      "loss": 1.0543,
+      "step": 5270
+    },
+    {
+      "epoch": 0.44916915210907543,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2099634843159654e-06,
+      "loss": 0.9676,
+      "step": 5271
+    },
+    {
+      "epoch": 0.4492543672773754,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2098558514146496e-06,
+      "loss": 0.5011,
+      "step": 5272
+    },
+    {
+      "epoch": 0.44933958244567535,
+      "grad_norm": 11.625,
+      "learning_rate": 2.2097482011677727e-06,
+      "loss": 0.6006,
+      "step": 5273
+    },
+    {
+      "epoch": 0.4494247976139753,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.2096405335772813e-06,
+      "loss": 0.8925,
+      "step": 5274
+    },
+    {
+      "epoch": 0.44951001278227526,
+      "grad_norm": 14.75,
+      "learning_rate": 2.20953284864512e-06,
+      "loss": 0.6267,
+      "step": 5275
+    },
+    {
+      "epoch": 0.4495952279505752,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2094251463732357e-06,
+      "loss": 0.8305,
+      "step": 5276
+    },
+    {
+      "epoch": 0.4496804431188752,
+      "grad_norm": 13.875,
+      "learning_rate": 2.209317426763574e-06,
+      "loss": 0.9639,
+      "step": 5277
+    },
+    {
+      "epoch": 0.44976565828717513,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2092096898180816e-06,
+      "loss": 0.7304,
+      "step": 5278
+    },
+    {
+      "epoch": 0.4498508734554751,
+      "grad_norm": 16.625,
+      "learning_rate": 2.209101935538706e-06,
+      "loss": 0.9641,
+      "step": 5279
+    },
+    {
+      "epoch": 0.44993608862377504,
+      "grad_norm": 33.75,
+      "learning_rate": 2.2089941639273936e-06,
+      "loss": 1.5391,
+      "step": 5280
+    },
+    {
+      "epoch": 0.450021303792075,
+      "grad_norm": 17.875,
+      "learning_rate": 2.208886374986092e-06,
+      "loss": 0.9008,
+      "step": 5281
+    },
+    {
+      "epoch": 0.45010651896037496,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.20877856871675e-06,
+      "loss": 0.8533,
+      "step": 5282
+    },
+    {
+      "epoch": 0.4501917341286749,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.208670745121315e-06,
+      "loss": 0.7125,
+      "step": 5283
+    },
+    {
+      "epoch": 0.45027694929697487,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2085629042017353e-06,
+      "loss": 0.8686,
+      "step": 5284
+    },
+    {
+      "epoch": 0.45036216446527483,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2084550459599606e-06,
+      "loss": 0.8119,
+      "step": 5285
+    },
+    {
+      "epoch": 0.4504473796335748,
+      "grad_norm": 21.875,
+      "learning_rate": 2.208347170397939e-06,
+      "loss": 1.0876,
+      "step": 5286
+    },
+    {
+      "epoch": 0.45053259480187474,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2082392775176205e-06,
+      "loss": 1.1659,
+      "step": 5287
+    },
+    {
+      "epoch": 0.4506178099701747,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2081313673209544e-06,
+      "loss": 0.6577,
+      "step": 5288
+    },
+    {
+      "epoch": 0.45070302513847466,
+      "grad_norm": 11.0,
+      "learning_rate": 2.208023439809891e-06,
+      "loss": 0.3374,
+      "step": 5289
+    },
+    {
+      "epoch": 0.4507882403067746,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.2079154949863806e-06,
+      "loss": 0.8558,
+      "step": 5290
+    },
+    {
+      "epoch": 0.45087345547507457,
+      "grad_norm": 18.5,
+      "learning_rate": 2.2078075328523744e-06,
+      "loss": 0.9127,
+      "step": 5291
+    },
+    {
+      "epoch": 0.4509586706433745,
+      "grad_norm": 25.75,
+      "learning_rate": 2.207699553409823e-06,
+      "loss": 1.2323,
+      "step": 5292
+    },
+    {
+      "epoch": 0.4510438858116745,
+      "grad_norm": 10.125,
+      "learning_rate": 2.207591556660677e-06,
+      "loss": 0.4612,
+      "step": 5293
+    },
+    {
+      "epoch": 0.45112910097997444,
+      "grad_norm": 15.125,
+      "learning_rate": 2.207483542606889e-06,
+      "loss": 0.605,
+      "step": 5294
+    },
+    {
+      "epoch": 0.4512143161482744,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.2073755112504105e-06,
+      "loss": 0.9879,
+      "step": 5295
+    },
+    {
+      "epoch": 0.45129953131657435,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.207267462593194e-06,
+      "loss": 0.5696,
+      "step": 5296
+    },
+    {
+      "epoch": 0.4513847464848743,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.207159396637192e-06,
+      "loss": 0.7613,
+      "step": 5297
+    },
+    {
+      "epoch": 0.45146996165317427,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.2070513133843568e-06,
+      "loss": 0.7534,
+      "step": 5298
+    },
+    {
+      "epoch": 0.4515551768214742,
+      "grad_norm": 21.75,
+      "learning_rate": 2.206943212836642e-06,
+      "loss": 1.1458,
+      "step": 5299
+    },
+    {
+      "epoch": 0.4516403919897742,
+      "grad_norm": 13.375,
+      "learning_rate": 2.206835094996001e-06,
+      "loss": 0.5133,
+      "step": 5300
+    },
+    {
+      "epoch": 0.45172560715807414,
+      "grad_norm": 15.625,
+      "learning_rate": 2.2067269598643874e-06,
+      "loss": 0.8986,
+      "step": 5301
+    },
+    {
+      "epoch": 0.4518108223263741,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2066188074437562e-06,
+      "loss": 0.6559,
+      "step": 5302
+    },
+    {
+      "epoch": 0.45189603749467405,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2065106377360614e-06,
+      "loss": 0.6126,
+      "step": 5303
+    },
+    {
+      "epoch": 0.451981252662974,
+      "grad_norm": 16.125,
+      "learning_rate": 2.206402450743257e-06,
+      "loss": 0.7747,
+      "step": 5304
+    },
+    {
+      "epoch": 0.45206646783127397,
+      "grad_norm": 16.25,
+      "learning_rate": 2.2062942464672984e-06,
+      "loss": 0.9428,
+      "step": 5305
+    },
+    {
+      "epoch": 0.4521516829995739,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2061860249101414e-06,
+      "loss": 0.9978,
+      "step": 5306
+    },
+    {
+      "epoch": 0.4522368981678739,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2060777860737415e-06,
+      "loss": 0.5861,
+      "step": 5307
+    },
+    {
+      "epoch": 0.45232211333617384,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2059695299600543e-06,
+      "loss": 0.836,
+      "step": 5308
+    },
+    {
+      "epoch": 0.4524073285044738,
+      "grad_norm": 16.625,
+      "learning_rate": 2.205861256571037e-06,
+      "loss": 0.8986,
+      "step": 5309
+    },
+    {
+      "epoch": 0.45249254367277375,
+      "grad_norm": 19.125,
+      "learning_rate": 2.205752965908645e-06,
+      "loss": 0.7699,
+      "step": 5310
+    },
+    {
+      "epoch": 0.4525777588410737,
+      "grad_norm": 18.625,
+      "learning_rate": 2.205644657974836e-06,
+      "loss": 0.5925,
+      "step": 5311
+    },
+    {
+      "epoch": 0.45266297400937366,
+      "grad_norm": 9.75,
+      "learning_rate": 2.205536332771567e-06,
+      "loss": 0.3757,
+      "step": 5312
+    },
+    {
+      "epoch": 0.4527481891776736,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.205427990300796e-06,
+      "loss": 0.4661,
+      "step": 5313
+    },
+    {
+      "epoch": 0.4528334043459736,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.20531963056448e-06,
+      "loss": 0.5236,
+      "step": 5314
+    },
+    {
+      "epoch": 0.45291861951427353,
+      "grad_norm": 17.5,
+      "learning_rate": 2.2052112535645777e-06,
+      "loss": 1.0352,
+      "step": 5315
+    },
+    {
+      "epoch": 0.4530038346825735,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2051028593030474e-06,
+      "loss": 0.8625,
+      "step": 5316
+    },
+    {
+      "epoch": 0.45308904985087345,
+      "grad_norm": 15.375,
+      "learning_rate": 2.204994447781848e-06,
+      "loss": 0.6187,
+      "step": 5317
+    },
+    {
+      "epoch": 0.4531742650191734,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.2048860190029393e-06,
+      "loss": 0.6228,
+      "step": 5318
+    },
+    {
+      "epoch": 0.45325948018747336,
+      "grad_norm": 17.625,
+      "learning_rate": 2.204777572968279e-06,
+      "loss": 0.6301,
+      "step": 5319
+    },
+    {
+      "epoch": 0.4533446953557733,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2046691096798285e-06,
+      "loss": 1.1456,
+      "step": 5320
+    },
+    {
+      "epoch": 0.4534299105240733,
+      "grad_norm": 19.625,
+      "learning_rate": 2.2045606291395472e-06,
+      "loss": 0.897,
+      "step": 5321
+    },
+    {
+      "epoch": 0.45351512569237323,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.204452131349395e-06,
+      "loss": 0.5846,
+      "step": 5322
+    },
+    {
+      "epoch": 0.4536003408606732,
+      "grad_norm": 19.875,
+      "learning_rate": 2.2043436163113334e-06,
+      "loss": 0.7718,
+      "step": 5323
+    },
+    {
+      "epoch": 0.45368555602897315,
+      "grad_norm": 18.625,
+      "learning_rate": 2.204235084027323e-06,
+      "loss": 0.6552,
+      "step": 5324
+    },
+    {
+      "epoch": 0.4537707711972731,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.204126534499325e-06,
+      "loss": 0.3773,
+      "step": 5325
+    },
+    {
+      "epoch": 0.45385598636557306,
+      "grad_norm": 16.75,
+      "learning_rate": 2.204017967729301e-06,
+      "loss": 0.8311,
+      "step": 5326
+    },
+    {
+      "epoch": 0.453941201533873,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.203909383719213e-06,
+      "loss": 1.0692,
+      "step": 5327
+    },
+    {
+      "epoch": 0.454026416702173,
+      "grad_norm": 16.875,
+      "learning_rate": 2.2038007824710232e-06,
+      "loss": 0.342,
+      "step": 5328
+    },
+    {
+      "epoch": 0.45411163187047293,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.203692163986694e-06,
+      "loss": 0.3911,
+      "step": 5329
+    },
+    {
+      "epoch": 0.4541968470387729,
+      "grad_norm": 11.25,
+      "learning_rate": 2.2035835282681888e-06,
+      "loss": 0.4331,
+      "step": 5330
+    },
+    {
+      "epoch": 0.45428206220707285,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.2034748753174703e-06,
+      "loss": 0.2653,
+      "step": 5331
+    },
+    {
+      "epoch": 0.4543672773753728,
+      "grad_norm": 16.5,
+      "learning_rate": 2.203366205136502e-06,
+      "loss": 0.8055,
+      "step": 5332
+    },
+    {
+      "epoch": 0.45445249254367276,
+      "grad_norm": 17.875,
+      "learning_rate": 2.2032575177272473e-06,
+      "loss": 1.0661,
+      "step": 5333
+    },
+    {
+      "epoch": 0.4545377077119727,
+      "grad_norm": 14.625,
+      "learning_rate": 2.2031488130916706e-06,
+      "loss": 0.9893,
+      "step": 5334
+    },
+    {
+      "epoch": 0.4546229228802727,
+      "grad_norm": 13.0,
+      "learning_rate": 2.203040091231737e-06,
+      "loss": 0.6344,
+      "step": 5335
+    },
+    {
+      "epoch": 0.45470813804857263,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.20293135214941e-06,
+      "loss": 0.4345,
+      "step": 5336
+    },
+    {
+      "epoch": 0.4547933532168726,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.202822595846656e-06,
+      "loss": 0.5901,
+      "step": 5337
+    },
+    {
+      "epoch": 0.45487856838517254,
+      "grad_norm": 28.0,
+      "learning_rate": 2.2027138223254386e-06,
+      "loss": 1.2241,
+      "step": 5338
+    },
+    {
+      "epoch": 0.4549637835534725,
+      "grad_norm": 19.375,
+      "learning_rate": 2.2026050315877253e-06,
+      "loss": 0.8944,
+      "step": 5339
+    },
+    {
+      "epoch": 0.45504899872177246,
+      "grad_norm": 23.0,
+      "learning_rate": 2.20249622363548e-06,
+      "loss": 0.8549,
+      "step": 5340
+    },
+    {
+      "epoch": 0.4551342138900724,
+      "grad_norm": 21.75,
+      "learning_rate": 2.2023873984706713e-06,
+      "loss": 1.0132,
+      "step": 5341
+    },
+    {
+      "epoch": 0.45521942905837237,
+      "grad_norm": 16.375,
+      "learning_rate": 2.2022785560952643e-06,
+      "loss": 0.8309,
+      "step": 5342
+    },
+    {
+      "epoch": 0.45530464422667233,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.202169696511226e-06,
+      "loss": 0.9,
+      "step": 5343
+    },
+    {
+      "epoch": 0.4553898593949723,
+      "grad_norm": 17.375,
+      "learning_rate": 2.2020608197205234e-06,
+      "loss": 0.6411,
+      "step": 5344
+    },
+    {
+      "epoch": 0.45547507456327224,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.201951925725125e-06,
+      "loss": 0.6685,
+      "step": 5345
+    },
+    {
+      "epoch": 0.4555602897315722,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.201843014526998e-06,
+      "loss": 0.7465,
+      "step": 5346
+    },
+    {
+      "epoch": 0.45564550489987216,
+      "grad_norm": 19.5,
+      "learning_rate": 2.2017340861281106e-06,
+      "loss": 0.7149,
+      "step": 5347
+    },
+    {
+      "epoch": 0.4557307200681721,
+      "grad_norm": 12.375,
+      "learning_rate": 2.201625140530431e-06,
+      "loss": 0.5987,
+      "step": 5348
+    },
+    {
+      "epoch": 0.45581593523647207,
+      "grad_norm": 18.0,
+      "learning_rate": 2.201516177735928e-06,
+      "loss": 0.9223,
+      "step": 5349
+    },
+    {
+      "epoch": 0.455901150404772,
+      "grad_norm": 11.25,
+      "learning_rate": 2.201407197746571e-06,
+      "loss": 0.5194,
+      "step": 5350
+    },
+    {
+      "epoch": 0.455986365573072,
+      "grad_norm": 20.0,
+      "learning_rate": 2.2012982005643296e-06,
+      "loss": 1.2225,
+      "step": 5351
+    },
+    {
+      "epoch": 0.45607158074137194,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2011891861911725e-06,
+      "loss": 0.7037,
+      "step": 5352
+    },
+    {
+      "epoch": 0.4561567959096719,
+      "grad_norm": 15.875,
+      "learning_rate": 2.2010801546290712e-06,
+      "loss": 0.6425,
+      "step": 5353
+    },
+    {
+      "epoch": 0.45624201107797185,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2009711058799943e-06,
+      "loss": 0.5349,
+      "step": 5354
+    },
+    {
+      "epoch": 0.4563272262462718,
+      "grad_norm": 24.0,
+      "learning_rate": 2.200862039945914e-06,
+      "loss": 1.1827,
+      "step": 5355
+    },
+    {
+      "epoch": 0.45641244141457177,
+      "grad_norm": 12.75,
+      "learning_rate": 2.2007529568287995e-06,
+      "loss": 0.7241,
+      "step": 5356
+    },
+    {
+      "epoch": 0.4564976565828717,
+      "grad_norm": 18.0,
+      "learning_rate": 2.200643856530624e-06,
+      "loss": 0.511,
+      "step": 5357
+    },
+    {
+      "epoch": 0.4565828717511717,
+      "grad_norm": 15.0,
+      "learning_rate": 2.200534739053357e-06,
+      "loss": 0.7864,
+      "step": 5358
+    },
+    {
+      "epoch": 0.45666808691947164,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2004256043989723e-06,
+      "loss": 0.8832,
+      "step": 5359
+    },
+    {
+      "epoch": 0.45675330208777165,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.200316452569441e-06,
+      "loss": 0.6291,
+      "step": 5360
+    },
+    {
+      "epoch": 0.4568385172560716,
+      "grad_norm": 17.125,
+      "learning_rate": 2.2002072835667353e-06,
+      "loss": 0.8866,
+      "step": 5361
+    },
+    {
+      "epoch": 0.45692373242437156,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.2000980973928294e-06,
+      "loss": 0.7363,
+      "step": 5362
+    },
+    {
+      "epoch": 0.4570089475926715,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.199988894049695e-06,
+      "loss": 0.8368,
+      "step": 5363
+    },
+    {
+      "epoch": 0.4570941627609715,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1998796735393056e-06,
+      "loss": 0.929,
+      "step": 5364
+    },
+    {
+      "epoch": 0.45717937792927144,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.1997704358636357e-06,
+      "loss": 0.5921,
+      "step": 5365
+    },
+    {
+      "epoch": 0.4572645930975714,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1996611810246592e-06,
+      "loss": 0.7622,
+      "step": 5366
+    },
+    {
+      "epoch": 0.45734980826587135,
+      "grad_norm": 15.0,
+      "learning_rate": 2.1995519090243497e-06,
+      "loss": 0.4529,
+      "step": 5367
+    },
+    {
+      "epoch": 0.4574350234341713,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.1994426198646828e-06,
+      "loss": 0.7882,
+      "step": 5368
+    },
+    {
+      "epoch": 0.45752023860247126,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.1993333135476325e-06,
+      "loss": 0.4896,
+      "step": 5369
+    },
+    {
+      "epoch": 0.4576054537707712,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.199223990075175e-06,
+      "loss": 0.469,
+      "step": 5370
+    },
+    {
+      "epoch": 0.4576906689390712,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1991146494492853e-06,
+      "loss": 0.7332,
+      "step": 5371
+    },
+    {
+      "epoch": 0.45777588410737113,
+      "grad_norm": 12.75,
+      "learning_rate": 2.1990052916719397e-06,
+      "loss": 0.6996,
+      "step": 5372
+    },
+    {
+      "epoch": 0.4578610992756711,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.198895916745114e-06,
+      "loss": 0.7664,
+      "step": 5373
+    },
+    {
+      "epoch": 0.45794631444397105,
+      "grad_norm": 32.5,
+      "learning_rate": 2.198786524670785e-06,
+      "loss": 0.9411,
+      "step": 5374
+    },
+    {
+      "epoch": 0.458031529612271,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.1986771154509293e-06,
+      "loss": 0.9394,
+      "step": 5375
+    },
+    {
+      "epoch": 0.45811674478057096,
+      "grad_norm": 13.625,
+      "learning_rate": 2.1985676890875243e-06,
+      "loss": 0.6836,
+      "step": 5376
+    },
+    {
+      "epoch": 0.4582019599488709,
+      "grad_norm": 15.625,
+      "learning_rate": 2.198458245582547e-06,
+      "loss": 0.5596,
+      "step": 5377
+    },
+    {
+      "epoch": 0.4582871751171709,
+      "grad_norm": 12.125,
+      "learning_rate": 2.198348784937976e-06,
+      "loss": 0.6925,
+      "step": 5378
+    },
+    {
+      "epoch": 0.45837239028547083,
+      "grad_norm": 13.375,
+      "learning_rate": 2.1982393071557884e-06,
+      "loss": 0.7237,
+      "step": 5379
+    },
+    {
+      "epoch": 0.4584576054537708,
+      "grad_norm": 15.125,
+      "learning_rate": 2.198129812237963e-06,
+      "loss": 0.9014,
+      "step": 5380
+    },
+    {
+      "epoch": 0.45854282062207075,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1980203001864784e-06,
+      "loss": 0.7805,
+      "step": 5381
+    },
+    {
+      "epoch": 0.4586280357903707,
+      "grad_norm": 18.875,
+      "learning_rate": 2.1979107710033135e-06,
+      "loss": 0.6593,
+      "step": 5382
+    },
+    {
+      "epoch": 0.45871325095867066,
+      "grad_norm": 17.625,
+      "learning_rate": 2.1978012246904485e-06,
+      "loss": 0.7923,
+      "step": 5383
+    },
+    {
+      "epoch": 0.4587984661269706,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.197691661249862e-06,
+      "loss": 0.5387,
+      "step": 5384
+    },
+    {
+      "epoch": 0.4588836812952706,
+      "grad_norm": 13.125,
+      "learning_rate": 2.197582080683534e-06,
+      "loss": 0.6742,
+      "step": 5385
+    },
+    {
+      "epoch": 0.45896889646357053,
+      "grad_norm": 13.125,
+      "learning_rate": 2.197472482993445e-06,
+      "loss": 0.6558,
+      "step": 5386
+    },
+    {
+      "epoch": 0.4590541116318705,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.197362868181576e-06,
+      "loss": 0.735,
+      "step": 5387
+    },
+    {
+      "epoch": 0.45913932680017044,
+      "grad_norm": 13.0,
+      "learning_rate": 2.1972532362499064e-06,
+      "loss": 0.712,
+      "step": 5388
+    },
+    {
+      "epoch": 0.4592245419684704,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.197143587200419e-06,
+      "loss": 0.5596,
+      "step": 5389
+    },
+    {
+      "epoch": 0.45930975713677036,
+      "grad_norm": 18.875,
+      "learning_rate": 2.197033921035094e-06,
+      "loss": 1.0656,
+      "step": 5390
+    },
+    {
+      "epoch": 0.4593949723050703,
+      "grad_norm": 9.8125,
+      "learning_rate": 2.196924237755914e-06,
+      "loss": 0.9935,
+      "step": 5391
+    },
+    {
+      "epoch": 0.45948018747337027,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.1968145373648607e-06,
+      "loss": 0.6395,
+      "step": 5392
+    },
+    {
+      "epoch": 0.45956540264167023,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1967048198639167e-06,
+      "loss": 0.8226,
+      "step": 5393
+    },
+    {
+      "epoch": 0.4596506178099702,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.196595085255064e-06,
+      "loss": 0.4611,
+      "step": 5394
+    },
+    {
+      "epoch": 0.45973583297827014,
+      "grad_norm": 14.625,
+      "learning_rate": 2.1964853335402873e-06,
+      "loss": 0.6716,
+      "step": 5395
+    },
+    {
+      "epoch": 0.4598210481465701,
+      "grad_norm": 11.875,
+      "learning_rate": 2.1963755647215684e-06,
+      "loss": 0.3015,
+      "step": 5396
+    },
+    {
+      "epoch": 0.45990626331487006,
+      "grad_norm": 9.625,
+      "learning_rate": 2.1962657788008915e-06,
+      "loss": 0.3835,
+      "step": 5397
+    },
+    {
+      "epoch": 0.45999147848317,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.1961559757802405e-06,
+      "loss": 0.4582,
+      "step": 5398
+    },
+    {
+      "epoch": 0.46007669365146997,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.1960461556615993e-06,
+      "loss": 0.8695,
+      "step": 5399
+    },
+    {
+      "epoch": 0.4601619088197699,
+      "grad_norm": 10.75,
+      "learning_rate": 2.195936318446953e-06,
+      "loss": 0.5271,
+      "step": 5400
+    },
+    {
+      "epoch": 0.4602471239880699,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.195826464138286e-06,
+      "loss": 0.6792,
+      "step": 5401
+    },
+    {
+      "epoch": 0.46033233915636984,
+      "grad_norm": 10.375,
+      "learning_rate": 2.1957165927375836e-06,
+      "loss": 0.458,
+      "step": 5402
+    },
+    {
+      "epoch": 0.4604175543246698,
+      "grad_norm": 13.0,
+      "learning_rate": 2.1956067042468317e-06,
+      "loss": 0.6812,
+      "step": 5403
+    },
+    {
+      "epoch": 0.46050276949296975,
+      "grad_norm": 17.25,
+      "learning_rate": 2.1954967986680155e-06,
+      "loss": 0.7756,
+      "step": 5404
+    },
+    {
+      "epoch": 0.4605879846612697,
+      "grad_norm": 13.875,
+      "learning_rate": 2.1953868760031218e-06,
+      "loss": 0.8689,
+      "step": 5405
+    },
+    {
+      "epoch": 0.46067319982956967,
+      "grad_norm": 16.375,
+      "learning_rate": 2.1952769362541363e-06,
+      "loss": 0.6738,
+      "step": 5406
+    },
+    {
+      "epoch": 0.4607584149978696,
+      "grad_norm": 10.375,
+      "learning_rate": 2.195166979423046e-06,
+      "loss": 0.442,
+      "step": 5407
+    },
+    {
+      "epoch": 0.4608436301661696,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1950570055118376e-06,
+      "loss": 0.7507,
+      "step": 5408
+    },
+    {
+      "epoch": 0.46092884533446954,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1949470145224995e-06,
+      "loss": 0.6132,
+      "step": 5409
+    },
+    {
+      "epoch": 0.4610140605027695,
+      "grad_norm": 11.75,
+      "learning_rate": 2.194837006457018e-06,
+      "loss": 0.8065,
+      "step": 5410
+    },
+    {
+      "epoch": 0.46109927567106945,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1947269813173823e-06,
+      "loss": 0.6637,
+      "step": 5411
+    },
+    {
+      "epoch": 0.4611844908393694,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.1946169391055795e-06,
+      "loss": 0.4672,
+      "step": 5412
+    },
+    {
+      "epoch": 0.46126970600766937,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.194506879823599e-06,
+      "loss": 0.4659,
+      "step": 5413
+    },
+    {
+      "epoch": 0.4613549211759693,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1943968034734292e-06,
+      "loss": 0.7119,
+      "step": 5414
+    },
+    {
+      "epoch": 0.4614401363442693,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.1942867100570597e-06,
+      "loss": 0.8455,
+      "step": 5415
+    },
+    {
+      "epoch": 0.46152535151256924,
+      "grad_norm": 13.125,
+      "learning_rate": 2.1941765995764796e-06,
+      "loss": 0.8413,
+      "step": 5416
+    },
+    {
+      "epoch": 0.4616105666808692,
+      "grad_norm": 13.25,
+      "learning_rate": 2.194066472033679e-06,
+      "loss": 0.7852,
+      "step": 5417
+    },
+    {
+      "epoch": 0.46169578184916915,
+      "grad_norm": 9.875,
+      "learning_rate": 2.193956327430648e-06,
+      "loss": 0.3056,
+      "step": 5418
+    },
+    {
+      "epoch": 0.4617809970174691,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1938461657693767e-06,
+      "loss": 0.7517,
+      "step": 5419
+    },
+    {
+      "epoch": 0.46186621218576907,
+      "grad_norm": 33.0,
+      "learning_rate": 2.1937359870518567e-06,
+      "loss": 0.5952,
+      "step": 5420
+    },
+    {
+      "epoch": 0.461951427354069,
+      "grad_norm": 15.5,
+      "learning_rate": 2.193625791280078e-06,
+      "loss": 0.5737,
+      "step": 5421
+    },
+    {
+      "epoch": 0.462036642522369,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.193515578456032e-06,
+      "loss": 0.8829,
+      "step": 5422
+    },
+    {
+      "epoch": 0.46212185769066894,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1934053485817113e-06,
+      "loss": 0.715,
+      "step": 5423
+    },
+    {
+      "epoch": 0.4622070728589689,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.193295101659107e-06,
+      "loss": 0.985,
+      "step": 5424
+    },
+    {
+      "epoch": 0.46229228802726885,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.1931848376902115e-06,
+      "loss": 0.4106,
+      "step": 5425
+    },
+    {
+      "epoch": 0.4623775031955688,
+      "grad_norm": 19.0,
+      "learning_rate": 2.193074556677018e-06,
+      "loss": 0.8552,
+      "step": 5426
+    },
+    {
+      "epoch": 0.46246271836386876,
+      "grad_norm": 14.375,
+      "learning_rate": 2.1929642586215186e-06,
+      "loss": 0.7395,
+      "step": 5427
+    },
+    {
+      "epoch": 0.4625479335321687,
+      "grad_norm": 14.625,
+      "learning_rate": 2.192853943525707e-06,
+      "loss": 0.6328,
+      "step": 5428
+    },
+    {
+      "epoch": 0.4626331487004687,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.1927436113915762e-06,
+      "loss": 0.3064,
+      "step": 5429
+    },
+    {
+      "epoch": 0.46271836386876863,
+      "grad_norm": 23.375,
+      "learning_rate": 2.1926332622211206e-06,
+      "loss": 0.9315,
+      "step": 5430
+    },
+    {
+      "epoch": 0.4628035790370686,
+      "grad_norm": 13.375,
+      "learning_rate": 2.192522896016334e-06,
+      "loss": 0.6818,
+      "step": 5431
+    },
+    {
+      "epoch": 0.46288879420536855,
+      "grad_norm": 14.875,
+      "learning_rate": 2.192412512779211e-06,
+      "loss": 0.6938,
+      "step": 5432
+    },
+    {
+      "epoch": 0.4629740093736685,
+      "grad_norm": 13.875,
+      "learning_rate": 2.192302112511746e-06,
+      "loss": 0.6912,
+      "step": 5433
+    },
+    {
+      "epoch": 0.46305922454196846,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1921916952159345e-06,
+      "loss": 0.651,
+      "step": 5434
+    },
+    {
+      "epoch": 0.4631444397102684,
+      "grad_norm": 12.875,
+      "learning_rate": 2.1920812608937713e-06,
+      "loss": 0.59,
+      "step": 5435
+    },
+    {
+      "epoch": 0.4632296548785684,
+      "grad_norm": 11.875,
+      "learning_rate": 2.1919708095472526e-06,
+      "loss": 0.6347,
+      "step": 5436
+    },
+    {
+      "epoch": 0.46331487004686833,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1918603411783743e-06,
+      "loss": 0.5928,
+      "step": 5437
+    },
+    {
+      "epoch": 0.4634000852151683,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.191749855789132e-06,
+      "loss": 0.7762,
+      "step": 5438
+    },
+    {
+      "epoch": 0.46348530038346825,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1916393533815234e-06,
+      "loss": 0.8205,
+      "step": 5439
+    },
+    {
+      "epoch": 0.4635705155517682,
+      "grad_norm": 12.25,
+      "learning_rate": 2.1915288339575445e-06,
+      "loss": 0.2623,
+      "step": 5440
+    },
+    {
+      "epoch": 0.46365573072006816,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1914182975191927e-06,
+      "loss": 0.3928,
+      "step": 5441
+    },
+    {
+      "epoch": 0.4637409458883681,
+      "grad_norm": 12.5,
+      "learning_rate": 2.191307744068466e-06,
+      "loss": 0.6305,
+      "step": 5442
+    },
+    {
+      "epoch": 0.4638261610566681,
+      "grad_norm": 15.625,
+      "learning_rate": 2.1911971736073617e-06,
+      "loss": 0.6561,
+      "step": 5443
+    },
+    {
+      "epoch": 0.46391137622496803,
+      "grad_norm": 19.125,
+      "learning_rate": 2.191086586137878e-06,
+      "loss": 0.9618,
+      "step": 5444
+    },
+    {
+      "epoch": 0.463996591393268,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.190975981662013e-06,
+      "loss": 0.3177,
+      "step": 5445
+    },
+    {
+      "epoch": 0.46408180656156794,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.190865360181766e-06,
+      "loss": 0.8093,
+      "step": 5446
+    },
+    {
+      "epoch": 0.4641670217298679,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.190754721699136e-06,
+      "loss": 0.7158,
+      "step": 5447
+    },
+    {
+      "epoch": 0.46425223689816786,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1906440662161215e-06,
+      "loss": 0.5968,
+      "step": 5448
+    },
+    {
+      "epoch": 0.4643374520664678,
+      "grad_norm": 19.75,
+      "learning_rate": 2.1905333937347235e-06,
+      "loss": 1.0111,
+      "step": 5449
+    },
+    {
+      "epoch": 0.46442266723476777,
+      "grad_norm": 14.875,
+      "learning_rate": 2.190422704256941e-06,
+      "loss": 0.7625,
+      "step": 5450
+    },
+    {
+      "epoch": 0.46450788240306773,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.190311997784775e-06,
+      "loss": 0.7636,
+      "step": 5451
+    },
+    {
+      "epoch": 0.4645930975713677,
+      "grad_norm": 16.375,
+      "learning_rate": 2.1902012743202254e-06,
+      "loss": 1.1334,
+      "step": 5452
+    },
+    {
+      "epoch": 0.46467831273966764,
+      "grad_norm": 13.125,
+      "learning_rate": 2.190090533865293e-06,
+      "loss": 0.7281,
+      "step": 5453
+    },
+    {
+      "epoch": 0.4647635279079676,
+      "grad_norm": 21.25,
+      "learning_rate": 2.18997977642198e-06,
+      "loss": 0.9653,
+      "step": 5454
+    },
+    {
+      "epoch": 0.46484874307626756,
+      "grad_norm": 11.75,
+      "learning_rate": 2.189869001992287e-06,
+      "loss": 0.4201,
+      "step": 5455
+    },
+    {
+      "epoch": 0.4649339582445675,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1897582105782156e-06,
+      "loss": 0.6026,
+      "step": 5456
+    },
+    {
+      "epoch": 0.46501917341286747,
+      "grad_norm": 11.875,
+      "learning_rate": 2.1896474021817686e-06,
+      "loss": 0.5422,
+      "step": 5457
+    },
+    {
+      "epoch": 0.4651043885811674,
+      "grad_norm": 15.5,
+      "learning_rate": 2.189536576804948e-06,
+      "loss": 0.4348,
+      "step": 5458
+    },
+    {
+      "epoch": 0.4651896037494674,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.189425734449757e-06,
+      "loss": 0.5475,
+      "step": 5459
+    },
+    {
+      "epoch": 0.46527481891776734,
+      "grad_norm": 19.5,
+      "learning_rate": 2.189314875118198e-06,
+      "loss": 0.68,
+      "step": 5460
+    },
+    {
+      "epoch": 0.4653600340860673,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.189203998812275e-06,
+      "loss": 0.7498,
+      "step": 5461
+    },
+    {
+      "epoch": 0.46544524925436725,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.189093105533991e-06,
+      "loss": 1.0508,
+      "step": 5462
+    },
+    {
+      "epoch": 0.4655304644226672,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.18898219528535e-06,
+      "loss": 0.6773,
+      "step": 5463
+    },
+    {
+      "epoch": 0.46561567959096717,
+      "grad_norm": 16.0,
+      "learning_rate": 2.188871268068357e-06,
+      "loss": 0.7554,
+      "step": 5464
+    },
+    {
+      "epoch": 0.4657008947592671,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.188760323885016e-06,
+      "loss": 0.6219,
+      "step": 5465
+    },
+    {
+      "epoch": 0.4657861099275671,
+      "grad_norm": 11.875,
+      "learning_rate": 2.188649362737332e-06,
+      "loss": 0.5974,
+      "step": 5466
+    },
+    {
+      "epoch": 0.46587132509586704,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.18853838462731e-06,
+      "loss": 0.5459,
+      "step": 5467
+    },
+    {
+      "epoch": 0.465956540264167,
+      "grad_norm": 28.375,
+      "learning_rate": 2.1884273895569553e-06,
+      "loss": 0.9978,
+      "step": 5468
+    },
+    {
+      "epoch": 0.46604175543246695,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.1883163775282744e-06,
+      "loss": 0.4482,
+      "step": 5469
+    },
+    {
+      "epoch": 0.4661269706007669,
+      "grad_norm": 16.875,
+      "learning_rate": 2.188205348543273e-06,
+      "loss": 0.6727,
+      "step": 5470
+    },
+    {
+      "epoch": 0.46621218576906687,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1880943026039575e-06,
+      "loss": 0.5859,
+      "step": 5471
+    },
+    {
+      "epoch": 0.4662974009373668,
+      "grad_norm": 12.875,
+      "learning_rate": 2.1879832397123347e-06,
+      "loss": 0.4882,
+      "step": 5472
+    },
+    {
+      "epoch": 0.46638261610566684,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1878721598704112e-06,
+      "loss": 0.9089,
+      "step": 5473
+    },
+    {
+      "epoch": 0.4664678312739668,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.187761063080195e-06,
+      "loss": 0.659,
+      "step": 5474
+    },
+    {
+      "epoch": 0.46655304644226675,
+      "grad_norm": 18.125,
+      "learning_rate": 2.187649949343693e-06,
+      "loss": 1.1222,
+      "step": 5475
+    },
+    {
+      "epoch": 0.4666382616105667,
+      "grad_norm": 24.0,
+      "learning_rate": 2.187538818662914e-06,
+      "loss": 1.0194,
+      "step": 5476
+    },
+    {
+      "epoch": 0.46672347677886666,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.1874276710398657e-06,
+      "loss": 0.8416,
+      "step": 5477
+    },
+    {
+      "epoch": 0.4668086919471666,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1873165064765566e-06,
+      "loss": 0.5463,
+      "step": 5478
+    },
+    {
+      "epoch": 0.4668939071154666,
+      "grad_norm": 21.875,
+      "learning_rate": 2.1872053249749957e-06,
+      "loss": 1.1472,
+      "step": 5479
+    },
+    {
+      "epoch": 0.46697912228376653,
+      "grad_norm": 13.125,
+      "learning_rate": 2.187094126537192e-06,
+      "loss": 0.6365,
+      "step": 5480
+    },
+    {
+      "epoch": 0.4670643374520665,
+      "grad_norm": 18.125,
+      "learning_rate": 2.1869829111651553e-06,
+      "loss": 1.2176,
+      "step": 5481
+    },
+    {
+      "epoch": 0.46714955262036645,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.1868716788608952e-06,
+      "loss": 0.5142,
+      "step": 5482
+    },
+    {
+      "epoch": 0.4672347677886664,
+      "grad_norm": 17.625,
+      "learning_rate": 2.186760429626422e-06,
+      "loss": 1.1302,
+      "step": 5483
+    },
+    {
+      "epoch": 0.46731998295696636,
+      "grad_norm": 12.375,
+      "learning_rate": 2.1866491634637456e-06,
+      "loss": 0.5261,
+      "step": 5484
+    },
+    {
+      "epoch": 0.4674051981252663,
+      "grad_norm": 15.0,
+      "learning_rate": 2.1865378803748764e-06,
+      "loss": 0.9297,
+      "step": 5485
+    },
+    {
+      "epoch": 0.4674904132935663,
+      "grad_norm": 13.25,
+      "learning_rate": 2.186426580361827e-06,
+      "loss": 0.7438,
+      "step": 5486
+    },
+    {
+      "epoch": 0.46757562846186623,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.1863152634266066e-06,
+      "loss": 0.8134,
+      "step": 5487
+    },
+    {
+      "epoch": 0.4676608436301662,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1862039295712285e-06,
+      "loss": 0.5644,
+      "step": 5488
+    },
+    {
+      "epoch": 0.46774605879846615,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.186092578797704e-06,
+      "loss": 0.5421,
+      "step": 5489
+    },
+    {
+      "epoch": 0.4678312739667661,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.1859812111080446e-06,
+      "loss": 0.45,
+      "step": 5490
+    },
+    {
+      "epoch": 0.46791648913506606,
+      "grad_norm": 15.75,
+      "learning_rate": 2.1858698265042643e-06,
+      "loss": 0.6468,
+      "step": 5491
+    },
+    {
+      "epoch": 0.468001704303366,
+      "grad_norm": 21.125,
+      "learning_rate": 2.1857584249883746e-06,
+      "loss": 0.8439,
+      "step": 5492
+    },
+    {
+      "epoch": 0.468086919471666,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1856470065623896e-06,
+      "loss": 0.8426,
+      "step": 5493
+    },
+    {
+      "epoch": 0.46817213463996593,
+      "grad_norm": 17.375,
+      "learning_rate": 2.185535571228322e-06,
+      "loss": 0.9854,
+      "step": 5494
+    },
+    {
+      "epoch": 0.4682573498082659,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1854241189881862e-06,
+      "loss": 0.6358,
+      "step": 5495
+    },
+    {
+      "epoch": 0.46834256497656584,
+      "grad_norm": 19.625,
+      "learning_rate": 2.185312649843996e-06,
+      "loss": 0.9284,
+      "step": 5496
+    },
+    {
+      "epoch": 0.4684277801448658,
+      "grad_norm": 14.5,
+      "learning_rate": 2.1852011637977656e-06,
+      "loss": 0.7636,
+      "step": 5497
+    },
+    {
+      "epoch": 0.46851299531316576,
+      "grad_norm": 27.375,
+      "learning_rate": 2.1850896608515093e-06,
+      "loss": 0.5462,
+      "step": 5498
+    },
+    {
+      "epoch": 0.4685982104814657,
+      "grad_norm": 10.875,
+      "learning_rate": 2.184978141007243e-06,
+      "loss": 0.4294,
+      "step": 5499
+    },
+    {
+      "epoch": 0.4686834256497657,
+      "grad_norm": 10.375,
+      "learning_rate": 2.1848666042669814e-06,
+      "loss": 0.4634,
+      "step": 5500
+    },
+    {
+      "epoch": 0.46876864081806563,
+      "grad_norm": 12.625,
+      "learning_rate": 2.18475505063274e-06,
+      "loss": 0.5324,
+      "step": 5501
+    },
+    {
+      "epoch": 0.4688538559863656,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.1846434801065347e-06,
+      "loss": 0.3975,
+      "step": 5502
+    },
+    {
+      "epoch": 0.46893907115466554,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.1845318926903823e-06,
+      "loss": 0.7668,
+      "step": 5503
+    },
+    {
+      "epoch": 0.4690242863229655,
+      "grad_norm": 14.125,
+      "learning_rate": 2.184420288386299e-06,
+      "loss": 0.7345,
+      "step": 5504
+    },
+    {
+      "epoch": 0.46910950149126546,
+      "grad_norm": 11.125,
+      "learning_rate": 2.184308667196301e-06,
+      "loss": 0.4143,
+      "step": 5505
+    },
+    {
+      "epoch": 0.4691947166595654,
+      "grad_norm": 13.0,
+      "learning_rate": 2.184197029122406e-06,
+      "loss": 0.7244,
+      "step": 5506
+    },
+    {
+      "epoch": 0.46927993182786537,
+      "grad_norm": 17.25,
+      "learning_rate": 2.1840853741666314e-06,
+      "loss": 1.0593,
+      "step": 5507
+    },
+    {
+      "epoch": 0.4693651469961653,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.183973702330994e-06,
+      "loss": 0.8069,
+      "step": 5508
+    },
+    {
+      "epoch": 0.4694503621644653,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.183862013617514e-06,
+      "loss": 0.7345,
+      "step": 5509
+    },
+    {
+      "epoch": 0.46953557733276524,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.1837503080282074e-06,
+      "loss": 0.9162,
+      "step": 5510
+    },
+    {
+      "epoch": 0.4696207925010652,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.1836385855650936e-06,
+      "loss": 0.5146,
+      "step": 5511
+    },
+    {
+      "epoch": 0.46970600766936516,
+      "grad_norm": 11.625,
+      "learning_rate": 2.1835268462301928e-06,
+      "loss": 0.556,
+      "step": 5512
+    },
+    {
+      "epoch": 0.4697912228376651,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.183415090025522e-06,
+      "loss": 0.5341,
+      "step": 5513
+    },
+    {
+      "epoch": 0.46987643800596507,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.1833033169531027e-06,
+      "loss": 0.6581,
+      "step": 5514
+    },
+    {
+      "epoch": 0.469961653174265,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1831915270149535e-06,
+      "loss": 0.6958,
+      "step": 5515
+    },
+    {
+      "epoch": 0.470046868342565,
+      "grad_norm": 11.75,
+      "learning_rate": 2.1830797202130953e-06,
+      "loss": 0.54,
+      "step": 5516
+    },
+    {
+      "epoch": 0.47013208351086494,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.182967896549548e-06,
+      "loss": 0.529,
+      "step": 5517
+    },
+    {
+      "epoch": 0.4702172986791649,
+      "grad_norm": 11.25,
+      "learning_rate": 2.1828560560263327e-06,
+      "loss": 0.3441,
+      "step": 5518
+    },
+    {
+      "epoch": 0.47030251384746485,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1827441986454708e-06,
+      "loss": 0.6511,
+      "step": 5519
+    },
+    {
+      "epoch": 0.4703877290157648,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1826323244089827e-06,
+      "loss": 0.8134,
+      "step": 5520
+    },
+    {
+      "epoch": 0.47047294418406477,
+      "grad_norm": 17.875,
+      "learning_rate": 2.182520433318891e-06,
+      "loss": 0.8867,
+      "step": 5521
+    },
+    {
+      "epoch": 0.4705581593523647,
+      "grad_norm": 13.75,
+      "learning_rate": 2.182408525377217e-06,
+      "loss": 0.8701,
+      "step": 5522
+    },
+    {
+      "epoch": 0.4706433745206647,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.1822966005859835e-06,
+      "loss": 0.4297,
+      "step": 5523
+    },
+    {
+      "epoch": 0.47072858968896464,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1821846589472133e-06,
+      "loss": 0.537,
+      "step": 5524
+    },
+    {
+      "epoch": 0.4708138048572646,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.1820727004629284e-06,
+      "loss": 0.7734,
+      "step": 5525
+    },
+    {
+      "epoch": 0.47089902002556455,
+      "grad_norm": 14.0,
+      "learning_rate": 2.181960725135153e-06,
+      "loss": 0.7162,
+      "step": 5526
+    },
+    {
+      "epoch": 0.4709842351938645,
+      "grad_norm": 11.5,
+      "learning_rate": 2.18184873296591e-06,
+      "loss": 0.5148,
+      "step": 5527
+    },
+    {
+      "epoch": 0.47106945036216447,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.181736723957223e-06,
+      "loss": 0.6375,
+      "step": 5528
+    },
+    {
+      "epoch": 0.4711546655304644,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1816246981111168e-06,
+      "loss": 0.7674,
+      "step": 5529
+    },
+    {
+      "epoch": 0.4712398806987644,
+      "grad_norm": 11.625,
+      "learning_rate": 2.1815126554296156e-06,
+      "loss": 0.5394,
+      "step": 5530
+    },
+    {
+      "epoch": 0.47132509586706434,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.181400595914744e-06,
+      "loss": 0.9407,
+      "step": 5531
+    },
+    {
+      "epoch": 0.4714103110353643,
+      "grad_norm": 16.625,
+      "learning_rate": 2.1812885195685265e-06,
+      "loss": 0.8713,
+      "step": 5532
+    },
+    {
+      "epoch": 0.47149552620366425,
+      "grad_norm": 19.0,
+      "learning_rate": 2.181176426392989e-06,
+      "loss": 0.7733,
+      "step": 5533
+    },
+    {
+      "epoch": 0.4715807413719642,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.181064316390158e-06,
+      "loss": 0.5568,
+      "step": 5534
+    },
+    {
+      "epoch": 0.47166595654026416,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1809521895620577e-06,
+      "loss": 0.853,
+      "step": 5535
+    },
+    {
+      "epoch": 0.4717511717085641,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1808400459107157e-06,
+      "loss": 0.5345,
+      "step": 5536
+    },
+    {
+      "epoch": 0.4718363868768641,
+      "grad_norm": 26.5,
+      "learning_rate": 2.180727885438158e-06,
+      "loss": 1.1742,
+      "step": 5537
+    },
+    {
+      "epoch": 0.47192160204516403,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.180615708146411e-06,
+      "loss": 0.2815,
+      "step": 5538
+    },
+    {
+      "epoch": 0.472006817213464,
+      "grad_norm": 17.125,
+      "learning_rate": 2.1805035140375028e-06,
+      "loss": 0.9562,
+      "step": 5539
+    },
+    {
+      "epoch": 0.47209203238176395,
+      "grad_norm": 18.75,
+      "learning_rate": 2.1803913031134603e-06,
+      "loss": 1.208,
+      "step": 5540
+    },
+    {
+      "epoch": 0.4721772475500639,
+      "grad_norm": 11.875,
+      "learning_rate": 2.180279075376311e-06,
+      "loss": 0.6441,
+      "step": 5541
+    },
+    {
+      "epoch": 0.47226246271836386,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1801668308280836e-06,
+      "loss": 0.7921,
+      "step": 5542
+    },
+    {
+      "epoch": 0.4723476778866638,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.1800545694708063e-06,
+      "loss": 0.2958,
+      "step": 5543
+    },
+    {
+      "epoch": 0.4724328930549638,
+      "grad_norm": 14.125,
+      "learning_rate": 2.179942291306508e-06,
+      "loss": 0.8886,
+      "step": 5544
+    },
+    {
+      "epoch": 0.47251810822326373,
+      "grad_norm": 16.125,
+      "learning_rate": 2.179829996337217e-06,
+      "loss": 0.7413,
+      "step": 5545
+    },
+    {
+      "epoch": 0.4726033233915637,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.179717684564963e-06,
+      "loss": 0.5756,
+      "step": 5546
+    },
+    {
+      "epoch": 0.47268853855986365,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.1796053559917755e-06,
+      "loss": 0.5664,
+      "step": 5547
+    },
+    {
+      "epoch": 0.4727737537281636,
+      "grad_norm": 15.75,
+      "learning_rate": 2.179493010619684e-06,
+      "loss": 0.6699,
+      "step": 5548
+    },
+    {
+      "epoch": 0.47285896889646356,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.1793806484507197e-06,
+      "loss": 0.4686,
+      "step": 5549
+    },
+    {
+      "epoch": 0.4729441840647635,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1792682694869127e-06,
+      "loss": 0.7772,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4730293992330635,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.1791558737302933e-06,
+      "loss": 0.7444,
+      "step": 5551
+    },
+    {
+      "epoch": 0.47311461440136343,
+      "grad_norm": 15.375,
+      "learning_rate": 2.1790434611828933e-06,
+      "loss": 0.9241,
+      "step": 5552
+    },
+    {
+      "epoch": 0.4731998295696634,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.178931031846743e-06,
+      "loss": 0.7668,
+      "step": 5553
+    },
+    {
+      "epoch": 0.47328504473796335,
+      "grad_norm": 14.625,
+      "learning_rate": 2.1788185857238754e-06,
+      "loss": 0.8363,
+      "step": 5554
+    },
+    {
+      "epoch": 0.4733702599062633,
+      "grad_norm": 12.375,
+      "learning_rate": 2.1787061228163216e-06,
+      "loss": 0.3811,
+      "step": 5555
+    },
+    {
+      "epoch": 0.47345547507456326,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1785936431261144e-06,
+      "loss": 0.8861,
+      "step": 5556
+    },
+    {
+      "epoch": 0.4735406902428632,
+      "grad_norm": 13.625,
+      "learning_rate": 2.1784811466552863e-06,
+      "loss": 0.6705,
+      "step": 5557
+    },
+    {
+      "epoch": 0.4736259054111632,
+      "grad_norm": 12.25,
+      "learning_rate": 2.17836863340587e-06,
+      "loss": 0.3512,
+      "step": 5558
+    },
+    {
+      "epoch": 0.47371112057946313,
+      "grad_norm": 12.375,
+      "learning_rate": 2.1782561033798994e-06,
+      "loss": 0.7501,
+      "step": 5559
+    },
+    {
+      "epoch": 0.4737963357477631,
+      "grad_norm": 18.5,
+      "learning_rate": 2.1781435565794074e-06,
+      "loss": 1.0052,
+      "step": 5560
+    },
+    {
+      "epoch": 0.47388155091606304,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.178030993006428e-06,
+      "loss": 0.7937,
+      "step": 5561
+    },
+    {
+      "epoch": 0.473966766084363,
+      "grad_norm": 12.25,
+      "learning_rate": 2.1779184126629954e-06,
+      "loss": 0.6188,
+      "step": 5562
+    },
+    {
+      "epoch": 0.47405198125266296,
+      "grad_norm": 12.625,
+      "learning_rate": 2.1778058155511437e-06,
+      "loss": 0.2805,
+      "step": 5563
+    },
+    {
+      "epoch": 0.4741371964209629,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.1776932016729085e-06,
+      "loss": 0.6485,
+      "step": 5564
+    },
+    {
+      "epoch": 0.47422241158926287,
+      "grad_norm": 17.625,
+      "learning_rate": 2.177580571030324e-06,
+      "loss": 0.9021,
+      "step": 5565
+    },
+    {
+      "epoch": 0.47430762675756283,
+      "grad_norm": 11.0,
+      "learning_rate": 2.1774679236254255e-06,
+      "loss": 0.4493,
+      "step": 5566
+    },
+    {
+      "epoch": 0.4743928419258628,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1773552594602495e-06,
+      "loss": 0.7791,
+      "step": 5567
+    },
+    {
+      "epoch": 0.47447805709416274,
+      "grad_norm": 11.625,
+      "learning_rate": 2.177242578536831e-06,
+      "loss": 0.6009,
+      "step": 5568
+    },
+    {
+      "epoch": 0.4745632722624627,
+      "grad_norm": 18.0,
+      "learning_rate": 2.1771298808572065e-06,
+      "loss": 0.8438,
+      "step": 5569
+    },
+    {
+      "epoch": 0.47464848743076266,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1770171664234134e-06,
+      "loss": 0.5735,
+      "step": 5570
+    },
+    {
+      "epoch": 0.4747337025990626,
+      "grad_norm": 14.625,
+      "learning_rate": 2.1769044352374877e-06,
+      "loss": 0.607,
+      "step": 5571
+    },
+    {
+      "epoch": 0.47481891776736257,
+      "grad_norm": 11.75,
+      "learning_rate": 2.1767916873014667e-06,
+      "loss": 0.5811,
+      "step": 5572
+    },
+    {
+      "epoch": 0.4749041329356625,
+      "grad_norm": 14.625,
+      "learning_rate": 2.1766789226173877e-06,
+      "loss": 0.7509,
+      "step": 5573
+    },
+    {
+      "epoch": 0.4749893481039625,
+      "grad_norm": 9.5,
+      "learning_rate": 2.176566141187289e-06,
+      "loss": 0.5692,
+      "step": 5574
+    },
+    {
+      "epoch": 0.47507456327226244,
+      "grad_norm": 13.625,
+      "learning_rate": 2.1764533430132087e-06,
+      "loss": 0.7993,
+      "step": 5575
+    },
+    {
+      "epoch": 0.4751597784405624,
+      "grad_norm": 21.125,
+      "learning_rate": 2.1763405280971846e-06,
+      "loss": 1.1098,
+      "step": 5576
+    },
+    {
+      "epoch": 0.47524499360886235,
+      "grad_norm": 14.5,
+      "learning_rate": 2.1762276964412556e-06,
+      "loss": 0.5298,
+      "step": 5577
+    },
+    {
+      "epoch": 0.4753302087771623,
+      "grad_norm": 15.5,
+      "learning_rate": 2.1761148480474604e-06,
+      "loss": 0.9666,
+      "step": 5578
+    },
+    {
+      "epoch": 0.47541542394546227,
+      "grad_norm": 12.5,
+      "learning_rate": 2.176001982917839e-06,
+      "loss": 0.4802,
+      "step": 5579
+    },
+    {
+      "epoch": 0.4755006391137622,
+      "grad_norm": 17.0,
+      "learning_rate": 2.1758891010544307e-06,
+      "loss": 1.1795,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4755858542820622,
+      "grad_norm": 25.25,
+      "learning_rate": 2.175776202459275e-06,
+      "loss": 0.6754,
+      "step": 5581
+    },
+    {
+      "epoch": 0.47567106945036214,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.175663287134413e-06,
+      "loss": 0.8893,
+      "step": 5582
+    },
+    {
+      "epoch": 0.4757562846186621,
+      "grad_norm": 20.0,
+      "learning_rate": 2.175550355081884e-06,
+      "loss": 0.8174,
+      "step": 5583
+    },
+    {
+      "epoch": 0.47584149978696205,
+      "grad_norm": 10.75,
+      "learning_rate": 2.1754374063037294e-06,
+      "loss": 0.3816,
+      "step": 5584
+    },
+    {
+      "epoch": 0.475926714955262,
+      "grad_norm": 12.375,
+      "learning_rate": 2.17532444080199e-06,
+      "loss": 0.6221,
+      "step": 5585
+    },
+    {
+      "epoch": 0.476011930123562,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.175211458578708e-06,
+      "loss": 0.7942,
+      "step": 5586
+    },
+    {
+      "epoch": 0.476097145291862,
+      "grad_norm": 16.75,
+      "learning_rate": 2.175098459635924e-06,
+      "loss": 1.0592,
+      "step": 5587
+    },
+    {
+      "epoch": 0.47618236046016194,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.1749854439756813e-06,
+      "loss": 0.4611,
+      "step": 5588
+    },
+    {
+      "epoch": 0.4762675756284619,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1748724116000213e-06,
+      "loss": 0.7988,
+      "step": 5589
+    },
+    {
+      "epoch": 0.47635279079676185,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1747593625109863e-06,
+      "loss": 0.5668,
+      "step": 5590
+    },
+    {
+      "epoch": 0.4764380059650618,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.17464629671062e-06,
+      "loss": 0.406,
+      "step": 5591
+    },
+    {
+      "epoch": 0.47652322113336176,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.1745332142009654e-06,
+      "loss": 0.7513,
+      "step": 5592
+    },
+    {
+      "epoch": 0.4766084363016617,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.174420114984066e-06,
+      "loss": 0.8797,
+      "step": 5593
+    },
+    {
+      "epoch": 0.4766936514699617,
+      "grad_norm": 12.75,
+      "learning_rate": 2.1743069990619653e-06,
+      "loss": 0.5748,
+      "step": 5594
+    },
+    {
+      "epoch": 0.47677886663826163,
+      "grad_norm": 17.0,
+      "learning_rate": 2.174193866436708e-06,
+      "loss": 1.0065,
+      "step": 5595
+    },
+    {
+      "epoch": 0.4768640818065616,
+      "grad_norm": 16.25,
+      "learning_rate": 2.174080717110338e-06,
+      "loss": 1.0059,
+      "step": 5596
+    },
+    {
+      "epoch": 0.47694929697486155,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1739675510849002e-06,
+      "loss": 0.7129,
+      "step": 5597
+    },
+    {
+      "epoch": 0.4770345121431615,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.1738543683624396e-06,
+      "loss": 0.8302,
+      "step": 5598
+    },
+    {
+      "epoch": 0.47711972731146146,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1737411689450015e-06,
+      "loss": 0.6182,
+      "step": 5599
+    },
+    {
+      "epoch": 0.4772049424797614,
+      "grad_norm": 19.5,
+      "learning_rate": 2.1736279528346315e-06,
+      "loss": 0.9148,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4772901576480614,
+      "grad_norm": 28.0,
+      "learning_rate": 2.1735147200333757e-06,
+      "loss": 1.5502,
+      "step": 5601
+    },
+    {
+      "epoch": 0.47737537281636133,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1734014705432804e-06,
+      "loss": 0.7342,
+      "step": 5602
+    },
+    {
+      "epoch": 0.4774605879846613,
+      "grad_norm": 12.0,
+      "learning_rate": 2.1732882043663917e-06,
+      "loss": 0.7289,
+      "step": 5603
+    },
+    {
+      "epoch": 0.47754580315296125,
+      "grad_norm": 15.625,
+      "learning_rate": 2.173174921504757e-06,
+      "loss": 0.7083,
+      "step": 5604
+    },
+    {
+      "epoch": 0.4776310183212612,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.173061621960423e-06,
+      "loss": 0.6174,
+      "step": 5605
+    },
+    {
+      "epoch": 0.47771623348956116,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1729483057354374e-06,
+      "loss": 1.0593,
+      "step": 5606
+    },
+    {
+      "epoch": 0.4778014486578611,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1728349728318478e-06,
+      "loss": 0.6123,
+      "step": 5607
+    },
+    {
+      "epoch": 0.4778866638261611,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.172721623251702e-06,
+      "loss": 1.1667,
+      "step": 5608
+    },
+    {
+      "epoch": 0.47797187899446103,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.1726082569970493e-06,
+      "loss": 0.6684,
+      "step": 5609
+    },
+    {
+      "epoch": 0.478057094162761,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.1724948740699367e-06,
+      "loss": 0.4906,
+      "step": 5610
+    },
+    {
+      "epoch": 0.47814230933106094,
+      "grad_norm": 20.5,
+      "learning_rate": 2.172381474472415e-06,
+      "loss": 0.9473,
+      "step": 5611
+    },
+    {
+      "epoch": 0.4782275244993609,
+      "grad_norm": 22.875,
+      "learning_rate": 2.1722680582065315e-06,
+      "loss": 1.3324,
+      "step": 5612
+    },
+    {
+      "epoch": 0.47831273966766086,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1721546252743376e-06,
+      "loss": 1.0274,
+      "step": 5613
+    },
+    {
+      "epoch": 0.4783979548359608,
+      "grad_norm": 16.25,
+      "learning_rate": 2.172041175677882e-06,
+      "loss": 0.8132,
+      "step": 5614
+    },
+    {
+      "epoch": 0.47848317000426077,
+      "grad_norm": 20.125,
+      "learning_rate": 2.171927709419215e-06,
+      "loss": 1.0301,
+      "step": 5615
+    },
+    {
+      "epoch": 0.47856838517256073,
+      "grad_norm": 10.25,
+      "learning_rate": 2.1718142265003875e-06,
+      "loss": 0.7921,
+      "step": 5616
+    },
+    {
+      "epoch": 0.4786536003408607,
+      "grad_norm": 20.375,
+      "learning_rate": 2.17170072692345e-06,
+      "loss": 0.7704,
+      "step": 5617
+    },
+    {
+      "epoch": 0.47873881550916064,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1715872106904535e-06,
+      "loss": 0.7501,
+      "step": 5618
+    },
+    {
+      "epoch": 0.4788240306774606,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.171473677803449e-06,
+      "loss": 0.3499,
+      "step": 5619
+    },
+    {
+      "epoch": 0.47890924584576056,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.171360128264489e-06,
+      "loss": 1.0791,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4789944610140605,
+      "grad_norm": 12.625,
+      "learning_rate": 2.171246562075625e-06,
+      "loss": 0.7571,
+      "step": 5621
+    },
+    {
+      "epoch": 0.47907967618236047,
+      "grad_norm": 15.75,
+      "learning_rate": 2.1711329792389088e-06,
+      "loss": 0.7993,
+      "step": 5622
+    },
+    {
+      "epoch": 0.4791648913506604,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.171019379756394e-06,
+      "loss": 0.7689,
+      "step": 5623
+    },
+    {
+      "epoch": 0.4792501065189604,
+      "grad_norm": 19.25,
+      "learning_rate": 2.1709057636301325e-06,
+      "loss": 0.9547,
+      "step": 5624
+    },
+    {
+      "epoch": 0.47933532168726034,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.1707921308621775e-06,
+      "loss": 0.9161,
+      "step": 5625
+    },
+    {
+      "epoch": 0.4794205368555603,
+      "grad_norm": 17.75,
+      "learning_rate": 2.170678481454583e-06,
+      "loss": 1.0343,
+      "step": 5626
+    },
+    {
+      "epoch": 0.47950575202386025,
+      "grad_norm": 13.625,
+      "learning_rate": 2.1705648154094028e-06,
+      "loss": 0.5476,
+      "step": 5627
+    },
+    {
+      "epoch": 0.4795909671921602,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.17045113272869e-06,
+      "loss": 0.3372,
+      "step": 5628
+    },
+    {
+      "epoch": 0.47967618236046017,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.1703374334145006e-06,
+      "loss": 1.1081,
+      "step": 5629
+    },
+    {
+      "epoch": 0.4797613975287601,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.1702237174688877e-06,
+      "loss": 0.5139,
+      "step": 5630
+    },
+    {
+      "epoch": 0.4798466126970601,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.1701099848939073e-06,
+      "loss": 0.4015,
+      "step": 5631
+    },
+    {
+      "epoch": 0.47993182786536004,
+      "grad_norm": 13.625,
+      "learning_rate": 2.169996235691614e-06,
+      "loss": 0.4946,
+      "step": 5632
+    },
+    {
+      "epoch": 0.48001704303366,
+      "grad_norm": 16.25,
+      "learning_rate": 2.169882469864063e-06,
+      "loss": 0.6849,
+      "step": 5633
+    },
+    {
+      "epoch": 0.48010225820195995,
+      "grad_norm": 19.0,
+      "learning_rate": 2.1697686874133114e-06,
+      "loss": 0.9169,
+      "step": 5634
+    },
+    {
+      "epoch": 0.4801874733702599,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.169654888341415e-06,
+      "loss": 0.6028,
+      "step": 5635
+    },
+    {
+      "epoch": 0.48027268853855987,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1695410726504295e-06,
+      "loss": 0.7547,
+      "step": 5636
+    },
+    {
+      "epoch": 0.4803579037068598,
+      "grad_norm": 17.125,
+      "learning_rate": 2.1694272403424123e-06,
+      "loss": 1.2083,
+      "step": 5637
+    },
+    {
+      "epoch": 0.4804431188751598,
+      "grad_norm": 16.5,
+      "learning_rate": 2.16931339141942e-06,
+      "loss": 0.7565,
+      "step": 5638
+    },
+    {
+      "epoch": 0.48052833404345974,
+      "grad_norm": 19.5,
+      "learning_rate": 2.1691995258835106e-06,
+      "loss": 0.8016,
+      "step": 5639
+    },
+    {
+      "epoch": 0.4806135492117597,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1690856437367414e-06,
+      "loss": 0.6593,
+      "step": 5640
+    },
+    {
+      "epoch": 0.48069876438005965,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1689717449811702e-06,
+      "loss": 0.9464,
+      "step": 5641
+    },
+    {
+      "epoch": 0.4807839795483596,
+      "grad_norm": 15.5,
+      "learning_rate": 2.1688578296188557e-06,
+      "loss": 1.011,
+      "step": 5642
+    },
+    {
+      "epoch": 0.48086919471665956,
+      "grad_norm": 16.5,
+      "learning_rate": 2.168743897651856e-06,
+      "loss": 0.8007,
+      "step": 5643
+    },
+    {
+      "epoch": 0.4809544098849595,
+      "grad_norm": 11.25,
+      "learning_rate": 2.1686299490822304e-06,
+      "loss": 0.5918,
+      "step": 5644
+    },
+    {
+      "epoch": 0.4810396250532595,
+      "grad_norm": 16.5,
+      "learning_rate": 2.1685159839120378e-06,
+      "loss": 0.9377,
+      "step": 5645
+    },
+    {
+      "epoch": 0.48112484022155944,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.168402002143338e-06,
+      "loss": 0.624,
+      "step": 5646
+    },
+    {
+      "epoch": 0.4812100553898594,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.16828800377819e-06,
+      "loss": 0.685,
+      "step": 5647
+    },
+    {
+      "epoch": 0.48129527055815935,
+      "grad_norm": 12.75,
+      "learning_rate": 2.1681739888186547e-06,
+      "loss": 0.4311,
+      "step": 5648
+    },
+    {
+      "epoch": 0.4813804857264593,
+      "grad_norm": 17.5,
+      "learning_rate": 2.1680599572667925e-06,
+      "loss": 0.7615,
+      "step": 5649
+    },
+    {
+      "epoch": 0.48146570089475926,
+      "grad_norm": 14.625,
+      "learning_rate": 2.167945909124663e-06,
+      "loss": 0.9038,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4815509160630592,
+      "grad_norm": 14.25,
+      "learning_rate": 2.167831844394328e-06,
+      "loss": 0.5393,
+      "step": 5651
+    },
+    {
+      "epoch": 0.4816361312313592,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.167717763077849e-06,
+      "loss": 0.3941,
+      "step": 5652
+    },
+    {
+      "epoch": 0.48172134639965913,
+      "grad_norm": 16.5,
+      "learning_rate": 2.167603665177287e-06,
+      "loss": 1.1596,
+      "step": 5653
+    },
+    {
+      "epoch": 0.4818065615679591,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1674895506947034e-06,
+      "loss": 0.6373,
+      "step": 5654
+    },
+    {
+      "epoch": 0.48189177673625905,
+      "grad_norm": 13.75,
+      "learning_rate": 2.167375419632162e-06,
+      "loss": 0.4344,
+      "step": 5655
+    },
+    {
+      "epoch": 0.481976991904559,
+      "grad_norm": 20.25,
+      "learning_rate": 2.1672612719917237e-06,
+      "loss": 1.0426,
+      "step": 5656
+    },
+    {
+      "epoch": 0.48206220707285896,
+      "grad_norm": 20.5,
+      "learning_rate": 2.1671471077754517e-06,
+      "loss": 0.9335,
+      "step": 5657
+    },
+    {
+      "epoch": 0.4821474222411589,
+      "grad_norm": 14.0,
+      "learning_rate": 2.1670329269854097e-06,
+      "loss": 0.8758,
+      "step": 5658
+    },
+    {
+      "epoch": 0.4822326374094589,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1669187296236606e-06,
+      "loss": 0.8131,
+      "step": 5659
+    },
+    {
+      "epoch": 0.48231785257775883,
+      "grad_norm": 11.625,
+      "learning_rate": 2.166804515692268e-06,
+      "loss": 0.5703,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4824030677460588,
+      "grad_norm": 17.5,
+      "learning_rate": 2.166690285193295e-06,
+      "loss": 0.6691,
+      "step": 5661
+    },
+    {
+      "epoch": 0.48248828291435875,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1665760381288077e-06,
+      "loss": 0.785,
+      "step": 5662
+    },
+    {
+      "epoch": 0.4825734980826587,
+      "grad_norm": 17.875,
+      "learning_rate": 2.1664617745008696e-06,
+      "loss": 0.3756,
+      "step": 5663
+    },
+    {
+      "epoch": 0.48265871325095866,
+      "grad_norm": 37.75,
+      "learning_rate": 2.1663474943115455e-06,
+      "loss": 1.4977,
+      "step": 5664
+    },
+    {
+      "epoch": 0.4827439284192586,
+      "grad_norm": 13.0,
+      "learning_rate": 2.1662331975629006e-06,
+      "loss": 0.8138,
+      "step": 5665
+    },
+    {
+      "epoch": 0.4828291435875586,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.1661188842570006e-06,
+      "loss": 0.8917,
+      "step": 5666
+    },
+    {
+      "epoch": 0.48291435875585853,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.166004554395911e-06,
+      "loss": 0.6281,
+      "step": 5667
+    },
+    {
+      "epoch": 0.4829995739241585,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.165890207981698e-06,
+      "loss": 0.4999,
+      "step": 5668
+    },
+    {
+      "epoch": 0.48308478909245844,
+      "grad_norm": 15.875,
+      "learning_rate": 2.165775845016428e-06,
+      "loss": 0.7781,
+      "step": 5669
+    },
+    {
+      "epoch": 0.4831700042607584,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1656614655021675e-06,
+      "loss": 0.9867,
+      "step": 5670
+    },
+    {
+      "epoch": 0.48325521942905836,
+      "grad_norm": 17.0,
+      "learning_rate": 2.165547069440984e-06,
+      "loss": 0.6159,
+      "step": 5671
+    },
+    {
+      "epoch": 0.4833404345973583,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.1654326568349436e-06,
+      "loss": 0.5578,
+      "step": 5672
+    },
+    {
+      "epoch": 0.48342564976565827,
+      "grad_norm": 11.5,
+      "learning_rate": 2.1653182276861153e-06,
+      "loss": 0.555,
+      "step": 5673
+    },
+    {
+      "epoch": 0.48351086493395823,
+      "grad_norm": 23.5,
+      "learning_rate": 2.1652037819965657e-06,
+      "loss": 0.9754,
+      "step": 5674
+    },
+    {
+      "epoch": 0.4835960801022582,
+      "grad_norm": 32.5,
+      "learning_rate": 2.1650893197683636e-06,
+      "loss": 1.3365,
+      "step": 5675
+    },
+    {
+      "epoch": 0.48368129527055814,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.164974841003577e-06,
+      "loss": 0.6482,
+      "step": 5676
+    },
+    {
+      "epoch": 0.4837665104388581,
+      "grad_norm": 19.125,
+      "learning_rate": 2.1648603457042756e-06,
+      "loss": 1.0686,
+      "step": 5677
+    },
+    {
+      "epoch": 0.48385172560715806,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.1647458338725273e-06,
+      "loss": 0.3889,
+      "step": 5678
+    },
+    {
+      "epoch": 0.483936940775458,
+      "grad_norm": 17.125,
+      "learning_rate": 2.164631305510402e-06,
+      "loss": 0.5939,
+      "step": 5679
+    },
+    {
+      "epoch": 0.48402215594375797,
+      "grad_norm": 21.375,
+      "learning_rate": 2.1645167606199693e-06,
+      "loss": 0.9593,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4841073711120579,
+      "grad_norm": 36.0,
+      "learning_rate": 2.1644021992032994e-06,
+      "loss": 1.0025,
+      "step": 5681
+    },
+    {
+      "epoch": 0.4841925862803579,
+      "grad_norm": 20.75,
+      "learning_rate": 2.164287621262462e-06,
+      "loss": 0.9502,
+      "step": 5682
+    },
+    {
+      "epoch": 0.48427780144865784,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1641730267995277e-06,
+      "loss": 0.6198,
+      "step": 5683
+    },
+    {
+      "epoch": 0.4843630166169578,
+      "grad_norm": 10.625,
+      "learning_rate": 2.164058415816568e-06,
+      "loss": 0.4153,
+      "step": 5684
+    },
+    {
+      "epoch": 0.48444823178525775,
+      "grad_norm": 14.0,
+      "learning_rate": 2.163943788315653e-06,
+      "loss": 0.3983,
+      "step": 5685
+    },
+    {
+      "epoch": 0.4845334469535577,
+      "grad_norm": 21.375,
+      "learning_rate": 2.163829144298855e-06,
+      "loss": 1.0344,
+      "step": 5686
+    },
+    {
+      "epoch": 0.48461866212185767,
+      "grad_norm": 11.5,
+      "learning_rate": 2.1637144837682456e-06,
+      "loss": 0.5683,
+      "step": 5687
+    },
+    {
+      "epoch": 0.4847038772901576,
+      "grad_norm": 17.125,
+      "learning_rate": 2.163599806725896e-06,
+      "loss": 0.879,
+      "step": 5688
+    },
+    {
+      "epoch": 0.4847890924584576,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.16348511317388e-06,
+      "loss": 0.3956,
+      "step": 5689
+    },
+    {
+      "epoch": 0.48487430762675754,
+      "grad_norm": 17.375,
+      "learning_rate": 2.163370403114269e-06,
+      "loss": 0.7775,
+      "step": 5690
+    },
+    {
+      "epoch": 0.4849595227950575,
+      "grad_norm": 15.5,
+      "learning_rate": 2.163255676549137e-06,
+      "loss": 0.6783,
+      "step": 5691
+    },
+    {
+      "epoch": 0.48504473796335745,
+      "grad_norm": 15.75,
+      "learning_rate": 2.163140933480556e-06,
+      "loss": 0.6896,
+      "step": 5692
+    },
+    {
+      "epoch": 0.4851299531316574,
+      "grad_norm": 25.125,
+      "learning_rate": 2.1630261739105996e-06,
+      "loss": 1.416,
+      "step": 5693
+    },
+    {
+      "epoch": 0.48521516829995737,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.162911397841343e-06,
+      "loss": 0.4953,
+      "step": 5694
+    },
+    {
+      "epoch": 0.4853003834682573,
+      "grad_norm": 19.375,
+      "learning_rate": 2.162796605274859e-06,
+      "loss": 1.0947,
+      "step": 5695
+    },
+    {
+      "epoch": 0.4853855986365573,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.162681796213223e-06,
+      "loss": 0.8249,
+      "step": 5696
+    },
+    {
+      "epoch": 0.48547081380485724,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1625669706585085e-06,
+      "loss": 0.8246,
+      "step": 5697
+    },
+    {
+      "epoch": 0.4855560289731572,
+      "grad_norm": 21.375,
+      "learning_rate": 2.1624521286127915e-06,
+      "loss": 1.2245,
+      "step": 5698
+    },
+    {
+      "epoch": 0.4856412441414572,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.1623372700781463e-06,
+      "loss": 0.5612,
+      "step": 5699
+    },
+    {
+      "epoch": 0.48572645930975716,
+      "grad_norm": 20.5,
+      "learning_rate": 2.1622223950566503e-06,
+      "loss": 1.0498,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4858116744780571,
+      "grad_norm": 13.5,
+      "learning_rate": 2.1621075035503777e-06,
+      "loss": 0.6437,
+      "step": 5701
+    },
+    {
+      "epoch": 0.4858968896463571,
+      "grad_norm": 18.0,
+      "learning_rate": 2.161992595561405e-06,
+      "loss": 0.6918,
+      "step": 5702
+    },
+    {
+      "epoch": 0.48598210481465703,
+      "grad_norm": 16.375,
+      "learning_rate": 2.16187767109181e-06,
+      "loss": 0.8423,
+      "step": 5703
+    },
+    {
+      "epoch": 0.486067319982957,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1617627301436677e-06,
+      "loss": 0.8333,
+      "step": 5704
+    },
+    {
+      "epoch": 0.48615253515125695,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.1616477727190563e-06,
+      "loss": 0.69,
+      "step": 5705
+    },
+    {
+      "epoch": 0.4862377503195569,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.161532798820053e-06,
+      "loss": 0.5914,
+      "step": 5706
+    },
+    {
+      "epoch": 0.48632296548785686,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.161417808448735e-06,
+      "loss": 0.7777,
+      "step": 5707
+    },
+    {
+      "epoch": 0.4864081806561568,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1613028016071814e-06,
+      "loss": 0.7,
+      "step": 5708
+    },
+    {
+      "epoch": 0.4864933958244568,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1611877782974696e-06,
+      "loss": 0.7747,
+      "step": 5709
+    },
+    {
+      "epoch": 0.48657861099275673,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.161072738521678e-06,
+      "loss": 0.9512,
+      "step": 5710
+    },
+    {
+      "epoch": 0.4866638261610567,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.160957682281886e-06,
+      "loss": 0.7135,
+      "step": 5711
+    },
+    {
+      "epoch": 0.48674904132935665,
+      "grad_norm": 9.4375,
+      "learning_rate": 2.1608426095801727e-06,
+      "loss": 0.376,
+      "step": 5712
+    },
+    {
+      "epoch": 0.4868342564976566,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.1607275204186174e-06,
+      "loss": 0.7216,
+      "step": 5713
+    },
+    {
+      "epoch": 0.48691947166595656,
+      "grad_norm": 12.75,
+      "learning_rate": 2.1606124147993e-06,
+      "loss": 0.4498,
+      "step": 5714
+    },
+    {
+      "epoch": 0.4870046868342565,
+      "grad_norm": 11.875,
+      "learning_rate": 2.160497292724301e-06,
+      "loss": 0.609,
+      "step": 5715
+    },
+    {
+      "epoch": 0.4870899020025565,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1603821541956997e-06,
+      "loss": 0.5074,
+      "step": 5716
+    },
+    {
+      "epoch": 0.48717511717085643,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1602669992155776e-06,
+      "loss": 0.6741,
+      "step": 5717
+    },
+    {
+      "epoch": 0.4872603323391564,
+      "grad_norm": 16.0,
+      "learning_rate": 2.160151827786016e-06,
+      "loss": 0.9134,
+      "step": 5718
+    },
+    {
+      "epoch": 0.48734554750745634,
+      "grad_norm": 17.75,
+      "learning_rate": 2.160036639909095e-06,
+      "loss": 1.1331,
+      "step": 5719
+    },
+    {
+      "epoch": 0.4874307626757563,
+      "grad_norm": 21.25,
+      "learning_rate": 2.159921435586897e-06,
+      "loss": 0.7145,
+      "step": 5720
+    },
+    {
+      "epoch": 0.48751597784405626,
+      "grad_norm": 14.875,
+      "learning_rate": 2.1598062148215037e-06,
+      "loss": 0.9386,
+      "step": 5721
+    },
+    {
+      "epoch": 0.4876011930123562,
+      "grad_norm": 25.375,
+      "learning_rate": 2.159690977614997e-06,
+      "loss": 1.2735,
+      "step": 5722
+    },
+    {
+      "epoch": 0.4876864081806562,
+      "grad_norm": 12.125,
+      "learning_rate": 2.1595757239694603e-06,
+      "loss": 0.659,
+      "step": 5723
+    },
+    {
+      "epoch": 0.48777162334895613,
+      "grad_norm": 15.75,
+      "learning_rate": 2.159460453886975e-06,
+      "loss": 0.748,
+      "step": 5724
+    },
+    {
+      "epoch": 0.4878568385172561,
+      "grad_norm": 16.625,
+      "learning_rate": 2.159345167369625e-06,
+      "loss": 0.9094,
+      "step": 5725
+    },
+    {
+      "epoch": 0.48794205368555604,
+      "grad_norm": 14.75,
+      "learning_rate": 2.159229864419493e-06,
+      "loss": 0.8414,
+      "step": 5726
+    },
+    {
+      "epoch": 0.488027268853856,
+      "grad_norm": 14.375,
+      "learning_rate": 2.1591145450386636e-06,
+      "loss": 0.2189,
+      "step": 5727
+    },
+    {
+      "epoch": 0.48811248402215596,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.15899920922922e-06,
+      "loss": 0.4003,
+      "step": 5728
+    },
+    {
+      "epoch": 0.4881976991904559,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.1588838569932466e-06,
+      "loss": 0.5217,
+      "step": 5729
+    },
+    {
+      "epoch": 0.48828291435875587,
+      "grad_norm": 16.5,
+      "learning_rate": 2.158768488332828e-06,
+      "loss": 0.9054,
+      "step": 5730
+    },
+    {
+      "epoch": 0.4883681295270558,
+      "grad_norm": 12.375,
+      "learning_rate": 2.1586531032500488e-06,
+      "loss": 0.536,
+      "step": 5731
+    },
+    {
+      "epoch": 0.4884533446953558,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.1585377017469946e-06,
+      "loss": 0.7898,
+      "step": 5732
+    },
+    {
+      "epoch": 0.48853855986365574,
+      "grad_norm": 13.625,
+      "learning_rate": 2.15842228382575e-06,
+      "loss": 0.4113,
+      "step": 5733
+    },
+    {
+      "epoch": 0.4886237750319557,
+      "grad_norm": 19.625,
+      "learning_rate": 2.1583068494884016e-06,
+      "loss": 0.8966,
+      "step": 5734
+    },
+    {
+      "epoch": 0.48870899020025566,
+      "grad_norm": 17.5,
+      "learning_rate": 2.158191398737035e-06,
+      "loss": 1.003,
+      "step": 5735
+    },
+    {
+      "epoch": 0.4887942053685556,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1580759315737367e-06,
+      "loss": 0.9048,
+      "step": 5736
+    },
+    {
+      "epoch": 0.48887942053685557,
+      "grad_norm": 12.875,
+      "learning_rate": 2.157960448000593e-06,
+      "loss": 0.5135,
+      "step": 5737
+    },
+    {
+      "epoch": 0.4889646357051555,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1578449480196913e-06,
+      "loss": 0.7703,
+      "step": 5738
+    },
+    {
+      "epoch": 0.4890498508734555,
+      "grad_norm": 21.375,
+      "learning_rate": 2.157729431633118e-06,
+      "loss": 0.7969,
+      "step": 5739
+    },
+    {
+      "epoch": 0.48913506604175544,
+      "grad_norm": 17.875,
+      "learning_rate": 2.1576138988429614e-06,
+      "loss": 1.0738,
+      "step": 5740
+    },
+    {
+      "epoch": 0.4892202812100554,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1574983496513086e-06,
+      "loss": 0.8701,
+      "step": 5741
+    },
+    {
+      "epoch": 0.48930549637835535,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.1573827840602483e-06,
+      "loss": 0.8098,
+      "step": 5742
+    },
+    {
+      "epoch": 0.4893907115466553,
+      "grad_norm": 17.625,
+      "learning_rate": 2.157267202071869e-06,
+      "loss": 1.2905,
+      "step": 5743
+    },
+    {
+      "epoch": 0.48947592671495527,
+      "grad_norm": 15.875,
+      "learning_rate": 2.1571516036882585e-06,
+      "loss": 0.9116,
+      "step": 5744
+    },
+    {
+      "epoch": 0.4895611418832552,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1570359889115065e-06,
+      "loss": 0.7612,
+      "step": 5745
+    },
+    {
+      "epoch": 0.4896463570515552,
+      "grad_norm": 14.375,
+      "learning_rate": 2.156920357743702e-06,
+      "loss": 0.5782,
+      "step": 5746
+    },
+    {
+      "epoch": 0.48973157221985514,
+      "grad_norm": 14.0,
+      "learning_rate": 2.156804710186935e-06,
+      "loss": 0.5309,
+      "step": 5747
+    },
+    {
+      "epoch": 0.4898167873881551,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1566890462432944e-06,
+      "loss": 0.6784,
+      "step": 5748
+    },
+    {
+      "epoch": 0.48990200255645505,
+      "grad_norm": 13.0,
+      "learning_rate": 2.156573365914872e-06,
+      "loss": 0.6522,
+      "step": 5749
+    },
+    {
+      "epoch": 0.489987217724755,
+      "grad_norm": 14.625,
+      "learning_rate": 2.1564576692037564e-06,
+      "loss": 0.9226,
+      "step": 5750
+    },
+    {
+      "epoch": 0.49007243289305497,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1563419561120393e-06,
+      "loss": 0.6082,
+      "step": 5751
+    },
+    {
+      "epoch": 0.4901576480613549,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.1562262266418115e-06,
+      "loss": 0.492,
+      "step": 5752
+    },
+    {
+      "epoch": 0.4902428632296549,
+      "grad_norm": 12.875,
+      "learning_rate": 2.1561104807951648e-06,
+      "loss": 0.6654,
+      "step": 5753
+    },
+    {
+      "epoch": 0.49032807839795484,
+      "grad_norm": 13.875,
+      "learning_rate": 2.155994718574191e-06,
+      "loss": 0.6855,
+      "step": 5754
+    },
+    {
+      "epoch": 0.4904132935662548,
+      "grad_norm": 24.125,
+      "learning_rate": 2.155878939980981e-06,
+      "loss": 0.7849,
+      "step": 5755
+    },
+    {
+      "epoch": 0.49049850873455475,
+      "grad_norm": 15.0,
+      "learning_rate": 2.1557631450176274e-06,
+      "loss": 0.831,
+      "step": 5756
+    },
+    {
+      "epoch": 0.4905837239028547,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1556473336862236e-06,
+      "loss": 0.4793,
+      "step": 5757
+    },
+    {
+      "epoch": 0.49066893907115466,
+      "grad_norm": 17.25,
+      "learning_rate": 2.1555315059888616e-06,
+      "loss": 0.9881,
+      "step": 5758
+    },
+    {
+      "epoch": 0.4907541542394546,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1554156619276346e-06,
+      "loss": 0.8504,
+      "step": 5759
+    },
+    {
+      "epoch": 0.4908393694077546,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.155299801504636e-06,
+      "loss": 0.8327,
+      "step": 5760
+    },
+    {
+      "epoch": 0.49092458457605453,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.15518392472196e-06,
+      "loss": 0.8131,
+      "step": 5761
+    },
+    {
+      "epoch": 0.4910097997443545,
+      "grad_norm": 11.25,
+      "learning_rate": 2.1550680315816996e-06,
+      "loss": 0.6647,
+      "step": 5762
+    },
+    {
+      "epoch": 0.49109501491265445,
+      "grad_norm": 26.125,
+      "learning_rate": 2.15495212208595e-06,
+      "loss": 0.7175,
+      "step": 5763
+    },
+    {
+      "epoch": 0.4911802300809544,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.154836196236806e-06,
+      "loss": 0.6945,
+      "step": 5764
+    },
+    {
+      "epoch": 0.49126544524925436,
+      "grad_norm": 15.375,
+      "learning_rate": 2.1547202540363617e-06,
+      "loss": 0.7726,
+      "step": 5765
+    },
+    {
+      "epoch": 0.4913506604175543,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1546042954867124e-06,
+      "loss": 0.9963,
+      "step": 5766
+    },
+    {
+      "epoch": 0.4914358755858543,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.154488320589954e-06,
+      "loss": 0.9346,
+      "step": 5767
+    },
+    {
+      "epoch": 0.49152109075415423,
+      "grad_norm": 19.125,
+      "learning_rate": 2.1543723293481826e-06,
+      "loss": 1.0091,
+      "step": 5768
+    },
+    {
+      "epoch": 0.4916063059224542,
+      "grad_norm": 17.875,
+      "learning_rate": 2.1542563217634936e-06,
+      "loss": 0.541,
+      "step": 5769
+    },
+    {
+      "epoch": 0.49169152109075415,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1541402978379835e-06,
+      "loss": 0.8654,
+      "step": 5770
+    },
+    {
+      "epoch": 0.4917767362590541,
+      "grad_norm": 24.625,
+      "learning_rate": 2.1540242575737486e-06,
+      "loss": 1.3225,
+      "step": 5771
+    },
+    {
+      "epoch": 0.49186195142735406,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.1539082009728863e-06,
+      "loss": 0.3247,
+      "step": 5772
+    },
+    {
+      "epoch": 0.491947166595654,
+      "grad_norm": 18.75,
+      "learning_rate": 2.1537921280374946e-06,
+      "loss": 0.6543,
+      "step": 5773
+    },
+    {
+      "epoch": 0.492032381763954,
+      "grad_norm": 20.875,
+      "learning_rate": 2.15367603876967e-06,
+      "loss": 1.0828,
+      "step": 5774
+    },
+    {
+      "epoch": 0.49211759693225393,
+      "grad_norm": 14.5,
+      "learning_rate": 2.15355993317151e-06,
+      "loss": 0.9117,
+      "step": 5775
+    },
+    {
+      "epoch": 0.4922028121005539,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.153443811245114e-06,
+      "loss": 0.7582,
+      "step": 5776
+    },
+    {
+      "epoch": 0.49228802726885384,
+      "grad_norm": 14.875,
+      "learning_rate": 2.15332767299258e-06,
+      "loss": 0.9515,
+      "step": 5777
+    },
+    {
+      "epoch": 0.4923732424371538,
+      "grad_norm": 18.25,
+      "learning_rate": 2.1532115184160063e-06,
+      "loss": 0.9592,
+      "step": 5778
+    },
+    {
+      "epoch": 0.49245845760545376,
+      "grad_norm": 24.25,
+      "learning_rate": 2.1530953475174926e-06,
+      "loss": 0.9465,
+      "step": 5779
+    },
+    {
+      "epoch": 0.4925436727737537,
+      "grad_norm": 12.5,
+      "learning_rate": 2.1529791602991375e-06,
+      "loss": 0.5654,
+      "step": 5780
+    },
+    {
+      "epoch": 0.4926288879420537,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.152862956763041e-06,
+      "loss": 0.5259,
+      "step": 5781
+    },
+    {
+      "epoch": 0.49271410311035363,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.152746736911303e-06,
+      "loss": 0.5468,
+      "step": 5782
+    },
+    {
+      "epoch": 0.4927993182786536,
+      "grad_norm": 21.375,
+      "learning_rate": 2.1526305007460235e-06,
+      "loss": 0.8994,
+      "step": 5783
+    },
+    {
+      "epoch": 0.49288453344695354,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1525142482693033e-06,
+      "loss": 0.8481,
+      "step": 5784
+    },
+    {
+      "epoch": 0.4929697486152535,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.152397979483243e-06,
+      "loss": 0.839,
+      "step": 5785
+    },
+    {
+      "epoch": 0.49305496378355346,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1522816943899443e-06,
+      "loss": 0.7765,
+      "step": 5786
+    },
+    {
+      "epoch": 0.4931401789518534,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1521653929915077e-06,
+      "loss": 0.757,
+      "step": 5787
+    },
+    {
+      "epoch": 0.49322539412015337,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.152049075290035e-06,
+      "loss": 0.4022,
+      "step": 5788
+    },
+    {
+      "epoch": 0.49331060928845333,
+      "grad_norm": 11.75,
+      "learning_rate": 2.151932741287629e-06,
+      "loss": 0.473,
+      "step": 5789
+    },
+    {
+      "epoch": 0.4933958244567533,
+      "grad_norm": 25.0,
+      "learning_rate": 2.151816390986391e-06,
+      "loss": 0.8495,
+      "step": 5790
+    },
+    {
+      "epoch": 0.49348103962505324,
+      "grad_norm": 10.125,
+      "learning_rate": 2.151700024388424e-06,
+      "loss": 0.4126,
+      "step": 5791
+    },
+    {
+      "epoch": 0.4935662547933532,
+      "grad_norm": 17.75,
+      "learning_rate": 2.1515836414958308e-06,
+      "loss": 0.6038,
+      "step": 5792
+    },
+    {
+      "epoch": 0.49365146996165316,
+      "grad_norm": 25.125,
+      "learning_rate": 2.1514672423107146e-06,
+      "loss": 1.2276,
+      "step": 5793
+    },
+    {
+      "epoch": 0.4937366851299531,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1513508268351787e-06,
+      "loss": 0.656,
+      "step": 5794
+    },
+    {
+      "epoch": 0.49382190029825307,
+      "grad_norm": 13.25,
+      "learning_rate": 2.151234395071327e-06,
+      "loss": 0.426,
+      "step": 5795
+    },
+    {
+      "epoch": 0.493907115466553,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.151117947021264e-06,
+      "loss": 0.3618,
+      "step": 5796
+    },
+    {
+      "epoch": 0.493992330634853,
+      "grad_norm": 15.25,
+      "learning_rate": 2.151001482687093e-06,
+      "loss": 0.7949,
+      "step": 5797
+    },
+    {
+      "epoch": 0.49407754580315294,
+      "grad_norm": 33.5,
+      "learning_rate": 2.1508850020709196e-06,
+      "loss": 0.962,
+      "step": 5798
+    },
+    {
+      "epoch": 0.4941627609714529,
+      "grad_norm": 14.125,
+      "learning_rate": 2.150768505174848e-06,
+      "loss": 1.0273,
+      "step": 5799
+    },
+    {
+      "epoch": 0.49424797613975285,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1506519920009838e-06,
+      "loss": 0.6742,
+      "step": 5800
+    },
+    {
+      "epoch": 0.4943331913080528,
+      "grad_norm": 17.75,
+      "learning_rate": 2.1505354625514325e-06,
+      "loss": 0.7906,
+      "step": 5801
+    },
+    {
+      "epoch": 0.49441840647635277,
+      "grad_norm": 13.125,
+      "learning_rate": 2.1504189168283e-06,
+      "loss": 0.749,
+      "step": 5802
+    },
+    {
+      "epoch": 0.4945036216446527,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.150302354833692e-06,
+      "loss": 0.6195,
+      "step": 5803
+    },
+    {
+      "epoch": 0.4945888368129527,
+      "grad_norm": 17.0,
+      "learning_rate": 2.150185776569715e-06,
+      "loss": 0.7822,
+      "step": 5804
+    },
+    {
+      "epoch": 0.49467405198125264,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1500691820384762e-06,
+      "loss": 0.7311,
+      "step": 5805
+    },
+    {
+      "epoch": 0.4947592671495526,
+      "grad_norm": 14.0,
+      "learning_rate": 2.1499525712420818e-06,
+      "loss": 0.6468,
+      "step": 5806
+    },
+    {
+      "epoch": 0.49484448231785255,
+      "grad_norm": 11.625,
+      "learning_rate": 2.1498359441826395e-06,
+      "loss": 0.473,
+      "step": 5807
+    },
+    {
+      "epoch": 0.4949296974861525,
+      "grad_norm": 19.625,
+      "learning_rate": 2.149719300862257e-06,
+      "loss": 0.763,
+      "step": 5808
+    },
+    {
+      "epoch": 0.49501491265445247,
+      "grad_norm": 20.25,
+      "learning_rate": 2.149602641283042e-06,
+      "loss": 0.6283,
+      "step": 5809
+    },
+    {
+      "epoch": 0.4951001278227524,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1494859654471027e-06,
+      "loss": 0.8069,
+      "step": 5810
+    },
+    {
+      "epoch": 0.4951853429910524,
+      "grad_norm": 13.625,
+      "learning_rate": 2.1493692733565474e-06,
+      "loss": 0.7092,
+      "step": 5811
+    },
+    {
+      "epoch": 0.4952705581593524,
+      "grad_norm": 13.25,
+      "learning_rate": 2.149252565013485e-06,
+      "loss": 0.8032,
+      "step": 5812
+    },
+    {
+      "epoch": 0.49535577332765235,
+      "grad_norm": 23.25,
+      "learning_rate": 2.1491358404200245e-06,
+      "loss": 1.2107,
+      "step": 5813
+    },
+    {
+      "epoch": 0.4954409884959523,
+      "grad_norm": 15.875,
+      "learning_rate": 2.149019099578275e-06,
+      "loss": 0.7964,
+      "step": 5814
+    },
+    {
+      "epoch": 0.49552620366425226,
+      "grad_norm": 13.25,
+      "learning_rate": 2.1489023424903466e-06,
+      "loss": 0.7263,
+      "step": 5815
+    },
+    {
+      "epoch": 0.4956114188325522,
+      "grad_norm": 17.375,
+      "learning_rate": 2.1487855691583487e-06,
+      "loss": 0.859,
+      "step": 5816
+    },
+    {
+      "epoch": 0.4956966340008522,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.148668779584392e-06,
+      "loss": 0.871,
+      "step": 5817
+    },
+    {
+      "epoch": 0.49578184916915213,
+      "grad_norm": 25.0,
+      "learning_rate": 2.1485519737705868e-06,
+      "loss": 0.8851,
+      "step": 5818
+    },
+    {
+      "epoch": 0.4958670643374521,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1484351517190435e-06,
+      "loss": 0.7336,
+      "step": 5819
+    },
+    {
+      "epoch": 0.49595227950575205,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.148318313431874e-06,
+      "loss": 0.6069,
+      "step": 5820
+    },
+    {
+      "epoch": 0.496037494674052,
+      "grad_norm": 21.25,
+      "learning_rate": 2.148201458911189e-06,
+      "loss": 0.9736,
+      "step": 5821
+    },
+    {
+      "epoch": 0.49612270984235196,
+      "grad_norm": 20.125,
+      "learning_rate": 2.1480845881591e-06,
+      "loss": 1.0662,
+      "step": 5822
+    },
+    {
+      "epoch": 0.4962079250106519,
+      "grad_norm": 13.875,
+      "learning_rate": 2.14796770117772e-06,
+      "loss": 0.8106,
+      "step": 5823
+    },
+    {
+      "epoch": 0.4962931401789519,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.147850797969161e-06,
+      "loss": 0.4575,
+      "step": 5824
+    },
+    {
+      "epoch": 0.49637835534725183,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.1477338785355345e-06,
+      "loss": 0.6992,
+      "step": 5825
+    },
+    {
+      "epoch": 0.4964635705155518,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.1476169428789545e-06,
+      "loss": 0.645,
+      "step": 5826
+    },
+    {
+      "epoch": 0.49654878568385175,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.1474999910015334e-06,
+      "loss": 0.823,
+      "step": 5827
+    },
+    {
+      "epoch": 0.4966340008521517,
+      "grad_norm": 12.5,
+      "learning_rate": 2.147383022905385e-06,
+      "loss": 0.5045,
+      "step": 5828
+    },
+    {
+      "epoch": 0.49671921602045166,
+      "grad_norm": 21.875,
+      "learning_rate": 2.1472660385926232e-06,
+      "loss": 1.1154,
+      "step": 5829
+    },
+    {
+      "epoch": 0.4968044311887516,
+      "grad_norm": 17.25,
+      "learning_rate": 2.1471490380653617e-06,
+      "loss": 0.6726,
+      "step": 5830
+    },
+    {
+      "epoch": 0.4968896463570516,
+      "grad_norm": 13.875,
+      "learning_rate": 2.147032021325715e-06,
+      "loss": 0.6526,
+      "step": 5831
+    },
+    {
+      "epoch": 0.49697486152535153,
+      "grad_norm": 14.625,
+      "learning_rate": 2.146914988375798e-06,
+      "loss": 0.7688,
+      "step": 5832
+    },
+    {
+      "epoch": 0.4970600766936515,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.1467979392177245e-06,
+      "loss": 0.9518,
+      "step": 5833
+    },
+    {
+      "epoch": 0.49714529186195144,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.1466808738536107e-06,
+      "loss": 0.9798,
+      "step": 5834
+    },
+    {
+      "epoch": 0.4972305070302514,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.146563792285572e-06,
+      "loss": 0.4721,
+      "step": 5835
+    },
+    {
+      "epoch": 0.49731572219855136,
+      "grad_norm": 12.25,
+      "learning_rate": 2.1464466945157237e-06,
+      "loss": 0.6144,
+      "step": 5836
+    },
+    {
+      "epoch": 0.4974009373668513,
+      "grad_norm": 12.75,
+      "learning_rate": 2.1463295805461828e-06,
+      "loss": 0.7683,
+      "step": 5837
+    },
+    {
+      "epoch": 0.49748615253515127,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1462124503790645e-06,
+      "loss": 0.8325,
+      "step": 5838
+    },
+    {
+      "epoch": 0.49757136770345123,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.146095304016486e-06,
+      "loss": 0.8617,
+      "step": 5839
+    },
+    {
+      "epoch": 0.4976565828717512,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1459781414605645e-06,
+      "loss": 0.9438,
+      "step": 5840
+    },
+    {
+      "epoch": 0.49774179804005114,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1458609627134166e-06,
+      "loss": 0.8293,
+      "step": 5841
+    },
+    {
+      "epoch": 0.4978270132083511,
+      "grad_norm": 13.5,
+      "learning_rate": 2.1457437677771605e-06,
+      "loss": 0.6026,
+      "step": 5842
+    },
+    {
+      "epoch": 0.49791222837665106,
+      "grad_norm": 9.875,
+      "learning_rate": 2.1456265566539137e-06,
+      "loss": 0.3548,
+      "step": 5843
+    },
+    {
+      "epoch": 0.497997443544951,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1455093293457947e-06,
+      "loss": 0.707,
+      "step": 5844
+    },
+    {
+      "epoch": 0.49808265871325097,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1453920858549216e-06,
+      "loss": 0.953,
+      "step": 5845
+    },
+    {
+      "epoch": 0.4981678738815509,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1452748261834127e-06,
+      "loss": 0.8664,
+      "step": 5846
+    },
+    {
+      "epoch": 0.4982530890498509,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.145157550333388e-06,
+      "loss": 0.6303,
+      "step": 5847
+    },
+    {
+      "epoch": 0.49833830421815084,
+      "grad_norm": 23.75,
+      "learning_rate": 2.1450402583069655e-06,
+      "loss": 0.5487,
+      "step": 5848
+    },
+    {
+      "epoch": 0.4984235193864508,
+      "grad_norm": 17.5,
+      "learning_rate": 2.1449229501062662e-06,
+      "loss": 0.6514,
+      "step": 5849
+    },
+    {
+      "epoch": 0.49850873455475075,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.1448056257334087e-06,
+      "loss": 0.3962,
+      "step": 5850
+    },
+    {
+      "epoch": 0.4985939497230507,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1446882851905144e-06,
+      "loss": 0.5081,
+      "step": 5851
+    },
+    {
+      "epoch": 0.49867916489135067,
+      "grad_norm": 14.5,
+      "learning_rate": 2.144570928479703e-06,
+      "loss": 0.696,
+      "step": 5852
+    },
+    {
+      "epoch": 0.4987643800596506,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.144453555603095e-06,
+      "loss": 0.6514,
+      "step": 5853
+    },
+    {
+      "epoch": 0.4988495952279506,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.1443361665628115e-06,
+      "loss": 0.7258,
+      "step": 5854
+    },
+    {
+      "epoch": 0.49893481039625054,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.144218761360975e-06,
+      "loss": 0.9006,
+      "step": 5855
+    },
+    {
+      "epoch": 0.4990200255645505,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.144101339999706e-06,
+      "loss": 0.669,
+      "step": 5856
+    },
+    {
+      "epoch": 0.49910524073285045,
+      "grad_norm": 15.25,
+      "learning_rate": 2.143983902481126e-06,
+      "loss": 0.7959,
+      "step": 5857
+    },
+    {
+      "epoch": 0.4991904559011504,
+      "grad_norm": 12.625,
+      "learning_rate": 2.1438664488073585e-06,
+      "loss": 0.8162,
+      "step": 5858
+    },
+    {
+      "epoch": 0.49927567106945037,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1437489789805257e-06,
+      "loss": 0.6857,
+      "step": 5859
+    },
+    {
+      "epoch": 0.4993608862377503,
+      "grad_norm": 11.0,
+      "learning_rate": 2.1436314930027494e-06,
+      "loss": 0.4244,
+      "step": 5860
+    },
+    {
+      "epoch": 0.4994461014060503,
+      "grad_norm": 14.0,
+      "learning_rate": 2.1435139908761535e-06,
+      "loss": 0.5832,
+      "step": 5861
+    },
+    {
+      "epoch": 0.49953131657435024,
+      "grad_norm": 13.125,
+      "learning_rate": 2.1433964726028617e-06,
+      "loss": 0.5866,
+      "step": 5862
+    },
+    {
+      "epoch": 0.4996165317426502,
+      "grad_norm": 18.625,
+      "learning_rate": 2.1432789381849974e-06,
+      "loss": 0.9645,
+      "step": 5863
+    },
+    {
+      "epoch": 0.49970174691095015,
+      "grad_norm": 25.875,
+      "learning_rate": 2.143161387624684e-06,
+      "loss": 0.6659,
+      "step": 5864
+    },
+    {
+      "epoch": 0.4997869620792501,
+      "grad_norm": 13.125,
+      "learning_rate": 2.1430438209240467e-06,
+      "loss": 0.7143,
+      "step": 5865
+    },
+    {
+      "epoch": 0.49987217724755006,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.142926238085209e-06,
+      "loss": 0.5149,
+      "step": 5866
+    },
+    {
+      "epoch": 0.49995739241585,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.1428086391102965e-06,
+      "loss": 0.8181,
+      "step": 5867
+    },
+    {
+      "epoch": 0.50004260758415,
+      "grad_norm": 11.25,
+      "learning_rate": 2.1426910240014335e-06,
+      "loss": 0.2639,
+      "step": 5868
+    },
+    {
+      "epoch": 0.5001278227524499,
+      "grad_norm": 10.625,
+      "learning_rate": 2.1425733927607466e-06,
+      "loss": 0.3995,
+      "step": 5869
+    },
+    {
+      "epoch": 0.5002130379207499,
+      "grad_norm": 15.5,
+      "learning_rate": 2.142455745390361e-06,
+      "loss": 0.8227,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5002982530890498,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.1423380818924028e-06,
+      "loss": 0.7189,
+      "step": 5871
+    },
+    {
+      "epoch": 0.5003834682573498,
+      "grad_norm": 29.5,
+      "learning_rate": 2.142220402268998e-06,
+      "loss": 0.8472,
+      "step": 5872
+    },
+    {
+      "epoch": 0.5004686834256498,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.142102706522273e-06,
+      "loss": 0.5844,
+      "step": 5873
+    },
+    {
+      "epoch": 0.5005538985939497,
+      "grad_norm": 18.375,
+      "learning_rate": 2.1419849946543554e-06,
+      "loss": 1.0016,
+      "step": 5874
+    },
+    {
+      "epoch": 0.5006391137622497,
+      "grad_norm": 18.25,
+      "learning_rate": 2.1418672666673716e-06,
+      "loss": 1.0531,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5007243289305496,
+      "grad_norm": 14.25,
+      "learning_rate": 2.1417495225634497e-06,
+      "loss": 0.7746,
+      "step": 5876
+    },
+    {
+      "epoch": 0.5008095440988496,
+      "grad_norm": 16.5,
+      "learning_rate": 2.1416317623447176e-06,
+      "loss": 0.9421,
+      "step": 5877
+    },
+    {
+      "epoch": 0.5008947592671495,
+      "grad_norm": 21.375,
+      "learning_rate": 2.141513986013302e-06,
+      "loss": 0.5886,
+      "step": 5878
+    },
+    {
+      "epoch": 0.5009799744354495,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.141396193571333e-06,
+      "loss": 0.3807,
+      "step": 5879
+    },
+    {
+      "epoch": 0.5010651896037495,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.141278385020938e-06,
+      "loss": 0.5249,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5011504047720494,
+      "grad_norm": 18.625,
+      "learning_rate": 2.1411605603642466e-06,
+      "loss": 0.9046,
+      "step": 5881
+    },
+    {
+      "epoch": 0.5012356199403494,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1410427196033877e-06,
+      "loss": 0.5197,
+      "step": 5882
+    },
+    {
+      "epoch": 0.5013208351086493,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.1409248627404905e-06,
+      "loss": 0.4229,
+      "step": 5883
+    },
+    {
+      "epoch": 0.5014060502769493,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1408069897776857e-06,
+      "loss": 0.9829,
+      "step": 5884
+    },
+    {
+      "epoch": 0.5014912654452492,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1406891007171023e-06,
+      "loss": 0.9808,
+      "step": 5885
+    },
+    {
+      "epoch": 0.5015764806135492,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.1405711955608714e-06,
+      "loss": 0.5924,
+      "step": 5886
+    },
+    {
+      "epoch": 0.5016616957818492,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.1404532743111235e-06,
+      "loss": 0.532,
+      "step": 5887
+    },
+    {
+      "epoch": 0.5017469109501491,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.1403353369699894e-06,
+      "loss": 0.4992,
+      "step": 5888
+    },
+    {
+      "epoch": 0.5018321261184491,
+      "grad_norm": 23.5,
+      "learning_rate": 2.1402173835396006e-06,
+      "loss": 1.0601,
+      "step": 5889
+    },
+    {
+      "epoch": 0.501917341286749,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.1400994140220883e-06,
+      "loss": 0.5125,
+      "step": 5890
+    },
+    {
+      "epoch": 0.502002556455049,
+      "grad_norm": 20.25,
+      "learning_rate": 2.1399814284195848e-06,
+      "loss": 0.8205,
+      "step": 5891
+    },
+    {
+      "epoch": 0.5020877716233489,
+      "grad_norm": 12.75,
+      "learning_rate": 2.139863426734222e-06,
+      "loss": 0.535,
+      "step": 5892
+    },
+    {
+      "epoch": 0.5021729867916489,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.139745408968132e-06,
+      "loss": 0.3571,
+      "step": 5893
+    },
+    {
+      "epoch": 0.5022582019599489,
+      "grad_norm": 22.125,
+      "learning_rate": 2.1396273751234475e-06,
+      "loss": 1.0411,
+      "step": 5894
+    },
+    {
+      "epoch": 0.5023434171282488,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.1395093252023022e-06,
+      "loss": 0.9942,
+      "step": 5895
+    },
+    {
+      "epoch": 0.5024286322965488,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.1393912592068285e-06,
+      "loss": 0.8437,
+      "step": 5896
+    },
+    {
+      "epoch": 0.5025138474648487,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1392731771391607e-06,
+      "loss": 0.4644,
+      "step": 5897
+    },
+    {
+      "epoch": 0.5025990626331487,
+      "grad_norm": 17.0,
+      "learning_rate": 2.1391550790014327e-06,
+      "loss": 0.7442,
+      "step": 5898
+    },
+    {
+      "epoch": 0.5026842778014486,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.1390369647957777e-06,
+      "loss": 1.0194,
+      "step": 5899
+    },
+    {
+      "epoch": 0.5027694929697486,
+      "grad_norm": 16.0,
+      "learning_rate": 2.138918834524331e-06,
+      "loss": 0.9234,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5028547081380486,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.138800688189227e-06,
+      "loss": 0.8514,
+      "step": 5901
+    },
+    {
+      "epoch": 0.5029399233063485,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.138682525792601e-06,
+      "loss": 0.7345,
+      "step": 5902
+    },
+    {
+      "epoch": 0.5030251384746485,
+      "grad_norm": 14.125,
+      "learning_rate": 2.138564347336588e-06,
+      "loss": 0.6518,
+      "step": 5903
+    },
+    {
+      "epoch": 0.5031103536429484,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.1384461528233235e-06,
+      "loss": 0.7844,
+      "step": 5904
+    },
+    {
+      "epoch": 0.5031955688112484,
+      "grad_norm": 14.375,
+      "learning_rate": 2.138327942254944e-06,
+      "loss": 0.7696,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5032807839795483,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.138209715633585e-06,
+      "loss": 0.3982,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5033659991478483,
+      "grad_norm": 13.25,
+      "learning_rate": 2.138091472961383e-06,
+      "loss": 0.8193,
+      "step": 5907
+    },
+    {
+      "epoch": 0.5034512143161483,
+      "grad_norm": 17.5,
+      "learning_rate": 2.1379732142404758e-06,
+      "loss": 0.7489,
+      "step": 5908
+    },
+    {
+      "epoch": 0.5035364294844482,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.1378549394729993e-06,
+      "loss": 0.7697,
+      "step": 5909
+    },
+    {
+      "epoch": 0.5036216446527482,
+      "grad_norm": 17.125,
+      "learning_rate": 2.1377366486610913e-06,
+      "loss": 1.1737,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5037068598210481,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.137618341806889e-06,
+      "loss": 0.7671,
+      "step": 5911
+    },
+    {
+      "epoch": 0.5037920749893481,
+      "grad_norm": 28.125,
+      "learning_rate": 2.1375000189125316e-06,
+      "loss": 0.918,
+      "step": 5912
+    },
+    {
+      "epoch": 0.503877290157648,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.1373816799801554e-06,
+      "loss": 0.8216,
+      "step": 5913
+    },
+    {
+      "epoch": 0.503962505325948,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.1372633250119005e-06,
+      "loss": 0.4578,
+      "step": 5914
+    },
+    {
+      "epoch": 0.504047720494248,
+      "grad_norm": 21.25,
+      "learning_rate": 2.137144954009905e-06,
+      "loss": 0.7276,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5041329356625479,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.137026566976308e-06,
+      "loss": 0.6661,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5042181508308479,
+      "grad_norm": 32.25,
+      "learning_rate": 2.136908163913249e-06,
+      "loss": 1.0548,
+      "step": 5917
+    },
+    {
+      "epoch": 0.5043033659991478,
+      "grad_norm": 21.25,
+      "learning_rate": 2.1367897448228676e-06,
+      "loss": 0.9312,
+      "step": 5918
+    },
+    {
+      "epoch": 0.5043885811674478,
+      "grad_norm": 23.625,
+      "learning_rate": 2.1366713097073037e-06,
+      "loss": 1.1637,
+      "step": 5919
+    },
+    {
+      "epoch": 0.5044737963357477,
+      "grad_norm": 24.0,
+      "learning_rate": 2.136552858568698e-06,
+      "loss": 1.2734,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5045590115040477,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.13643439140919e-06,
+      "loss": 0.6037,
+      "step": 5921
+    },
+    {
+      "epoch": 0.5046442266723477,
+      "grad_norm": 12.75,
+      "learning_rate": 2.136315908230922e-06,
+      "loss": 0.4343,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5047294418406476,
+      "grad_norm": 16.25,
+      "learning_rate": 2.136197409036034e-06,
+      "loss": 0.3127,
+      "step": 5923
+    },
+    {
+      "epoch": 0.5048146570089476,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.136078893826667e-06,
+      "loss": 0.5275,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5048998721772475,
+      "grad_norm": 17.0,
+      "learning_rate": 2.135960362604964e-06,
+      "loss": 0.7268,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5049850873455475,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.135841815373066e-06,
+      "loss": 0.6535,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5050703025138474,
+      "grad_norm": 18.375,
+      "learning_rate": 2.1357232521331157e-06,
+      "loss": 1.2585,
+      "step": 5927
+    },
+    {
+      "epoch": 0.5051555176821474,
+      "grad_norm": 14.125,
+      "learning_rate": 2.135604672887256e-06,
+      "loss": 0.7749,
+      "step": 5928
+    },
+    {
+      "epoch": 0.5052407328504473,
+      "grad_norm": 12.25,
+      "learning_rate": 2.135486077637629e-06,
+      "loss": 0.5934,
+      "step": 5929
+    },
+    {
+      "epoch": 0.5053259480187473,
+      "grad_norm": 9.875,
+      "learning_rate": 2.135367466386378e-06,
+      "loss": 0.3538,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5054111631870473,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.135248839135647e-06,
+      "loss": 0.6921,
+      "step": 5931
+    },
+    {
+      "epoch": 0.5054963783553472,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.1351301958875788e-06,
+      "loss": 0.7804,
+      "step": 5932
+    },
+    {
+      "epoch": 0.5055815935236472,
+      "grad_norm": 11.875,
+      "learning_rate": 2.135011536644318e-06,
+      "loss": 0.6082,
+      "step": 5933
+    },
+    {
+      "epoch": 0.5056668086919471,
+      "grad_norm": 11.625,
+      "learning_rate": 2.1348928614080093e-06,
+      "loss": 0.3377,
+      "step": 5934
+    },
+    {
+      "epoch": 0.5057520238602471,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.134774170180796e-06,
+      "loss": 0.9423,
+      "step": 5935
+    },
+    {
+      "epoch": 0.505837239028547,
+      "grad_norm": 23.125,
+      "learning_rate": 2.134655462964824e-06,
+      "loss": 0.7924,
+      "step": 5936
+    },
+    {
+      "epoch": 0.505922454196847,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1345367397622386e-06,
+      "loss": 0.7311,
+      "step": 5937
+    },
+    {
+      "epoch": 0.506007669365147,
+      "grad_norm": 16.625,
+      "learning_rate": 2.134418000575185e-06,
+      "loss": 1.3445,
+      "step": 5938
+    },
+    {
+      "epoch": 0.5060928845334469,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.1342992454058077e-06,
+      "loss": 0.4964,
+      "step": 5939
+    },
+    {
+      "epoch": 0.5061780997017469,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.1341804742562545e-06,
+      "loss": 0.9353,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5062633148700468,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.134061687128671e-06,
+      "loss": 0.8831,
+      "step": 5941
+    },
+    {
+      "epoch": 0.5063485300383468,
+      "grad_norm": 11.0,
+      "learning_rate": 2.1339428840252036e-06,
+      "loss": 0.595,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5064337452066467,
+      "grad_norm": 17.125,
+      "learning_rate": 2.1338240649479995e-06,
+      "loss": 0.6384,
+      "step": 5943
+    },
+    {
+      "epoch": 0.5065189603749467,
+      "grad_norm": 18.125,
+      "learning_rate": 2.133705229899206e-06,
+      "loss": 1.064,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5066041755432467,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1335863788809703e-06,
+      "loss": 0.9516,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5066893907115466,
+      "grad_norm": 16.375,
+      "learning_rate": 2.13346751189544e-06,
+      "loss": 0.5388,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5067746058798466,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1333486289447634e-06,
+      "loss": 0.3838,
+      "step": 5947
+    },
+    {
+      "epoch": 0.5068598210481465,
+      "grad_norm": 15.0,
+      "learning_rate": 2.133229730031089e-06,
+      "loss": 0.6363,
+      "step": 5948
+    },
+    {
+      "epoch": 0.5069450362164465,
+      "grad_norm": 34.25,
+      "learning_rate": 2.1331108151565655e-06,
+      "loss": 1.8226,
+      "step": 5949
+    },
+    {
+      "epoch": 0.5070302513847464,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1329918843233412e-06,
+      "loss": 0.7174,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5071154665530464,
+      "grad_norm": 23.875,
+      "learning_rate": 2.1328729375335655e-06,
+      "loss": 0.849,
+      "step": 5951
+    },
+    {
+      "epoch": 0.5072006817213464,
+      "grad_norm": 33.5,
+      "learning_rate": 2.1327539747893875e-06,
+      "loss": 0.9369,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5072858968896463,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.132634996092959e-06,
+      "loss": 0.5689,
+      "step": 5953
+    },
+    {
+      "epoch": 0.5073711120579463,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.1325160014464273e-06,
+      "loss": 0.3275,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5074563272262462,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1323969908519445e-06,
+      "loss": 0.7506,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5075415423945462,
+      "grad_norm": 11.125,
+      "learning_rate": 2.1322779643116603e-06,
+      "loss": 0.5071,
+      "step": 5956
+    },
+    {
+      "epoch": 0.5076267575628461,
+      "grad_norm": 20.375,
+      "learning_rate": 2.1321589218277266e-06,
+      "loss": 0.6119,
+      "step": 5957
+    },
+    {
+      "epoch": 0.5077119727311461,
+      "grad_norm": 13.75,
+      "learning_rate": 2.132039863402294e-06,
+      "loss": 0.912,
+      "step": 5958
+    },
+    {
+      "epoch": 0.5077971878994461,
+      "grad_norm": 12.25,
+      "learning_rate": 2.131920789037514e-06,
+      "loss": 0.5427,
+      "step": 5959
+    },
+    {
+      "epoch": 0.507882403067746,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.1318016987355385e-06,
+      "loss": 0.5217,
+      "step": 5960
+    },
+    {
+      "epoch": 0.507967618236046,
+      "grad_norm": 19.75,
+      "learning_rate": 2.1316825924985198e-06,
+      "loss": 0.5625,
+      "step": 5961
+    },
+    {
+      "epoch": 0.5080528334043459,
+      "grad_norm": 11.0,
+      "learning_rate": 2.1315634703286097e-06,
+      "loss": 0.565,
+      "step": 5962
+    },
+    {
+      "epoch": 0.5081380485726459,
+      "grad_norm": 13.375,
+      "learning_rate": 2.1314443322279615e-06,
+      "loss": 0.7379,
+      "step": 5963
+    },
+    {
+      "epoch": 0.5082232637409458,
+      "grad_norm": 16.125,
+      "learning_rate": 2.131325178198728e-06,
+      "loss": 0.6521,
+      "step": 5964
+    },
+    {
+      "epoch": 0.5083084789092458,
+      "grad_norm": 19.0,
+      "learning_rate": 2.131206008243062e-06,
+      "loss": 1.0253,
+      "step": 5965
+    },
+    {
+      "epoch": 0.5083936940775458,
+      "grad_norm": 14.875,
+      "learning_rate": 2.131086822363117e-06,
+      "loss": 0.4935,
+      "step": 5966
+    },
+    {
+      "epoch": 0.5084789092458457,
+      "grad_norm": 29.625,
+      "learning_rate": 2.1309676205610474e-06,
+      "loss": 1.0021,
+      "step": 5967
+    },
+    {
+      "epoch": 0.5085641244141457,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.1308484028390073e-06,
+      "loss": 0.7737,
+      "step": 5968
+    },
+    {
+      "epoch": 0.5086493395824456,
+      "grad_norm": 15.375,
+      "learning_rate": 2.1307291691991506e-06,
+      "loss": 0.7197,
+      "step": 5969
+    },
+    {
+      "epoch": 0.5087345547507456,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.130609919643632e-06,
+      "loss": 0.6636,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5088197699190455,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.130490654174607e-06,
+      "loss": 0.5925,
+      "step": 5971
+    },
+    {
+      "epoch": 0.5089049850873455,
+      "grad_norm": 13.75,
+      "learning_rate": 2.13037137279423e-06,
+      "loss": 0.9755,
+      "step": 5972
+    },
+    {
+      "epoch": 0.5089902002556455,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1302520755046572e-06,
+      "loss": 0.7805,
+      "step": 5973
+    },
+    {
+      "epoch": 0.5090754154239454,
+      "grad_norm": 15.5,
+      "learning_rate": 2.130132762308045e-06,
+      "loss": 0.9451,
+      "step": 5974
+    },
+    {
+      "epoch": 0.5091606305922454,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1300134332065476e-06,
+      "loss": 0.5044,
+      "step": 5975
+    },
+    {
+      "epoch": 0.5092458457605453,
+      "grad_norm": 21.125,
+      "learning_rate": 2.129894088202323e-06,
+      "loss": 1.1951,
+      "step": 5976
+    },
+    {
+      "epoch": 0.5093310609288453,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.1297747272975274e-06,
+      "loss": 0.4616,
+      "step": 5977
+    },
+    {
+      "epoch": 0.5094162760971452,
+      "grad_norm": 16.5,
+      "learning_rate": 2.129655350494318e-06,
+      "loss": 1.068,
+      "step": 5978
+    },
+    {
+      "epoch": 0.5095014912654452,
+      "grad_norm": 12.75,
+      "learning_rate": 2.1295359577948515e-06,
+      "loss": 0.4224,
+      "step": 5979
+    },
+    {
+      "epoch": 0.5095867064337452,
+      "grad_norm": 14.375,
+      "learning_rate": 2.1294165492012856e-06,
+      "loss": 0.9224,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5096719216020452,
+      "grad_norm": 10.875,
+      "learning_rate": 2.129297124715779e-06,
+      "loss": 0.4246,
+      "step": 5981
+    },
+    {
+      "epoch": 0.5097571367703452,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.129177684340489e-06,
+      "loss": 0.7747,
+      "step": 5982
+    },
+    {
+      "epoch": 0.5098423519386451,
+      "grad_norm": 16.0,
+      "learning_rate": 2.129058228077574e-06,
+      "loss": 0.7793,
+      "step": 5983
+    },
+    {
+      "epoch": 0.5099275671069451,
+      "grad_norm": 18.375,
+      "learning_rate": 2.1289387559291927e-06,
+      "loss": 0.492,
+      "step": 5984
+    },
+    {
+      "epoch": 0.510012782275245,
+      "grad_norm": 21.875,
+      "learning_rate": 2.1288192678975046e-06,
+      "loss": 0.8946,
+      "step": 5985
+    },
+    {
+      "epoch": 0.510097997443545,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1286997639846683e-06,
+      "loss": 0.7928,
+      "step": 5986
+    },
+    {
+      "epoch": 0.510183212611845,
+      "grad_norm": 10.125,
+      "learning_rate": 2.1285802441928437e-06,
+      "loss": 0.4138,
+      "step": 5987
+    },
+    {
+      "epoch": 0.5102684277801449,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.128460708524191e-06,
+      "loss": 0.8671,
+      "step": 5988
+    },
+    {
+      "epoch": 0.5103536429484449,
+      "grad_norm": 16.375,
+      "learning_rate": 2.1283411569808695e-06,
+      "loss": 0.9474,
+      "step": 5989
+    },
+    {
+      "epoch": 0.5104388581167448,
+      "grad_norm": 22.125,
+      "learning_rate": 2.1282215895650405e-06,
+      "loss": 0.9463,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5105240732850448,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.128102006278864e-06,
+      "loss": 1.1134,
+      "step": 5991
+    },
+    {
+      "epoch": 0.5106092884533447,
+      "grad_norm": 19.5,
+      "learning_rate": 2.1279824071245007e-06,
+      "loss": 0.8526,
+      "step": 5992
+    },
+    {
+      "epoch": 0.5106945036216447,
+      "grad_norm": 20.125,
+      "learning_rate": 2.1278627921041128e-06,
+      "loss": 0.8309,
+      "step": 5993
+    },
+    {
+      "epoch": 0.5107797187899447,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.127743161219862e-06,
+      "loss": 0.3082,
+      "step": 5994
+    },
+    {
+      "epoch": 0.5108649339582446,
+      "grad_norm": 11.125,
+      "learning_rate": 2.1276235144739087e-06,
+      "loss": 0.3851,
+      "step": 5995
+    },
+    {
+      "epoch": 0.5109501491265446,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.1275038518684168e-06,
+      "loss": 0.6475,
+      "step": 5996
+    },
+    {
+      "epoch": 0.5110353642948445,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.1273841734055478e-06,
+      "loss": 0.6217,
+      "step": 5997
+    },
+    {
+      "epoch": 0.5111205794631445,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.127264479087464e-06,
+      "loss": 0.7222,
+      "step": 5998
+    },
+    {
+      "epoch": 0.5112057946314444,
+      "grad_norm": 14.5,
+      "learning_rate": 2.1271447689163293e-06,
+      "loss": 0.5706,
+      "step": 5999
+    },
+    {
+      "epoch": 0.5112910097997444,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.1270250428943064e-06,
+      "loss": 0.5834,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5113762249680444,
+      "grad_norm": 18.0,
+      "learning_rate": 2.126905301023559e-06,
+      "loss": 0.9105,
+      "step": 6001
+    },
+    {
+      "epoch": 0.5114614401363443,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.126785543306252e-06,
+      "loss": 0.8711,
+      "step": 6002
+    },
+    {
+      "epoch": 0.5115466553046443,
+      "grad_norm": 15.875,
+      "learning_rate": 2.126665769744547e-06,
+      "loss": 0.7354,
+      "step": 6003
+    },
+    {
+      "epoch": 0.5116318704729442,
+      "grad_norm": 15.75,
+      "learning_rate": 2.126545980340611e-06,
+      "loss": 0.615,
+      "step": 6004
+    },
+    {
+      "epoch": 0.5117170856412442,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.1264261750966073e-06,
+      "loss": 0.5414,
+      "step": 6005
+    },
+    {
+      "epoch": 0.5118023008095441,
+      "grad_norm": 19.625,
+      "learning_rate": 2.1263063540147018e-06,
+      "loss": 0.7888,
+      "step": 6006
+    },
+    {
+      "epoch": 0.5118875159778441,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.1261865170970587e-06,
+      "loss": 0.5355,
+      "step": 6007
+    },
+    {
+      "epoch": 0.5119727311461441,
+      "grad_norm": 13.25,
+      "learning_rate": 2.1260666643458445e-06,
+      "loss": 0.552,
+      "step": 6008
+    },
+    {
+      "epoch": 0.512057946314444,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1259467957632244e-06,
+      "loss": 0.9799,
+      "step": 6009
+    },
+    {
+      "epoch": 0.512143161482744,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.1258269113513653e-06,
+      "loss": 0.6855,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5122283766510439,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.125707011112433e-06,
+      "loss": 0.5838,
+      "step": 6011
+    },
+    {
+      "epoch": 0.5123135918193439,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.125587095048594e-06,
+      "loss": 0.6412,
+      "step": 6012
+    },
+    {
+      "epoch": 0.5123988069876438,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.125467163162016e-06,
+      "loss": 0.5421,
+      "step": 6013
+    },
+    {
+      "epoch": 0.5124840221559438,
+      "grad_norm": 19.875,
+      "learning_rate": 2.1253472154548665e-06,
+      "loss": 0.6201,
+      "step": 6014
+    },
+    {
+      "epoch": 0.5125692373242438,
+      "grad_norm": 15.0,
+      "learning_rate": 2.1252272519293123e-06,
+      "loss": 0.6306,
+      "step": 6015
+    },
+    {
+      "epoch": 0.5126544524925437,
+      "grad_norm": 15.5,
+      "learning_rate": 2.125107272587521e-06,
+      "loss": 0.4924,
+      "step": 6016
+    },
+    {
+      "epoch": 0.5127396676608437,
+      "grad_norm": 21.25,
+      "learning_rate": 2.1249872774316623e-06,
+      "loss": 0.8329,
+      "step": 6017
+    },
+    {
+      "epoch": 0.5128248828291436,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.124867266463903e-06,
+      "loss": 0.7385,
+      "step": 6018
+    },
+    {
+      "epoch": 0.5129100979974436,
+      "grad_norm": 14.0,
+      "learning_rate": 2.1247472396864123e-06,
+      "loss": 0.7668,
+      "step": 6019
+    },
+    {
+      "epoch": 0.5129953131657435,
+      "grad_norm": 14.875,
+      "learning_rate": 2.1246271971013596e-06,
+      "loss": 0.9232,
+      "step": 6020
+    },
+    {
+      "epoch": 0.5130805283340435,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.124507138710914e-06,
+      "loss": 1.0421,
+      "step": 6021
+    },
+    {
+      "epoch": 0.5131657435023435,
+      "grad_norm": 19.125,
+      "learning_rate": 2.124387064517245e-06,
+      "loss": 0.8987,
+      "step": 6022
+    },
+    {
+      "epoch": 0.5132509586706434,
+      "grad_norm": 17.0,
+      "learning_rate": 2.1242669745225223e-06,
+      "loss": 0.9868,
+      "step": 6023
+    },
+    {
+      "epoch": 0.5133361738389434,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.124146868728916e-06,
+      "loss": 0.7766,
+      "step": 6024
+    },
+    {
+      "epoch": 0.5134213890072433,
+      "grad_norm": 16.0,
+      "learning_rate": 2.124026747138597e-06,
+      "loss": 0.9565,
+      "step": 6025
+    },
+    {
+      "epoch": 0.5135066041755433,
+      "grad_norm": 16.375,
+      "learning_rate": 2.1239066097537363e-06,
+      "loss": 0.9806,
+      "step": 6026
+    },
+    {
+      "epoch": 0.5135918193438432,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.123786456576504e-06,
+      "loss": 0.3605,
+      "step": 6027
+    },
+    {
+      "epoch": 0.5136770345121432,
+      "grad_norm": 19.25,
+      "learning_rate": 2.123666287609072e-06,
+      "loss": 0.8089,
+      "step": 6028
+    },
+    {
+      "epoch": 0.5137622496804432,
+      "grad_norm": 17.875,
+      "learning_rate": 2.1235461028536112e-06,
+      "loss": 0.8724,
+      "step": 6029
+    },
+    {
+      "epoch": 0.5138474648487431,
+      "grad_norm": 15.625,
+      "learning_rate": 2.123425902312294e-06,
+      "loss": 0.736,
+      "step": 6030
+    },
+    {
+      "epoch": 0.5139326800170431,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1233056859872925e-06,
+      "loss": 0.8992,
+      "step": 6031
+    },
+    {
+      "epoch": 0.514017895185343,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.123185453880779e-06,
+      "loss": 1.0508,
+      "step": 6032
+    },
+    {
+      "epoch": 0.514103110353643,
+      "grad_norm": 11.625,
+      "learning_rate": 2.1230652059949265e-06,
+      "loss": 0.5298,
+      "step": 6033
+    },
+    {
+      "epoch": 0.5141883255219429,
+      "grad_norm": 11.625,
+      "learning_rate": 2.1229449423319075e-06,
+      "loss": 0.7364,
+      "step": 6034
+    },
+    {
+      "epoch": 0.5142735406902429,
+      "grad_norm": 17.5,
+      "learning_rate": 2.122824662893896e-06,
+      "loss": 0.864,
+      "step": 6035
+    },
+    {
+      "epoch": 0.5143587558585428,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.122704367683065e-06,
+      "loss": 0.9077,
+      "step": 6036
+    },
+    {
+      "epoch": 0.5144439710268428,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.122584056701588e-06,
+      "loss": 0.8837,
+      "step": 6037
+    },
+    {
+      "epoch": 0.5145291861951428,
+      "grad_norm": 23.0,
+      "learning_rate": 2.1224637299516403e-06,
+      "loss": 1.0585,
+      "step": 6038
+    },
+    {
+      "epoch": 0.5146144013634427,
+      "grad_norm": 17.0,
+      "learning_rate": 2.1223433874353956e-06,
+      "loss": 1.0097,
+      "step": 6039
+    },
+    {
+      "epoch": 0.5146996165317427,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1222230291550284e-06,
+      "loss": 0.9164,
+      "step": 6040
+    },
+    {
+      "epoch": 0.5147848317000426,
+      "grad_norm": 16.125,
+      "learning_rate": 2.122102655112714e-06,
+      "loss": 0.6209,
+      "step": 6041
+    },
+    {
+      "epoch": 0.5148700468683426,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.121982265310628e-06,
+      "loss": 0.6848,
+      "step": 6042
+    },
+    {
+      "epoch": 0.5149552620366425,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.1218618597509453e-06,
+      "loss": 0.503,
+      "step": 6043
+    },
+    {
+      "epoch": 0.5150404772049425,
+      "grad_norm": 15.125,
+      "learning_rate": 2.121741438435842e-06,
+      "loss": 0.8655,
+      "step": 6044
+    },
+    {
+      "epoch": 0.5151256923732425,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.121621001367494e-06,
+      "loss": 0.9471,
+      "step": 6045
+    },
+    {
+      "epoch": 0.5152109075415424,
+      "grad_norm": 11.25,
+      "learning_rate": 2.121500548548079e-06,
+      "loss": 0.4627,
+      "step": 6046
+    },
+    {
+      "epoch": 0.5152961227098424,
+      "grad_norm": 21.125,
+      "learning_rate": 2.121380079979772e-06,
+      "loss": 0.9475,
+      "step": 6047
+    },
+    {
+      "epoch": 0.5153813378781423,
+      "grad_norm": 16.125,
+      "learning_rate": 2.121259595664751e-06,
+      "loss": 0.8405,
+      "step": 6048
+    },
+    {
+      "epoch": 0.5154665530464423,
+      "grad_norm": 19.625,
+      "learning_rate": 2.1211390956051927e-06,
+      "loss": 1.0412,
+      "step": 6049
+    },
+    {
+      "epoch": 0.5155517682147422,
+      "grad_norm": 16.625,
+      "learning_rate": 2.121018579803275e-06,
+      "loss": 0.9891,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5156369833830422,
+      "grad_norm": 14.25,
+      "learning_rate": 2.1208980482611763e-06,
+      "loss": 0.6184,
+      "step": 6051
+    },
+    {
+      "epoch": 0.5157221985513422,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1207775009810736e-06,
+      "loss": 1.0893,
+      "step": 6052
+    },
+    {
+      "epoch": 0.5158074137196421,
+      "grad_norm": 14.0,
+      "learning_rate": 2.1206569379651456e-06,
+      "loss": 0.6063,
+      "step": 6053
+    },
+    {
+      "epoch": 0.5158926288879421,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1205363592155713e-06,
+      "loss": 1.0436,
+      "step": 6054
+    },
+    {
+      "epoch": 0.515977844056242,
+      "grad_norm": 16.625,
+      "learning_rate": 2.12041576473453e-06,
+      "loss": 0.7825,
+      "step": 6055
+    },
+    {
+      "epoch": 0.516063059224542,
+      "grad_norm": 18.125,
+      "learning_rate": 2.1202951545242006e-06,
+      "loss": 0.8175,
+      "step": 6056
+    },
+    {
+      "epoch": 0.5161482743928419,
+      "grad_norm": 12.75,
+      "learning_rate": 2.120174528586762e-06,
+      "loss": 0.3533,
+      "step": 6057
+    },
+    {
+      "epoch": 0.5162334895611419,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.120053886924395e-06,
+      "loss": 0.498,
+      "step": 6058
+    },
+    {
+      "epoch": 0.5163187047294419,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1199332295392796e-06,
+      "loss": 1.0237,
+      "step": 6059
+    },
+    {
+      "epoch": 0.5164039198977418,
+      "grad_norm": 25.125,
+      "learning_rate": 2.1198125564335956e-06,
+      "loss": 1.0128,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5164891350660418,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.119691867609524e-06,
+      "loss": 0.6675,
+      "step": 6061
+    },
+    {
+      "epoch": 0.5165743502343417,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1195711630692463e-06,
+      "loss": 0.7392,
+      "step": 6062
+    },
+    {
+      "epoch": 0.5166595654026417,
+      "grad_norm": 17.625,
+      "learning_rate": 2.1194504428149425e-06,
+      "loss": 0.7134,
+      "step": 6063
+    },
+    {
+      "epoch": 0.5167447805709416,
+      "grad_norm": 22.25,
+      "learning_rate": 2.1193297068487958e-06,
+      "loss": 0.8278,
+      "step": 6064
+    },
+    {
+      "epoch": 0.5168299957392416,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.119208955172986e-06,
+      "loss": 0.7356,
+      "step": 6065
+    },
+    {
+      "epoch": 0.5169152109075416,
+      "grad_norm": 25.375,
+      "learning_rate": 2.119088187789697e-06,
+      "loss": 1.1296,
+      "step": 6066
+    },
+    {
+      "epoch": 0.5170004260758415,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.1189674047011106e-06,
+      "loss": 0.3042,
+      "step": 6067
+    },
+    {
+      "epoch": 0.5170856412441415,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.1188466059094092e-06,
+      "loss": 0.4142,
+      "step": 6068
+    },
+    {
+      "epoch": 0.5171708564124414,
+      "grad_norm": 18.375,
+      "learning_rate": 2.118725791416776e-06,
+      "loss": 0.8698,
+      "step": 6069
+    },
+    {
+      "epoch": 0.5172560715807414,
+      "grad_norm": 17.125,
+      "learning_rate": 2.118604961225394e-06,
+      "loss": 0.7608,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5173412867490413,
+      "grad_norm": 17.0,
+      "learning_rate": 2.1184841153374475e-06,
+      "loss": 0.8325,
+      "step": 6071
+    },
+    {
+      "epoch": 0.5174265019173413,
+      "grad_norm": 13.625,
+      "learning_rate": 2.118363253755119e-06,
+      "loss": 0.4737,
+      "step": 6072
+    },
+    {
+      "epoch": 0.5175117170856413,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.1182423764805937e-06,
+      "loss": 0.8842,
+      "step": 6073
+    },
+    {
+      "epoch": 0.5175969322539412,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.118121483516055e-06,
+      "loss": 0.6919,
+      "step": 6074
+    },
+    {
+      "epoch": 0.5176821474222412,
+      "grad_norm": 14.0,
+      "learning_rate": 2.118000574863689e-06,
+      "loss": 0.7782,
+      "step": 6075
+    },
+    {
+      "epoch": 0.5177673625905411,
+      "grad_norm": 20.375,
+      "learning_rate": 2.1178796505256794e-06,
+      "loss": 1.2393,
+      "step": 6076
+    },
+    {
+      "epoch": 0.5178525777588411,
+      "grad_norm": 13.0,
+      "learning_rate": 2.1177587105042114e-06,
+      "loss": 0.6245,
+      "step": 6077
+    },
+    {
+      "epoch": 0.517937792927141,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1176377548014717e-06,
+      "loss": 0.6985,
+      "step": 6078
+    },
+    {
+      "epoch": 0.518023008095441,
+      "grad_norm": 10.875,
+      "learning_rate": 2.1175167834196443e-06,
+      "loss": 0.2675,
+      "step": 6079
+    },
+    {
+      "epoch": 0.518108223263741,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.117395796360917e-06,
+      "loss": 0.4387,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5181934384320409,
+      "grad_norm": 16.125,
+      "learning_rate": 2.117274793627475e-06,
+      "loss": 0.7577,
+      "step": 6081
+    },
+    {
+      "epoch": 0.5182786536003409,
+      "grad_norm": 23.375,
+      "learning_rate": 2.117153775221506e-06,
+      "loss": 1.2516,
+      "step": 6082
+    },
+    {
+      "epoch": 0.5183638687686408,
+      "grad_norm": 16.625,
+      "learning_rate": 2.1170327411451957e-06,
+      "loss": 0.7569,
+      "step": 6083
+    },
+    {
+      "epoch": 0.5184490839369408,
+      "grad_norm": 15.625,
+      "learning_rate": 2.116911691400732e-06,
+      "loss": 0.8209,
+      "step": 6084
+    },
+    {
+      "epoch": 0.5185342991052407,
+      "grad_norm": 29.75,
+      "learning_rate": 2.116790625990302e-06,
+      "loss": 0.9579,
+      "step": 6085
+    },
+    {
+      "epoch": 0.5186195142735407,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1166695449160946e-06,
+      "loss": 0.7305,
+      "step": 6086
+    },
+    {
+      "epoch": 0.5187047294418407,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.116548448180297e-06,
+      "loss": 0.4928,
+      "step": 6087
+    },
+    {
+      "epoch": 0.5187899446101406,
+      "grad_norm": 14.75,
+      "learning_rate": 2.1164273357850965e-06,
+      "loss": 0.6753,
+      "step": 6088
+    },
+    {
+      "epoch": 0.5188751597784406,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.1163062077326836e-06,
+      "loss": 0.6788,
+      "step": 6089
+    },
+    {
+      "epoch": 0.5189603749467405,
+      "grad_norm": 20.875,
+      "learning_rate": 2.116185064025246e-06,
+      "loss": 1.0482,
+      "step": 6090
+    },
+    {
+      "epoch": 0.5190455901150405,
+      "grad_norm": 18.75,
+      "learning_rate": 2.1160639046649733e-06,
+      "loss": 0.6192,
+      "step": 6091
+    },
+    {
+      "epoch": 0.5191308052833404,
+      "grad_norm": 17.75,
+      "learning_rate": 2.1159427296540554e-06,
+      "loss": 1.3317,
+      "step": 6092
+    },
+    {
+      "epoch": 0.5192160204516404,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.115821538994681e-06,
+      "loss": 0.4594,
+      "step": 6093
+    },
+    {
+      "epoch": 0.5193012356199403,
+      "grad_norm": 17.625,
+      "learning_rate": 2.1157003326890413e-06,
+      "loss": 0.6357,
+      "step": 6094
+    },
+    {
+      "epoch": 0.5193864507882403,
+      "grad_norm": 10.875,
+      "learning_rate": 2.115579110739326e-06,
+      "loss": 0.439,
+      "step": 6095
+    },
+    {
+      "epoch": 0.5194716659565403,
+      "grad_norm": 16.25,
+      "learning_rate": 2.115457873147725e-06,
+      "loss": 0.8373,
+      "step": 6096
+    },
+    {
+      "epoch": 0.5195568811248402,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1153366199164306e-06,
+      "loss": 0.8145,
+      "step": 6097
+    },
+    {
+      "epoch": 0.5196420962931402,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.1152153510476333e-06,
+      "loss": 1.0841,
+      "step": 6098
+    },
+    {
+      "epoch": 0.5197273114614401,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1150940665435247e-06,
+      "loss": 0.5388,
+      "step": 6099
+    },
+    {
+      "epoch": 0.5198125266297401,
+      "grad_norm": 16.125,
+      "learning_rate": 2.114972766406296e-06,
+      "loss": 0.7404,
+      "step": 6100
+    },
+    {
+      "epoch": 0.51989774179804,
+      "grad_norm": 17.125,
+      "learning_rate": 2.11485145063814e-06,
+      "loss": 0.9856,
+      "step": 6101
+    },
+    {
+      "epoch": 0.51998295696634,
+      "grad_norm": 17.5,
+      "learning_rate": 2.1147301192412485e-06,
+      "loss": 0.845,
+      "step": 6102
+    },
+    {
+      "epoch": 0.52006817213464,
+      "grad_norm": 18.5,
+      "learning_rate": 2.1146087722178138e-06,
+      "loss": 0.9277,
+      "step": 6103
+    },
+    {
+      "epoch": 0.5201533873029399,
+      "grad_norm": 18.0,
+      "learning_rate": 2.1144874095700297e-06,
+      "loss": 0.7724,
+      "step": 6104
+    },
+    {
+      "epoch": 0.5202386024712399,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.1143660313000887e-06,
+      "loss": 0.8382,
+      "step": 6105
+    },
+    {
+      "epoch": 0.5203238176395398,
+      "grad_norm": 15.75,
+      "learning_rate": 2.114244637410184e-06,
+      "loss": 0.7275,
+      "step": 6106
+    },
+    {
+      "epoch": 0.5204090328078398,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.1141232279025097e-06,
+      "loss": 0.5651,
+      "step": 6107
+    },
+    {
+      "epoch": 0.5204942479761397,
+      "grad_norm": 17.625,
+      "learning_rate": 2.1140018027792597e-06,
+      "loss": 0.6045,
+      "step": 6108
+    },
+    {
+      "epoch": 0.5205794631444397,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.1138803620426285e-06,
+      "loss": 0.5992,
+      "step": 6109
+    },
+    {
+      "epoch": 0.5206646783127397,
+      "grad_norm": 15.875,
+      "learning_rate": 2.1137589056948104e-06,
+      "loss": 0.5883,
+      "step": 6110
+    },
+    {
+      "epoch": 0.5207498934810396,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.1136374337380005e-06,
+      "loss": 1.1205,
+      "step": 6111
+    },
+    {
+      "epoch": 0.5208351086493396,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.1135159461743937e-06,
+      "loss": 0.863,
+      "step": 6112
+    },
+    {
+      "epoch": 0.5209203238176395,
+      "grad_norm": 14.875,
+      "learning_rate": 2.1133944430061853e-06,
+      "loss": 0.6164,
+      "step": 6113
+    },
+    {
+      "epoch": 0.5210055389859395,
+      "grad_norm": 17.25,
+      "learning_rate": 2.1132729242355707e-06,
+      "loss": 0.672,
+      "step": 6114
+    },
+    {
+      "epoch": 0.5210907541542394,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1131513898647466e-06,
+      "loss": 0.7632,
+      "step": 6115
+    },
+    {
+      "epoch": 0.5211759693225394,
+      "grad_norm": 15.125,
+      "learning_rate": 2.113029839895909e-06,
+      "loss": 0.8079,
+      "step": 6116
+    },
+    {
+      "epoch": 0.5212611844908394,
+      "grad_norm": 15.125,
+      "learning_rate": 2.112908274331254e-06,
+      "loss": 0.6959,
+      "step": 6117
+    },
+    {
+      "epoch": 0.5213463996591393,
+      "grad_norm": 19.0,
+      "learning_rate": 2.112786693172979e-06,
+      "loss": 0.6458,
+      "step": 6118
+    },
+    {
+      "epoch": 0.5214316148274393,
+      "grad_norm": 13.75,
+      "learning_rate": 2.112665096423281e-06,
+      "loss": 0.4932,
+      "step": 6119
+    },
+    {
+      "epoch": 0.5215168299957392,
+      "grad_norm": 22.625,
+      "learning_rate": 2.112543484084357e-06,
+      "loss": 1.2836,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5216020451640392,
+      "grad_norm": 19.0,
+      "learning_rate": 2.1124218561584053e-06,
+      "loss": 1.1012,
+      "step": 6121
+    },
+    {
+      "epoch": 0.5216872603323391,
+      "grad_norm": 17.25,
+      "learning_rate": 2.112300212647623e-06,
+      "loss": 1.1612,
+      "step": 6122
+    },
+    {
+      "epoch": 0.5217724755006391,
+      "grad_norm": 21.375,
+      "learning_rate": 2.1121785535542093e-06,
+      "loss": 0.5436,
+      "step": 6123
+    },
+    {
+      "epoch": 0.5218576906689391,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1120568788803617e-06,
+      "loss": 0.5055,
+      "step": 6124
+    },
+    {
+      "epoch": 0.521942905837239,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.1119351886282797e-06,
+      "loss": 0.6316,
+      "step": 6125
+    },
+    {
+      "epoch": 0.522028121005539,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1118134828001618e-06,
+      "loss": 0.7559,
+      "step": 6126
+    },
+    {
+      "epoch": 0.5221133361738389,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1116917613982072e-06,
+      "loss": 0.568,
+      "step": 6127
+    },
+    {
+      "epoch": 0.5221985513421389,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.1115700244246167e-06,
+      "loss": 0.6198,
+      "step": 6128
+    },
+    {
+      "epoch": 0.5222837665104388,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1114482718815895e-06,
+      "loss": 0.5334,
+      "step": 6129
+    },
+    {
+      "epoch": 0.5223689816787388,
+      "grad_norm": 23.75,
+      "learning_rate": 2.1113265037713253e-06,
+      "loss": 0.7541,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5224541968470388,
+      "grad_norm": 16.625,
+      "learning_rate": 2.1112047200960256e-06,
+      "loss": 0.7181,
+      "step": 6131
+    },
+    {
+      "epoch": 0.5225394120153387,
+      "grad_norm": 16.625,
+      "learning_rate": 2.1110829208578905e-06,
+      "loss": 0.7677,
+      "step": 6132
+    },
+    {
+      "epoch": 0.5226246271836387,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1109611060591207e-06,
+      "loss": 0.7057,
+      "step": 6133
+    },
+    {
+      "epoch": 0.5227098423519386,
+      "grad_norm": 16.625,
+      "learning_rate": 2.110839275701918e-06,
+      "loss": 0.726,
+      "step": 6134
+    },
+    {
+      "epoch": 0.5227950575202386,
+      "grad_norm": 12.625,
+      "learning_rate": 2.1107174297884843e-06,
+      "loss": 0.4921,
+      "step": 6135
+    },
+    {
+      "epoch": 0.5228802726885385,
+      "grad_norm": 14.0,
+      "learning_rate": 2.110595568321021e-06,
+      "loss": 0.7788,
+      "step": 6136
+    },
+    {
+      "epoch": 0.5229654878568385,
+      "grad_norm": 17.75,
+      "learning_rate": 2.1104736913017304e-06,
+      "loss": 1.323,
+      "step": 6137
+    },
+    {
+      "epoch": 0.5230507030251385,
+      "grad_norm": 20.875,
+      "learning_rate": 2.110351798732815e-06,
+      "loss": 0.7678,
+      "step": 6138
+    },
+    {
+      "epoch": 0.5231359181934384,
+      "grad_norm": 19.75,
+      "learning_rate": 2.110229890616477e-06,
+      "loss": 1.2404,
+      "step": 6139
+    },
+    {
+      "epoch": 0.5232211333617384,
+      "grad_norm": 20.0,
+      "learning_rate": 2.11010796695492e-06,
+      "loss": 0.6425,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5233063485300383,
+      "grad_norm": 14.5,
+      "learning_rate": 2.1099860277503475e-06,
+      "loss": 0.7397,
+      "step": 6141
+    },
+    {
+      "epoch": 0.5233915636983383,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1098640730049623e-06,
+      "loss": 0.7646,
+      "step": 6142
+    },
+    {
+      "epoch": 0.5234767788666382,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.109742102720969e-06,
+      "loss": 0.4993,
+      "step": 6143
+    },
+    {
+      "epoch": 0.5235619940349382,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1096201169005716e-06,
+      "loss": 0.614,
+      "step": 6144
+    },
+    {
+      "epoch": 0.5236472092032382,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.1094981155459736e-06,
+      "loss": 0.5377,
+      "step": 6145
+    },
+    {
+      "epoch": 0.5237324243715381,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1093760986593804e-06,
+      "loss": 0.8189,
+      "step": 6146
+    },
+    {
+      "epoch": 0.5238176395398381,
+      "grad_norm": 14.75,
+      "learning_rate": 2.109254066242997e-06,
+      "loss": 0.7903,
+      "step": 6147
+    },
+    {
+      "epoch": 0.523902854708138,
+      "grad_norm": 15.625,
+      "learning_rate": 2.109132018299029e-06,
+      "loss": 0.8813,
+      "step": 6148
+    },
+    {
+      "epoch": 0.523988069876438,
+      "grad_norm": 9.1875,
+      "learning_rate": 2.1090099548296814e-06,
+      "loss": 0.3579,
+      "step": 6149
+    },
+    {
+      "epoch": 0.5240732850447379,
+      "grad_norm": 16.5,
+      "learning_rate": 2.10888787583716e-06,
+      "loss": 0.6973,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5241585002130379,
+      "grad_norm": 16.0,
+      "learning_rate": 2.108765781323671e-06,
+      "loss": 0.6114,
+      "step": 6151
+    },
+    {
+      "epoch": 0.5242437153813378,
+      "grad_norm": 16.375,
+      "learning_rate": 2.1086436712914206e-06,
+      "loss": 0.9928,
+      "step": 6152
+    },
+    {
+      "epoch": 0.5243289305496378,
+      "grad_norm": 12.5,
+      "learning_rate": 2.1085215457426157e-06,
+      "loss": 0.3729,
+      "step": 6153
+    },
+    {
+      "epoch": 0.5244141457179378,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.108399404679463e-06,
+      "loss": 0.9815,
+      "step": 6154
+    },
+    {
+      "epoch": 0.5244993608862377,
+      "grad_norm": 16.125,
+      "learning_rate": 2.10827724810417e-06,
+      "loss": 0.9905,
+      "step": 6155
+    },
+    {
+      "epoch": 0.5245845760545377,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1081550760189444e-06,
+      "loss": 0.7105,
+      "step": 6156
+    },
+    {
+      "epoch": 0.5246697912228376,
+      "grad_norm": 14.375,
+      "learning_rate": 2.108032888425994e-06,
+      "loss": 0.552,
+      "step": 6157
+    },
+    {
+      "epoch": 0.5247550063911376,
+      "grad_norm": 13.25,
+      "learning_rate": 2.1079106853275255e-06,
+      "loss": 0.5625,
+      "step": 6158
+    },
+    {
+      "epoch": 0.5248402215594375,
+      "grad_norm": 25.25,
+      "learning_rate": 2.1077884667257483e-06,
+      "loss": 1.145,
+      "step": 6159
+    },
+    {
+      "epoch": 0.5249254367277375,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.1076662326228713e-06,
+      "loss": 0.3689,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5250106518960375,
+      "grad_norm": 14.25,
+      "learning_rate": 2.107543983021103e-06,
+      "loss": 0.5504,
+      "step": 6161
+    },
+    {
+      "epoch": 0.5250958670643374,
+      "grad_norm": 14.875,
+      "learning_rate": 2.1074217179226525e-06,
+      "loss": 0.6711,
+      "step": 6162
+    },
+    {
+      "epoch": 0.5251810822326374,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1072994373297297e-06,
+      "loss": 0.5921,
+      "step": 6163
+    },
+    {
+      "epoch": 0.5252662974009373,
+      "grad_norm": 13.375,
+      "learning_rate": 2.1071771412445434e-06,
+      "loss": 0.6126,
+      "step": 6164
+    },
+    {
+      "epoch": 0.5253515125692373,
+      "grad_norm": 9.25,
+      "learning_rate": 2.1070548296693043e-06,
+      "loss": 0.2824,
+      "step": 6165
+    },
+    {
+      "epoch": 0.5254367277375372,
+      "grad_norm": 11.375,
+      "learning_rate": 2.1069325026062232e-06,
+      "loss": 0.4675,
+      "step": 6166
+    },
+    {
+      "epoch": 0.5255219429058372,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1068101600575096e-06,
+      "loss": 0.8072,
+      "step": 6167
+    },
+    {
+      "epoch": 0.5256071580741372,
+      "grad_norm": 18.125,
+      "learning_rate": 2.106687802025375e-06,
+      "loss": 1.2718,
+      "step": 6168
+    },
+    {
+      "epoch": 0.5256923732424371,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.10656542851203e-06,
+      "loss": 0.5753,
+      "step": 6169
+    },
+    {
+      "epoch": 0.5257775884107371,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1064430395196867e-06,
+      "loss": 1.0124,
+      "step": 6170
+    },
+    {
+      "epoch": 0.525862803579037,
+      "grad_norm": 11.25,
+      "learning_rate": 2.106320635050556e-06,
+      "loss": 0.5928,
+      "step": 6171
+    },
+    {
+      "epoch": 0.525948018747337,
+      "grad_norm": 13.25,
+      "learning_rate": 2.106198215106851e-06,
+      "loss": 0.728,
+      "step": 6172
+    },
+    {
+      "epoch": 0.5260332339156369,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.106075779690783e-06,
+      "loss": 0.3916,
+      "step": 6173
+    },
+    {
+      "epoch": 0.5261184490839369,
+      "grad_norm": 15.125,
+      "learning_rate": 2.1059533288045648e-06,
+      "loss": 0.9074,
+      "step": 6174
+    },
+    {
+      "epoch": 0.5262036642522369,
+      "grad_norm": 18.5,
+      "learning_rate": 2.105830862450409e-06,
+      "loss": 1.1213,
+      "step": 6175
+    },
+    {
+      "epoch": 0.5262888794205368,
+      "grad_norm": 16.25,
+      "learning_rate": 2.105708380630529e-06,
+      "loss": 0.7609,
+      "step": 6176
+    },
+    {
+      "epoch": 0.5263740945888368,
+      "grad_norm": 15.25,
+      "learning_rate": 2.105585883347138e-06,
+      "loss": 0.7624,
+      "step": 6177
+    },
+    {
+      "epoch": 0.5264593097571367,
+      "grad_norm": 25.625,
+      "learning_rate": 2.1054633706024504e-06,
+      "loss": 0.9504,
+      "step": 6178
+    },
+    {
+      "epoch": 0.5265445249254367,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1053408423986786e-06,
+      "loss": 0.9271,
+      "step": 6179
+    },
+    {
+      "epoch": 0.5266297400937366,
+      "grad_norm": 13.5,
+      "learning_rate": 2.1052182987380383e-06,
+      "loss": 0.8987,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5267149552620366,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.1050957396227427e-06,
+      "loss": 0.5143,
+      "step": 6181
+    },
+    {
+      "epoch": 0.5268001704303366,
+      "grad_norm": 12.5,
+      "learning_rate": 2.104973165055008e-06,
+      "loss": 0.4736,
+      "step": 6182
+    },
+    {
+      "epoch": 0.5268853855986365,
+      "grad_norm": 15.0,
+      "learning_rate": 2.104850575037048e-06,
+      "loss": 0.9557,
+      "step": 6183
+    },
+    {
+      "epoch": 0.5269706007669365,
+      "grad_norm": 11.5,
+      "learning_rate": 2.1047279695710786e-06,
+      "loss": 0.7713,
+      "step": 6184
+    },
+    {
+      "epoch": 0.5270558159352364,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1046053486593153e-06,
+      "loss": 1.0148,
+      "step": 6185
+    },
+    {
+      "epoch": 0.5271410311035364,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1044827123039738e-06,
+      "loss": 0.6012,
+      "step": 6186
+    },
+    {
+      "epoch": 0.5272262462718363,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.1043600605072708e-06,
+      "loss": 0.7012,
+      "step": 6187
+    },
+    {
+      "epoch": 0.5273114614401363,
+      "grad_norm": 18.0,
+      "learning_rate": 2.1042373932714217e-06,
+      "loss": 1.245,
+      "step": 6188
+    },
+    {
+      "epoch": 0.5273966766084363,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.104114710598645e-06,
+      "loss": 0.4105,
+      "step": 6189
+    },
+    {
+      "epoch": 0.5274818917767362,
+      "grad_norm": 18.375,
+      "learning_rate": 2.103992012491156e-06,
+      "loss": 0.7026,
+      "step": 6190
+    },
+    {
+      "epoch": 0.5275671069450362,
+      "grad_norm": 17.5,
+      "learning_rate": 2.1038692989511724e-06,
+      "loss": 0.7081,
+      "step": 6191
+    },
+    {
+      "epoch": 0.5276523221133361,
+      "grad_norm": 16.875,
+      "learning_rate": 2.1037465699809117e-06,
+      "loss": 0.5986,
+      "step": 6192
+    },
+    {
+      "epoch": 0.5277375372816361,
+      "grad_norm": 18.625,
+      "learning_rate": 2.1036238255825926e-06,
+      "loss": 1.2049,
+      "step": 6193
+    },
+    {
+      "epoch": 0.527822752449936,
+      "grad_norm": 20.125,
+      "learning_rate": 2.1035010657584322e-06,
+      "loss": 0.7627,
+      "step": 6194
+    },
+    {
+      "epoch": 0.527907967618236,
+      "grad_norm": 15.375,
+      "learning_rate": 2.1033782905106493e-06,
+      "loss": 0.8761,
+      "step": 6195
+    },
+    {
+      "epoch": 0.527993182786536,
+      "grad_norm": 19.0,
+      "learning_rate": 2.1032554998414627e-06,
+      "loss": 0.9158,
+      "step": 6196
+    },
+    {
+      "epoch": 0.5280783979548359,
+      "grad_norm": 12.375,
+      "learning_rate": 2.103132693753091e-06,
+      "loss": 0.6206,
+      "step": 6197
+    },
+    {
+      "epoch": 0.5281636131231359,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.103009872247754e-06,
+      "loss": 0.2813,
+      "step": 6198
+    },
+    {
+      "epoch": 0.5282488282914358,
+      "grad_norm": 13.625,
+      "learning_rate": 2.1028870353276708e-06,
+      "loss": 0.7611,
+      "step": 6199
+    },
+    {
+      "epoch": 0.5283340434597358,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.1027641829950607e-06,
+      "loss": 0.6344,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5284192586280357,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.102641315252145e-06,
+      "loss": 0.5089,
+      "step": 6201
+    },
+    {
+      "epoch": 0.5285044737963357,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.1025184321011427e-06,
+      "loss": 0.7062,
+      "step": 6202
+    },
+    {
+      "epoch": 0.5285896889646357,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.102395533544275e-06,
+      "loss": 0.548,
+      "step": 6203
+    },
+    {
+      "epoch": 0.5286749041329356,
+      "grad_norm": 10.125,
+      "learning_rate": 2.1022726195837633e-06,
+      "loss": 0.5138,
+      "step": 6204
+    },
+    {
+      "epoch": 0.5287601193012356,
+      "grad_norm": 12.5,
+      "learning_rate": 2.1021496902218284e-06,
+      "loss": 0.557,
+      "step": 6205
+    },
+    {
+      "epoch": 0.5288453344695355,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.1020267454606918e-06,
+      "loss": 0.7341,
+      "step": 6206
+    },
+    {
+      "epoch": 0.5289305496378356,
+      "grad_norm": 19.5,
+      "learning_rate": 2.101903785302575e-06,
+      "loss": 1.0213,
+      "step": 6207
+    },
+    {
+      "epoch": 0.5290157648061355,
+      "grad_norm": 15.375,
+      "learning_rate": 2.1017808097496996e-06,
+      "loss": 0.6237,
+      "step": 6208
+    },
+    {
+      "epoch": 0.5291009799744355,
+      "grad_norm": 13.25,
+      "learning_rate": 2.101657818804289e-06,
+      "loss": 0.6032,
+      "step": 6209
+    },
+    {
+      "epoch": 0.5291861951427355,
+      "grad_norm": 23.5,
+      "learning_rate": 2.1015348124685653e-06,
+      "loss": 1.1062,
+      "step": 6210
+    },
+    {
+      "epoch": 0.5292714103110354,
+      "grad_norm": 10.25,
+      "learning_rate": 2.101411790744751e-06,
+      "loss": 0.3556,
+      "step": 6211
+    },
+    {
+      "epoch": 0.5293566254793354,
+      "grad_norm": 43.5,
+      "learning_rate": 2.101288753635069e-06,
+      "loss": 0.6076,
+      "step": 6212
+    },
+    {
+      "epoch": 0.5294418406476353,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.101165701141744e-06,
+      "loss": 0.9004,
+      "step": 6213
+    },
+    {
+      "epoch": 0.5295270558159353,
+      "grad_norm": 13.375,
+      "learning_rate": 2.1010426332669986e-06,
+      "loss": 0.7381,
+      "step": 6214
+    },
+    {
+      "epoch": 0.5296122709842352,
+      "grad_norm": 12.75,
+      "learning_rate": 2.100919550013057e-06,
+      "loss": 0.6725,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5296974861525352,
+      "grad_norm": 15.0,
+      "learning_rate": 2.1007964513821434e-06,
+      "loss": 0.624,
+      "step": 6216
+    },
+    {
+      "epoch": 0.5297827013208352,
+      "grad_norm": 11.25,
+      "learning_rate": 2.1006733373764822e-06,
+      "loss": 0.4557,
+      "step": 6217
+    },
+    {
+      "epoch": 0.5298679164891351,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.100550207998299e-06,
+      "loss": 1.0441,
+      "step": 6218
+    },
+    {
+      "epoch": 0.5299531316574351,
+      "grad_norm": 15.875,
+      "learning_rate": 2.1004270632498174e-06,
+      "loss": 0.9227,
+      "step": 6219
+    },
+    {
+      "epoch": 0.530038346825735,
+      "grad_norm": 13.75,
+      "learning_rate": 2.100303903133264e-06,
+      "loss": 0.6257,
+      "step": 6220
+    },
+    {
+      "epoch": 0.530123561994035,
+      "grad_norm": 18.875,
+      "learning_rate": 2.1001807276508642e-06,
+      "loss": 0.9663,
+      "step": 6221
+    },
+    {
+      "epoch": 0.5302087771623349,
+      "grad_norm": 13.875,
+      "learning_rate": 2.1000575368048436e-06,
+      "loss": 0.6762,
+      "step": 6222
+    },
+    {
+      "epoch": 0.5302939923306349,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.0999343305974287e-06,
+      "loss": 1.0094,
+      "step": 6223
+    },
+    {
+      "epoch": 0.5303792074989349,
+      "grad_norm": 14.875,
+      "learning_rate": 2.0998111090308453e-06,
+      "loss": 0.675,
+      "step": 6224
+    },
+    {
+      "epoch": 0.5304644226672348,
+      "grad_norm": 19.0,
+      "learning_rate": 2.0996878721073212e-06,
+      "loss": 0.8701,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5305496378355348,
+      "grad_norm": 9.875,
+      "learning_rate": 2.0995646198290825e-06,
+      "loss": 0.4679,
+      "step": 6226
+    },
+    {
+      "epoch": 0.5306348530038347,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0994413521983567e-06,
+      "loss": 0.8929,
+      "step": 6227
+    },
+    {
+      "epoch": 0.5307200681721347,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0993180692173713e-06,
+      "loss": 0.7287,
+      "step": 6228
+    },
+    {
+      "epoch": 0.5308052833404346,
+      "grad_norm": 19.0,
+      "learning_rate": 2.099194770888355e-06,
+      "loss": 0.894,
+      "step": 6229
+    },
+    {
+      "epoch": 0.5308904985087346,
+      "grad_norm": 16.125,
+      "learning_rate": 2.099071457213535e-06,
+      "loss": 1.2041,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5309757136770346,
+      "grad_norm": 18.125,
+      "learning_rate": 2.0989481281951395e-06,
+      "loss": 0.7065,
+      "step": 6231
+    },
+    {
+      "epoch": 0.5310609288453345,
+      "grad_norm": 18.25,
+      "learning_rate": 2.098824783835398e-06,
+      "loss": 0.7604,
+      "step": 6232
+    },
+    {
+      "epoch": 0.5311461440136345,
+      "grad_norm": 12.875,
+      "learning_rate": 2.0987014241365396e-06,
+      "loss": 0.727,
+      "step": 6233
+    },
+    {
+      "epoch": 0.5312313591819344,
+      "grad_norm": 11.5,
+      "learning_rate": 2.0985780491007923e-06,
+      "loss": 0.5016,
+      "step": 6234
+    },
+    {
+      "epoch": 0.5313165743502344,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0984546587303863e-06,
+      "loss": 0.9232,
+      "step": 6235
+    },
+    {
+      "epoch": 0.5314017895185343,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.098331253027552e-06,
+      "loss": 0.5898,
+      "step": 6236
+    },
+    {
+      "epoch": 0.5314870046868343,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0982078319945185e-06,
+      "loss": 0.5249,
+      "step": 6237
+    },
+    {
+      "epoch": 0.5315722198551343,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.098084395633517e-06,
+      "loss": 0.6885,
+      "step": 6238
+    },
+    {
+      "epoch": 0.5316574350234342,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.097960943946777e-06,
+      "loss": 0.728,
+      "step": 6239
+    },
+    {
+      "epoch": 0.5317426501917342,
+      "grad_norm": 17.375,
+      "learning_rate": 2.09783747693653e-06,
+      "loss": 0.9854,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5318278653600341,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.097713994605008e-06,
+      "loss": 0.7116,
+      "step": 6241
+    },
+    {
+      "epoch": 0.5319130805283341,
+      "grad_norm": 21.125,
+      "learning_rate": 2.0975904969544413e-06,
+      "loss": 0.9684,
+      "step": 6242
+    },
+    {
+      "epoch": 0.531998295696634,
+      "grad_norm": 12.625,
+      "learning_rate": 2.0974669839870615e-06,
+      "loss": 0.666,
+      "step": 6243
+    },
+    {
+      "epoch": 0.532083510864934,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0973434557051016e-06,
+      "loss": 0.6634,
+      "step": 6244
+    },
+    {
+      "epoch": 0.532168726033234,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0972199121107934e-06,
+      "loss": 0.6413,
+      "step": 6245
+    },
+    {
+      "epoch": 0.5322539412015339,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.0970963532063694e-06,
+      "loss": 0.729,
+      "step": 6246
+    },
+    {
+      "epoch": 0.5323391563698339,
+      "grad_norm": 9.625,
+      "learning_rate": 2.0969727789940623e-06,
+      "loss": 0.3331,
+      "step": 6247
+    },
+    {
+      "epoch": 0.5324243715381338,
+      "grad_norm": 14.125,
+      "learning_rate": 2.096849189476106e-06,
+      "loss": 0.757,
+      "step": 6248
+    },
+    {
+      "epoch": 0.5325095867064338,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0967255846547326e-06,
+      "loss": 0.686,
+      "step": 6249
+    },
+    {
+      "epoch": 0.5325948018747337,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0966019645321767e-06,
+      "loss": 0.9446,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5326800170430337,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.0964783291106714e-06,
+      "loss": 0.616,
+      "step": 6251
+    },
+    {
+      "epoch": 0.5327652322113337,
+      "grad_norm": 18.5,
+      "learning_rate": 2.0963546783924525e-06,
+      "loss": 0.9818,
+      "step": 6252
+    },
+    {
+      "epoch": 0.5328504473796336,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.096231012379753e-06,
+      "loss": 0.5134,
+      "step": 6253
+    },
+    {
+      "epoch": 0.5329356625479336,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.096107331074808e-06,
+      "loss": 0.4026,
+      "step": 6254
+    },
+    {
+      "epoch": 0.5330208777162335,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0959836344798527e-06,
+      "loss": 0.8716,
+      "step": 6255
+    },
+    {
+      "epoch": 0.5331060928845335,
+      "grad_norm": 14.875,
+      "learning_rate": 2.095859922597123e-06,
+      "loss": 0.4164,
+      "step": 6256
+    },
+    {
+      "epoch": 0.5331913080528334,
+      "grad_norm": 16.125,
+      "learning_rate": 2.095736195428853e-06,
+      "loss": 0.8771,
+      "step": 6257
+    },
+    {
+      "epoch": 0.5332765232211334,
+      "grad_norm": 13.625,
+      "learning_rate": 2.0956124529772805e-06,
+      "loss": 0.8138,
+      "step": 6258
+    },
+    {
+      "epoch": 0.5333617383894333,
+      "grad_norm": 16.625,
+      "learning_rate": 2.09548869524464e-06,
+      "loss": 1.078,
+      "step": 6259
+    },
+    {
+      "epoch": 0.5334469535577333,
+      "grad_norm": 18.125,
+      "learning_rate": 2.0953649222331686e-06,
+      "loss": 1.2338,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5335321687260333,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.0952411339451027e-06,
+      "loss": 0.4029,
+      "step": 6261
+    },
+    {
+      "epoch": 0.5336173838943332,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0951173303826806e-06,
+      "loss": 0.5247,
+      "step": 6262
+    },
+    {
+      "epoch": 0.5337025990626332,
+      "grad_norm": 11.5,
+      "learning_rate": 2.0949935115481376e-06,
+      "loss": 0.5547,
+      "step": 6263
+    },
+    {
+      "epoch": 0.5337878142309331,
+      "grad_norm": 24.875,
+      "learning_rate": 2.0948696774437123e-06,
+      "loss": 0.9971,
+      "step": 6264
+    },
+    {
+      "epoch": 0.5338730293992331,
+      "grad_norm": 13.125,
+      "learning_rate": 2.0947458280716427e-06,
+      "loss": 0.5835,
+      "step": 6265
+    },
+    {
+      "epoch": 0.533958244567533,
+      "grad_norm": 13.75,
+      "learning_rate": 2.094621963434166e-06,
+      "loss": 0.8904,
+      "step": 6266
+    },
+    {
+      "epoch": 0.534043459735833,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0944980835335214e-06,
+      "loss": 0.9664,
+      "step": 6267
+    },
+    {
+      "epoch": 0.534128674904133,
+      "grad_norm": 21.75,
+      "learning_rate": 2.094374188371947e-06,
+      "loss": 1.1452,
+      "step": 6268
+    },
+    {
+      "epoch": 0.5342138900724329,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0942502779516825e-06,
+      "loss": 0.9308,
+      "step": 6269
+    },
+    {
+      "epoch": 0.5342991052407329,
+      "grad_norm": 17.75,
+      "learning_rate": 2.0941263522749662e-06,
+      "loss": 0.4857,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5343843204090328,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.094002411344038e-06,
+      "loss": 0.4573,
+      "step": 6271
+    },
+    {
+      "epoch": 0.5344695355773328,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.093878455161137e-06,
+      "loss": 0.644,
+      "step": 6272
+    },
+    {
+      "epoch": 0.5345547507456327,
+      "grad_norm": 14.625,
+      "learning_rate": 2.093754483728504e-06,
+      "loss": 0.8602,
+      "step": 6273
+    },
+    {
+      "epoch": 0.5346399659139327,
+      "grad_norm": 18.625,
+      "learning_rate": 2.093630497048379e-06,
+      "loss": 0.9489,
+      "step": 6274
+    },
+    {
+      "epoch": 0.5347251810822327,
+      "grad_norm": 10.0,
+      "learning_rate": 2.0935064951230026e-06,
+      "loss": 0.3423,
+      "step": 6275
+    },
+    {
+      "epoch": 0.5348103962505326,
+      "grad_norm": 12.375,
+      "learning_rate": 2.093382477954616e-06,
+      "loss": 0.6262,
+      "step": 6276
+    },
+    {
+      "epoch": 0.5348956114188326,
+      "grad_norm": 16.0,
+      "learning_rate": 2.0932584455454594e-06,
+      "loss": 0.2831,
+      "step": 6277
+    },
+    {
+      "epoch": 0.5349808265871325,
+      "grad_norm": 13.25,
+      "learning_rate": 2.093134397897775e-06,
+      "loss": 0.5382,
+      "step": 6278
+    },
+    {
+      "epoch": 0.5350660417554325,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.093010335013804e-06,
+      "loss": 0.6745,
+      "step": 6279
+    },
+    {
+      "epoch": 0.5351512569237324,
+      "grad_norm": 30.25,
+      "learning_rate": 2.0928862568957886e-06,
+      "loss": 0.8815,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5352364720920324,
+      "grad_norm": 14.875,
+      "learning_rate": 2.092762163545971e-06,
+      "loss": 1.1011,
+      "step": 6281
+    },
+    {
+      "epoch": 0.5353216872603324,
+      "grad_norm": 14.625,
+      "learning_rate": 2.092638054966594e-06,
+      "loss": 0.5702,
+      "step": 6282
+    },
+    {
+      "epoch": 0.5354069024286323,
+      "grad_norm": 26.75,
+      "learning_rate": 2.0925139311598996e-06,
+      "loss": 0.805,
+      "step": 6283
+    },
+    {
+      "epoch": 0.5354921175969323,
+      "grad_norm": 15.375,
+      "learning_rate": 2.092389792128131e-06,
+      "loss": 0.7906,
+      "step": 6284
+    },
+    {
+      "epoch": 0.5355773327652322,
+      "grad_norm": 20.75,
+      "learning_rate": 2.0922656378735324e-06,
+      "loss": 0.9157,
+      "step": 6285
+    },
+    {
+      "epoch": 0.5356625479335322,
+      "grad_norm": 19.375,
+      "learning_rate": 2.0921414683983464e-06,
+      "loss": 0.4856,
+      "step": 6286
+    },
+    {
+      "epoch": 0.5357477631018321,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.092017283704817e-06,
+      "loss": 0.4744,
+      "step": 6287
+    },
+    {
+      "epoch": 0.5358329782701321,
+      "grad_norm": 13.375,
+      "learning_rate": 2.091893083795189e-06,
+      "loss": 0.4649,
+      "step": 6288
+    },
+    {
+      "epoch": 0.5359181934384321,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.091768868671706e-06,
+      "loss": 0.3992,
+      "step": 6289
+    },
+    {
+      "epoch": 0.536003408606732,
+      "grad_norm": 24.625,
+      "learning_rate": 2.091644638336614e-06,
+      "loss": 1.0316,
+      "step": 6290
+    },
+    {
+      "epoch": 0.536088623775032,
+      "grad_norm": 15.75,
+      "learning_rate": 2.0915203927921563e-06,
+      "loss": 0.856,
+      "step": 6291
+    },
+    {
+      "epoch": 0.5361738389433319,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0913961320405783e-06,
+      "loss": 1.061,
+      "step": 6292
+    },
+    {
+      "epoch": 0.5362590541116319,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.091271856084127e-06,
+      "loss": 0.5562,
+      "step": 6293
+    },
+    {
+      "epoch": 0.5363442692799318,
+      "grad_norm": 22.125,
+      "learning_rate": 2.091147564925047e-06,
+      "loss": 1.3397,
+      "step": 6294
+    },
+    {
+      "epoch": 0.5364294844482318,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.091023258565585e-06,
+      "loss": 0.7106,
+      "step": 6295
+    },
+    {
+      "epoch": 0.5365146996165318,
+      "grad_norm": 22.0,
+      "learning_rate": 2.0908989370079867e-06,
+      "loss": 0.6138,
+      "step": 6296
+    },
+    {
+      "epoch": 0.5365999147848317,
+      "grad_norm": 29.625,
+      "learning_rate": 2.090774600254499e-06,
+      "loss": 1.3506,
+      "step": 6297
+    },
+    {
+      "epoch": 0.5366851299531317,
+      "grad_norm": 12.625,
+      "learning_rate": 2.090650248307369e-06,
+      "loss": 0.6136,
+      "step": 6298
+    },
+    {
+      "epoch": 0.5367703451214316,
+      "grad_norm": 12.25,
+      "learning_rate": 2.0905258811688432e-06,
+      "loss": 0.3978,
+      "step": 6299
+    },
+    {
+      "epoch": 0.5368555602897316,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.09040149884117e-06,
+      "loss": 0.7418,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5369407754580315,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0902771013265964e-06,
+      "loss": 0.5526,
+      "step": 6301
+    },
+    {
+      "epoch": 0.5370259906263315,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.090152688627371e-06,
+      "loss": 0.6491,
+      "step": 6302
+    },
+    {
+      "epoch": 0.5371112057946315,
+      "grad_norm": 14.0,
+      "learning_rate": 2.0900282607457413e-06,
+      "loss": 0.7835,
+      "step": 6303
+    },
+    {
+      "epoch": 0.5371964209629314,
+      "grad_norm": 13.0,
+      "learning_rate": 2.089903817683956e-06,
+      "loss": 0.3832,
+      "step": 6304
+    },
+    {
+      "epoch": 0.5372816361312314,
+      "grad_norm": 20.125,
+      "learning_rate": 2.089779359444264e-06,
+      "loss": 0.9013,
+      "step": 6305
+    },
+    {
+      "epoch": 0.5373668512995313,
+      "grad_norm": 20.5,
+      "learning_rate": 2.0896548860289153e-06,
+      "loss": 0.9356,
+      "step": 6306
+    },
+    {
+      "epoch": 0.5374520664678313,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.0895303974401582e-06,
+      "loss": 0.5911,
+      "step": 6307
+    },
+    {
+      "epoch": 0.5375372816361312,
+      "grad_norm": 22.625,
+      "learning_rate": 2.0894058936802424e-06,
+      "loss": 1.011,
+      "step": 6308
+    },
+    {
+      "epoch": 0.5376224968044312,
+      "grad_norm": 15.125,
+      "learning_rate": 2.089281374751418e-06,
+      "loss": 0.855,
+      "step": 6309
+    },
+    {
+      "epoch": 0.5377077119727312,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0891568406559355e-06,
+      "loss": 0.8929,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5377929271410311,
+      "grad_norm": 11.5,
+      "learning_rate": 2.089032291396045e-06,
+      "loss": 0.3536,
+      "step": 6311
+    },
+    {
+      "epoch": 0.5378781423093311,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0889077269739967e-06,
+      "loss": 0.6752,
+      "step": 6312
+    },
+    {
+      "epoch": 0.537963357477631,
+      "grad_norm": 20.375,
+      "learning_rate": 2.0887831473920426e-06,
+      "loss": 1.3051,
+      "step": 6313
+    },
+    {
+      "epoch": 0.538048572645931,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0886585526524335e-06,
+      "loss": 0.6454,
+      "step": 6314
+    },
+    {
+      "epoch": 0.5381337878142309,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.088533942757421e-06,
+      "loss": 0.4837,
+      "step": 6315
+    },
+    {
+      "epoch": 0.5382190029825309,
+      "grad_norm": 22.75,
+      "learning_rate": 2.0884093177092575e-06,
+      "loss": 0.9437,
+      "step": 6316
+    },
+    {
+      "epoch": 0.5383042181508308,
+      "grad_norm": 29.75,
+      "learning_rate": 2.088284677510194e-06,
+      "loss": 1.1829,
+      "step": 6317
+    },
+    {
+      "epoch": 0.5383894333191308,
+      "grad_norm": 21.625,
+      "learning_rate": 2.088160022162483e-06,
+      "loss": 0.9178,
+      "step": 6318
+    },
+    {
+      "epoch": 0.5384746484874308,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.0880353516683782e-06,
+      "loss": 0.5022,
+      "step": 6319
+    },
+    {
+      "epoch": 0.5385598636557307,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.087910666030132e-06,
+      "loss": 0.7753,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5386450788240307,
+      "grad_norm": 14.5,
+      "learning_rate": 2.087785965249997e-06,
+      "loss": 0.7404,
+      "step": 6321
+    },
+    {
+      "epoch": 0.5387302939923306,
+      "grad_norm": 15.625,
+      "learning_rate": 2.087661249330227e-06,
+      "loss": 0.6213,
+      "step": 6322
+    },
+    {
+      "epoch": 0.5388155091606306,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0875365182730767e-06,
+      "loss": 0.596,
+      "step": 6323
+    },
+    {
+      "epoch": 0.5389007243289305,
+      "grad_norm": 18.25,
+      "learning_rate": 2.087411772080799e-06,
+      "loss": 1.1188,
+      "step": 6324
+    },
+    {
+      "epoch": 0.5389859394972305,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0872870107556483e-06,
+      "loss": 0.776,
+      "step": 6325
+    },
+    {
+      "epoch": 0.5390711546655305,
+      "grad_norm": 22.75,
+      "learning_rate": 2.0871622342998794e-06,
+      "loss": 1.0287,
+      "step": 6326
+    },
+    {
+      "epoch": 0.5391563698338304,
+      "grad_norm": 10.625,
+      "learning_rate": 2.087037442715747e-06,
+      "loss": 0.5038,
+      "step": 6327
+    },
+    {
+      "epoch": 0.5392415850021304,
+      "grad_norm": 15.375,
+      "learning_rate": 2.0869126360055065e-06,
+      "loss": 0.9662,
+      "step": 6328
+    },
+    {
+      "epoch": 0.5393268001704303,
+      "grad_norm": 10.75,
+      "learning_rate": 2.086787814171413e-06,
+      "loss": 0.4304,
+      "step": 6329
+    },
+    {
+      "epoch": 0.5394120153387303,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.086662977215722e-06,
+      "loss": 0.6223,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5394972305070302,
+      "grad_norm": 9.1875,
+      "learning_rate": 2.08653812514069e-06,
+      "loss": 0.3308,
+      "step": 6331
+    },
+    {
+      "epoch": 0.5395824456753302,
+      "grad_norm": 15.625,
+      "learning_rate": 2.086413257948573e-06,
+      "loss": 0.7595,
+      "step": 6332
+    },
+    {
+      "epoch": 0.5396676608436302,
+      "grad_norm": 14.5,
+      "learning_rate": 2.086288375641627e-06,
+      "loss": 0.8933,
+      "step": 6333
+    },
+    {
+      "epoch": 0.5397528760119301,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0861634782221093e-06,
+      "loss": 0.7803,
+      "step": 6334
+    },
+    {
+      "epoch": 0.5398380911802301,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.0860385656922767e-06,
+      "loss": 0.4623,
+      "step": 6335
+    },
+    {
+      "epoch": 0.53992330634853,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.0859136380543866e-06,
+      "loss": 0.4133,
+      "step": 6336
+    },
+    {
+      "epoch": 0.54000852151683,
+      "grad_norm": 13.875,
+      "learning_rate": 2.0857886953106965e-06,
+      "loss": 0.7697,
+      "step": 6337
+    },
+    {
+      "epoch": 0.5400937366851299,
+      "grad_norm": 13.875,
+      "learning_rate": 2.085663737463464e-06,
+      "loss": 0.7458,
+      "step": 6338
+    },
+    {
+      "epoch": 0.5401789518534299,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.0855387645149475e-06,
+      "loss": 0.5537,
+      "step": 6339
+    },
+    {
+      "epoch": 0.5402641670217299,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.0854137764674055e-06,
+      "loss": 0.4072,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5403493821900298,
+      "grad_norm": 11.625,
+      "learning_rate": 2.085288773323096e-06,
+      "loss": 0.5515,
+      "step": 6341
+    },
+    {
+      "epoch": 0.5404345973583298,
+      "grad_norm": 16.375,
+      "learning_rate": 2.085163755084279e-06,
+      "loss": 0.6574,
+      "step": 6342
+    },
+    {
+      "epoch": 0.5405198125266297,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0850387217532127e-06,
+      "loss": 0.8315,
+      "step": 6343
+    },
+    {
+      "epoch": 0.5406050276949297,
+      "grad_norm": 21.125,
+      "learning_rate": 2.0849136733321573e-06,
+      "loss": 1.0187,
+      "step": 6344
+    },
+    {
+      "epoch": 0.5406902428632296,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0847886098233724e-06,
+      "loss": 0.7853,
+      "step": 6345
+    },
+    {
+      "epoch": 0.5407754580315296,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.084663531229117e-06,
+      "loss": 0.6363,
+      "step": 6346
+    },
+    {
+      "epoch": 0.5408606731998296,
+      "grad_norm": 20.75,
+      "learning_rate": 2.084538437551653e-06,
+      "loss": 1.0188,
+      "step": 6347
+    },
+    {
+      "epoch": 0.5409458883681295,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.0844133287932404e-06,
+      "loss": 0.7096,
+      "step": 6348
+    },
+    {
+      "epoch": 0.5410311035364295,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.0842882049561398e-06,
+      "loss": 0.6152,
+      "step": 6349
+    },
+    {
+      "epoch": 0.5411163187047294,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0841630660426116e-06,
+      "loss": 0.9194,
+      "step": 6350
+    },
+    {
+      "epoch": 0.5412015338730294,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.084037912054919e-06,
+      "loss": 0.887,
+      "step": 6351
+    },
+    {
+      "epoch": 0.5412867490413293,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.083912742995322e-06,
+      "loss": 0.6242,
+      "step": 6352
+    },
+    {
+      "epoch": 0.5413719642096293,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.0837875588660834e-06,
+      "loss": 0.7515,
+      "step": 6353
+    },
+    {
+      "epoch": 0.5414571793779293,
+      "grad_norm": 14.125,
+      "learning_rate": 2.083662359669465e-06,
+      "loss": 0.6306,
+      "step": 6354
+    },
+    {
+      "epoch": 0.5415423945462292,
+      "grad_norm": 11.875,
+      "learning_rate": 2.08353714540773e-06,
+      "loss": 0.5489,
+      "step": 6355
+    },
+    {
+      "epoch": 0.5416276097145292,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.08341191608314e-06,
+      "loss": 0.6301,
+      "step": 6356
+    },
+    {
+      "epoch": 0.5417128248828291,
+      "grad_norm": 14.625,
+      "learning_rate": 2.083286671697959e-06,
+      "loss": 1.1312,
+      "step": 6357
+    },
+    {
+      "epoch": 0.5417980400511291,
+      "grad_norm": 15.125,
+      "learning_rate": 2.0831614122544496e-06,
+      "loss": 0.8191,
+      "step": 6358
+    },
+    {
+      "epoch": 0.541883255219429,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0830361377548757e-06,
+      "loss": 0.7818,
+      "step": 6359
+    },
+    {
+      "epoch": 0.541968470387729,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.0829108482015015e-06,
+      "loss": 0.2524,
+      "step": 6360
+    },
+    {
+      "epoch": 0.542053685556029,
+      "grad_norm": 25.875,
+      "learning_rate": 2.0827855435965907e-06,
+      "loss": 1.318,
+      "step": 6361
+    },
+    {
+      "epoch": 0.5421389007243289,
+      "grad_norm": 17.75,
+      "learning_rate": 2.082660223942407e-06,
+      "loss": 0.9532,
+      "step": 6362
+    },
+    {
+      "epoch": 0.5422241158926289,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.082534889241216e-06,
+      "loss": 0.8932,
+      "step": 6363
+    },
+    {
+      "epoch": 0.5423093310609288,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.0824095394952828e-06,
+      "loss": 0.8502,
+      "step": 6364
+    },
+    {
+      "epoch": 0.5423945462292288,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.082284174706872e-06,
+      "loss": 0.6244,
+      "step": 6365
+    },
+    {
+      "epoch": 0.5424797613975287,
+      "grad_norm": 18.0,
+      "learning_rate": 2.082158794878249e-06,
+      "loss": 0.9709,
+      "step": 6366
+    },
+    {
+      "epoch": 0.5425649765658287,
+      "grad_norm": 19.25,
+      "learning_rate": 2.0820334000116803e-06,
+      "loss": 1.0619,
+      "step": 6367
+    },
+    {
+      "epoch": 0.5426501917341287,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0819079901094306e-06,
+      "loss": 0.7625,
+      "step": 6368
+    },
+    {
+      "epoch": 0.5427354069024286,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0817825651737674e-06,
+      "loss": 0.9346,
+      "step": 6369
+    },
+    {
+      "epoch": 0.5428206220707286,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.0816571252069565e-06,
+      "loss": 0.6954,
+      "step": 6370
+    },
+    {
+      "epoch": 0.5429058372390285,
+      "grad_norm": 14.75,
+      "learning_rate": 2.081531670211266e-06,
+      "loss": 0.8664,
+      "step": 6371
+    },
+    {
+      "epoch": 0.5429910524073285,
+      "grad_norm": 15.875,
+      "learning_rate": 2.0814062001889613e-06,
+      "loss": 0.775,
+      "step": 6372
+    },
+    {
+      "epoch": 0.5430762675756284,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.0812807151423103e-06,
+      "loss": 0.7374,
+      "step": 6373
+    },
+    {
+      "epoch": 0.5431614827439284,
+      "grad_norm": 17.5,
+      "learning_rate": 2.0811552150735813e-06,
+      "loss": 0.9329,
+      "step": 6374
+    },
+    {
+      "epoch": 0.5432466979122283,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.081029699985042e-06,
+      "loss": 0.8339,
+      "step": 6375
+    },
+    {
+      "epoch": 0.5433319130805283,
+      "grad_norm": 18.125,
+      "learning_rate": 2.0809041698789598e-06,
+      "loss": 0.7687,
+      "step": 6376
+    },
+    {
+      "epoch": 0.5434171282488283,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.080778624757604e-06,
+      "loss": 0.7427,
+      "step": 6377
+    },
+    {
+      "epoch": 0.5435023434171282,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.080653064623243e-06,
+      "loss": 0.908,
+      "step": 6378
+    },
+    {
+      "epoch": 0.5435875585854282,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0805274894781458e-06,
+      "loss": 0.9505,
+      "step": 6379
+    },
+    {
+      "epoch": 0.5436727737537281,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.0804018993245816e-06,
+      "loss": 0.307,
+      "step": 6380
+    },
+    {
+      "epoch": 0.5437579889220281,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.0802762941648203e-06,
+      "loss": 1.0888,
+      "step": 6381
+    },
+    {
+      "epoch": 0.543843204090328,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0801506740011312e-06,
+      "loss": 0.9752,
+      "step": 6382
+    },
+    {
+      "epoch": 0.543928419258628,
+      "grad_norm": 14.625,
+      "learning_rate": 2.0800250388357847e-06,
+      "loss": 0.7256,
+      "step": 6383
+    },
+    {
+      "epoch": 0.544013634426928,
+      "grad_norm": 17.375,
+      "learning_rate": 2.079899388671051e-06,
+      "loss": 1.1044,
+      "step": 6384
+    },
+    {
+      "epoch": 0.5440988495952279,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.0797737235092015e-06,
+      "loss": 0.6349,
+      "step": 6385
+    },
+    {
+      "epoch": 0.5441840647635279,
+      "grad_norm": 16.0,
+      "learning_rate": 2.0796480433525057e-06,
+      "loss": 0.7979,
+      "step": 6386
+    },
+    {
+      "epoch": 0.5442692799318278,
+      "grad_norm": 66.5,
+      "learning_rate": 2.0795223482032357e-06,
+      "loss": 1.5932,
+      "step": 6387
+    },
+    {
+      "epoch": 0.5443544951001278,
+      "grad_norm": 12.625,
+      "learning_rate": 2.079396638063663e-06,
+      "loss": 0.7549,
+      "step": 6388
+    },
+    {
+      "epoch": 0.5444397102684277,
+      "grad_norm": 23.875,
+      "learning_rate": 2.0792709129360583e-06,
+      "loss": 1.1931,
+      "step": 6389
+    },
+    {
+      "epoch": 0.5445249254367277,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.079145172822695e-06,
+      "loss": 0.7229,
+      "step": 6390
+    },
+    {
+      "epoch": 0.5446101406050277,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0790194177258436e-06,
+      "loss": 0.7531,
+      "step": 6391
+    },
+    {
+      "epoch": 0.5446953557733276,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.078893647647779e-06,
+      "loss": 0.759,
+      "step": 6392
+    },
+    {
+      "epoch": 0.5447805709416276,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.078767862590772e-06,
+      "loss": 0.6109,
+      "step": 6393
+    },
+    {
+      "epoch": 0.5448657861099275,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0786420625570963e-06,
+      "loss": 0.9057,
+      "step": 6394
+    },
+    {
+      "epoch": 0.5449510012782275,
+      "grad_norm": 17.375,
+      "learning_rate": 2.0785162475490254e-06,
+      "loss": 0.8169,
+      "step": 6395
+    },
+    {
+      "epoch": 0.5450362164465274,
+      "grad_norm": 19.875,
+      "learning_rate": 2.078390417568833e-06,
+      "loss": 0.3474,
+      "step": 6396
+    },
+    {
+      "epoch": 0.5451214316148274,
+      "grad_norm": 25.375,
+      "learning_rate": 2.0782645726187926e-06,
+      "loss": 0.8701,
+      "step": 6397
+    },
+    {
+      "epoch": 0.5452066467831274,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.078138712701178e-06,
+      "loss": 0.5955,
+      "step": 6398
+    },
+    {
+      "epoch": 0.5452918619514273,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0780128378182645e-06,
+      "loss": 0.6416,
+      "step": 6399
+    },
+    {
+      "epoch": 0.5453770771197273,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0778869479723267e-06,
+      "loss": 0.9216,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5454622922880272,
+      "grad_norm": 16.25,
+      "learning_rate": 2.077761043165639e-06,
+      "loss": 0.8919,
+      "step": 6401
+    },
+    {
+      "epoch": 0.5455475074563272,
+      "grad_norm": 13.75,
+      "learning_rate": 2.077635123400477e-06,
+      "loss": 0.7583,
+      "step": 6402
+    },
+    {
+      "epoch": 0.5456327226246271,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0775091886791158e-06,
+      "loss": 0.7405,
+      "step": 6403
+    },
+    {
+      "epoch": 0.5457179377929271,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0773832390038316e-06,
+      "loss": 0.8922,
+      "step": 6404
+    },
+    {
+      "epoch": 0.5458031529612271,
+      "grad_norm": 21.375,
+      "learning_rate": 2.0772572743769002e-06,
+      "loss": 1.1329,
+      "step": 6405
+    },
+    {
+      "epoch": 0.545888368129527,
+      "grad_norm": 15.25,
+      "learning_rate": 2.077131294800598e-06,
+      "loss": 0.904,
+      "step": 6406
+    },
+    {
+      "epoch": 0.545973583297827,
+      "grad_norm": 18.0,
+      "learning_rate": 2.0770053002772016e-06,
+      "loss": 0.8525,
+      "step": 6407
+    },
+    {
+      "epoch": 0.5460587984661269,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.0768792908089876e-06,
+      "loss": 0.5698,
+      "step": 6408
+    },
+    {
+      "epoch": 0.5461440136344269,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0767532663982332e-06,
+      "loss": 0.9145,
+      "step": 6409
+    },
+    {
+      "epoch": 0.5462292288027268,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.0766272270472165e-06,
+      "loss": 0.9785,
+      "step": 6410
+    },
+    {
+      "epoch": 0.5463144439710268,
+      "grad_norm": 13.875,
+      "learning_rate": 2.076501172758214e-06,
+      "loss": 0.7995,
+      "step": 6411
+    },
+    {
+      "epoch": 0.5463996591393268,
+      "grad_norm": 18.0,
+      "learning_rate": 2.0763751035335047e-06,
+      "loss": 0.9431,
+      "step": 6412
+    },
+    {
+      "epoch": 0.5464848743076267,
+      "grad_norm": 17.375,
+      "learning_rate": 2.076249019375366e-06,
+      "loss": 0.8848,
+      "step": 6413
+    },
+    {
+      "epoch": 0.5465700894759267,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.0761229202860765e-06,
+      "loss": 0.7104,
+      "step": 6414
+    },
+    {
+      "epoch": 0.5466553046442266,
+      "grad_norm": 20.375,
+      "learning_rate": 2.0759968062679154e-06,
+      "loss": 0.9082,
+      "step": 6415
+    },
+    {
+      "epoch": 0.5467405198125266,
+      "grad_norm": 19.25,
+      "learning_rate": 2.0758706773231614e-06,
+      "loss": 0.8975,
+      "step": 6416
+    },
+    {
+      "epoch": 0.5468257349808265,
+      "grad_norm": 19.0,
+      "learning_rate": 2.075744533454094e-06,
+      "loss": 0.9932,
+      "step": 6417
+    },
+    {
+      "epoch": 0.5469109501491265,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.075618374662992e-06,
+      "loss": 0.8165,
+      "step": 6418
+    },
+    {
+      "epoch": 0.5469961653174265,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.0754922009521357e-06,
+      "loss": 0.4637,
+      "step": 6419
+    },
+    {
+      "epoch": 0.5470813804857264,
+      "grad_norm": 31.5,
+      "learning_rate": 2.075366012323806e-06,
+      "loss": 0.9428,
+      "step": 6420
+    },
+    {
+      "epoch": 0.5471665956540264,
+      "grad_norm": 29.5,
+      "learning_rate": 2.0752398087802817e-06,
+      "loss": 0.8586,
+      "step": 6421
+    },
+    {
+      "epoch": 0.5472518108223263,
+      "grad_norm": 17.5,
+      "learning_rate": 2.0751135903238446e-06,
+      "loss": 0.2941,
+      "step": 6422
+    },
+    {
+      "epoch": 0.5473370259906263,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.0749873569567753e-06,
+      "loss": 0.6178,
+      "step": 6423
+    },
+    {
+      "epoch": 0.5474222411589262,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.0748611086813548e-06,
+      "loss": 0.5331,
+      "step": 6424
+    },
+    {
+      "epoch": 0.5475074563272262,
+      "grad_norm": 15.125,
+      "learning_rate": 2.074734845499864e-06,
+      "loss": 0.7394,
+      "step": 6425
+    },
+    {
+      "epoch": 0.5475926714955262,
+      "grad_norm": 15.25,
+      "learning_rate": 2.0746085674145856e-06,
+      "loss": 0.8424,
+      "step": 6426
+    },
+    {
+      "epoch": 0.5476778866638261,
+      "grad_norm": 20.0,
+      "learning_rate": 2.0744822744278017e-06,
+      "loss": 1.0727,
+      "step": 6427
+    },
+    {
+      "epoch": 0.5477631018321261,
+      "grad_norm": 19.125,
+      "learning_rate": 2.0743559665417933e-06,
+      "loss": 0.4943,
+      "step": 6428
+    },
+    {
+      "epoch": 0.547848317000426,
+      "grad_norm": 16.5,
+      "learning_rate": 2.074229643758844e-06,
+      "loss": 0.9638,
+      "step": 6429
+    },
+    {
+      "epoch": 0.547933532168726,
+      "grad_norm": 14.875,
+      "learning_rate": 2.074103306081236e-06,
+      "loss": 0.9248,
+      "step": 6430
+    },
+    {
+      "epoch": 0.5480187473370259,
+      "grad_norm": 13.375,
+      "learning_rate": 2.0739769535112527e-06,
+      "loss": 0.6848,
+      "step": 6431
+    },
+    {
+      "epoch": 0.548103962505326,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0738505860511776e-06,
+      "loss": 0.7043,
+      "step": 6432
+    },
+    {
+      "epoch": 0.548189177673626,
+      "grad_norm": 16.625,
+      "learning_rate": 2.0737242037032935e-06,
+      "loss": 0.7008,
+      "step": 6433
+    },
+    {
+      "epoch": 0.5482743928419259,
+      "grad_norm": 12.125,
+      "learning_rate": 2.0735978064698846e-06,
+      "loss": 0.6398,
+      "step": 6434
+    },
+    {
+      "epoch": 0.5483596080102259,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0734713943532355e-06,
+      "loss": 0.8238,
+      "step": 6435
+    },
+    {
+      "epoch": 0.5484448231785258,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.0733449673556305e-06,
+      "loss": 0.7854,
+      "step": 6436
+    },
+    {
+      "epoch": 0.5485300383468258,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.0732185254793534e-06,
+      "loss": 1.0776,
+      "step": 6437
+    },
+    {
+      "epoch": 0.5486152535151257,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.07309206872669e-06,
+      "loss": 0.8542,
+      "step": 6438
+    },
+    {
+      "epoch": 0.5487004686834257,
+      "grad_norm": 10.875,
+      "learning_rate": 2.0729655970999252e-06,
+      "loss": 0.4787,
+      "step": 6439
+    },
+    {
+      "epoch": 0.5487856838517257,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0728391106013448e-06,
+      "loss": 0.6915,
+      "step": 6440
+    },
+    {
+      "epoch": 0.5488708990200256,
+      "grad_norm": 17.125,
+      "learning_rate": 2.072712609233234e-06,
+      "loss": 0.864,
+      "step": 6441
+    },
+    {
+      "epoch": 0.5489561141883256,
+      "grad_norm": 19.125,
+      "learning_rate": 2.0725860929978787e-06,
+      "loss": 0.6895,
+      "step": 6442
+    },
+    {
+      "epoch": 0.5490413293566255,
+      "grad_norm": 12.625,
+      "learning_rate": 2.072459561897566e-06,
+      "loss": 0.5052,
+      "step": 6443
+    },
+    {
+      "epoch": 0.5491265445249255,
+      "grad_norm": 15.25,
+      "learning_rate": 2.0723330159345815e-06,
+      "loss": 0.9757,
+      "step": 6444
+    },
+    {
+      "epoch": 0.5492117596932254,
+      "grad_norm": 16.25,
+      "learning_rate": 2.072206455111213e-06,
+      "loss": 0.8129,
+      "step": 6445
+    },
+    {
+      "epoch": 0.5492969748615254,
+      "grad_norm": 17.5,
+      "learning_rate": 2.072079879429747e-06,
+      "loss": 0.8724,
+      "step": 6446
+    },
+    {
+      "epoch": 0.5493821900298254,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.0719532888924705e-06,
+      "loss": 0.5526,
+      "step": 6447
+    },
+    {
+      "epoch": 0.5494674051981253,
+      "grad_norm": 16.5,
+      "learning_rate": 2.071826683501672e-06,
+      "loss": 0.9774,
+      "step": 6448
+    },
+    {
+      "epoch": 0.5495526203664253,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.071700063259639e-06,
+      "loss": 0.4172,
+      "step": 6449
+    },
+    {
+      "epoch": 0.5496378355347252,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.071573428168659e-06,
+      "loss": 0.3816,
+      "step": 6450
+    },
+    {
+      "epoch": 0.5497230507030252,
+      "grad_norm": 16.0,
+      "learning_rate": 2.071446778231021e-06,
+      "loss": 0.7898,
+      "step": 6451
+    },
+    {
+      "epoch": 0.5498082658713251,
+      "grad_norm": 11.25,
+      "learning_rate": 2.0713201134490143e-06,
+      "loss": 0.5958,
+      "step": 6452
+    },
+    {
+      "epoch": 0.5498934810396251,
+      "grad_norm": 24.375,
+      "learning_rate": 2.0711934338249266e-06,
+      "loss": 1.0397,
+      "step": 6453
+    },
+    {
+      "epoch": 0.549978696207925,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.0710667393610483e-06,
+      "loss": 0.8518,
+      "step": 6454
+    },
+    {
+      "epoch": 0.550063911376225,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.070940030059668e-06,
+      "loss": 0.7177,
+      "step": 6455
+    },
+    {
+      "epoch": 0.550149126544525,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.070813305923076e-06,
+      "loss": 1.0817,
+      "step": 6456
+    },
+    {
+      "epoch": 0.5502343417128249,
+      "grad_norm": 16.75,
+      "learning_rate": 2.0706865669535627e-06,
+      "loss": 0.9834,
+      "step": 6457
+    },
+    {
+      "epoch": 0.5503195568811249,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.070559813153417e-06,
+      "loss": 0.5247,
+      "step": 6458
+    },
+    {
+      "epoch": 0.5504047720494248,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0704330445249306e-06,
+      "loss": 0.5587,
+      "step": 6459
+    },
+    {
+      "epoch": 0.5504899872177248,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.0703062610703944e-06,
+      "loss": 0.6146,
+      "step": 6460
+    },
+    {
+      "epoch": 0.5505752023860248,
+      "grad_norm": 16.75,
+      "learning_rate": 2.070179462792099e-06,
+      "loss": 0.3817,
+      "step": 6461
+    },
+    {
+      "epoch": 0.5506604175543247,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0700526496923355e-06,
+      "loss": 1.0213,
+      "step": 6462
+    },
+    {
+      "epoch": 0.5507456327226247,
+      "grad_norm": 19.375,
+      "learning_rate": 2.069925821773397e-06,
+      "loss": 1.2363,
+      "step": 6463
+    },
+    {
+      "epoch": 0.5508308478909246,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0697989790375734e-06,
+      "loss": 0.8194,
+      "step": 6464
+    },
+    {
+      "epoch": 0.5509160630592246,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0696721214871586e-06,
+      "loss": 0.3935,
+      "step": 6465
+    },
+    {
+      "epoch": 0.5510012782275245,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.069545249124444e-06,
+      "loss": 0.5125,
+      "step": 6466
+    },
+    {
+      "epoch": 0.5510864933958245,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.069418361951722e-06,
+      "loss": 0.4013,
+      "step": 6467
+    },
+    {
+      "epoch": 0.5511717085641245,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.069291459971287e-06,
+      "loss": 0.5735,
+      "step": 6468
+    },
+    {
+      "epoch": 0.5512569237324244,
+      "grad_norm": 19.5,
+      "learning_rate": 2.069164543185432e-06,
+      "loss": 0.5107,
+      "step": 6469
+    },
+    {
+      "epoch": 0.5513421389007244,
+      "grad_norm": 24.125,
+      "learning_rate": 2.069037611596449e-06,
+      "loss": 0.5876,
+      "step": 6470
+    },
+    {
+      "epoch": 0.5514273540690243,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0689106652066327e-06,
+      "loss": 0.5836,
+      "step": 6471
+    },
+    {
+      "epoch": 0.5515125692373243,
+      "grad_norm": 35.0,
+      "learning_rate": 2.0687837040182775e-06,
+      "loss": 1.0261,
+      "step": 6472
+    },
+    {
+      "epoch": 0.5515977844056242,
+      "grad_norm": 17.375,
+      "learning_rate": 2.068656728033678e-06,
+      "loss": 0.525,
+      "step": 6473
+    },
+    {
+      "epoch": 0.5516829995739242,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.068529737255127e-06,
+      "loss": 0.2721,
+      "step": 6474
+    },
+    {
+      "epoch": 0.5517682147422242,
+      "grad_norm": 11.0,
+      "learning_rate": 2.0684027316849212e-06,
+      "loss": 0.371,
+      "step": 6475
+    },
+    {
+      "epoch": 0.5518534299105241,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.0682757113253554e-06,
+      "loss": 0.374,
+      "step": 6476
+    },
+    {
+      "epoch": 0.5519386450788241,
+      "grad_norm": 17.75,
+      "learning_rate": 2.068148676178724e-06,
+      "loss": 0.8038,
+      "step": 6477
+    },
+    {
+      "epoch": 0.552023860247124,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0680216262473237e-06,
+      "loss": 0.8735,
+      "step": 6478
+    },
+    {
+      "epoch": 0.552109075415424,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.06789456153345e-06,
+      "loss": 0.7367,
+      "step": 6479
+    },
+    {
+      "epoch": 0.5521942905837239,
+      "grad_norm": 10.0,
+      "learning_rate": 2.067767482039399e-06,
+      "loss": 0.302,
+      "step": 6480
+    },
+    {
+      "epoch": 0.5522795057520239,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0676403877674677e-06,
+      "loss": 0.9204,
+      "step": 6481
+    },
+    {
+      "epoch": 0.5523647209203238,
+      "grad_norm": 17.5,
+      "learning_rate": 2.067513278719952e-06,
+      "loss": 0.6165,
+      "step": 6482
+    },
+    {
+      "epoch": 0.5524499360886238,
+      "grad_norm": 25.125,
+      "learning_rate": 2.067386154899149e-06,
+      "loss": 0.969,
+      "step": 6483
+    },
+    {
+      "epoch": 0.5525351512569238,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0672590163073566e-06,
+      "loss": 0.919,
+      "step": 6484
+    },
+    {
+      "epoch": 0.5526203664252237,
+      "grad_norm": 14.0,
+      "learning_rate": 2.067131862946872e-06,
+      "loss": 0.6028,
+      "step": 6485
+    },
+    {
+      "epoch": 0.5527055815935237,
+      "grad_norm": 18.625,
+      "learning_rate": 2.067004694819993e-06,
+      "loss": 0.8,
+      "step": 6486
+    },
+    {
+      "epoch": 0.5527907967618236,
+      "grad_norm": 13.5,
+      "learning_rate": 2.0668775119290176e-06,
+      "loss": 0.6528,
+      "step": 6487
+    },
+    {
+      "epoch": 0.5528760119301236,
+      "grad_norm": 18.125,
+      "learning_rate": 2.0667503142762435e-06,
+      "loss": 0.7239,
+      "step": 6488
+    },
+    {
+      "epoch": 0.5529612270984235,
+      "grad_norm": 12.125,
+      "learning_rate": 2.0666231018639707e-06,
+      "loss": 0.7263,
+      "step": 6489
+    },
+    {
+      "epoch": 0.5530464422667235,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.066495874694497e-06,
+      "loss": 0.6555,
+      "step": 6490
+    },
+    {
+      "epoch": 0.5531316574350235,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0663686327701215e-06,
+      "loss": 0.5581,
+      "step": 6491
+    },
+    {
+      "epoch": 0.5532168726033234,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.0662413760931444e-06,
+      "loss": 0.7057,
+      "step": 6492
+    },
+    {
+      "epoch": 0.5533020877716234,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.0661141046658644e-06,
+      "loss": 0.528,
+      "step": 6493
+    },
+    {
+      "epoch": 0.5533873029399233,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.0659868184905825e-06,
+      "loss": 0.554,
+      "step": 6494
+    },
+    {
+      "epoch": 0.5534725181082233,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.0658595175695975e-06,
+      "loss": 0.9603,
+      "step": 6495
+    },
+    {
+      "epoch": 0.5535577332765232,
+      "grad_norm": 11.5,
+      "learning_rate": 2.0657322019052108e-06,
+      "loss": 0.4121,
+      "step": 6496
+    },
+    {
+      "epoch": 0.5536429484448232,
+      "grad_norm": 10.5,
+      "learning_rate": 2.0656048714997233e-06,
+      "loss": 0.4158,
+      "step": 6497
+    },
+    {
+      "epoch": 0.5537281636131232,
+      "grad_norm": 18.75,
+      "learning_rate": 2.0654775263554353e-06,
+      "loss": 0.738,
+      "step": 6498
+    },
+    {
+      "epoch": 0.5538133787814231,
+      "grad_norm": 18.5,
+      "learning_rate": 2.0653501664746485e-06,
+      "loss": 0.8041,
+      "step": 6499
+    },
+    {
+      "epoch": 0.5538985939497231,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.065222791859664e-06,
+      "loss": 0.5907,
+      "step": 6500
+    },
+    {
+      "epoch": 0.553983809118023,
+      "grad_norm": 13.125,
+      "learning_rate": 2.0650954025127844e-06,
+      "loss": 0.5922,
+      "step": 6501
+    },
+    {
+      "epoch": 0.554069024286323,
+      "grad_norm": 16.0,
+      "learning_rate": 2.0649679984363108e-06,
+      "loss": 0.8694,
+      "step": 6502
+    },
+    {
+      "epoch": 0.5541542394546229,
+      "grad_norm": 14.0,
+      "learning_rate": 2.064840579632546e-06,
+      "loss": 0.5095,
+      "step": 6503
+    },
+    {
+      "epoch": 0.5542394546229229,
+      "grad_norm": 21.0,
+      "learning_rate": 2.064713146103793e-06,
+      "loss": 0.5398,
+      "step": 6504
+    },
+    {
+      "epoch": 0.5543246697912229,
+      "grad_norm": 17.75,
+      "learning_rate": 2.064585697852354e-06,
+      "loss": 0.7407,
+      "step": 6505
+    },
+    {
+      "epoch": 0.5544098849595228,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.0644582348805316e-06,
+      "loss": 0.6482,
+      "step": 6506
+    },
+    {
+      "epoch": 0.5544951001278228,
+      "grad_norm": 22.0,
+      "learning_rate": 2.0643307571906306e-06,
+      "loss": 1.1116,
+      "step": 6507
+    },
+    {
+      "epoch": 0.5545803152961227,
+      "grad_norm": 10.75,
+      "learning_rate": 2.0642032647849535e-06,
+      "loss": 0.3904,
+      "step": 6508
+    },
+    {
+      "epoch": 0.5546655304644227,
+      "grad_norm": 16.5,
+      "learning_rate": 2.064075757665805e-06,
+      "loss": 0.6855,
+      "step": 6509
+    },
+    {
+      "epoch": 0.5547507456327226,
+      "grad_norm": 13.125,
+      "learning_rate": 2.063948235835489e-06,
+      "loss": 0.56,
+      "step": 6510
+    },
+    {
+      "epoch": 0.5548359608010226,
+      "grad_norm": 17.75,
+      "learning_rate": 2.0638206992963098e-06,
+      "loss": 1.0035,
+      "step": 6511
+    },
+    {
+      "epoch": 0.5549211759693226,
+      "grad_norm": 16.75,
+      "learning_rate": 2.063693148050572e-06,
+      "loss": 0.7331,
+      "step": 6512
+    },
+    {
+      "epoch": 0.5550063911376225,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.0635655821005813e-06,
+      "loss": 0.8202,
+      "step": 6513
+    },
+    {
+      "epoch": 0.5550916063059225,
+      "grad_norm": 12.375,
+      "learning_rate": 2.063438001448642e-06,
+      "loss": 0.4944,
+      "step": 6514
+    },
+    {
+      "epoch": 0.5551768214742224,
+      "grad_norm": 12.25,
+      "learning_rate": 2.06331040609706e-06,
+      "loss": 0.6298,
+      "step": 6515
+    },
+    {
+      "epoch": 0.5552620366425224,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0631827960481414e-06,
+      "loss": 0.8438,
+      "step": 6516
+    },
+    {
+      "epoch": 0.5553472518108223,
+      "grad_norm": 20.25,
+      "learning_rate": 2.0630551713041915e-06,
+      "loss": 0.7935,
+      "step": 6517
+    },
+    {
+      "epoch": 0.5554324669791223,
+      "grad_norm": 22.375,
+      "learning_rate": 2.062927531867518e-06,
+      "loss": 1.115,
+      "step": 6518
+    },
+    {
+      "epoch": 0.5555176821474223,
+      "grad_norm": 28.25,
+      "learning_rate": 2.0627998777404252e-06,
+      "loss": 1.2354,
+      "step": 6519
+    },
+    {
+      "epoch": 0.5556028973157222,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.0626722089252223e-06,
+      "loss": 1.0337,
+      "step": 6520
+    },
+    {
+      "epoch": 0.5556881124840222,
+      "grad_norm": 12.875,
+      "learning_rate": 2.062544525424215e-06,
+      "loss": 0.6304,
+      "step": 6521
+    },
+    {
+      "epoch": 0.5557733276523221,
+      "grad_norm": 14.5,
+      "learning_rate": 2.062416827239711e-06,
+      "loss": 0.7317,
+      "step": 6522
+    },
+    {
+      "epoch": 0.5558585428206221,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.0622891143740182e-06,
+      "loss": 0.5654,
+      "step": 6523
+    },
+    {
+      "epoch": 0.555943757988922,
+      "grad_norm": 17.75,
+      "learning_rate": 2.062161386829444e-06,
+      "loss": 0.8162,
+      "step": 6524
+    },
+    {
+      "epoch": 0.556028973157222,
+      "grad_norm": 14.0,
+      "learning_rate": 2.062033644608297e-06,
+      "loss": 0.9634,
+      "step": 6525
+    },
+    {
+      "epoch": 0.556114188325522,
+      "grad_norm": 12.375,
+      "learning_rate": 2.0619058877128858e-06,
+      "loss": 0.4097,
+      "step": 6526
+    },
+    {
+      "epoch": 0.5561994034938219,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.0617781161455183e-06,
+      "loss": 0.8011,
+      "step": 6527
+    },
+    {
+      "epoch": 0.5562846186621219,
+      "grad_norm": 12.5,
+      "learning_rate": 2.061650329908504e-06,
+      "loss": 0.6,
+      "step": 6528
+    },
+    {
+      "epoch": 0.5563698338304218,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0615225290041525e-06,
+      "loss": 0.3198,
+      "step": 6529
+    },
+    {
+      "epoch": 0.5564550489987218,
+      "grad_norm": 41.5,
+      "learning_rate": 2.0613947134347723e-06,
+      "loss": 1.2413,
+      "step": 6530
+    },
+    {
+      "epoch": 0.5565402641670217,
+      "grad_norm": 22.625,
+      "learning_rate": 2.0612668832026737e-06,
+      "loss": 1.0818,
+      "step": 6531
+    },
+    {
+      "epoch": 0.5566254793353217,
+      "grad_norm": 17.125,
+      "learning_rate": 2.061139038310167e-06,
+      "loss": 0.8843,
+      "step": 6532
+    },
+    {
+      "epoch": 0.5567106945036217,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0610111787595623e-06,
+      "loss": 0.8752,
+      "step": 6533
+    },
+    {
+      "epoch": 0.5567959096719216,
+      "grad_norm": 18.625,
+      "learning_rate": 2.06088330455317e-06,
+      "loss": 0.8696,
+      "step": 6534
+    },
+    {
+      "epoch": 0.5568811248402216,
+      "grad_norm": 18.75,
+      "learning_rate": 2.060755415693301e-06,
+      "loss": 1.0898,
+      "step": 6535
+    },
+    {
+      "epoch": 0.5569663400085215,
+      "grad_norm": 29.0,
+      "learning_rate": 2.060627512182266e-06,
+      "loss": 0.8938,
+      "step": 6536
+    },
+    {
+      "epoch": 0.5570515551768215,
+      "grad_norm": 12.75,
+      "learning_rate": 2.060499594022377e-06,
+      "loss": 0.6198,
+      "step": 6537
+    },
+    {
+      "epoch": 0.5571367703451214,
+      "grad_norm": 24.0,
+      "learning_rate": 2.0603716612159458e-06,
+      "loss": 0.743,
+      "step": 6538
+    },
+    {
+      "epoch": 0.5572219855134214,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.060243713765283e-06,
+      "loss": 0.3479,
+      "step": 6539
+    },
+    {
+      "epoch": 0.5573072006817213,
+      "grad_norm": 33.25,
+      "learning_rate": 2.0601157516727018e-06,
+      "loss": 0.789,
+      "step": 6540
+    },
+    {
+      "epoch": 0.5573924158500213,
+      "grad_norm": 27.875,
+      "learning_rate": 2.0599877749405146e-06,
+      "loss": 0.8828,
+      "step": 6541
+    },
+    {
+      "epoch": 0.5574776310183213,
+      "grad_norm": 13.0,
+      "learning_rate": 2.059859783571033e-06,
+      "loss": 0.7172,
+      "step": 6542
+    },
+    {
+      "epoch": 0.5575628461866212,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0597317775665714e-06,
+      "loss": 0.6149,
+      "step": 6543
+    },
+    {
+      "epoch": 0.5576480613549212,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.0596037569294423e-06,
+      "loss": 1.1744,
+      "step": 6544
+    },
+    {
+      "epoch": 0.5577332765232211,
+      "grad_norm": 14.25,
+      "learning_rate": 2.0594757216619595e-06,
+      "loss": 0.7805,
+      "step": 6545
+    },
+    {
+      "epoch": 0.5578184916915211,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.059347671766436e-06,
+      "loss": 0.7751,
+      "step": 6546
+    },
+    {
+      "epoch": 0.557903706859821,
+      "grad_norm": 14.25,
+      "learning_rate": 2.0592196072451857e-06,
+      "loss": 0.6645,
+      "step": 6547
+    },
+    {
+      "epoch": 0.557988922028121,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0590915281005243e-06,
+      "loss": 0.8242,
+      "step": 6548
+    },
+    {
+      "epoch": 0.558074137196421,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.0589634343347646e-06,
+      "loss": 0.9557,
+      "step": 6549
+    },
+    {
+      "epoch": 0.5581593523647209,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.058835325950223e-06,
+      "loss": 0.6361,
+      "step": 6550
+    },
+    {
+      "epoch": 0.5582445675330209,
+      "grad_norm": 16.0,
+      "learning_rate": 2.0587072029492134e-06,
+      "loss": 1.0102,
+      "step": 6551
+    },
+    {
+      "epoch": 0.5583297827013208,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.058579065334051e-06,
+      "loss": 0.5505,
+      "step": 6552
+    },
+    {
+      "epoch": 0.5584149978696208,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.0584509131070522e-06,
+      "loss": 0.3969,
+      "step": 6553
+    },
+    {
+      "epoch": 0.5585002130379207,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.0583227462705326e-06,
+      "loss": 0.689,
+      "step": 6554
+    },
+    {
+      "epoch": 0.5585854282062207,
+      "grad_norm": 18.75,
+      "learning_rate": 2.0581945648268077e-06,
+      "loss": 0.892,
+      "step": 6555
+    },
+    {
+      "epoch": 0.5586706433745207,
+      "grad_norm": 13.5,
+      "learning_rate": 2.0580663687781944e-06,
+      "loss": 0.6407,
+      "step": 6556
+    },
+    {
+      "epoch": 0.5587558585428206,
+      "grad_norm": 13.875,
+      "learning_rate": 2.057938158127009e-06,
+      "loss": 0.5784,
+      "step": 6557
+    },
+    {
+      "epoch": 0.5588410737111206,
+      "grad_norm": 11.75,
+      "learning_rate": 2.057809932875569e-06,
+      "loss": 0.5397,
+      "step": 6558
+    },
+    {
+      "epoch": 0.5589262888794205,
+      "grad_norm": 19.875,
+      "learning_rate": 2.0576816930261917e-06,
+      "loss": 0.9131,
+      "step": 6559
+    },
+    {
+      "epoch": 0.5590115040477205,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.0575534385811932e-06,
+      "loss": 0.5312,
+      "step": 6560
+    },
+    {
+      "epoch": 0.5590967192160204,
+      "grad_norm": 15.5,
+      "learning_rate": 2.057425169542892e-06,
+      "loss": 0.9385,
+      "step": 6561
+    },
+    {
+      "epoch": 0.5591819343843204,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.0572968859136066e-06,
+      "loss": 0.7566,
+      "step": 6562
+    },
+    {
+      "epoch": 0.5592671495526204,
+      "grad_norm": 16.875,
+      "learning_rate": 2.0571685876956542e-06,
+      "loss": 0.6952,
+      "step": 6563
+    },
+    {
+      "epoch": 0.5593523647209203,
+      "grad_norm": 20.0,
+      "learning_rate": 2.057040274891354e-06,
+      "loss": 1.1375,
+      "step": 6564
+    },
+    {
+      "epoch": 0.5594375798892203,
+      "grad_norm": 17.625,
+      "learning_rate": 2.056911947503025e-06,
+      "loss": 1.2299,
+      "step": 6565
+    },
+    {
+      "epoch": 0.5595227950575202,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.056783605532985e-06,
+      "loss": 0.3876,
+      "step": 6566
+    },
+    {
+      "epoch": 0.5596080102258202,
+      "grad_norm": 13.5,
+      "learning_rate": 2.056655248983554e-06,
+      "loss": 0.4853,
+      "step": 6567
+    },
+    {
+      "epoch": 0.5596932253941201,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0565268778570518e-06,
+      "loss": 0.9482,
+      "step": 6568
+    },
+    {
+      "epoch": 0.5597784405624201,
+      "grad_norm": 12.125,
+      "learning_rate": 2.0563984921557973e-06,
+      "loss": 0.5212,
+      "step": 6569
+    },
+    {
+      "epoch": 0.5598636557307201,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.0562700918821117e-06,
+      "loss": 0.7779,
+      "step": 6570
+    },
+    {
+      "epoch": 0.55994887089902,
+      "grad_norm": 14.0,
+      "learning_rate": 2.0561416770383147e-06,
+      "loss": 0.9288,
+      "step": 6571
+    },
+    {
+      "epoch": 0.56003408606732,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.056013247626727e-06,
+      "loss": 1.0304,
+      "step": 6572
+    },
+    {
+      "epoch": 0.5601193012356199,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.0558848036496693e-06,
+      "loss": 0.7677,
+      "step": 6573
+    },
+    {
+      "epoch": 0.5602045164039199,
+      "grad_norm": 22.0,
+      "learning_rate": 2.0557563451094627e-06,
+      "loss": 0.4116,
+      "step": 6574
+    },
+    {
+      "epoch": 0.5602897315722198,
+      "grad_norm": 22.375,
+      "learning_rate": 2.055627872008429e-06,
+      "loss": 1.3887,
+      "step": 6575
+    },
+    {
+      "epoch": 0.5603749467405198,
+      "grad_norm": 20.875,
+      "learning_rate": 2.055499384348889e-06,
+      "loss": 1.071,
+      "step": 6576
+    },
+    {
+      "epoch": 0.5604601619088198,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.0553708821331657e-06,
+      "loss": 0.7193,
+      "step": 6577
+    },
+    {
+      "epoch": 0.5605453770771197,
+      "grad_norm": 17.75,
+      "learning_rate": 2.0552423653635805e-06,
+      "loss": 0.8723,
+      "step": 6578
+    },
+    {
+      "epoch": 0.5606305922454197,
+      "grad_norm": 16.5,
+      "learning_rate": 2.055113834042456e-06,
+      "loss": 0.5388,
+      "step": 6579
+    },
+    {
+      "epoch": 0.5607158074137196,
+      "grad_norm": 11.25,
+      "learning_rate": 2.0549852881721148e-06,
+      "loss": 0.4838,
+      "step": 6580
+    },
+    {
+      "epoch": 0.5608010225820196,
+      "grad_norm": 15.125,
+      "learning_rate": 2.05485672775488e-06,
+      "loss": 0.8816,
+      "step": 6581
+    },
+    {
+      "epoch": 0.5608862377503195,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.0547281527930753e-06,
+      "loss": 0.5521,
+      "step": 6582
+    },
+    {
+      "epoch": 0.5609714529186195,
+      "grad_norm": 19.375,
+      "learning_rate": 2.054599563289023e-06,
+      "loss": 1.0904,
+      "step": 6583
+    },
+    {
+      "epoch": 0.5610566680869195,
+      "grad_norm": 13.375,
+      "learning_rate": 2.0544709592450473e-06,
+      "loss": 0.6103,
+      "step": 6584
+    },
+    {
+      "epoch": 0.5611418832552194,
+      "grad_norm": 16.0,
+      "learning_rate": 2.054342340663473e-06,
+      "loss": 0.9819,
+      "step": 6585
+    },
+    {
+      "epoch": 0.5612270984235194,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.0542137075466234e-06,
+      "loss": 0.8006,
+      "step": 6586
+    },
+    {
+      "epoch": 0.5613123135918193,
+      "grad_norm": 14.875,
+      "learning_rate": 2.054085059896824e-06,
+      "loss": 0.698,
+      "step": 6587
+    },
+    {
+      "epoch": 0.5613975287601193,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0539563977163983e-06,
+      "loss": 0.7512,
+      "step": 6588
+    },
+    {
+      "epoch": 0.5614827439284192,
+      "grad_norm": 19.375,
+      "learning_rate": 2.0538277210076722e-06,
+      "loss": 1.2143,
+      "step": 6589
+    },
+    {
+      "epoch": 0.5615679590967192,
+      "grad_norm": 31.5,
+      "learning_rate": 2.0536990297729715e-06,
+      "loss": 1.0487,
+      "step": 6590
+    },
+    {
+      "epoch": 0.5616531742650192,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0535703240146207e-06,
+      "loss": 1.0768,
+      "step": 6591
+    },
+    {
+      "epoch": 0.5617383894333191,
+      "grad_norm": 13.75,
+      "learning_rate": 2.053441603734946e-06,
+      "loss": 0.8349,
+      "step": 6592
+    },
+    {
+      "epoch": 0.5618236046016191,
+      "grad_norm": 13.25,
+      "learning_rate": 2.0533128689362738e-06,
+      "loss": 0.6898,
+      "step": 6593
+    },
+    {
+      "epoch": 0.561908819769919,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.05318411962093e-06,
+      "loss": 0.6294,
+      "step": 6594
+    },
+    {
+      "epoch": 0.561994034938219,
+      "grad_norm": 22.25,
+      "learning_rate": 2.0530553557912417e-06,
+      "loss": 1.0585,
+      "step": 6595
+    },
+    {
+      "epoch": 0.5620792501065189,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.0529265774495357e-06,
+      "loss": 0.6274,
+      "step": 6596
+    },
+    {
+      "epoch": 0.5621644652748189,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0527977845981383e-06,
+      "loss": 0.5907,
+      "step": 6597
+    },
+    {
+      "epoch": 0.5622496804431188,
+      "grad_norm": 13.125,
+      "learning_rate": 2.052668977239378e-06,
+      "loss": 0.4528,
+      "step": 6598
+    },
+    {
+      "epoch": 0.5623348956114188,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.052540155375583e-06,
+      "loss": 0.7443,
+      "step": 6599
+    },
+    {
+      "epoch": 0.5624201107797188,
+      "grad_norm": 14.0,
+      "learning_rate": 2.0524113190090795e-06,
+      "loss": 0.7788,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5625053259480187,
+      "grad_norm": 12.375,
+      "learning_rate": 2.0522824681421967e-06,
+      "loss": 0.5758,
+      "step": 6601
+    },
+    {
+      "epoch": 0.5625905411163187,
+      "grad_norm": 18.5,
+      "learning_rate": 2.0521536027772633e-06,
+      "loss": 0.8529,
+      "step": 6602
+    },
+    {
+      "epoch": 0.5626757562846186,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.0520247229166075e-06,
+      "loss": 0.489,
+      "step": 6603
+    },
+    {
+      "epoch": 0.5627609714529186,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0518958285625585e-06,
+      "loss": 0.9931,
+      "step": 6604
+    },
+    {
+      "epoch": 0.5628461866212185,
+      "grad_norm": 16.125,
+      "learning_rate": 2.051766919717445e-06,
+      "loss": 0.8584,
+      "step": 6605
+    },
+    {
+      "epoch": 0.5629314017895185,
+      "grad_norm": 21.625,
+      "learning_rate": 2.0516379963835977e-06,
+      "loss": 0.7719,
+      "step": 6606
+    },
+    {
+      "epoch": 0.5630166169578185,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0515090585633453e-06,
+      "loss": 1.024,
+      "step": 6607
+    },
+    {
+      "epoch": 0.5631018321261184,
+      "grad_norm": 12.375,
+      "learning_rate": 2.0513801062590184e-06,
+      "loss": 0.4838,
+      "step": 6608
+    },
+    {
+      "epoch": 0.5631870472944184,
+      "grad_norm": 12.25,
+      "learning_rate": 2.051251139472947e-06,
+      "loss": 0.6111,
+      "step": 6609
+    },
+    {
+      "epoch": 0.5632722624627183,
+      "grad_norm": 13.5,
+      "learning_rate": 2.051122158207462e-06,
+      "loss": 0.5268,
+      "step": 6610
+    },
+    {
+      "epoch": 0.5633574776310183,
+      "grad_norm": 11.25,
+      "learning_rate": 2.050993162464894e-06,
+      "loss": 0.4595,
+      "step": 6611
+    },
+    {
+      "epoch": 0.5634426927993182,
+      "grad_norm": 12.625,
+      "learning_rate": 2.0508641522475734e-06,
+      "loss": 0.3808,
+      "step": 6612
+    },
+    {
+      "epoch": 0.5635279079676182,
+      "grad_norm": 11.875,
+      "learning_rate": 2.050735127557833e-06,
+      "loss": 0.525,
+      "step": 6613
+    },
+    {
+      "epoch": 0.5636131231359182,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.0506060883980034e-06,
+      "loss": 0.6565,
+      "step": 6614
+    },
+    {
+      "epoch": 0.5636983383042181,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0504770347704167e-06,
+      "loss": 1.0895,
+      "step": 6615
+    },
+    {
+      "epoch": 0.5637835534725181,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.050347966677405e-06,
+      "loss": 0.4161,
+      "step": 6616
+    },
+    {
+      "epoch": 0.563868768640818,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.0502188841213012e-06,
+      "loss": 0.9376,
+      "step": 6617
+    },
+    {
+      "epoch": 0.563953983809118,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.0500897871044366e-06,
+      "loss": 0.7401,
+      "step": 6618
+    },
+    {
+      "epoch": 0.5640391989774179,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.049960675629146e-06,
+      "loss": 0.6603,
+      "step": 6619
+    },
+    {
+      "epoch": 0.5641244141457179,
+      "grad_norm": 16.5,
+      "learning_rate": 2.049831549697761e-06,
+      "loss": 0.6978,
+      "step": 6620
+    },
+    {
+      "epoch": 0.5642096293140179,
+      "grad_norm": 11.75,
+      "learning_rate": 2.0497024093126155e-06,
+      "loss": 0.4691,
+      "step": 6621
+    },
+    {
+      "epoch": 0.5642948444823178,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.0495732544760435e-06,
+      "loss": 0.8976,
+      "step": 6622
+    },
+    {
+      "epoch": 0.5643800596506178,
+      "grad_norm": 18.75,
+      "learning_rate": 2.049444085190379e-06,
+      "loss": 0.2405,
+      "step": 6623
+    },
+    {
+      "epoch": 0.5644652748189177,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.0493149014579557e-06,
+      "loss": 0.2641,
+      "step": 6624
+    },
+    {
+      "epoch": 0.5645504899872177,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.049185703281108e-06,
+      "loss": 0.4319,
+      "step": 6625
+    },
+    {
+      "epoch": 0.5646357051555176,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.0490564906621717e-06,
+      "loss": 0.4829,
+      "step": 6626
+    },
+    {
+      "epoch": 0.5647209203238176,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0489272636034808e-06,
+      "loss": 0.7447,
+      "step": 6627
+    },
+    {
+      "epoch": 0.5648061354921176,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0487980221073707e-06,
+      "loss": 0.9176,
+      "step": 6628
+    },
+    {
+      "epoch": 0.5648913506604175,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0486687661761773e-06,
+      "loss": 1.0748,
+      "step": 6629
+    },
+    {
+      "epoch": 0.5649765658287175,
+      "grad_norm": 17.375,
+      "learning_rate": 2.0485394958122357e-06,
+      "loss": 0.6985,
+      "step": 6630
+    },
+    {
+      "epoch": 0.5650617809970174,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.048410211017883e-06,
+      "loss": 0.9069,
+      "step": 6631
+    },
+    {
+      "epoch": 0.5651469961653174,
+      "grad_norm": 15.5,
+      "learning_rate": 2.048280911795455e-06,
+      "loss": 0.538,
+      "step": 6632
+    },
+    {
+      "epoch": 0.5652322113336173,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.0481515981472877e-06,
+      "loss": 0.3747,
+      "step": 6633
+    },
+    {
+      "epoch": 0.5653174265019173,
+      "grad_norm": 10.375,
+      "learning_rate": 2.0480222700757186e-06,
+      "loss": 0.434,
+      "step": 6634
+    },
+    {
+      "epoch": 0.5654026416702173,
+      "grad_norm": 20.5,
+      "learning_rate": 2.0478929275830845e-06,
+      "loss": 0.8986,
+      "step": 6635
+    },
+    {
+      "epoch": 0.5654878568385172,
+      "grad_norm": 11.625,
+      "learning_rate": 2.047763570671723e-06,
+      "loss": 0.4706,
+      "step": 6636
+    },
+    {
+      "epoch": 0.5655730720068172,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0476341993439718e-06,
+      "loss": 1.1335,
+      "step": 6637
+    },
+    {
+      "epoch": 0.5656582871751171,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.047504813602168e-06,
+      "loss": 1.0077,
+      "step": 6638
+    },
+    {
+      "epoch": 0.5657435023434171,
+      "grad_norm": 15.5,
+      "learning_rate": 2.047375413448651e-06,
+      "loss": 0.6693,
+      "step": 6639
+    },
+    {
+      "epoch": 0.565828717511717,
+      "grad_norm": 14.5,
+      "learning_rate": 2.047245998885758e-06,
+      "loss": 0.9674,
+      "step": 6640
+    },
+    {
+      "epoch": 0.565913932680017,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.047116569915828e-06,
+      "loss": 0.5483,
+      "step": 6641
+    },
+    {
+      "epoch": 0.565999147848317,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.0469871265412e-06,
+      "loss": 0.8026,
+      "step": 6642
+    },
+    {
+      "epoch": 0.5660843630166169,
+      "grad_norm": 19.75,
+      "learning_rate": 2.0468576687642135e-06,
+      "loss": 0.9924,
+      "step": 6643
+    },
+    {
+      "epoch": 0.5661695781849169,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.0467281965872073e-06,
+      "loss": 0.6448,
+      "step": 6644
+    },
+    {
+      "epoch": 0.5662547933532168,
+      "grad_norm": 16.625,
+      "learning_rate": 2.046598710012522e-06,
+      "loss": 0.8509,
+      "step": 6645
+    },
+    {
+      "epoch": 0.5663400085215168,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0464692090424963e-06,
+      "loss": 0.5845,
+      "step": 6646
+    },
+    {
+      "epoch": 0.5664252236898167,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.046339693679471e-06,
+      "loss": 0.8635,
+      "step": 6647
+    },
+    {
+      "epoch": 0.5665104388581167,
+      "grad_norm": 15.25,
+      "learning_rate": 2.046210163925787e-06,
+      "loss": 0.8912,
+      "step": 6648
+    },
+    {
+      "epoch": 0.5665956540264167,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0460806197837845e-06,
+      "loss": 0.6407,
+      "step": 6649
+    },
+    {
+      "epoch": 0.5666808691947166,
+      "grad_norm": 16.25,
+      "learning_rate": 2.045951061255805e-06,
+      "loss": 0.7689,
+      "step": 6650
+    },
+    {
+      "epoch": 0.5667660843630166,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0458214883441897e-06,
+      "loss": 0.8556,
+      "step": 6651
+    },
+    {
+      "epoch": 0.5668512995313165,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0456919010512793e-06,
+      "loss": 0.4139,
+      "step": 6652
+    },
+    {
+      "epoch": 0.5669365146996165,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0455622993794162e-06,
+      "loss": 0.83,
+      "step": 6653
+    },
+    {
+      "epoch": 0.5670217298679164,
+      "grad_norm": 14.25,
+      "learning_rate": 2.0454326833309426e-06,
+      "loss": 1.0854,
+      "step": 6654
+    },
+    {
+      "epoch": 0.5671069450362164,
+      "grad_norm": 19.375,
+      "learning_rate": 2.0453030529082e-06,
+      "loss": 0.8774,
+      "step": 6655
+    },
+    {
+      "epoch": 0.5671921602045163,
+      "grad_norm": 19.25,
+      "learning_rate": 2.045173408113532e-06,
+      "loss": 0.7203,
+      "step": 6656
+    },
+    {
+      "epoch": 0.5672773753728163,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0450437489492806e-06,
+      "loss": 0.932,
+      "step": 6657
+    },
+    {
+      "epoch": 0.5673625905411164,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.0449140754177893e-06,
+      "loss": 0.4206,
+      "step": 6658
+    },
+    {
+      "epoch": 0.5674478057094163,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.0447843875214014e-06,
+      "loss": 0.7666,
+      "step": 6659
+    },
+    {
+      "epoch": 0.5675330208777163,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0446546852624603e-06,
+      "loss": 0.71,
+      "step": 6660
+    },
+    {
+      "epoch": 0.5676182360460162,
+      "grad_norm": 22.625,
+      "learning_rate": 2.04452496864331e-06,
+      "loss": 0.8318,
+      "step": 6661
+    },
+    {
+      "epoch": 0.5677034512143162,
+      "grad_norm": 13.5,
+      "learning_rate": 2.0443952376662946e-06,
+      "loss": 0.7203,
+      "step": 6662
+    },
+    {
+      "epoch": 0.5677886663826162,
+      "grad_norm": 12.0,
+      "learning_rate": 2.044265492333758e-06,
+      "loss": 0.5695,
+      "step": 6663
+    },
+    {
+      "epoch": 0.5678738815509161,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.044135732648046e-06,
+      "loss": 0.4586,
+      "step": 6664
+    },
+    {
+      "epoch": 0.5679590967192161,
+      "grad_norm": 15.5,
+      "learning_rate": 2.044005958611502e-06,
+      "loss": 0.9225,
+      "step": 6665
+    },
+    {
+      "epoch": 0.568044311887516,
+      "grad_norm": 10.5,
+      "learning_rate": 2.0438761702264722e-06,
+      "loss": 0.3583,
+      "step": 6666
+    },
+    {
+      "epoch": 0.568129527055816,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.043746367495302e-06,
+      "loss": 0.6217,
+      "step": 6667
+    },
+    {
+      "epoch": 0.5682147422241159,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0436165504203366e-06,
+      "loss": 0.9287,
+      "step": 6668
+    },
+    {
+      "epoch": 0.5682999573924159,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.043486719003922e-06,
+      "loss": 0.4519,
+      "step": 6669
+    },
+    {
+      "epoch": 0.5683851725607159,
+      "grad_norm": 13.875,
+      "learning_rate": 2.043356873248404e-06,
+      "loss": 0.7129,
+      "step": 6670
+    },
+    {
+      "epoch": 0.5684703877290158,
+      "grad_norm": 14.0,
+      "learning_rate": 2.0432270131561304e-06,
+      "loss": 0.3966,
+      "step": 6671
+    },
+    {
+      "epoch": 0.5685556028973158,
+      "grad_norm": 19.0,
+      "learning_rate": 2.043097138729447e-06,
+      "loss": 1.1598,
+      "step": 6672
+    },
+    {
+      "epoch": 0.5686408180656157,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0429672499707e-06,
+      "loss": 0.8472,
+      "step": 6673
+    },
+    {
+      "epoch": 0.5687260332339157,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.0428373468822376e-06,
+      "loss": 0.3286,
+      "step": 6674
+    },
+    {
+      "epoch": 0.5688112484022156,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0427074294664074e-06,
+      "loss": 0.8259,
+      "step": 6675
+    },
+    {
+      "epoch": 0.5688964635705156,
+      "grad_norm": 13.375,
+      "learning_rate": 2.0425774977255567e-06,
+      "loss": 0.7435,
+      "step": 6676
+    },
+    {
+      "epoch": 0.5689816787388156,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.042447551662034e-06,
+      "loss": 0.6939,
+      "step": 6677
+    },
+    {
+      "epoch": 0.5690668939071155,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.0423175912781867e-06,
+      "loss": 0.6179,
+      "step": 6678
+    },
+    {
+      "epoch": 0.5691521090754155,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.0421876165763637e-06,
+      "loss": 0.7054,
+      "step": 6679
+    },
+    {
+      "epoch": 0.5692373242437154,
+      "grad_norm": 18.625,
+      "learning_rate": 2.042057627558914e-06,
+      "loss": 0.9556,
+      "step": 6680
+    },
+    {
+      "epoch": 0.5693225394120154,
+      "grad_norm": 18.75,
+      "learning_rate": 2.0419276242281866e-06,
+      "loss": 1.0551,
+      "step": 6681
+    },
+    {
+      "epoch": 0.5694077545803153,
+      "grad_norm": 20.25,
+      "learning_rate": 2.041797606586531e-06,
+      "loss": 0.8718,
+      "step": 6682
+    },
+    {
+      "epoch": 0.5694929697486153,
+      "grad_norm": 19.625,
+      "learning_rate": 2.0416675746362956e-06,
+      "loss": 1.1494,
+      "step": 6683
+    },
+    {
+      "epoch": 0.5695781849169153,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0415375283798316e-06,
+      "loss": 0.7423,
+      "step": 6684
+    },
+    {
+      "epoch": 0.5696634000852152,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.0414074678194884e-06,
+      "loss": 0.6575,
+      "step": 6685
+    },
+    {
+      "epoch": 0.5697486152535152,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0412773929576165e-06,
+      "loss": 0.8467,
+      "step": 6686
+    },
+    {
+      "epoch": 0.5698338304218151,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.0411473037965663e-06,
+      "loss": 0.6279,
+      "step": 6687
+    },
+    {
+      "epoch": 0.5699190455901151,
+      "grad_norm": 12.25,
+      "learning_rate": 2.0410172003386892e-06,
+      "loss": 0.4611,
+      "step": 6688
+    },
+    {
+      "epoch": 0.570004260758415,
+      "grad_norm": 11.0,
+      "learning_rate": 2.0408870825863354e-06,
+      "loss": 0.5633,
+      "step": 6689
+    },
+    {
+      "epoch": 0.570089475926715,
+      "grad_norm": 17.375,
+      "learning_rate": 2.0407569505418566e-06,
+      "loss": 0.6571,
+      "step": 6690
+    },
+    {
+      "epoch": 0.570174691095015,
+      "grad_norm": 21.625,
+      "learning_rate": 2.040626804207605e-06,
+      "loss": 0.6264,
+      "step": 6691
+    },
+    {
+      "epoch": 0.5702599062633149,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.040496643585932e-06,
+      "loss": 0.5343,
+      "step": 6692
+    },
+    {
+      "epoch": 0.5703451214316149,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.040366468679189e-06,
+      "loss": 0.5013,
+      "step": 6693
+    },
+    {
+      "epoch": 0.5704303365999148,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0402362794897294e-06,
+      "loss": 0.494,
+      "step": 6694
+    },
+    {
+      "epoch": 0.5705155517682148,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0401060760199063e-06,
+      "loss": 0.6601,
+      "step": 6695
+    },
+    {
+      "epoch": 0.5706007669365147,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.0399758582720715e-06,
+      "loss": 0.5622,
+      "step": 6696
+    },
+    {
+      "epoch": 0.5706859821048147,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.0398456262485784e-06,
+      "loss": 0.5121,
+      "step": 6697
+    },
+    {
+      "epoch": 0.5707711972731146,
+      "grad_norm": 26.25,
+      "learning_rate": 2.0397153799517806e-06,
+      "loss": 0.86,
+      "step": 6698
+    },
+    {
+      "epoch": 0.5708564124414146,
+      "grad_norm": 14.875,
+      "learning_rate": 2.039585119384032e-06,
+      "loss": 0.8094,
+      "step": 6699
+    },
+    {
+      "epoch": 0.5709416276097146,
+      "grad_norm": 13.5,
+      "learning_rate": 2.039454844547686e-06,
+      "loss": 0.4824,
+      "step": 6700
+    },
+    {
+      "epoch": 0.5710268427780145,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.039324555445097e-06,
+      "loss": 0.5616,
+      "step": 6701
+    },
+    {
+      "epoch": 0.5711120579463145,
+      "grad_norm": 16.375,
+      "learning_rate": 2.03919425207862e-06,
+      "loss": 1.2018,
+      "step": 6702
+    },
+    {
+      "epoch": 0.5711972731146144,
+      "grad_norm": 16.125,
+      "learning_rate": 2.039063934450609e-06,
+      "loss": 0.7871,
+      "step": 6703
+    },
+    {
+      "epoch": 0.5712824882829144,
+      "grad_norm": 13.125,
+      "learning_rate": 2.0389336025634194e-06,
+      "loss": 0.7451,
+      "step": 6704
+    },
+    {
+      "epoch": 0.5713677034512143,
+      "grad_norm": 20.25,
+      "learning_rate": 2.038803256419406e-06,
+      "loss": 0.7874,
+      "step": 6705
+    },
+    {
+      "epoch": 0.5714529186195143,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.0386728960209247e-06,
+      "loss": 0.3616,
+      "step": 6706
+    },
+    {
+      "epoch": 0.5715381337878143,
+      "grad_norm": 12.5,
+      "learning_rate": 2.0385425213703313e-06,
+      "loss": 0.5933,
+      "step": 6707
+    },
+    {
+      "epoch": 0.5716233489561142,
+      "grad_norm": 17.375,
+      "learning_rate": 2.038412132469981e-06,
+      "loss": 0.989,
+      "step": 6708
+    },
+    {
+      "epoch": 0.5717085641244142,
+      "grad_norm": 11.5,
+      "learning_rate": 2.0382817293222314e-06,
+      "loss": 0.5472,
+      "step": 6709
+    },
+    {
+      "epoch": 0.5717937792927141,
+      "grad_norm": 16.25,
+      "learning_rate": 2.038151311929438e-06,
+      "loss": 0.7496,
+      "step": 6710
+    },
+    {
+      "epoch": 0.5718789944610141,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.038020880293958e-06,
+      "loss": 0.9538,
+      "step": 6711
+    },
+    {
+      "epoch": 0.571964209629314,
+      "grad_norm": 13.875,
+      "learning_rate": 2.037890434418148e-06,
+      "loss": 0.8813,
+      "step": 6712
+    },
+    {
+      "epoch": 0.572049424797614,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.0377599743043653e-06,
+      "loss": 0.3264,
+      "step": 6713
+    },
+    {
+      "epoch": 0.572134639965914,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.0376294999549684e-06,
+      "loss": 1.0381,
+      "step": 6714
+    },
+    {
+      "epoch": 0.5722198551342139,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.0374990113723136e-06,
+      "loss": 0.9154,
+      "step": 6715
+    },
+    {
+      "epoch": 0.5723050703025139,
+      "grad_norm": 16.75,
+      "learning_rate": 2.0373685085587604e-06,
+      "loss": 0.7372,
+      "step": 6716
+    },
+    {
+      "epoch": 0.5723902854708138,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.0372379915166664e-06,
+      "loss": 0.9768,
+      "step": 6717
+    },
+    {
+      "epoch": 0.5724755006391138,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.03710746024839e-06,
+      "loss": 0.5258,
+      "step": 6718
+    },
+    {
+      "epoch": 0.5725607158074137,
+      "grad_norm": 13.0,
+      "learning_rate": 2.0369769147562908e-06,
+      "loss": 0.7403,
+      "step": 6719
+    },
+    {
+      "epoch": 0.5726459309757137,
+      "grad_norm": 20.0,
+      "learning_rate": 2.0368463550427267e-06,
+      "loss": 0.8875,
+      "step": 6720
+    },
+    {
+      "epoch": 0.5727311461440137,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.0367157811100587e-06,
+      "loss": 0.5844,
+      "step": 6721
+    },
+    {
+      "epoch": 0.5728163613123136,
+      "grad_norm": 12.5,
+      "learning_rate": 2.0365851929606444e-06,
+      "loss": 0.351,
+      "step": 6722
+    },
+    {
+      "epoch": 0.5729015764806136,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.0364545905968454e-06,
+      "loss": 0.895,
+      "step": 6723
+    },
+    {
+      "epoch": 0.5729867916489135,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0363239740210206e-06,
+      "loss": 0.7221,
+      "step": 6724
+    },
+    {
+      "epoch": 0.5730720068172135,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0361933432355312e-06,
+      "loss": 0.857,
+      "step": 6725
+    },
+    {
+      "epoch": 0.5731572219855134,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0360626982427375e-06,
+      "loss": 0.8438,
+      "step": 6726
+    },
+    {
+      "epoch": 0.5732424371538134,
+      "grad_norm": 20.875,
+      "learning_rate": 2.035932039045e-06,
+      "loss": 1.0177,
+      "step": 6727
+    },
+    {
+      "epoch": 0.5733276523221134,
+      "grad_norm": 16.5,
+      "learning_rate": 2.035801365644681e-06,
+      "loss": 0.9042,
+      "step": 6728
+    },
+    {
+      "epoch": 0.5734128674904133,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.0356706780441406e-06,
+      "loss": 0.8774,
+      "step": 6729
+    },
+    {
+      "epoch": 0.5734980826587133,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0355399762457408e-06,
+      "loss": 0.7066,
+      "step": 6730
+    },
+    {
+      "epoch": 0.5735832978270132,
+      "grad_norm": 14.375,
+      "learning_rate": 2.035409260251844e-06,
+      "loss": 0.617,
+      "step": 6731
+    },
+    {
+      "epoch": 0.5736685129953132,
+      "grad_norm": 20.125,
+      "learning_rate": 2.035278530064812e-06,
+      "loss": 0.7918,
+      "step": 6732
+    },
+    {
+      "epoch": 0.5737537281636131,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.035147785687007e-06,
+      "loss": 0.4654,
+      "step": 6733
+    },
+    {
+      "epoch": 0.5738389433319131,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0350170271207924e-06,
+      "loss": 0.5677,
+      "step": 6734
+    },
+    {
+      "epoch": 0.573924158500213,
+      "grad_norm": 19.25,
+      "learning_rate": 2.034886254368531e-06,
+      "loss": 0.9539,
+      "step": 6735
+    },
+    {
+      "epoch": 0.574009373668513,
+      "grad_norm": 13.125,
+      "learning_rate": 2.034755467432585e-06,
+      "loss": 0.4915,
+      "step": 6736
+    },
+    {
+      "epoch": 0.574094588836813,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.0346246663153193e-06,
+      "loss": 0.5336,
+      "step": 6737
+    },
+    {
+      "epoch": 0.5741798040051129,
+      "grad_norm": 22.25,
+      "learning_rate": 2.0344938510190964e-06,
+      "loss": 1.0443,
+      "step": 6738
+    },
+    {
+      "epoch": 0.5742650191734129,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.034363021546281e-06,
+      "loss": 0.7953,
+      "step": 6739
+    },
+    {
+      "epoch": 0.5743502343417128,
+      "grad_norm": 19.5,
+      "learning_rate": 2.034232177899237e-06,
+      "loss": 0.9649,
+      "step": 6740
+    },
+    {
+      "epoch": 0.5744354495100128,
+      "grad_norm": 18.625,
+      "learning_rate": 2.034101320080329e-06,
+      "loss": 0.8548,
+      "step": 6741
+    },
+    {
+      "epoch": 0.5745206646783128,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.0339704480919216e-06,
+      "loss": 0.7896,
+      "step": 6742
+    },
+    {
+      "epoch": 0.5746058798466127,
+      "grad_norm": 14.0,
+      "learning_rate": 2.0338395619363794e-06,
+      "loss": 0.7382,
+      "step": 6743
+    },
+    {
+      "epoch": 0.5746910950149127,
+      "grad_norm": 30.25,
+      "learning_rate": 2.0337086616160687e-06,
+      "loss": 0.9601,
+      "step": 6744
+    },
+    {
+      "epoch": 0.5747763101832126,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0335777471333545e-06,
+      "loss": 1.0277,
+      "step": 6745
+    },
+    {
+      "epoch": 0.5748615253515126,
+      "grad_norm": 15.25,
+      "learning_rate": 2.033446818490602e-06,
+      "loss": 0.6509,
+      "step": 6746
+    },
+    {
+      "epoch": 0.5749467405198125,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.033315875690178e-06,
+      "loss": 0.4963,
+      "step": 6747
+    },
+    {
+      "epoch": 0.5750319556881125,
+      "grad_norm": 18.375,
+      "learning_rate": 2.0331849187344483e-06,
+      "loss": 1.1608,
+      "step": 6748
+    },
+    {
+      "epoch": 0.5751171708564125,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0330539476257795e-06,
+      "loss": 0.8022,
+      "step": 6749
+    },
+    {
+      "epoch": 0.5752023860247124,
+      "grad_norm": 36.5,
+      "learning_rate": 2.0329229623665384e-06,
+      "loss": 1.1027,
+      "step": 6750
+    },
+    {
+      "epoch": 0.5752876011930124,
+      "grad_norm": 14.25,
+      "learning_rate": 2.032791962959092e-06,
+      "loss": 0.844,
+      "step": 6751
+    },
+    {
+      "epoch": 0.5753728163613123,
+      "grad_norm": 21.75,
+      "learning_rate": 2.032660949405808e-06,
+      "loss": 0.732,
+      "step": 6752
+    },
+    {
+      "epoch": 0.5754580315296123,
+      "grad_norm": 30.625,
+      "learning_rate": 2.0325299217090534e-06,
+      "loss": 1.2567,
+      "step": 6753
+    },
+    {
+      "epoch": 0.5755432466979122,
+      "grad_norm": 23.75,
+      "learning_rate": 2.0323988798711963e-06,
+      "loss": 0.8542,
+      "step": 6754
+    },
+    {
+      "epoch": 0.5756284618662122,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0322678238946047e-06,
+      "loss": 0.6581,
+      "step": 6755
+    },
+    {
+      "epoch": 0.5757136770345122,
+      "grad_norm": 13.75,
+      "learning_rate": 2.032136753781647e-06,
+      "loss": 0.7792,
+      "step": 6756
+    },
+    {
+      "epoch": 0.5757988922028121,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.032005669534692e-06,
+      "loss": 0.32,
+      "step": 6757
+    },
+    {
+      "epoch": 0.5758841073711121,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.031874571156107e-06,
+      "loss": 0.8013,
+      "step": 6758
+    },
+    {
+      "epoch": 0.575969322539412,
+      "grad_norm": 12.0,
+      "learning_rate": 2.031743458648263e-06,
+      "loss": 0.4304,
+      "step": 6759
+    },
+    {
+      "epoch": 0.576054537707712,
+      "grad_norm": 14.0,
+      "learning_rate": 2.031612332013529e-06,
+      "loss": 0.9376,
+      "step": 6760
+    },
+    {
+      "epoch": 0.5761397528760119,
+      "grad_norm": 22.25,
+      "learning_rate": 2.031481191254274e-06,
+      "loss": 1.0684,
+      "step": 6761
+    },
+    {
+      "epoch": 0.5762249680443119,
+      "grad_norm": 20.625,
+      "learning_rate": 2.0313500363728675e-06,
+      "loss": 0.6924,
+      "step": 6762
+    },
+    {
+      "epoch": 0.5763101832126118,
+      "grad_norm": 22.875,
+      "learning_rate": 2.0312188673716802e-06,
+      "loss": 0.9757,
+      "step": 6763
+    },
+    {
+      "epoch": 0.5763953983809118,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.0310876842530826e-06,
+      "loss": 1.0137,
+      "step": 6764
+    },
+    {
+      "epoch": 0.5764806135492118,
+      "grad_norm": 13.5,
+      "learning_rate": 2.0309564870194453e-06,
+      "loss": 0.6846,
+      "step": 6765
+    },
+    {
+      "epoch": 0.5765658287175117,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.0308252756731388e-06,
+      "loss": 0.8261,
+      "step": 6766
+    },
+    {
+      "epoch": 0.5766510438858117,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0306940502165343e-06,
+      "loss": 0.4896,
+      "step": 6767
+    },
+    {
+      "epoch": 0.5767362590541116,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0305628106520033e-06,
+      "loss": 0.7851,
+      "step": 6768
+    },
+    {
+      "epoch": 0.5768214742224116,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0304315569819176e-06,
+      "loss": 0.6587,
+      "step": 6769
+    },
+    {
+      "epoch": 0.5769066893907115,
+      "grad_norm": 17.375,
+      "learning_rate": 2.0303002892086487e-06,
+      "loss": 0.4979,
+      "step": 6770
+    },
+    {
+      "epoch": 0.5769919045590115,
+      "grad_norm": 11.75,
+      "learning_rate": 2.0301690073345687e-06,
+      "loss": 0.3979,
+      "step": 6771
+    },
+    {
+      "epoch": 0.5770771197273115,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.030037711362051e-06,
+      "loss": 0.558,
+      "step": 6772
+    },
+    {
+      "epoch": 0.5771623348956114,
+      "grad_norm": 22.875,
+      "learning_rate": 2.0299064012934665e-06,
+      "loss": 0.9738,
+      "step": 6773
+    },
+    {
+      "epoch": 0.5772475500639114,
+      "grad_norm": 16.375,
+      "learning_rate": 2.02977507713119e-06,
+      "loss": 0.8899,
+      "step": 6774
+    },
+    {
+      "epoch": 0.5773327652322113,
+      "grad_norm": 9.75,
+      "learning_rate": 2.0296437388775927e-06,
+      "loss": 0.3427,
+      "step": 6775
+    },
+    {
+      "epoch": 0.5774179804005113,
+      "grad_norm": 10.625,
+      "learning_rate": 2.0295123865350496e-06,
+      "loss": 0.2844,
+      "step": 6776
+    },
+    {
+      "epoch": 0.5775031955688112,
+      "grad_norm": 10.1875,
+      "learning_rate": 2.029381020105934e-06,
+      "loss": 0.4853,
+      "step": 6777
+    },
+    {
+      "epoch": 0.5775884107371112,
+      "grad_norm": 20.125,
+      "learning_rate": 2.029249639592619e-06,
+      "loss": 0.8696,
+      "step": 6778
+    },
+    {
+      "epoch": 0.5776736259054112,
+      "grad_norm": 13.25,
+      "learning_rate": 2.02911824499748e-06,
+      "loss": 0.5983,
+      "step": 6779
+    },
+    {
+      "epoch": 0.5777588410737111,
+      "grad_norm": 13.125,
+      "learning_rate": 2.028986836322891e-06,
+      "loss": 0.8721,
+      "step": 6780
+    },
+    {
+      "epoch": 0.5778440562420111,
+      "grad_norm": 14.625,
+      "learning_rate": 2.028855413571226e-06,
+      "loss": 0.6158,
+      "step": 6781
+    },
+    {
+      "epoch": 0.577929271410311,
+      "grad_norm": 22.25,
+      "learning_rate": 2.0287239767448608e-06,
+      "loss": 0.9306,
+      "step": 6782
+    },
+    {
+      "epoch": 0.578014486578611,
+      "grad_norm": 13.5,
+      "learning_rate": 2.02859252584617e-06,
+      "loss": 0.7195,
+      "step": 6783
+    },
+    {
+      "epoch": 0.5780997017469109,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0284610608775294e-06,
+      "loss": 0.6131,
+      "step": 6784
+    },
+    {
+      "epoch": 0.5781849169152109,
+      "grad_norm": 22.5,
+      "learning_rate": 2.0283295818413152e-06,
+      "loss": 1.3303,
+      "step": 6785
+    },
+    {
+      "epoch": 0.5782701320835109,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.028198088739902e-06,
+      "loss": 0.6883,
+      "step": 6786
+    },
+    {
+      "epoch": 0.5783553472518108,
+      "grad_norm": 18.375,
+      "learning_rate": 2.0280665815756672e-06,
+      "loss": 0.8534,
+      "step": 6787
+    },
+    {
+      "epoch": 0.5784405624201108,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.027935060350987e-06,
+      "loss": 0.5742,
+      "step": 6788
+    },
+    {
+      "epoch": 0.5785257775884107,
+      "grad_norm": 20.75,
+      "learning_rate": 2.0278035250682373e-06,
+      "loss": 0.8672,
+      "step": 6789
+    },
+    {
+      "epoch": 0.5786109927567107,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.027671975729796e-06,
+      "loss": 0.5317,
+      "step": 6790
+    },
+    {
+      "epoch": 0.5786962079250106,
+      "grad_norm": 14.125,
+      "learning_rate": 2.027540412338041e-06,
+      "loss": 0.7013,
+      "step": 6791
+    },
+    {
+      "epoch": 0.5787814230933106,
+      "grad_norm": 16.875,
+      "learning_rate": 2.027408834895348e-06,
+      "loss": 0.9326,
+      "step": 6792
+    },
+    {
+      "epoch": 0.5788666382616106,
+      "grad_norm": 9.875,
+      "learning_rate": 2.0272772434040955e-06,
+      "loss": 0.3284,
+      "step": 6793
+    },
+    {
+      "epoch": 0.5789518534299105,
+      "grad_norm": 17.0,
+      "learning_rate": 2.027145637866662e-06,
+      "loss": 0.9317,
+      "step": 6794
+    },
+    {
+      "epoch": 0.5790370685982105,
+      "grad_norm": 16.5,
+      "learning_rate": 2.027014018285425e-06,
+      "loss": 0.8467,
+      "step": 6795
+    },
+    {
+      "epoch": 0.5791222837665104,
+      "grad_norm": 22.875,
+      "learning_rate": 2.026882384662764e-06,
+      "loss": 1.1981,
+      "step": 6796
+    },
+    {
+      "epoch": 0.5792074989348104,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0267507370010564e-06,
+      "loss": 0.7329,
+      "step": 6797
+    },
+    {
+      "epoch": 0.5792927141031103,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.026619075302682e-06,
+      "loss": 0.7239,
+      "step": 6798
+    },
+    {
+      "epoch": 0.5793779292714103,
+      "grad_norm": 20.375,
+      "learning_rate": 2.0264873995700203e-06,
+      "loss": 1.0197,
+      "step": 6799
+    },
+    {
+      "epoch": 0.5794631444397103,
+      "grad_norm": 16.75,
+      "learning_rate": 2.0263557098054503e-06,
+      "loss": 0.992,
+      "step": 6800
+    },
+    {
+      "epoch": 0.5795483596080102,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.0262240060113517e-06,
+      "loss": 0.5332,
+      "step": 6801
+    },
+    {
+      "epoch": 0.5796335747763102,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0260922881901055e-06,
+      "loss": 0.488,
+      "step": 6802
+    },
+    {
+      "epoch": 0.5797187899446101,
+      "grad_norm": 13.875,
+      "learning_rate": 2.0259605563440904e-06,
+      "loss": 0.8797,
+      "step": 6803
+    },
+    {
+      "epoch": 0.5798040051129101,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.025828810475688e-06,
+      "loss": 0.5762,
+      "step": 6804
+    },
+    {
+      "epoch": 0.57988922028121,
+      "grad_norm": 15.625,
+      "learning_rate": 2.0256970505872793e-06,
+      "loss": 0.8089,
+      "step": 6805
+    },
+    {
+      "epoch": 0.57997443544951,
+      "grad_norm": 12.0,
+      "learning_rate": 2.0255652766812444e-06,
+      "loss": 0.3175,
+      "step": 6806
+    },
+    {
+      "epoch": 0.58005965061781,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0254334887599652e-06,
+      "loss": 0.9716,
+      "step": 6807
+    },
+    {
+      "epoch": 0.5801448657861099,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.025301686825823e-06,
+      "loss": 0.7905,
+      "step": 6808
+    },
+    {
+      "epoch": 0.5802300809544099,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.0251698708811996e-06,
+      "loss": 0.7279,
+      "step": 6809
+    },
+    {
+      "epoch": 0.5803152961227098,
+      "grad_norm": 24.25,
+      "learning_rate": 2.0250380409284777e-06,
+      "loss": 1.1662,
+      "step": 6810
+    },
+    {
+      "epoch": 0.5804005112910098,
+      "grad_norm": 14.125,
+      "learning_rate": 2.0249061969700386e-06,
+      "loss": 0.603,
+      "step": 6811
+    },
+    {
+      "epoch": 0.5804857264593097,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.0247743390082648e-06,
+      "loss": 0.6623,
+      "step": 6812
+    },
+    {
+      "epoch": 0.5805709416276097,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.0246424670455404e-06,
+      "loss": 0.8346,
+      "step": 6813
+    },
+    {
+      "epoch": 0.5806561567959097,
+      "grad_norm": 18.75,
+      "learning_rate": 2.0245105810842476e-06,
+      "loss": 0.6573,
+      "step": 6814
+    },
+    {
+      "epoch": 0.5807413719642096,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.024378681126769e-06,
+      "loss": 0.5768,
+      "step": 6815
+    },
+    {
+      "epoch": 0.5808265871325096,
+      "grad_norm": 14.375,
+      "learning_rate": 2.02424676717549e-06,
+      "loss": 0.7226,
+      "step": 6816
+    },
+    {
+      "epoch": 0.5809118023008095,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.0241148392327926e-06,
+      "loss": 0.8274,
+      "step": 6817
+    },
+    {
+      "epoch": 0.5809970174691095,
+      "grad_norm": 14.875,
+      "learning_rate": 2.0239828973010615e-06,
+      "loss": 0.8791,
+      "step": 6818
+    },
+    {
+      "epoch": 0.5810822326374094,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.023850941382682e-06,
+      "loss": 0.3899,
+      "step": 6819
+    },
+    {
+      "epoch": 0.5811674478057094,
+      "grad_norm": 12.25,
+      "learning_rate": 2.023718971480037e-06,
+      "loss": 0.8178,
+      "step": 6820
+    },
+    {
+      "epoch": 0.5812526629740093,
+      "grad_norm": 11.5,
+      "learning_rate": 2.0235869875955123e-06,
+      "loss": 0.3628,
+      "step": 6821
+    },
+    {
+      "epoch": 0.5813378781423093,
+      "grad_norm": 15.75,
+      "learning_rate": 2.0234549897314933e-06,
+      "loss": 0.6951,
+      "step": 6822
+    },
+    {
+      "epoch": 0.5814230933106093,
+      "grad_norm": 17.375,
+      "learning_rate": 2.0233229778903646e-06,
+      "loss": 0.7554,
+      "step": 6823
+    },
+    {
+      "epoch": 0.5815083084789092,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.023190952074512e-06,
+      "loss": 0.5659,
+      "step": 6824
+    },
+    {
+      "epoch": 0.5815935236472092,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0230589122863215e-06,
+      "loss": 1.1069,
+      "step": 6825
+    },
+    {
+      "epoch": 0.5816787388155091,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.022926858528179e-06,
+      "loss": 0.8917,
+      "step": 6826
+    },
+    {
+      "epoch": 0.5817639539838091,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.022794790802471e-06,
+      "loss": 0.656,
+      "step": 6827
+    },
+    {
+      "epoch": 0.581849169152109,
+      "grad_norm": 26.125,
+      "learning_rate": 2.0226627091115837e-06,
+      "loss": 1.0499,
+      "step": 6828
+    },
+    {
+      "epoch": 0.581934384320409,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0225306134579044e-06,
+      "loss": 0.6488,
+      "step": 6829
+    },
+    {
+      "epoch": 0.582019599488709,
+      "grad_norm": 42.5,
+      "learning_rate": 2.0223985038438202e-06,
+      "loss": 1.4108,
+      "step": 6830
+    },
+    {
+      "epoch": 0.5821048146570089,
+      "grad_norm": 20.75,
+      "learning_rate": 2.0222663802717185e-06,
+      "loss": 0.899,
+      "step": 6831
+    },
+    {
+      "epoch": 0.5821900298253089,
+      "grad_norm": 13.875,
+      "learning_rate": 2.0221342427439867e-06,
+      "loss": 0.6804,
+      "step": 6832
+    },
+    {
+      "epoch": 0.5822752449936088,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.022002091263012e-06,
+      "loss": 0.547,
+      "step": 6833
+    },
+    {
+      "epoch": 0.5823604601619088,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.0218699258311843e-06,
+      "loss": 0.5511,
+      "step": 6834
+    },
+    {
+      "epoch": 0.5824456753302087,
+      "grad_norm": 16.625,
+      "learning_rate": 2.0217377464508904e-06,
+      "loss": 0.9553,
+      "step": 6835
+    },
+    {
+      "epoch": 0.5825308904985087,
+      "grad_norm": 16.75,
+      "learning_rate": 2.021605553124519e-06,
+      "loss": 0.6424,
+      "step": 6836
+    },
+    {
+      "epoch": 0.5826161056668087,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.0214733458544595e-06,
+      "loss": 0.7372,
+      "step": 6837
+    },
+    {
+      "epoch": 0.5827013208351086,
+      "grad_norm": 19.375,
+      "learning_rate": 2.0213411246431014e-06,
+      "loss": 0.7079,
+      "step": 6838
+    },
+    {
+      "epoch": 0.5827865360034086,
+      "grad_norm": 12.375,
+      "learning_rate": 2.0212088894928328e-06,
+      "loss": 0.6697,
+      "step": 6839
+    },
+    {
+      "epoch": 0.5828717511717085,
+      "grad_norm": 19.125,
+      "learning_rate": 2.0210766404060443e-06,
+      "loss": 1.0168,
+      "step": 6840
+    },
+    {
+      "epoch": 0.5829569663400085,
+      "grad_norm": 16.75,
+      "learning_rate": 2.020944377385126e-06,
+      "loss": 0.689,
+      "step": 6841
+    },
+    {
+      "epoch": 0.5830421815083084,
+      "grad_norm": 12.75,
+      "learning_rate": 2.020812100432467e-06,
+      "loss": 0.3973,
+      "step": 6842
+    },
+    {
+      "epoch": 0.5831273966766084,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.0206798095504587e-06,
+      "loss": 0.2996,
+      "step": 6843
+    },
+    {
+      "epoch": 0.5832126118449084,
+      "grad_norm": 14.375,
+      "learning_rate": 2.020547504741491e-06,
+      "loss": 0.5681,
+      "step": 6844
+    },
+    {
+      "epoch": 0.5832978270132083,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0204151860079545e-06,
+      "loss": 0.9279,
+      "step": 6845
+    },
+    {
+      "epoch": 0.5833830421815083,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0202828533522415e-06,
+      "loss": 0.6386,
+      "step": 6846
+    },
+    {
+      "epoch": 0.5834682573498082,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.020150506776743e-06,
+      "loss": 0.7953,
+      "step": 6847
+    },
+    {
+      "epoch": 0.5835534725181082,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.02001814628385e-06,
+      "loss": 0.5392,
+      "step": 6848
+    },
+    {
+      "epoch": 0.5836386876864081,
+      "grad_norm": 18.0,
+      "learning_rate": 2.0198857718759545e-06,
+      "loss": 1.1605,
+      "step": 6849
+    },
+    {
+      "epoch": 0.5837239028547081,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.0197533835554496e-06,
+      "loss": 0.5234,
+      "step": 6850
+    },
+    {
+      "epoch": 0.5838091180230081,
+      "grad_norm": 15.75,
+      "learning_rate": 2.0196209813247265e-06,
+      "loss": 0.8265,
+      "step": 6851
+    },
+    {
+      "epoch": 0.583894333191308,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.0194885651861785e-06,
+      "loss": 0.7643,
+      "step": 6852
+    },
+    {
+      "epoch": 0.583979548359608,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.0193561351421985e-06,
+      "loss": 0.3946,
+      "step": 6853
+    },
+    {
+      "epoch": 0.5840647635279079,
+      "grad_norm": 18.875,
+      "learning_rate": 2.0192236911951794e-06,
+      "loss": 1.0265,
+      "step": 6854
+    },
+    {
+      "epoch": 0.5841499786962079,
+      "grad_norm": 18.375,
+      "learning_rate": 2.019091233347515e-06,
+      "loss": 1.0825,
+      "step": 6855
+    },
+    {
+      "epoch": 0.5842351938645078,
+      "grad_norm": 12.0,
+      "learning_rate": 2.018958761601598e-06,
+      "loss": 0.6365,
+      "step": 6856
+    },
+    {
+      "epoch": 0.5843204090328078,
+      "grad_norm": 13.75,
+      "learning_rate": 2.018826275959823e-06,
+      "loss": 0.4975,
+      "step": 6857
+    },
+    {
+      "epoch": 0.5844056242011078,
+      "grad_norm": 16.75,
+      "learning_rate": 2.0186937764245847e-06,
+      "loss": 0.8235,
+      "step": 6858
+    },
+    {
+      "epoch": 0.5844908393694077,
+      "grad_norm": 21.5,
+      "learning_rate": 2.0185612629982765e-06,
+      "loss": 0.6771,
+      "step": 6859
+    },
+    {
+      "epoch": 0.5845760545377077,
+      "grad_norm": 15.25,
+      "learning_rate": 2.018428735683293e-06,
+      "loss": 0.788,
+      "step": 6860
+    },
+    {
+      "epoch": 0.5846612697060076,
+      "grad_norm": 14.625,
+      "learning_rate": 2.01829619448203e-06,
+      "loss": 0.921,
+      "step": 6861
+    },
+    {
+      "epoch": 0.5847464848743076,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0181636393968823e-06,
+      "loss": 0.9471,
+      "step": 6862
+    },
+    {
+      "epoch": 0.5848317000426075,
+      "grad_norm": 14.125,
+      "learning_rate": 2.018031070430245e-06,
+      "loss": 0.5874,
+      "step": 6863
+    },
+    {
+      "epoch": 0.5849169152109075,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.017898487584514e-06,
+      "loss": 0.6185,
+      "step": 6864
+    },
+    {
+      "epoch": 0.5850021303792075,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0177658908620853e-06,
+      "loss": 1.0259,
+      "step": 6865
+    },
+    {
+      "epoch": 0.5850873455475074,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0176332802653542e-06,
+      "loss": 0.8108,
+      "step": 6866
+    },
+    {
+      "epoch": 0.5851725607158074,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0175006557967184e-06,
+      "loss": 1.0411,
+      "step": 6867
+    },
+    {
+      "epoch": 0.5852577758841073,
+      "grad_norm": 14.625,
+      "learning_rate": 2.017368017458574e-06,
+      "loss": 0.8712,
+      "step": 6868
+    },
+    {
+      "epoch": 0.5853429910524073,
+      "grad_norm": 12.0,
+      "learning_rate": 2.0172353652533177e-06,
+      "loss": 0.6059,
+      "step": 6869
+    },
+    {
+      "epoch": 0.5854282062207072,
+      "grad_norm": 17.25,
+      "learning_rate": 2.017102699183347e-06,
+      "loss": 0.8399,
+      "step": 6870
+    },
+    {
+      "epoch": 0.5855134213890072,
+      "grad_norm": 16.75,
+      "learning_rate": 2.016970019251059e-06,
+      "loss": 0.8368,
+      "step": 6871
+    },
+    {
+      "epoch": 0.5855986365573072,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.0168373254588514e-06,
+      "loss": 0.4935,
+      "step": 6872
+    },
+    {
+      "epoch": 0.5856838517256071,
+      "grad_norm": 10.5,
+      "learning_rate": 2.0167046178091225e-06,
+      "loss": 0.5049,
+      "step": 6873
+    },
+    {
+      "epoch": 0.5857690668939071,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.01657189630427e-06,
+      "loss": 0.7636,
+      "step": 6874
+    },
+    {
+      "epoch": 0.585854282062207,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.0164391609466928e-06,
+      "loss": 0.6882,
+      "step": 6875
+    },
+    {
+      "epoch": 0.585939497230507,
+      "grad_norm": 21.125,
+      "learning_rate": 2.0163064117387886e-06,
+      "loss": 1.1915,
+      "step": 6876
+    },
+    {
+      "epoch": 0.5860247123988069,
+      "grad_norm": 12.375,
+      "learning_rate": 2.0161736486829576e-06,
+      "loss": 0.5292,
+      "step": 6877
+    },
+    {
+      "epoch": 0.5861099275671069,
+      "grad_norm": 28.125,
+      "learning_rate": 2.0160408717815975e-06,
+      "loss": 0.9053,
+      "step": 6878
+    },
+    {
+      "epoch": 0.5861951427354068,
+      "grad_norm": 31.875,
+      "learning_rate": 2.0159080810371094e-06,
+      "loss": 0.8698,
+      "step": 6879
+    },
+    {
+      "epoch": 0.5862803579037068,
+      "grad_norm": 23.0,
+      "learning_rate": 2.0157752764518916e-06,
+      "loss": 0.7061,
+      "step": 6880
+    },
+    {
+      "epoch": 0.5863655730720068,
+      "grad_norm": 17.5,
+      "learning_rate": 2.015642458028345e-06,
+      "loss": 0.9361,
+      "step": 6881
+    },
+    {
+      "epoch": 0.5864507882403067,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.015509625768869e-06,
+      "loss": 0.8305,
+      "step": 6882
+    },
+    {
+      "epoch": 0.5865360034086067,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.015376779675864e-06,
+      "loss": 0.7644,
+      "step": 6883
+    },
+    {
+      "epoch": 0.5866212185769067,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.015243919751731e-06,
+      "loss": 0.5548,
+      "step": 6884
+    },
+    {
+      "epoch": 0.5867064337452067,
+      "grad_norm": 14.25,
+      "learning_rate": 2.0151110459988714e-06,
+      "loss": 0.96,
+      "step": 6885
+    },
+    {
+      "epoch": 0.5867916489135067,
+      "grad_norm": 12.125,
+      "learning_rate": 2.0149781584196853e-06,
+      "loss": 0.4868,
+      "step": 6886
+    },
+    {
+      "epoch": 0.5868768640818066,
+      "grad_norm": 23.375,
+      "learning_rate": 2.014845257016575e-06,
+      "loss": 0.7223,
+      "step": 6887
+    },
+    {
+      "epoch": 0.5869620792501066,
+      "grad_norm": 12.25,
+      "learning_rate": 2.0147123417919412e-06,
+      "loss": 0.5124,
+      "step": 6888
+    },
+    {
+      "epoch": 0.5870472944184065,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0145794127481868e-06,
+      "loss": 0.4794,
+      "step": 6889
+    },
+    {
+      "epoch": 0.5871325095867065,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0144464698877134e-06,
+      "loss": 0.9951,
+      "step": 6890
+    },
+    {
+      "epoch": 0.5872177247550064,
+      "grad_norm": 19.875,
+      "learning_rate": 2.0143135132129243e-06,
+      "loss": 0.7204,
+      "step": 6891
+    },
+    {
+      "epoch": 0.5873029399233064,
+      "grad_norm": 19.0,
+      "learning_rate": 2.0141805427262207e-06,
+      "loss": 0.8634,
+      "step": 6892
+    },
+    {
+      "epoch": 0.5873881550916064,
+      "grad_norm": 15.625,
+      "learning_rate": 2.014047558430006e-06,
+      "loss": 0.6356,
+      "step": 6893
+    },
+    {
+      "epoch": 0.5874733702599063,
+      "grad_norm": 17.75,
+      "learning_rate": 2.0139145603266843e-06,
+      "loss": 0.6298,
+      "step": 6894
+    },
+    {
+      "epoch": 0.5875585854282063,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0137815484186584e-06,
+      "loss": 0.8022,
+      "step": 6895
+    },
+    {
+      "epoch": 0.5876438005965062,
+      "grad_norm": 22.25,
+      "learning_rate": 2.0136485227083314e-06,
+      "loss": 1.3751,
+      "step": 6896
+    },
+    {
+      "epoch": 0.5877290157648062,
+      "grad_norm": 19.875,
+      "learning_rate": 2.0135154831981077e-06,
+      "loss": 0.6795,
+      "step": 6897
+    },
+    {
+      "epoch": 0.5878142309331061,
+      "grad_norm": 11.375,
+      "learning_rate": 2.0133824298903915e-06,
+      "loss": 0.5237,
+      "step": 6898
+    },
+    {
+      "epoch": 0.5878994461014061,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.0132493627875875e-06,
+      "loss": 0.5565,
+      "step": 6899
+    },
+    {
+      "epoch": 0.587984661269706,
+      "grad_norm": 12.5,
+      "learning_rate": 2.0131162818920997e-06,
+      "loss": 0.8173,
+      "step": 6900
+    },
+    {
+      "epoch": 0.588069876438006,
+      "grad_norm": 12.375,
+      "learning_rate": 2.012983187206333e-06,
+      "loss": 0.5747,
+      "step": 6901
+    },
+    {
+      "epoch": 0.588155091606306,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.012850078732693e-06,
+      "loss": 0.7632,
+      "step": 6902
+    },
+    {
+      "epoch": 0.5882403067746059,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0127169564735853e-06,
+      "loss": 0.8048,
+      "step": 6903
+    },
+    {
+      "epoch": 0.5883255219429059,
+      "grad_norm": 42.75,
+      "learning_rate": 2.0125838204314152e-06,
+      "loss": 1.3908,
+      "step": 6904
+    },
+    {
+      "epoch": 0.5884107371112058,
+      "grad_norm": 12.0,
+      "learning_rate": 2.0124506706085887e-06,
+      "loss": 0.4324,
+      "step": 6905
+    },
+    {
+      "epoch": 0.5884959522795058,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.0123175070075113e-06,
+      "loss": 0.6727,
+      "step": 6906
+    },
+    {
+      "epoch": 0.5885811674478058,
+      "grad_norm": 16.625,
+      "learning_rate": 2.012184329630591e-06,
+      "loss": 0.9828,
+      "step": 6907
+    },
+    {
+      "epoch": 0.5886663826161057,
+      "grad_norm": 14.875,
+      "learning_rate": 2.0120511384802327e-06,
+      "loss": 0.8603,
+      "step": 6908
+    },
+    {
+      "epoch": 0.5887515977844057,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.011917933558844e-06,
+      "loss": 0.5823,
+      "step": 6909
+    },
+    {
+      "epoch": 0.5888368129527056,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.011784714868832e-06,
+      "loss": 0.5983,
+      "step": 6910
+    },
+    {
+      "epoch": 0.5889220281210056,
+      "grad_norm": 15.75,
+      "learning_rate": 2.0116514824126047e-06,
+      "loss": 0.5843,
+      "step": 6911
+    },
+    {
+      "epoch": 0.5890072432893055,
+      "grad_norm": 16.125,
+      "learning_rate": 2.011518236192569e-06,
+      "loss": 0.7228,
+      "step": 6912
+    },
+    {
+      "epoch": 0.5890924584576055,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.0113849762111328e-06,
+      "loss": 0.5681,
+      "step": 6913
+    },
+    {
+      "epoch": 0.5891776736259055,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0112517024707047e-06,
+      "loss": 0.6344,
+      "step": 6914
+    },
+    {
+      "epoch": 0.5892628887942054,
+      "grad_norm": 13.375,
+      "learning_rate": 2.0111184149736925e-06,
+      "loss": 0.6513,
+      "step": 6915
+    },
+    {
+      "epoch": 0.5893481039625054,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.0109851137225055e-06,
+      "loss": 0.7996,
+      "step": 6916
+    },
+    {
+      "epoch": 0.5894333191308053,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.010851798719552e-06,
+      "loss": 0.5699,
+      "step": 6917
+    },
+    {
+      "epoch": 0.5895185342991053,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.010718469967242e-06,
+      "loss": 0.534,
+      "step": 6918
+    },
+    {
+      "epoch": 0.5896037494674052,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0105851274679834e-06,
+      "loss": 0.63,
+      "step": 6919
+    },
+    {
+      "epoch": 0.5896889646357052,
+      "grad_norm": 10.25,
+      "learning_rate": 2.0104517712241873e-06,
+      "loss": 0.4163,
+      "step": 6920
+    },
+    {
+      "epoch": 0.5897741798040051,
+      "grad_norm": 13.75,
+      "learning_rate": 2.0103184012382625e-06,
+      "loss": 0.6829,
+      "step": 6921
+    },
+    {
+      "epoch": 0.5898593949723051,
+      "grad_norm": 13.125,
+      "learning_rate": 2.01018501751262e-06,
+      "loss": 0.6144,
+      "step": 6922
+    },
+    {
+      "epoch": 0.5899446101406051,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.01005162004967e-06,
+      "loss": 0.7802,
+      "step": 6923
+    },
+    {
+      "epoch": 0.590029825308905,
+      "grad_norm": 23.0,
+      "learning_rate": 2.0099182088518225e-06,
+      "loss": 1.0607,
+      "step": 6924
+    },
+    {
+      "epoch": 0.590115040477205,
+      "grad_norm": 16.125,
+      "learning_rate": 2.009784783921489e-06,
+      "loss": 0.6496,
+      "step": 6925
+    },
+    {
+      "epoch": 0.5902002556455049,
+      "grad_norm": 14.375,
+      "learning_rate": 2.0096513452610805e-06,
+      "loss": 0.7623,
+      "step": 6926
+    },
+    {
+      "epoch": 0.5902854708138049,
+      "grad_norm": 12.0,
+      "learning_rate": 2.009517892873008e-06,
+      "loss": 0.2801,
+      "step": 6927
+    },
+    {
+      "epoch": 0.5903706859821048,
+      "grad_norm": 19.875,
+      "learning_rate": 2.009384426759684e-06,
+      "loss": 1.1026,
+      "step": 6928
+    },
+    {
+      "epoch": 0.5904559011504048,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.0092509469235193e-06,
+      "loss": 0.4951,
+      "step": 6929
+    },
+    {
+      "epoch": 0.5905411163187048,
+      "grad_norm": 18.625,
+      "learning_rate": 2.0091174533669267e-06,
+      "loss": 0.9784,
+      "step": 6930
+    },
+    {
+      "epoch": 0.5906263314870047,
+      "grad_norm": 15.25,
+      "learning_rate": 2.0089839460923183e-06,
+      "loss": 1.0034,
+      "step": 6931
+    },
+    {
+      "epoch": 0.5907115466553047,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.0088504251021073e-06,
+      "loss": 0.7972,
+      "step": 6932
+    },
+    {
+      "epoch": 0.5907967618236046,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.008716890398705e-06,
+      "loss": 0.7977,
+      "step": 6933
+    },
+    {
+      "epoch": 0.5908819769919046,
+      "grad_norm": 12.0,
+      "learning_rate": 2.0085833419845265e-06,
+      "loss": 0.3837,
+      "step": 6934
+    },
+    {
+      "epoch": 0.5909671921602045,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0084497798619835e-06,
+      "loss": 0.991,
+      "step": 6935
+    },
+    {
+      "epoch": 0.5910524073285045,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.0083162040334906e-06,
+      "loss": 0.5306,
+      "step": 6936
+    },
+    {
+      "epoch": 0.5911376224968045,
+      "grad_norm": 13.625,
+      "learning_rate": 2.0081826145014616e-06,
+      "loss": 0.8169,
+      "step": 6937
+    },
+    {
+      "epoch": 0.5912228376651044,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.00804901126831e-06,
+      "loss": 0.5817,
+      "step": 6938
+    },
+    {
+      "epoch": 0.5913080528334044,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0079153943364504e-06,
+      "loss": 0.6393,
+      "step": 6939
+    },
+    {
+      "epoch": 0.5913932680017043,
+      "grad_norm": 18.75,
+      "learning_rate": 2.007781763708298e-06,
+      "loss": 0.882,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5914784831700043,
+      "grad_norm": 13.25,
+      "learning_rate": 2.0076481193862667e-06,
+      "loss": 0.852,
+      "step": 6941
+    },
+    {
+      "epoch": 0.5915636983383042,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.0075144613727717e-06,
+      "loss": 0.7561,
+      "step": 6942
+    },
+    {
+      "epoch": 0.5916489135066042,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.007380789670229e-06,
+      "loss": 0.5374,
+      "step": 6943
+    },
+    {
+      "epoch": 0.5917341286749042,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.0072471042810542e-06,
+      "loss": 0.7475,
+      "step": 6944
+    },
+    {
+      "epoch": 0.5918193438432041,
+      "grad_norm": 15.625,
+      "learning_rate": 2.0071134052076625e-06,
+      "loss": 0.7733,
+      "step": 6945
+    },
+    {
+      "epoch": 0.5919045590115041,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.00697969245247e-06,
+      "loss": 1.0881,
+      "step": 6946
+    },
+    {
+      "epoch": 0.591989774179804,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0068459660178932e-06,
+      "loss": 1.1784,
+      "step": 6947
+    },
+    {
+      "epoch": 0.592074989348104,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.006712225906349e-06,
+      "loss": 0.4161,
+      "step": 6948
+    },
+    {
+      "epoch": 0.5921602045164039,
+      "grad_norm": 32.5,
+      "learning_rate": 2.006578472120254e-06,
+      "loss": 0.9841,
+      "step": 6949
+    },
+    {
+      "epoch": 0.5922454196847039,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.006444704662025e-06,
+      "loss": 0.6029,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5923306348530039,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0063109235340794e-06,
+      "loss": 0.6538,
+      "step": 6951
+    },
+    {
+      "epoch": 0.5924158500213038,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.0061771287388355e-06,
+      "loss": 0.7806,
+      "step": 6952
+    },
+    {
+      "epoch": 0.5925010651896038,
+      "grad_norm": 17.5,
+      "learning_rate": 2.00604332027871e-06,
+      "loss": 0.7478,
+      "step": 6953
+    },
+    {
+      "epoch": 0.5925862803579037,
+      "grad_norm": 16.875,
+      "learning_rate": 2.005909498156121e-06,
+      "loss": 0.9095,
+      "step": 6954
+    },
+    {
+      "epoch": 0.5926714955262037,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.005775662373488e-06,
+      "loss": 0.5839,
+      "step": 6955
+    },
+    {
+      "epoch": 0.5927567106945036,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.0056418129332283e-06,
+      "loss": 0.4359,
+      "step": 6956
+    },
+    {
+      "epoch": 0.5928419258628036,
+      "grad_norm": 19.0,
+      "learning_rate": 2.005507949837762e-06,
+      "loss": 0.7483,
+      "step": 6957
+    },
+    {
+      "epoch": 0.5929271410311036,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.005374073089506e-06,
+      "loss": 0.9208,
+      "step": 6958
+    },
+    {
+      "epoch": 0.5930123561994035,
+      "grad_norm": 16.5,
+      "learning_rate": 2.0052401826908817e-06,
+      "loss": 0.5269,
+      "step": 6959
+    },
+    {
+      "epoch": 0.5930975713677035,
+      "grad_norm": 16.125,
+      "learning_rate": 2.005106278644308e-06,
+      "loss": 0.4589,
+      "step": 6960
+    },
+    {
+      "epoch": 0.5931827865360034,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0049723609522043e-06,
+      "loss": 0.535,
+      "step": 6961
+    },
+    {
+      "epoch": 0.5932680017043034,
+      "grad_norm": 13.25,
+      "learning_rate": 2.004838429616991e-06,
+      "loss": 0.7606,
+      "step": 6962
+    },
+    {
+      "epoch": 0.5933532168726033,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.004704484641088e-06,
+      "loss": 0.5362,
+      "step": 6963
+    },
+    {
+      "epoch": 0.5934384320409033,
+      "grad_norm": 18.0,
+      "learning_rate": 2.0045705260269164e-06,
+      "loss": 1.0528,
+      "step": 6964
+    },
+    {
+      "epoch": 0.5935236472092033,
+      "grad_norm": 20.25,
+      "learning_rate": 2.0044365537768962e-06,
+      "loss": 1.1418,
+      "step": 6965
+    },
+    {
+      "epoch": 0.5936088623775032,
+      "grad_norm": 14.5,
+      "learning_rate": 2.004302567893449e-06,
+      "loss": 0.7125,
+      "step": 6966
+    },
+    {
+      "epoch": 0.5936940775458032,
+      "grad_norm": 34.75,
+      "learning_rate": 2.0041685683789957e-06,
+      "loss": 1.3935,
+      "step": 6967
+    },
+    {
+      "epoch": 0.5937792927141031,
+      "grad_norm": 12.125,
+      "learning_rate": 2.0040345552359587e-06,
+      "loss": 0.3361,
+      "step": 6968
+    },
+    {
+      "epoch": 0.5938645078824031,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0039005284667586e-06,
+      "loss": 1.0431,
+      "step": 6969
+    },
+    {
+      "epoch": 0.593949723050703,
+      "grad_norm": 12.375,
+      "learning_rate": 2.0037664880738174e-06,
+      "loss": 0.5184,
+      "step": 6970
+    },
+    {
+      "epoch": 0.594034938219003,
+      "grad_norm": 12.0,
+      "learning_rate": 2.003632434059558e-06,
+      "loss": 0.3794,
+      "step": 6971
+    },
+    {
+      "epoch": 0.594120153387303,
+      "grad_norm": 12.625,
+      "learning_rate": 2.003498366426403e-06,
+      "loss": 0.6815,
+      "step": 6972
+    },
+    {
+      "epoch": 0.5942053685556029,
+      "grad_norm": 11.875,
+      "learning_rate": 2.003364285176775e-06,
+      "loss": 0.8421,
+      "step": 6973
+    },
+    {
+      "epoch": 0.5942905837239029,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.003230190313096e-06,
+      "loss": 0.5783,
+      "step": 6974
+    },
+    {
+      "epoch": 0.5943757988922028,
+      "grad_norm": 12.75,
+      "learning_rate": 2.0030960818377906e-06,
+      "loss": 0.8218,
+      "step": 6975
+    },
+    {
+      "epoch": 0.5944610140605028,
+      "grad_norm": 16.75,
+      "learning_rate": 2.002961959753282e-06,
+      "loss": 1.0087,
+      "step": 6976
+    },
+    {
+      "epoch": 0.5945462292288027,
+      "grad_norm": 16.125,
+      "learning_rate": 2.002827824061993e-06,
+      "loss": 0.653,
+      "step": 6977
+    },
+    {
+      "epoch": 0.5946314443971027,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.0026936747663484e-06,
+      "loss": 0.5757,
+      "step": 6978
+    },
+    {
+      "epoch": 0.5947166595654026,
+      "grad_norm": 20.75,
+      "learning_rate": 2.002559511868772e-06,
+      "loss": 1.0626,
+      "step": 6979
+    },
+    {
+      "epoch": 0.5948018747337026,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.002425335371689e-06,
+      "loss": 0.7168,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5948870899020026,
+      "grad_norm": 16.0,
+      "learning_rate": 2.002291145277523e-06,
+      "loss": 0.769,
+      "step": 6981
+    },
+    {
+      "epoch": 0.5949723050703025,
+      "grad_norm": 19.25,
+      "learning_rate": 2.0021569415887e-06,
+      "loss": 0.717,
+      "step": 6982
+    },
+    {
+      "epoch": 0.5950575202386025,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.0020227243076444e-06,
+      "loss": 0.3966,
+      "step": 6983
+    },
+    {
+      "epoch": 0.5951427354069024,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.001888493436782e-06,
+      "loss": 0.6422,
+      "step": 6984
+    },
+    {
+      "epoch": 0.5952279505752024,
+      "grad_norm": 16.875,
+      "learning_rate": 2.0017542489785387e-06,
+      "loss": 0.6107,
+      "step": 6985
+    },
+    {
+      "epoch": 0.5953131657435023,
+      "grad_norm": 13.0,
+      "learning_rate": 2.00161999093534e-06,
+      "loss": 0.482,
+      "step": 6986
+    },
+    {
+      "epoch": 0.5953983809118023,
+      "grad_norm": 18.375,
+      "learning_rate": 2.001485719309612e-06,
+      "loss": 0.6762,
+      "step": 6987
+    },
+    {
+      "epoch": 0.5954835960801023,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.0013514341037817e-06,
+      "loss": 0.8348,
+      "step": 6988
+    },
+    {
+      "epoch": 0.5955688112484022,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0012171353202754e-06,
+      "loss": 1.065,
+      "step": 6989
+    },
+    {
+      "epoch": 0.5956540264167022,
+      "grad_norm": 15.875,
+      "learning_rate": 2.00108282296152e-06,
+      "loss": 0.7257,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5957392415850021,
+      "grad_norm": 10.9375,
+      "learning_rate": 2.0009484970299424e-06,
+      "loss": 0.3832,
+      "step": 6991
+    },
+    {
+      "epoch": 0.5958244567533021,
+      "grad_norm": 13.0,
+      "learning_rate": 2.0008141575279704e-06,
+      "loss": 0.6952,
+      "step": 6992
+    },
+    {
+      "epoch": 0.595909671921602,
+      "grad_norm": 16.0,
+      "learning_rate": 2.0006798044580324e-06,
+      "loss": 1.0096,
+      "step": 6993
+    },
+    {
+      "epoch": 0.595994887089902,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.0005454378225547e-06,
+      "loss": 0.7708,
+      "step": 6994
+    },
+    {
+      "epoch": 0.596080102258202,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.000411057623966e-06,
+      "loss": 0.7953,
+      "step": 6995
+    },
+    {
+      "epoch": 0.5961653174265019,
+      "grad_norm": 13.75,
+      "learning_rate": 2.000276663864695e-06,
+      "loss": 0.6034,
+      "step": 6996
+    },
+    {
+      "epoch": 0.5962505325948019,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.000142256547171e-06,
+      "loss": 0.6687,
+      "step": 6997
+    },
+    {
+      "epoch": 0.5963357477631018,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0000078356738215e-06,
+      "loss": 1.3052,
+      "step": 6998
+    },
+    {
+      "epoch": 0.5964209629314018,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.999873401247076e-06,
+      "loss": 0.6121,
+      "step": 6999
+    },
+    {
+      "epoch": 0.5965061780997017,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9997389532693644e-06,
+      "loss": 0.6854,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5965913932680017,
+      "grad_norm": 18.75,
+      "learning_rate": 1.999604491743116e-06,
+      "loss": 0.7471,
+      "step": 7001
+    },
+    {
+      "epoch": 0.5966766084363017,
+      "grad_norm": 16.875,
+      "learning_rate": 1.999470016670761e-06,
+      "loss": 0.6186,
+      "step": 7002
+    },
+    {
+      "epoch": 0.5967618236046016,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9993355280547287e-06,
+      "loss": 0.6835,
+      "step": 7003
+    },
+    {
+      "epoch": 0.5968470387729016,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9992010258974497e-06,
+      "loss": 0.9148,
+      "step": 7004
+    },
+    {
+      "epoch": 0.5969322539412015,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9990665102013553e-06,
+      "loss": 0.9025,
+      "step": 7005
+    },
+    {
+      "epoch": 0.5970174691095015,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9989319809688757e-06,
+      "loss": 0.4374,
+      "step": 7006
+    },
+    {
+      "epoch": 0.5971026842778014,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9987974382024418e-06,
+      "loss": 1.1291,
+      "step": 7007
+    },
+    {
+      "epoch": 0.5971878994461014,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9986628819044854e-06,
+      "loss": 0.6285,
+      "step": 7008
+    },
+    {
+      "epoch": 0.5972731146144014,
+      "grad_norm": 17.5,
+      "learning_rate": 1.998528312077438e-06,
+      "loss": 0.8571,
+      "step": 7009
+    },
+    {
+      "epoch": 0.5973583297827013,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.998393728723731e-06,
+      "loss": 0.7511,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5974435449510013,
+      "grad_norm": 21.125,
+      "learning_rate": 1.9982591318457968e-06,
+      "loss": 1.0017,
+      "step": 7011
+    },
+    {
+      "epoch": 0.5975287601193012,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9981245214460678e-06,
+      "loss": 1.0681,
+      "step": 7012
+    },
+    {
+      "epoch": 0.5976139752876012,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9979898975269767e-06,
+      "loss": 0.3837,
+      "step": 7013
+    },
+    {
+      "epoch": 0.5976991904559011,
+      "grad_norm": 11.25,
+      "learning_rate": 1.9978552600909555e-06,
+      "loss": 0.4004,
+      "step": 7014
+    },
+    {
+      "epoch": 0.5977844056242011,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9977206091404377e-06,
+      "loss": 0.6966,
+      "step": 7015
+    },
+    {
+      "epoch": 0.597869620792501,
+      "grad_norm": 25.75,
+      "learning_rate": 1.9975859446778567e-06,
+      "loss": 0.8567,
+      "step": 7016
+    },
+    {
+      "epoch": 0.597954835960801,
+      "grad_norm": 15.375,
+      "learning_rate": 1.997451266705646e-06,
+      "loss": 0.8897,
+      "step": 7017
+    },
+    {
+      "epoch": 0.598040051129101,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9973165752262395e-06,
+      "loss": 0.887,
+      "step": 7018
+    },
+    {
+      "epoch": 0.5981252662974009,
+      "grad_norm": 14.75,
+      "learning_rate": 1.997181870242071e-06,
+      "loss": 0.6838,
+      "step": 7019
+    },
+    {
+      "epoch": 0.5982104814657009,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.9970471517555743e-06,
+      "loss": 0.6087,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5982956966340008,
+      "grad_norm": 16.625,
+      "learning_rate": 1.996912419769185e-06,
+      "loss": 0.9808,
+      "step": 7021
+    },
+    {
+      "epoch": 0.5983809118023008,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9967776742853364e-06,
+      "loss": 0.4564,
+      "step": 7022
+    },
+    {
+      "epoch": 0.5984661269706008,
+      "grad_norm": 21.75,
+      "learning_rate": 1.996642915306465e-06,
+      "loss": 1.0245,
+      "step": 7023
+    },
+    {
+      "epoch": 0.5985513421389007,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.9965081428350046e-06,
+      "loss": 0.761,
+      "step": 7024
+    },
+    {
+      "epoch": 0.5986365573072007,
+      "grad_norm": 16.25,
+      "learning_rate": 1.9963733568733917e-06,
+      "loss": 1.0195,
+      "step": 7025
+    },
+    {
+      "epoch": 0.5987217724755006,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9962385574240616e-06,
+      "loss": 0.5404,
+      "step": 7026
+    },
+    {
+      "epoch": 0.5988069876438006,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.996103744489451e-06,
+      "loss": 0.6135,
+      "step": 7027
+    },
+    {
+      "epoch": 0.5988922028121005,
+      "grad_norm": 19.0,
+      "learning_rate": 1.995968918071995e-06,
+      "loss": 0.9434,
+      "step": 7028
+    },
+    {
+      "epoch": 0.5989774179804005,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9958340781741302e-06,
+      "loss": 0.9301,
+      "step": 7029
+    },
+    {
+      "epoch": 0.5990626331487005,
+      "grad_norm": 18.25,
+      "learning_rate": 1.995699224798294e-06,
+      "loss": 0.7505,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5991478483170004,
+      "grad_norm": 10.25,
+      "learning_rate": 1.995564357946923e-06,
+      "loss": 0.5414,
+      "step": 7031
+    },
+    {
+      "epoch": 0.5992330634853004,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.995429477622454e-06,
+      "loss": 0.9013,
+      "step": 7032
+    },
+    {
+      "epoch": 0.5993182786536003,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9952945838273256e-06,
+      "loss": 1.0507,
+      "step": 7033
+    },
+    {
+      "epoch": 0.5994034938219003,
+      "grad_norm": 11.875,
+      "learning_rate": 1.9951596765639734e-06,
+      "loss": 0.5178,
+      "step": 7034
+    },
+    {
+      "epoch": 0.5994887089902002,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9950247558348374e-06,
+      "loss": 0.6673,
+      "step": 7035
+    },
+    {
+      "epoch": 0.5995739241585002,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9948898216423546e-06,
+      "loss": 0.817,
+      "step": 7036
+    },
+    {
+      "epoch": 0.5996591393268002,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.994754873988964e-06,
+      "loss": 0.548,
+      "step": 7037
+    },
+    {
+      "epoch": 0.5997443544951001,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9946199128771036e-06,
+      "loss": 0.7274,
+      "step": 7038
+    },
+    {
+      "epoch": 0.5998295696634001,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.9944849383092123e-06,
+      "loss": 0.5219,
+      "step": 7039
+    },
+    {
+      "epoch": 0.5999147848317,
+      "grad_norm": 35.75,
+      "learning_rate": 1.99434995028773e-06,
+      "loss": 0.9749,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 23.125,
+      "learning_rate": 1.994214948815095e-06,
+      "loss": 0.8435,
+      "step": 7041
+    },
+    {
+      "epoch": 0.6000852151682999,
+      "grad_norm": 13.375,
+      "learning_rate": 1.994079933893748e-06,
+      "loss": 0.8088,
+      "step": 7042
+    },
+    {
+      "epoch": 0.6001704303365999,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9939449055261284e-06,
+      "loss": 0.513,
+      "step": 7043
+    },
+    {
+      "epoch": 0.6002556455048998,
+      "grad_norm": 18.625,
+      "learning_rate": 1.993809863714676e-06,
+      "loss": 1.2511,
+      "step": 7044
+    },
+    {
+      "epoch": 0.6003408606731998,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.993674808461832e-06,
+      "loss": 0.5832,
+      "step": 7045
+    },
+    {
+      "epoch": 0.6004260758414998,
+      "grad_norm": 14.0,
+      "learning_rate": 1.993539739770036e-06,
+      "loss": 0.7236,
+      "step": 7046
+    },
+    {
+      "epoch": 0.6005112910097997,
+      "grad_norm": 10.75,
+      "learning_rate": 1.993404657641729e-06,
+      "loss": 0.3398,
+      "step": 7047
+    },
+    {
+      "epoch": 0.6005965061780997,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9932695620793525e-06,
+      "loss": 0.9036,
+      "step": 7048
+    },
+    {
+      "epoch": 0.6006817213463996,
+      "grad_norm": 19.625,
+      "learning_rate": 1.993134453085348e-06,
+      "loss": 1.061,
+      "step": 7049
+    },
+    {
+      "epoch": 0.6007669365146996,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.992999330662156e-06,
+      "loss": 0.568,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6008521516829995,
+      "grad_norm": 10.75,
+      "learning_rate": 1.992864194812219e-06,
+      "loss": 0.411,
+      "step": 7051
+    },
+    {
+      "epoch": 0.6009373668512995,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.9927290455379792e-06,
+      "loss": 0.3216,
+      "step": 7052
+    },
+    {
+      "epoch": 0.6010225820195995,
+      "grad_norm": 25.375,
+      "learning_rate": 1.992593882841879e-06,
+      "loss": 0.9012,
+      "step": 7053
+    },
+    {
+      "epoch": 0.6011077971878994,
+      "grad_norm": 26.0,
+      "learning_rate": 1.9924587067263603e-06,
+      "loss": 1.1129,
+      "step": 7054
+    },
+    {
+      "epoch": 0.6011930123561994,
+      "grad_norm": 13.125,
+      "learning_rate": 1.992323517193866e-06,
+      "loss": 0.7143,
+      "step": 7055
+    },
+    {
+      "epoch": 0.6012782275244993,
+      "grad_norm": 18.625,
+      "learning_rate": 1.992188314246839e-06,
+      "loss": 0.7174,
+      "step": 7056
+    },
+    {
+      "epoch": 0.6013634426927993,
+      "grad_norm": 11.375,
+      "learning_rate": 1.9920530978877235e-06,
+      "loss": 0.4578,
+      "step": 7057
+    },
+    {
+      "epoch": 0.6014486578610992,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.991917868118962e-06,
+      "loss": 0.8299,
+      "step": 7058
+    },
+    {
+      "epoch": 0.6015338730293992,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.991782624942998e-06,
+      "loss": 0.9186,
+      "step": 7059
+    },
+    {
+      "epoch": 0.6016190881976992,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9916473683622773e-06,
+      "loss": 0.6989,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6017043033659991,
+      "grad_norm": 13.375,
+      "learning_rate": 1.991512098379242e-06,
+      "loss": 0.4594,
+      "step": 7061
+    },
+    {
+      "epoch": 0.6017895185342991,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.991376814996338e-06,
+      "loss": 0.5327,
+      "step": 7062
+    },
+    {
+      "epoch": 0.601874733702599,
+      "grad_norm": 16.125,
+      "learning_rate": 1.991241518216009e-06,
+      "loss": 1.1869,
+      "step": 7063
+    },
+    {
+      "epoch": 0.601959948870899,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9911062080407003e-06,
+      "loss": 0.703,
+      "step": 7064
+    },
+    {
+      "epoch": 0.6020451640391989,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9909708844728578e-06,
+      "loss": 0.7683,
+      "step": 7065
+    },
+    {
+      "epoch": 0.6021303792074989,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9908355475149257e-06,
+      "loss": 0.6584,
+      "step": 7066
+    },
+    {
+      "epoch": 0.6022155943757989,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9907001971693503e-06,
+      "loss": 0.9931,
+      "step": 7067
+    },
+    {
+      "epoch": 0.6023008095440988,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.990564833438578e-06,
+      "loss": 0.7852,
+      "step": 7068
+    },
+    {
+      "epoch": 0.6023860247123988,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9904294563250537e-06,
+      "loss": 0.533,
+      "step": 7069
+    },
+    {
+      "epoch": 0.6024712398806987,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9902940658312254e-06,
+      "loss": 0.7965,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6025564550489987,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9901586619595383e-06,
+      "loss": 1.1383,
+      "step": 7071
+    },
+    {
+      "epoch": 0.6026416702172986,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9900232447124403e-06,
+      "loss": 0.6484,
+      "step": 7072
+    },
+    {
+      "epoch": 0.6027268853855986,
+      "grad_norm": 19.75,
+      "learning_rate": 1.9898878140923777e-06,
+      "loss": 1.0655,
+      "step": 7073
+    },
+    {
+      "epoch": 0.6028121005538986,
+      "grad_norm": 20.625,
+      "learning_rate": 1.9897523701017983e-06,
+      "loss": 0.9423,
+      "step": 7074
+    },
+    {
+      "epoch": 0.6028973157221985,
+      "grad_norm": 14.375,
+      "learning_rate": 1.98961691274315e-06,
+      "loss": 0.7268,
+      "step": 7075
+    },
+    {
+      "epoch": 0.6029825308904985,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9894814420188808e-06,
+      "loss": 0.7933,
+      "step": 7076
+    },
+    {
+      "epoch": 0.6030677460587984,
+      "grad_norm": 20.75,
+      "learning_rate": 1.989345957931437e-06,
+      "loss": 0.9622,
+      "step": 7077
+    },
+    {
+      "epoch": 0.6031529612270984,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9892104604832698e-06,
+      "loss": 0.591,
+      "step": 7078
+    },
+    {
+      "epoch": 0.6032381763953983,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9890749496768253e-06,
+      "loss": 0.698,
+      "step": 7079
+    },
+    {
+      "epoch": 0.6033233915636983,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9889394255145534e-06,
+      "loss": 1.0469,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6034086067319983,
+      "grad_norm": 16.875,
+      "learning_rate": 1.988803887998903e-06,
+      "loss": 0.8548,
+      "step": 7081
+    },
+    {
+      "epoch": 0.6034938219002982,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.988668337132324e-06,
+      "loss": 0.6733,
+      "step": 7082
+    },
+    {
+      "epoch": 0.6035790370685982,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9885327729172642e-06,
+      "loss": 0.8537,
+      "step": 7083
+    },
+    {
+      "epoch": 0.6036642522368981,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9883971953561752e-06,
+      "loss": 0.8663,
+      "step": 7084
+    },
+    {
+      "epoch": 0.6037494674051981,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9882616044515064e-06,
+      "loss": 0.844,
+      "step": 7085
+    },
+    {
+      "epoch": 0.603834682573498,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9881260002057082e-06,
+      "loss": 0.8308,
+      "step": 7086
+    },
+    {
+      "epoch": 0.603919897741798,
+      "grad_norm": 15.75,
+      "learning_rate": 1.987990382621231e-06,
+      "loss": 0.8632,
+      "step": 7087
+    },
+    {
+      "epoch": 0.604005112910098,
+      "grad_norm": 17.125,
+      "learning_rate": 1.987854751700525e-06,
+      "loss": 0.6834,
+      "step": 7088
+    },
+    {
+      "epoch": 0.6040903280783979,
+      "grad_norm": 12.125,
+      "learning_rate": 1.9877191074460416e-06,
+      "loss": 0.7014,
+      "step": 7089
+    },
+    {
+      "epoch": 0.6041755432466979,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9875834498602325e-06,
+      "loss": 0.7654,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6042607584149978,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.987447778945549e-06,
+      "loss": 0.7517,
+      "step": 7091
+    },
+    {
+      "epoch": 0.6043459735832978,
+      "grad_norm": 14.25,
+      "learning_rate": 1.987312094704442e-06,
+      "loss": 1.0279,
+      "step": 7092
+    },
+    {
+      "epoch": 0.6044311887515977,
+      "grad_norm": 18.375,
+      "learning_rate": 1.987176397139364e-06,
+      "loss": 0.6282,
+      "step": 7093
+    },
+    {
+      "epoch": 0.6045164039198977,
+      "grad_norm": 18.125,
+      "learning_rate": 1.987040686252768e-06,
+      "loss": 0.845,
+      "step": 7094
+    },
+    {
+      "epoch": 0.6046016190881977,
+      "grad_norm": 11.875,
+      "learning_rate": 1.9869049620471048e-06,
+      "loss": 0.4352,
+      "step": 7095
+    },
+    {
+      "epoch": 0.6046868342564976,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9867692245248283e-06,
+      "loss": 0.4568,
+      "step": 7096
+    },
+    {
+      "epoch": 0.6047720494247976,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.986633473688391e-06,
+      "loss": 0.533,
+      "step": 7097
+    },
+    {
+      "epoch": 0.6048572645930975,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.986497709540246e-06,
+      "loss": 0.7173,
+      "step": 7098
+    },
+    {
+      "epoch": 0.6049424797613975,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.986361932082847e-06,
+      "loss": 0.7891,
+      "step": 7099
+    },
+    {
+      "epoch": 0.6050276949296974,
+      "grad_norm": 15.75,
+      "learning_rate": 1.986226141318647e-06,
+      "loss": 0.6882,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6051129100979974,
+      "grad_norm": 19.0,
+      "learning_rate": 1.986090337250101e-06,
+      "loss": 0.9855,
+      "step": 7101
+    },
+    {
+      "epoch": 0.6051981252662973,
+      "grad_norm": 19.75,
+      "learning_rate": 1.9859545198796615e-06,
+      "loss": 0.995,
+      "step": 7102
+    },
+    {
+      "epoch": 0.6052833404345973,
+      "grad_norm": 27.5,
+      "learning_rate": 1.9858186892097843e-06,
+      "loss": 1.2267,
+      "step": 7103
+    },
+    {
+      "epoch": 0.6053685556028973,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9856828452429234e-06,
+      "loss": 0.9777,
+      "step": 7104
+    },
+    {
+      "epoch": 0.6054537707711972,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9855469879815335e-06,
+      "loss": 0.7694,
+      "step": 7105
+    },
+    {
+      "epoch": 0.6055389859394972,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9854111174280706e-06,
+      "loss": 0.7356,
+      "step": 7106
+    },
+    {
+      "epoch": 0.6056242011077971,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9852752335849886e-06,
+      "loss": 0.6151,
+      "step": 7107
+    },
+    {
+      "epoch": 0.6057094162760971,
+      "grad_norm": 22.375,
+      "learning_rate": 1.9851393364547434e-06,
+      "loss": 1.1245,
+      "step": 7108
+    },
+    {
+      "epoch": 0.6057946314443972,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.985003426039792e-06,
+      "loss": 0.6953,
+      "step": 7109
+    },
+    {
+      "epoch": 0.6058798466126971,
+      "grad_norm": 18.0,
+      "learning_rate": 1.984867502342589e-06,
+      "loss": 0.6803,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6059650617809971,
+      "grad_norm": 10.0,
+      "learning_rate": 1.9847315653655916e-06,
+      "loss": 0.8856,
+      "step": 7111
+    },
+    {
+      "epoch": 0.606050276949297,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.984595615111256e-06,
+      "loss": 0.8024,
+      "step": 7112
+    },
+    {
+      "epoch": 0.606135492117597,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9844596515820388e-06,
+      "loss": 0.7841,
+      "step": 7113
+    },
+    {
+      "epoch": 0.6062207072858969,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.984323674780397e-06,
+      "loss": 0.8036,
+      "step": 7114
+    },
+    {
+      "epoch": 0.6063059224541969,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.984187684708788e-06,
+      "loss": 0.4548,
+      "step": 7115
+    },
+    {
+      "epoch": 0.6063911376224969,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.98405168136967e-06,
+      "loss": 0.7197,
+      "step": 7116
+    },
+    {
+      "epoch": 0.6064763527907968,
+      "grad_norm": 19.0,
+      "learning_rate": 1.983915664765499e-06,
+      "loss": 0.9,
+      "step": 7117
+    },
+    {
+      "epoch": 0.6065615679590968,
+      "grad_norm": 17.125,
+      "learning_rate": 1.983779634898734e-06,
+      "loss": 0.7093,
+      "step": 7118
+    },
+    {
+      "epoch": 0.6066467831273967,
+      "grad_norm": 24.0,
+      "learning_rate": 1.983643591771834e-06,
+      "loss": 1.4911,
+      "step": 7119
+    },
+    {
+      "epoch": 0.6067319982956967,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.983507535387256e-06,
+      "loss": 1.0796,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6068172134639966,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.983371465747459e-06,
+      "loss": 0.6264,
+      "step": 7121
+    },
+    {
+      "epoch": 0.6069024286322966,
+      "grad_norm": 11.75,
+      "learning_rate": 1.9832353828549025e-06,
+      "loss": 0.4159,
+      "step": 7122
+    },
+    {
+      "epoch": 0.6069876438005966,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9830992867120454e-06,
+      "loss": 0.833,
+      "step": 7123
+    },
+    {
+      "epoch": 0.6070728589688965,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9829631773213466e-06,
+      "loss": 0.5283,
+      "step": 7124
+    },
+    {
+      "epoch": 0.6071580741371965,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.982827054685266e-06,
+      "loss": 0.7466,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6072432893054964,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.982690918806264e-06,
+      "loss": 0.9566,
+      "step": 7126
+    },
+    {
+      "epoch": 0.6073285044737964,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.9825547696868e-06,
+      "loss": 0.5742,
+      "step": 7127
+    },
+    {
+      "epoch": 0.6074137196420963,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9824186073293354e-06,
+      "loss": 0.9767,
+      "step": 7128
+    },
+    {
+      "epoch": 0.6074989348103963,
+      "grad_norm": 17.875,
+      "learning_rate": 1.982282431736329e-06,
+      "loss": 0.8745,
+      "step": 7129
+    },
+    {
+      "epoch": 0.6075841499786963,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9821462429102435e-06,
+      "loss": 1.0922,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6076693651469962,
+      "grad_norm": 10.5,
+      "learning_rate": 1.9820100408535393e-06,
+      "loss": 0.4651,
+      "step": 7131
+    },
+    {
+      "epoch": 0.6077545803152962,
+      "grad_norm": 22.875,
+      "learning_rate": 1.981873825568677e-06,
+      "loss": 0.9883,
+      "step": 7132
+    },
+    {
+      "epoch": 0.6078397954835961,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9817375970581187e-06,
+      "loss": 0.3495,
+      "step": 7133
+    },
+    {
+      "epoch": 0.6079250106518961,
+      "grad_norm": 18.25,
+      "learning_rate": 1.9816013553243265e-06,
+      "loss": 1.1247,
+      "step": 7134
+    },
+    {
+      "epoch": 0.608010225820196,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.981465100369762e-06,
+      "loss": 0.785,
+      "step": 7135
+    },
+    {
+      "epoch": 0.608095440988496,
+      "grad_norm": 16.375,
+      "learning_rate": 1.981328832196888e-06,
+      "loss": 0.7935,
+      "step": 7136
+    },
+    {
+      "epoch": 0.608180656156796,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.981192550808166e-06,
+      "loss": 0.9892,
+      "step": 7137
+    },
+    {
+      "epoch": 0.6082658713250959,
+      "grad_norm": 15.0,
+      "learning_rate": 1.98105625620606e-06,
+      "loss": 0.8352,
+      "step": 7138
+    },
+    {
+      "epoch": 0.6083510864933959,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9809199483930317e-06,
+      "loss": 0.4018,
+      "step": 7139
+    },
+    {
+      "epoch": 0.6084363016616958,
+      "grad_norm": 13.875,
+      "learning_rate": 1.980783627371545e-06,
+      "loss": 0.7114,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6085215168299958,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9806472931440634e-06,
+      "loss": 0.3879,
+      "step": 7141
+    },
+    {
+      "epoch": 0.6086067319982957,
+      "grad_norm": 11.625,
+      "learning_rate": 1.9805109457130507e-06,
+      "loss": 0.5047,
+      "step": 7142
+    },
+    {
+      "epoch": 0.6086919471665957,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.98037458508097e-06,
+      "loss": 0.6552,
+      "step": 7143
+    },
+    {
+      "epoch": 0.6087771623348956,
+      "grad_norm": 25.25,
+      "learning_rate": 1.9802382112502867e-06,
+      "loss": 0.9809,
+      "step": 7144
+    },
+    {
+      "epoch": 0.6088623775031956,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9801018242234644e-06,
+      "loss": 0.8074,
+      "step": 7145
+    },
+    {
+      "epoch": 0.6089475926714956,
+      "grad_norm": 17.625,
+      "learning_rate": 1.979965424002968e-06,
+      "loss": 0.7705,
+      "step": 7146
+    },
+    {
+      "epoch": 0.6090328078397955,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9798290105912625e-06,
+      "loss": 0.961,
+      "step": 7147
+    },
+    {
+      "epoch": 0.6091180230080955,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9796925839908127e-06,
+      "loss": 0.6139,
+      "step": 7148
+    },
+    {
+      "epoch": 0.6092032381763954,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.979556144204084e-06,
+      "loss": 0.5328,
+      "step": 7149
+    },
+    {
+      "epoch": 0.6092884533446954,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.9794196912335422e-06,
+      "loss": 0.4705,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6093736685129953,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.9792832250816534e-06,
+      "loss": 0.3359,
+      "step": 7151
+    },
+    {
+      "epoch": 0.6094588836812953,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9791467457508836e-06,
+      "loss": 0.5211,
+      "step": 7152
+    },
+    {
+      "epoch": 0.6095440988495953,
+      "grad_norm": 88.0,
+      "learning_rate": 1.9790102532436984e-06,
+      "loss": 1.0034,
+      "step": 7153
+    },
+    {
+      "epoch": 0.6096293140178952,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9788737475625645e-06,
+      "loss": 0.8336,
+      "step": 7154
+    },
+    {
+      "epoch": 0.6097145291861952,
+      "grad_norm": 19.875,
+      "learning_rate": 1.9787372287099497e-06,
+      "loss": 1.0461,
+      "step": 7155
+    },
+    {
+      "epoch": 0.6097997443544951,
+      "grad_norm": 13.25,
+      "learning_rate": 1.97860069668832e-06,
+      "loss": 0.5638,
+      "step": 7156
+    },
+    {
+      "epoch": 0.6098849595227951,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9784641515001436e-06,
+      "loss": 0.3514,
+      "step": 7157
+    },
+    {
+      "epoch": 0.609970174691095,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9783275931478867e-06,
+      "loss": 0.6992,
+      "step": 7158
+    },
+    {
+      "epoch": 0.610055389859395,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9781910216340185e-06,
+      "loss": 0.8709,
+      "step": 7159
+    },
+    {
+      "epoch": 0.610140605027695,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.9780544369610055e-06,
+      "loss": 0.4385,
+      "step": 7160
+    },
+    {
+      "epoch": 0.6102258201959949,
+      "grad_norm": 22.625,
+      "learning_rate": 1.9779178391313174e-06,
+      "loss": 0.789,
+      "step": 7161
+    },
+    {
+      "epoch": 0.6103110353642949,
+      "grad_norm": 12.5,
+      "learning_rate": 1.977781228147422e-06,
+      "loss": 0.6892,
+      "step": 7162
+    },
+    {
+      "epoch": 0.6103962505325948,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9776446040117877e-06,
+      "loss": 0.7682,
+      "step": 7163
+    },
+    {
+      "epoch": 0.6104814657008948,
+      "grad_norm": 14.875,
+      "learning_rate": 1.977507966726883e-06,
+      "loss": 0.7456,
+      "step": 7164
+    },
+    {
+      "epoch": 0.6105666808691947,
+      "grad_norm": 18.75,
+      "learning_rate": 1.9773713162951787e-06,
+      "loss": 0.9256,
+      "step": 7165
+    },
+    {
+      "epoch": 0.6106518960374947,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.977234652719143e-06,
+      "loss": 0.913,
+      "step": 7166
+    },
+    {
+      "epoch": 0.6107371112057947,
+      "grad_norm": 17.25,
+      "learning_rate": 1.977097976001246e-06,
+      "loss": 0.9924,
+      "step": 7167
+    },
+    {
+      "epoch": 0.6108223263740946,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9769612861439573e-06,
+      "loss": 0.9682,
+      "step": 7168
+    },
+    {
+      "epoch": 0.6109075415423946,
+      "grad_norm": 11.625,
+      "learning_rate": 1.976824583149747e-06,
+      "loss": 0.4987,
+      "step": 7169
+    },
+    {
+      "epoch": 0.6109927567106945,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.976687867021086e-06,
+      "loss": 0.8199,
+      "step": 7170
+    },
+    {
+      "epoch": 0.6110779718789945,
+      "grad_norm": 10.625,
+      "learning_rate": 1.976551137760444e-06,
+      "loss": 0.5617,
+      "step": 7171
+    },
+    {
+      "epoch": 0.6111631870472944,
+      "grad_norm": 17.125,
+      "learning_rate": 1.976414395370293e-06,
+      "loss": 0.8479,
+      "step": 7172
+    },
+    {
+      "epoch": 0.6112484022155944,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9762776398531027e-06,
+      "loss": 0.9526,
+      "step": 7173
+    },
+    {
+      "epoch": 0.6113336173838944,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9761408712113455e-06,
+      "loss": 0.574,
+      "step": 7174
+    },
+    {
+      "epoch": 0.6114188325521943,
+      "grad_norm": 11.625,
+      "learning_rate": 1.9760040894474923e-06,
+      "loss": 0.5072,
+      "step": 7175
+    },
+    {
+      "epoch": 0.6115040477204943,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9758672945640157e-06,
+      "loss": 0.7083,
+      "step": 7176
+    },
+    {
+      "epoch": 0.6115892628887942,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.9757304865633864e-06,
+      "loss": 0.6307,
+      "step": 7177
+    },
+    {
+      "epoch": 0.6116744780570942,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.975593665448078e-06,
+      "loss": 0.3968,
+      "step": 7178
+    },
+    {
+      "epoch": 0.6117596932253941,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9754568312205625e-06,
+      "loss": 0.5954,
+      "step": 7179
+    },
+    {
+      "epoch": 0.6118449083936941,
+      "grad_norm": 17.375,
+      "learning_rate": 1.975319983883312e-06,
+      "loss": 0.9735,
+      "step": 7180
+    },
+    {
+      "epoch": 0.611930123561994,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9751831234388004e-06,
+      "loss": 0.7383,
+      "step": 7181
+    },
+    {
+      "epoch": 0.612015338730294,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9750462498895003e-06,
+      "loss": 0.708,
+      "step": 7182
+    },
+    {
+      "epoch": 0.612100553898594,
+      "grad_norm": 18.125,
+      "learning_rate": 1.974909363237886e-06,
+      "loss": 0.6898,
+      "step": 7183
+    },
+    {
+      "epoch": 0.6121857690668939,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.9747724634864297e-06,
+      "loss": 0.5799,
+      "step": 7184
+    },
+    {
+      "epoch": 0.6122709842351939,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9746355506376063e-06,
+      "loss": 0.3727,
+      "step": 7185
+    },
+    {
+      "epoch": 0.6123561994034938,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9744986246938902e-06,
+      "loss": 0.8564,
+      "step": 7186
+    },
+    {
+      "epoch": 0.6124414145717938,
+      "grad_norm": 23.125,
+      "learning_rate": 1.9743616856577557e-06,
+      "loss": 0.9636,
+      "step": 7187
+    },
+    {
+      "epoch": 0.6125266297400938,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9742247335316766e-06,
+      "loss": 0.7136,
+      "step": 7188
+    },
+    {
+      "epoch": 0.6126118449083937,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.974087768318128e-06,
+      "loss": 0.9755,
+      "step": 7189
+    },
+    {
+      "epoch": 0.6126970600766937,
+      "grad_norm": 15.0,
+      "learning_rate": 1.973950790019586e-06,
+      "loss": 0.7343,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6127822752449936,
+      "grad_norm": 20.75,
+      "learning_rate": 1.973813798638525e-06,
+      "loss": 0.9544,
+      "step": 7191
+    },
+    {
+      "epoch": 0.6128674904132936,
+      "grad_norm": 11.75,
+      "learning_rate": 1.973676794177421e-06,
+      "loss": 0.49,
+      "step": 7192
+    },
+    {
+      "epoch": 0.6129527055815935,
+      "grad_norm": 11.125,
+      "learning_rate": 1.9735397766387494e-06,
+      "loss": 0.3162,
+      "step": 7193
+    },
+    {
+      "epoch": 0.6130379207498935,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9734027460249862e-06,
+      "loss": 0.4049,
+      "step": 7194
+    },
+    {
+      "epoch": 0.6131231359181935,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9732657023386083e-06,
+      "loss": 0.7312,
+      "step": 7195
+    },
+    {
+      "epoch": 0.6132083510864934,
+      "grad_norm": 13.0,
+      "learning_rate": 1.973128645582092e-06,
+      "loss": 0.8564,
+      "step": 7196
+    },
+    {
+      "epoch": 0.6132935662547934,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9729915757579133e-06,
+      "loss": 0.8346,
+      "step": 7197
+    },
+    {
+      "epoch": 0.6133787814230933,
+      "grad_norm": 17.5,
+      "learning_rate": 1.97285449286855e-06,
+      "loss": 0.8359,
+      "step": 7198
+    },
+    {
+      "epoch": 0.6134639965913933,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9727173969164797e-06,
+      "loss": 0.7552,
+      "step": 7199
+    },
+    {
+      "epoch": 0.6135492117596932,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.972580287904179e-06,
+      "loss": 0.445,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6136344269279932,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9724431658341256e-06,
+      "loss": 0.9417,
+      "step": 7201
+    },
+    {
+      "epoch": 0.6137196420962931,
+      "grad_norm": 15.375,
+      "learning_rate": 1.972306030708798e-06,
+      "loss": 0.6247,
+      "step": 7202
+    },
+    {
+      "epoch": 0.6138048572645931,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9721688825306737e-06,
+      "loss": 0.5556,
+      "step": 7203
+    },
+    {
+      "epoch": 0.6138900724328931,
+      "grad_norm": 19.75,
+      "learning_rate": 1.972031721302232e-06,
+      "loss": 0.7019,
+      "step": 7204
+    },
+    {
+      "epoch": 0.613975287601193,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.971894547025951e-06,
+      "loss": 0.592,
+      "step": 7205
+    },
+    {
+      "epoch": 0.614060502769493,
+      "grad_norm": 12.625,
+      "learning_rate": 1.971757359704309e-06,
+      "loss": 0.4118,
+      "step": 7206
+    },
+    {
+      "epoch": 0.6141457179377929,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9716201593397863e-06,
+      "loss": 0.9746,
+      "step": 7207
+    },
+    {
+      "epoch": 0.6142309331060929,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9714829459348618e-06,
+      "loss": 0.5896,
+      "step": 7208
+    },
+    {
+      "epoch": 0.6143161482743928,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9713457194920143e-06,
+      "loss": 0.705,
+      "step": 7209
+    },
+    {
+      "epoch": 0.6144013634426928,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.971208480013725e-06,
+      "loss": 0.8955,
+      "step": 7210
+    },
+    {
+      "epoch": 0.6144865786109928,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9710712275024726e-06,
+      "loss": 0.9369,
+      "step": 7211
+    },
+    {
+      "epoch": 0.6145717937792927,
+      "grad_norm": 23.625,
+      "learning_rate": 1.9709339619607385e-06,
+      "loss": 1.138,
+      "step": 7212
+    },
+    {
+      "epoch": 0.6146570089475927,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9707966833910025e-06,
+      "loss": 0.8803,
+      "step": 7213
+    },
+    {
+      "epoch": 0.6147422241158926,
+      "grad_norm": 12.875,
+      "learning_rate": 1.9706593917957457e-06,
+      "loss": 0.3852,
+      "step": 7214
+    },
+    {
+      "epoch": 0.6148274392841926,
+      "grad_norm": 11.75,
+      "learning_rate": 1.970522087177449e-06,
+      "loss": 0.3889,
+      "step": 7215
+    },
+    {
+      "epoch": 0.6149126544524925,
+      "grad_norm": 11.875,
+      "learning_rate": 1.970384769538594e-06,
+      "loss": 0.5226,
+      "step": 7216
+    },
+    {
+      "epoch": 0.6149978696207925,
+      "grad_norm": 14.75,
+      "learning_rate": 1.970247438881661e-06,
+      "loss": 0.6125,
+      "step": 7217
+    },
+    {
+      "epoch": 0.6150830847890925,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.970110095209133e-06,
+      "loss": 0.7232,
+      "step": 7218
+    },
+    {
+      "epoch": 0.6151682999573924,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.9699727385234915e-06,
+      "loss": 0.4787,
+      "step": 7219
+    },
+    {
+      "epoch": 0.6152535151256924,
+      "grad_norm": 12.75,
+      "learning_rate": 1.969835368827219e-06,
+      "loss": 0.3575,
+      "step": 7220
+    },
+    {
+      "epoch": 0.6153387302939923,
+      "grad_norm": 17.375,
+      "learning_rate": 1.969697986122797e-06,
+      "loss": 0.9611,
+      "step": 7221
+    },
+    {
+      "epoch": 0.6154239454622923,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9695605904127085e-06,
+      "loss": 0.8128,
+      "step": 7222
+    },
+    {
+      "epoch": 0.6155091606305922,
+      "grad_norm": 24.5,
+      "learning_rate": 1.969423181699437e-06,
+      "loss": 1.3074,
+      "step": 7223
+    },
+    {
+      "epoch": 0.6155943757988922,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.969285759985465e-06,
+      "loss": 0.7217,
+      "step": 7224
+    },
+    {
+      "epoch": 0.6156795909671922,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.969148325273276e-06,
+      "loss": 0.8687,
+      "step": 7225
+    },
+    {
+      "epoch": 0.6157648061354921,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9690108775653535e-06,
+      "loss": 0.771,
+      "step": 7226
+    },
+    {
+      "epoch": 0.6158500213037921,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9688734168641816e-06,
+      "loss": 0.8843,
+      "step": 7227
+    },
+    {
+      "epoch": 0.615935236472092,
+      "grad_norm": 17.625,
+      "learning_rate": 1.968735943172244e-06,
+      "loss": 0.6593,
+      "step": 7228
+    },
+    {
+      "epoch": 0.616020451640392,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9685984564920253e-06,
+      "loss": 1.0138,
+      "step": 7229
+    },
+    {
+      "epoch": 0.6161056668086919,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9684609568260097e-06,
+      "loss": 0.9872,
+      "step": 7230
+    },
+    {
+      "epoch": 0.6161908819769919,
+      "grad_norm": 9.25,
+      "learning_rate": 1.9683234441766824e-06,
+      "loss": 0.2184,
+      "step": 7231
+    },
+    {
+      "epoch": 0.6162760971452919,
+      "grad_norm": 12.375,
+      "learning_rate": 1.968185918546528e-06,
+      "loss": 0.4289,
+      "step": 7232
+    },
+    {
+      "epoch": 0.6163613123135918,
+      "grad_norm": 25.875,
+      "learning_rate": 1.968048379938032e-06,
+      "loss": 1.0873,
+      "step": 7233
+    },
+    {
+      "epoch": 0.6164465274818918,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9679108283536796e-06,
+      "loss": 0.9169,
+      "step": 7234
+    },
+    {
+      "epoch": 0.6165317426501917,
+      "grad_norm": 10.125,
+      "learning_rate": 1.967773263795957e-06,
+      "loss": 0.4318,
+      "step": 7235
+    },
+    {
+      "epoch": 0.6166169578184917,
+      "grad_norm": 14.25,
+      "learning_rate": 1.967635686267349e-06,
+      "loss": 0.6577,
+      "step": 7236
+    },
+    {
+      "epoch": 0.6167021729867916,
+      "grad_norm": 12.125,
+      "learning_rate": 1.967498095770343e-06,
+      "loss": 0.5881,
+      "step": 7237
+    },
+    {
+      "epoch": 0.6167873881550916,
+      "grad_norm": 13.75,
+      "learning_rate": 1.967360492307425e-06,
+      "loss": 0.5848,
+      "step": 7238
+    },
+    {
+      "epoch": 0.6168726033233916,
+      "grad_norm": 18.375,
+      "learning_rate": 1.967222875881081e-06,
+      "loss": 1.015,
+      "step": 7239
+    },
+    {
+      "epoch": 0.6169578184916915,
+      "grad_norm": 18.875,
+      "learning_rate": 1.967085246493799e-06,
+      "loss": 0.959,
+      "step": 7240
+    },
+    {
+      "epoch": 0.6170430336599915,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9669476041480652e-06,
+      "loss": 0.4899,
+      "step": 7241
+    },
+    {
+      "epoch": 0.6171282488282914,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9668099488463672e-06,
+      "loss": 0.833,
+      "step": 7242
+    },
+    {
+      "epoch": 0.6172134639965914,
+      "grad_norm": 18.75,
+      "learning_rate": 1.966672280591193e-06,
+      "loss": 0.8397,
+      "step": 7243
+    },
+    {
+      "epoch": 0.6172986791648913,
+      "grad_norm": 16.125,
+      "learning_rate": 1.96653459938503e-06,
+      "loss": 0.8139,
+      "step": 7244
+    },
+    {
+      "epoch": 0.6173838943331913,
+      "grad_norm": 16.75,
+      "learning_rate": 1.966396905230366e-06,
+      "loss": 0.9423,
+      "step": 7245
+    },
+    {
+      "epoch": 0.6174691095014913,
+      "grad_norm": 13.0,
+      "learning_rate": 1.966259198129689e-06,
+      "loss": 0.3414,
+      "step": 7246
+    },
+    {
+      "epoch": 0.6175543246697912,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.966121478085489e-06,
+      "loss": 0.7184,
+      "step": 7247
+    },
+    {
+      "epoch": 0.6176395398380912,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9659837451002532e-06,
+      "loss": 1.2353,
+      "step": 7248
+    },
+    {
+      "epoch": 0.6177247550063911,
+      "grad_norm": 22.5,
+      "learning_rate": 1.9658459991764713e-06,
+      "loss": 0.743,
+      "step": 7249
+    },
+    {
+      "epoch": 0.6178099701746911,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9657082403166323e-06,
+      "loss": 0.3962,
+      "step": 7250
+    },
+    {
+      "epoch": 0.617895185342991,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9655704685232257e-06,
+      "loss": 0.6572,
+      "step": 7251
+    },
+    {
+      "epoch": 0.617980400511291,
+      "grad_norm": 21.5,
+      "learning_rate": 1.9654326837987414e-06,
+      "loss": 1.2073,
+      "step": 7252
+    },
+    {
+      "epoch": 0.618065615679591,
+      "grad_norm": 20.875,
+      "learning_rate": 1.965294886145669e-06,
+      "loss": 1.2191,
+      "step": 7253
+    },
+    {
+      "epoch": 0.6181508308478909,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9651570755664983e-06,
+      "loss": 0.8051,
+      "step": 7254
+    },
+    {
+      "epoch": 0.6182360460161909,
+      "grad_norm": 12.375,
+      "learning_rate": 1.965019252063721e-06,
+      "loss": 0.5261,
+      "step": 7255
+    },
+    {
+      "epoch": 0.6183212611844908,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.964881415639826e-06,
+      "loss": 0.9768,
+      "step": 7256
+    },
+    {
+      "epoch": 0.6184064763527908,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.964743566297305e-06,
+      "loss": 0.4253,
+      "step": 7257
+    },
+    {
+      "epoch": 0.6184916915210907,
+      "grad_norm": 12.25,
+      "learning_rate": 1.964605704038649e-06,
+      "loss": 0.4746,
+      "step": 7258
+    },
+    {
+      "epoch": 0.6185769066893907,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.9644678288663497e-06,
+      "loss": 0.6493,
+      "step": 7259
+    },
+    {
+      "epoch": 0.6186621218576907,
+      "grad_norm": 11.875,
+      "learning_rate": 1.964329940782898e-06,
+      "loss": 0.5322,
+      "step": 7260
+    },
+    {
+      "epoch": 0.6187473370259906,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9641920397907864e-06,
+      "loss": 0.6982,
+      "step": 7261
+    },
+    {
+      "epoch": 0.6188325521942906,
+      "grad_norm": 12.125,
+      "learning_rate": 1.9640541258925062e-06,
+      "loss": 0.4661,
+      "step": 7262
+    },
+    {
+      "epoch": 0.6189177673625905,
+      "grad_norm": 14.5,
+      "learning_rate": 1.96391619909055e-06,
+      "loss": 0.9326,
+      "step": 7263
+    },
+    {
+      "epoch": 0.6190029825308905,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9637782593874097e-06,
+      "loss": 0.4321,
+      "step": 7264
+    },
+    {
+      "epoch": 0.6190881976991904,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.963640306785579e-06,
+      "loss": 0.7314,
+      "step": 7265
+    },
+    {
+      "epoch": 0.6191734128674904,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9635023412875497e-06,
+      "loss": 0.442,
+      "step": 7266
+    },
+    {
+      "epoch": 0.6192586280357903,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9633643628958163e-06,
+      "loss": 0.8016,
+      "step": 7267
+    },
+    {
+      "epoch": 0.6193438432040903,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9632263716128713e-06,
+      "loss": 0.8831,
+      "step": 7268
+    },
+    {
+      "epoch": 0.6194290583723903,
+      "grad_norm": 10.375,
+      "learning_rate": 1.9630883674412086e-06,
+      "loss": 0.2588,
+      "step": 7269
+    },
+    {
+      "epoch": 0.6195142735406902,
+      "grad_norm": 17.25,
+      "learning_rate": 1.962950350383322e-06,
+      "loss": 0.9381,
+      "step": 7270
+    },
+    {
+      "epoch": 0.6195994887089902,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9628123204417055e-06,
+      "loss": 0.928,
+      "step": 7271
+    },
+    {
+      "epoch": 0.6196847038772901,
+      "grad_norm": 16.875,
+      "learning_rate": 1.962674277618854e-06,
+      "loss": 0.8597,
+      "step": 7272
+    },
+    {
+      "epoch": 0.6197699190455901,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9625362219172616e-06,
+      "loss": 0.9568,
+      "step": 7273
+    },
+    {
+      "epoch": 0.61985513421389,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9623981533394226e-06,
+      "loss": 0.798,
+      "step": 7274
+    },
+    {
+      "epoch": 0.61994034938219,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9622600718878328e-06,
+      "loss": 0.6489,
+      "step": 7275
+    },
+    {
+      "epoch": 0.62002556455049,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9621219775649873e-06,
+      "loss": 0.6409,
+      "step": 7276
+    },
+    {
+      "epoch": 0.6201107797187899,
+      "grad_norm": 15.25,
+      "learning_rate": 1.961983870373382e-06,
+      "loss": 0.775,
+      "step": 7277
+    },
+    {
+      "epoch": 0.6201959948870899,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.961845750315512e-06,
+      "loss": 0.5108,
+      "step": 7278
+    },
+    {
+      "epoch": 0.6202812100553898,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.961707617393873e-06,
+      "loss": 0.4268,
+      "step": 7279
+    },
+    {
+      "epoch": 0.6203664252236898,
+      "grad_norm": 23.25,
+      "learning_rate": 1.9615694716109622e-06,
+      "loss": 1.4864,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6204516403919897,
+      "grad_norm": 13.125,
+      "learning_rate": 1.961431312969275e-06,
+      "loss": 0.7875,
+      "step": 7281
+    },
+    {
+      "epoch": 0.6205368555602897,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9612931414713092e-06,
+      "loss": 0.6852,
+      "step": 7282
+    },
+    {
+      "epoch": 0.6206220707285897,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9611549571195608e-06,
+      "loss": 0.8346,
+      "step": 7283
+    },
+    {
+      "epoch": 0.6207072858968896,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9610167599165268e-06,
+      "loss": 0.6892,
+      "step": 7284
+    },
+    {
+      "epoch": 0.6207925010651896,
+      "grad_norm": 13.625,
+      "learning_rate": 1.960878549864706e-06,
+      "loss": 0.6599,
+      "step": 7285
+    },
+    {
+      "epoch": 0.6208777162334895,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9607403269665943e-06,
+      "loss": 1.1817,
+      "step": 7286
+    },
+    {
+      "epoch": 0.6209629314017895,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.96060209122469e-06,
+      "loss": 0.7801,
+      "step": 7287
+    },
+    {
+      "epoch": 0.6210481465700894,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.960463842641491e-06,
+      "loss": 0.6956,
+      "step": 7288
+    },
+    {
+      "epoch": 0.6211333617383894,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.960325581219497e-06,
+      "loss": 0.7612,
+      "step": 7289
+    },
+    {
+      "epoch": 0.6212185769066894,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9601873069612047e-06,
+      "loss": 0.4868,
+      "step": 7290
+    },
+    {
+      "epoch": 0.6213037920749893,
+      "grad_norm": 17.375,
+      "learning_rate": 1.960049019869114e-06,
+      "loss": 0.7407,
+      "step": 7291
+    },
+    {
+      "epoch": 0.6213890072432893,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9599107199457234e-06,
+      "loss": 0.8536,
+      "step": 7292
+    },
+    {
+      "epoch": 0.6214742224115892,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9597724071935324e-06,
+      "loss": 0.927,
+      "step": 7293
+    },
+    {
+      "epoch": 0.6215594375798892,
+      "grad_norm": 17.25,
+      "learning_rate": 1.95963408161504e-06,
+      "loss": 0.7374,
+      "step": 7294
+    },
+    {
+      "epoch": 0.6216446527481891,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.959495743212746e-06,
+      "loss": 1.0294,
+      "step": 7295
+    },
+    {
+      "epoch": 0.6217298679164891,
+      "grad_norm": 9.25,
+      "learning_rate": 1.9593573919891506e-06,
+      "loss": 0.2492,
+      "step": 7296
+    },
+    {
+      "epoch": 0.621815083084789,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.959219027946754e-06,
+      "loss": 0.8218,
+      "step": 7297
+    },
+    {
+      "epoch": 0.621900298253089,
+      "grad_norm": 13.125,
+      "learning_rate": 1.959080651088056e-06,
+      "loss": 0.6199,
+      "step": 7298
+    },
+    {
+      "epoch": 0.621985513421389,
+      "grad_norm": 14.25,
+      "learning_rate": 1.958942261415558e-06,
+      "loss": 0.6136,
+      "step": 7299
+    },
+    {
+      "epoch": 0.6220707285896889,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9588038589317604e-06,
+      "loss": 0.7423,
+      "step": 7300
+    },
+    {
+      "epoch": 0.6221559437579889,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9586654436391644e-06,
+      "loss": 0.7883,
+      "step": 7301
+    },
+    {
+      "epoch": 0.6222411589262888,
+      "grad_norm": 14.0,
+      "learning_rate": 1.958527015540271e-06,
+      "loss": 0.7382,
+      "step": 7302
+    },
+    {
+      "epoch": 0.6223263740945888,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.958388574637582e-06,
+      "loss": 0.6697,
+      "step": 7303
+    },
+    {
+      "epoch": 0.6224115892628888,
+      "grad_norm": 23.375,
+      "learning_rate": 1.958250120933599e-06,
+      "loss": 1.1391,
+      "step": 7304
+    },
+    {
+      "epoch": 0.6224968044311887,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9581116544308243e-06,
+      "loss": 0.8311,
+      "step": 7305
+    },
+    {
+      "epoch": 0.6225820195994887,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9579731751317597e-06,
+      "loss": 0.5279,
+      "step": 7306
+    },
+    {
+      "epoch": 0.6226672347677886,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9578346830389083e-06,
+      "loss": 0.6692,
+      "step": 7307
+    },
+    {
+      "epoch": 0.6227524499360886,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.957696178154772e-06,
+      "loss": 0.7927,
+      "step": 7308
+    },
+    {
+      "epoch": 0.6228376651043885,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9575576604818544e-06,
+      "loss": 0.6831,
+      "step": 7309
+    },
+    {
+      "epoch": 0.6229228802726885,
+      "grad_norm": 40.25,
+      "learning_rate": 1.957419130022658e-06,
+      "loss": 0.8341,
+      "step": 7310
+    },
+    {
+      "epoch": 0.6230080954409885,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9572805867796866e-06,
+      "loss": 1.093,
+      "step": 7311
+    },
+    {
+      "epoch": 0.6230933106092884,
+      "grad_norm": 23.125,
+      "learning_rate": 1.9571420307554446e-06,
+      "loss": 0.8997,
+      "step": 7312
+    },
+    {
+      "epoch": 0.6231785257775884,
+      "grad_norm": 18.625,
+      "learning_rate": 1.957003461952434e-06,
+      "loss": 1.2009,
+      "step": 7313
+    },
+    {
+      "epoch": 0.6232637409458883,
+      "grad_norm": 11.75,
+      "learning_rate": 1.9568648803731603e-06,
+      "loss": 0.4673,
+      "step": 7314
+    },
+    {
+      "epoch": 0.6233489561141883,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9567262860201273e-06,
+      "loss": 0.7243,
+      "step": 7315
+    },
+    {
+      "epoch": 0.6234341712824882,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.95658767889584e-06,
+      "loss": 1.0163,
+      "step": 7316
+    },
+    {
+      "epoch": 0.6235193864507882,
+      "grad_norm": 14.5,
+      "learning_rate": 1.9564490590028026e-06,
+      "loss": 0.8731,
+      "step": 7317
+    },
+    {
+      "epoch": 0.6236046016190882,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9563104263435203e-06,
+      "loss": 0.8048,
+      "step": 7318
+    },
+    {
+      "epoch": 0.6236898167873881,
+      "grad_norm": 28.375,
+      "learning_rate": 1.9561717809204984e-06,
+      "loss": 0.8914,
+      "step": 7319
+    },
+    {
+      "epoch": 0.6237750319556881,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.956033122736242e-06,
+      "loss": 0.3814,
+      "step": 7320
+    },
+    {
+      "epoch": 0.623860247123988,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9558944517932574e-06,
+      "loss": 0.8612,
+      "step": 7321
+    },
+    {
+      "epoch": 0.623945462292288,
+      "grad_norm": 30.25,
+      "learning_rate": 1.9557557680940503e-06,
+      "loss": 1.2453,
+      "step": 7322
+    },
+    {
+      "epoch": 0.6240306774605879,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9556170716411266e-06,
+      "loss": 0.5864,
+      "step": 7323
+    },
+    {
+      "epoch": 0.6241158926288879,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9554783624369926e-06,
+      "loss": 0.3767,
+      "step": 7324
+    },
+    {
+      "epoch": 0.6242011077971878,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9553396404841555e-06,
+      "loss": 0.6376,
+      "step": 7325
+    },
+    {
+      "epoch": 0.6242863229654878,
+      "grad_norm": 13.0,
+      "learning_rate": 1.9552009057851218e-06,
+      "loss": 0.6214,
+      "step": 7326
+    },
+    {
+      "epoch": 0.6243715381337878,
+      "grad_norm": 12.875,
+      "learning_rate": 1.9550621583423985e-06,
+      "loss": 0.7112,
+      "step": 7327
+    },
+    {
+      "epoch": 0.6244567533020877,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.954923398158493e-06,
+      "loss": 0.8004,
+      "step": 7328
+    },
+    {
+      "epoch": 0.6245419684703877,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.954784625235913e-06,
+      "loss": 0.4881,
+      "step": 7329
+    },
+    {
+      "epoch": 0.6246271836386876,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.954645839577166e-06,
+      "loss": 0.55,
+      "step": 7330
+    },
+    {
+      "epoch": 0.6247123988069876,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.9545070411847604e-06,
+      "loss": 0.4759,
+      "step": 7331
+    },
+    {
+      "epoch": 0.6247976139752875,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9543682300612037e-06,
+      "loss": 0.4297,
+      "step": 7332
+    },
+    {
+      "epoch": 0.6248828291435875,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.954229406209005e-06,
+      "loss": 0.8195,
+      "step": 7333
+    },
+    {
+      "epoch": 0.6249680443118875,
+      "grad_norm": 11.125,
+      "learning_rate": 1.954090569630673e-06,
+      "loss": 0.5054,
+      "step": 7334
+    },
+    {
+      "epoch": 0.6250532594801875,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9539517203287157e-06,
+      "loss": 0.7437,
+      "step": 7335
+    },
+    {
+      "epoch": 0.6251384746484875,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.9538128583056435e-06,
+      "loss": 0.5343,
+      "step": 7336
+    },
+    {
+      "epoch": 0.6252236898167874,
+      "grad_norm": 9.75,
+      "learning_rate": 1.953673983563965e-06,
+      "loss": 0.2756,
+      "step": 7337
+    },
+    {
+      "epoch": 0.6253089049850874,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9535350961061903e-06,
+      "loss": 0.9079,
+      "step": 7338
+    },
+    {
+      "epoch": 0.6253941201533874,
+      "grad_norm": 30.125,
+      "learning_rate": 1.9533961959348284e-06,
+      "loss": 1.1096,
+      "step": 7339
+    },
+    {
+      "epoch": 0.6254793353216873,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.9532572830523904e-06,
+      "loss": 0.8761,
+      "step": 7340
+    },
+    {
+      "epoch": 0.6255645504899873,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.953118357461386e-06,
+      "loss": 0.7214,
+      "step": 7341
+    },
+    {
+      "epoch": 0.6256497656582872,
+      "grad_norm": 11.625,
+      "learning_rate": 1.952979419164326e-06,
+      "loss": 0.4724,
+      "step": 7342
+    },
+    {
+      "epoch": 0.6257349808265872,
+      "grad_norm": 11.375,
+      "learning_rate": 1.9528404681637205e-06,
+      "loss": 0.4248,
+      "step": 7343
+    },
+    {
+      "epoch": 0.6258201959948871,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9527015044620814e-06,
+      "loss": 0.9084,
+      "step": 7344
+    },
+    {
+      "epoch": 0.6259054111631871,
+      "grad_norm": 17.25,
+      "learning_rate": 1.952562528061919e-06,
+      "loss": 0.897,
+      "step": 7345
+    },
+    {
+      "epoch": 0.625990626331487,
+      "grad_norm": 25.125,
+      "learning_rate": 1.952423538965746e-06,
+      "loss": 0.9288,
+      "step": 7346
+    },
+    {
+      "epoch": 0.626075841499787,
+      "grad_norm": 74.5,
+      "learning_rate": 1.952284537176073e-06,
+      "loss": 0.8996,
+      "step": 7347
+    },
+    {
+      "epoch": 0.626161056668087,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9521455226954124e-06,
+      "loss": 0.6105,
+      "step": 7348
+    },
+    {
+      "epoch": 0.6262462718363869,
+      "grad_norm": 9.125,
+      "learning_rate": 1.9520064955262756e-06,
+      "loss": 0.6165,
+      "step": 7349
+    },
+    {
+      "epoch": 0.6263314870046869,
+      "grad_norm": 17.0,
+      "learning_rate": 1.951867455671176e-06,
+      "loss": 0.8216,
+      "step": 7350
+    },
+    {
+      "epoch": 0.6264167021729868,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.951728403132625e-06,
+      "loss": 0.528,
+      "step": 7351
+    },
+    {
+      "epoch": 0.6265019173412868,
+      "grad_norm": 13.5,
+      "learning_rate": 1.951589337913137e-06,
+      "loss": 0.582,
+      "step": 7352
+    },
+    {
+      "epoch": 0.6265871325095868,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9514502600152236e-06,
+      "loss": 0.7162,
+      "step": 7353
+    },
+    {
+      "epoch": 0.6266723476778867,
+      "grad_norm": 18.75,
+      "learning_rate": 1.951311169441399e-06,
+      "loss": 0.9798,
+      "step": 7354
+    },
+    {
+      "epoch": 0.6267575628461867,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9511720661941757e-06,
+      "loss": 0.8901,
+      "step": 7355
+    },
+    {
+      "epoch": 0.6268427780144866,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.9510329502760687e-06,
+      "loss": 0.5513,
+      "step": 7356
+    },
+    {
+      "epoch": 0.6269279931827866,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9508938216895913e-06,
+      "loss": 0.643,
+      "step": 7357
+    },
+    {
+      "epoch": 0.6270132083510865,
+      "grad_norm": 11.25,
+      "learning_rate": 1.9507546804372574e-06,
+      "loss": 0.6086,
+      "step": 7358
+    },
+    {
+      "epoch": 0.6270984235193865,
+      "grad_norm": 16.25,
+      "learning_rate": 1.950615526521582e-06,
+      "loss": 1.0132,
+      "step": 7359
+    },
+    {
+      "epoch": 0.6271836386876865,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9504763599450792e-06,
+      "loss": 0.6273,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6272688538559864,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9503371807102646e-06,
+      "loss": 0.5556,
+      "step": 7361
+    },
+    {
+      "epoch": 0.6273540690242864,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.9501979888196533e-06,
+      "loss": 0.6749,
+      "step": 7362
+    },
+    {
+      "epoch": 0.6274392841925863,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9500587842757594e-06,
+      "loss": 0.6568,
+      "step": 7363
+    },
+    {
+      "epoch": 0.6275244993608863,
+      "grad_norm": 11.375,
+      "learning_rate": 1.9499195670810993e-06,
+      "loss": 0.608,
+      "step": 7364
+    },
+    {
+      "epoch": 0.6276097145291862,
+      "grad_norm": 18.25,
+      "learning_rate": 1.949780337238189e-06,
+      "loss": 0.8802,
+      "step": 7365
+    },
+    {
+      "epoch": 0.6276949296974862,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9496410947495447e-06,
+      "loss": 0.5076,
+      "step": 7366
+    },
+    {
+      "epoch": 0.6277801448657861,
+      "grad_norm": 13.125,
+      "learning_rate": 1.949501839617682e-06,
+      "loss": 0.6984,
+      "step": 7367
+    },
+    {
+      "epoch": 0.6278653600340861,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9493625718451176e-06,
+      "loss": 0.9003,
+      "step": 7368
+    },
+    {
+      "epoch": 0.6279505752023861,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9492232914343685e-06,
+      "loss": 0.6419,
+      "step": 7369
+    },
+    {
+      "epoch": 0.628035790370686,
+      "grad_norm": 13.375,
+      "learning_rate": 1.949083998387951e-06,
+      "loss": 0.5162,
+      "step": 7370
+    },
+    {
+      "epoch": 0.628121005538986,
+      "grad_norm": 11.875,
+      "learning_rate": 1.948944692708383e-06,
+      "loss": 0.4872,
+      "step": 7371
+    },
+    {
+      "epoch": 0.6282062207072859,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9488053743981814e-06,
+      "loss": 0.8946,
+      "step": 7372
+    },
+    {
+      "epoch": 0.6282914358755859,
+      "grad_norm": 23.25,
+      "learning_rate": 1.9486660434598638e-06,
+      "loss": 1.0278,
+      "step": 7373
+    },
+    {
+      "epoch": 0.6283766510438858,
+      "grad_norm": 16.125,
+      "learning_rate": 1.948526699895949e-06,
+      "loss": 0.929,
+      "step": 7374
+    },
+    {
+      "epoch": 0.6284618662121858,
+      "grad_norm": 22.375,
+      "learning_rate": 1.9483873437089536e-06,
+      "loss": 0.8232,
+      "step": 7375
+    },
+    {
+      "epoch": 0.6285470813804858,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.948247974901397e-06,
+      "loss": 0.815,
+      "step": 7376
+    },
+    {
+      "epoch": 0.6286322965487857,
+      "grad_norm": 12.0,
+      "learning_rate": 1.948108593475797e-06,
+      "loss": 0.3755,
+      "step": 7377
+    },
+    {
+      "epoch": 0.6287175117170857,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9479691994346727e-06,
+      "loss": 0.6934,
+      "step": 7378
+    },
+    {
+      "epoch": 0.6288027268853856,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9478297927805433e-06,
+      "loss": 0.7856,
+      "step": 7379
+    },
+    {
+      "epoch": 0.6288879420536856,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.947690373515928e-06,
+      "loss": 0.7932,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6289731572219855,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9475509416433457e-06,
+      "loss": 0.7672,
+      "step": 7381
+    },
+    {
+      "epoch": 0.6290583723902855,
+      "grad_norm": 10.5,
+      "learning_rate": 1.947411497165316e-06,
+      "loss": 0.3226,
+      "step": 7382
+    },
+    {
+      "epoch": 0.6291435875585855,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.9472720400843604e-06,
+      "loss": 0.5668,
+      "step": 7383
+    },
+    {
+      "epoch": 0.6292288027268854,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9471325704029973e-06,
+      "loss": 0.6698,
+      "step": 7384
+    },
+    {
+      "epoch": 0.6293140178951854,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9469930881237474e-06,
+      "loss": 0.929,
+      "step": 7385
+    },
+    {
+      "epoch": 0.6293992330634853,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.946853593249132e-06,
+      "loss": 0.3636,
+      "step": 7386
+    },
+    {
+      "epoch": 0.6294844482317853,
+      "grad_norm": 16.5,
+      "learning_rate": 1.946714085781671e-06,
+      "loss": 0.7777,
+      "step": 7387
+    },
+    {
+      "epoch": 0.6295696634000852,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.946574565723886e-06,
+      "loss": 1.119,
+      "step": 7388
+    },
+    {
+      "epoch": 0.6296548785683852,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9464350330782985e-06,
+      "loss": 0.7391,
+      "step": 7389
+    },
+    {
+      "epoch": 0.6297400937366852,
+      "grad_norm": 19.25,
+      "learning_rate": 1.9462954878474295e-06,
+      "loss": 0.9352,
+      "step": 7390
+    },
+    {
+      "epoch": 0.6298253089049851,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.9461559300338005e-06,
+      "loss": 0.6283,
+      "step": 7391
+    },
+    {
+      "epoch": 0.6299105240732851,
+      "grad_norm": 15.75,
+      "learning_rate": 1.946016359639934e-06,
+      "loss": 0.4493,
+      "step": 7392
+    },
+    {
+      "epoch": 0.629995739241585,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9458767766683517e-06,
+      "loss": 0.5668,
+      "step": 7393
+    },
+    {
+      "epoch": 0.630080954409885,
+      "grad_norm": 14.25,
+      "learning_rate": 1.945737181121577e-06,
+      "loss": 0.4691,
+      "step": 7394
+    },
+    {
+      "epoch": 0.6301661695781849,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9455975730021308e-06,
+      "loss": 0.7759,
+      "step": 7395
+    },
+    {
+      "epoch": 0.6302513847464849,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9454579523125376e-06,
+      "loss": 0.7293,
+      "step": 7396
+    },
+    {
+      "epoch": 0.6303365999147849,
+      "grad_norm": 15.125,
+      "learning_rate": 1.94531831905532e-06,
+      "loss": 0.926,
+      "step": 7397
+    },
+    {
+      "epoch": 0.6304218150830848,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9451786732330006e-06,
+      "loss": 0.8237,
+      "step": 7398
+    },
+    {
+      "epoch": 0.6305070302513848,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9450390148481043e-06,
+      "loss": 1.0485,
+      "step": 7399
+    },
+    {
+      "epoch": 0.6305922454196847,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.944899343903154e-06,
+      "loss": 0.6306,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6306774605879847,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9447596604006735e-06,
+      "loss": 0.8967,
+      "step": 7401
+    },
+    {
+      "epoch": 0.6307626757562846,
+      "grad_norm": 14.375,
+      "learning_rate": 1.944619964343187e-06,
+      "loss": 0.561,
+      "step": 7402
+    },
+    {
+      "epoch": 0.6308478909245846,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9444802557332195e-06,
+      "loss": 0.6972,
+      "step": 7403
+    },
+    {
+      "epoch": 0.6309331060928846,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9443405345732957e-06,
+      "loss": 0.4426,
+      "step": 7404
+    },
+    {
+      "epoch": 0.6310183212611845,
+      "grad_norm": 22.375,
+      "learning_rate": 1.9442008008659395e-06,
+      "loss": 0.7587,
+      "step": 7405
+    },
+    {
+      "epoch": 0.6311035364294845,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.944061054613677e-06,
+      "loss": 0.7279,
+      "step": 7406
+    },
+    {
+      "epoch": 0.6311887515977844,
+      "grad_norm": 15.875,
+      "learning_rate": 1.943921295819034e-06,
+      "loss": 0.5185,
+      "step": 7407
+    },
+    {
+      "epoch": 0.6312739667660844,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.943781524484535e-06,
+      "loss": 0.7107,
+      "step": 7408
+    },
+    {
+      "epoch": 0.6313591819343843,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.943641740612706e-06,
+      "loss": 0.6492,
+      "step": 7409
+    },
+    {
+      "epoch": 0.6314443971026843,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9435019442060733e-06,
+      "loss": 0.5004,
+      "step": 7410
+    },
+    {
+      "epoch": 0.6315296122709843,
+      "grad_norm": 22.125,
+      "learning_rate": 1.9433621352671634e-06,
+      "loss": 1.1887,
+      "step": 7411
+    },
+    {
+      "epoch": 0.6316148274392842,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9432223137985024e-06,
+      "loss": 0.4793,
+      "step": 7412
+    },
+    {
+      "epoch": 0.6317000426075842,
+      "grad_norm": 10.0,
+      "learning_rate": 1.943082479802617e-06,
+      "loss": 0.4498,
+      "step": 7413
+    },
+    {
+      "epoch": 0.6317852577758841,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.942942633282034e-06,
+      "loss": 0.8601,
+      "step": 7414
+    },
+    {
+      "epoch": 0.6318704729441841,
+      "grad_norm": 12.75,
+      "learning_rate": 1.9428027742392813e-06,
+      "loss": 0.4812,
+      "step": 7415
+    },
+    {
+      "epoch": 0.631955688112484,
+      "grad_norm": 13.25,
+      "learning_rate": 1.942662902676885e-06,
+      "loss": 0.7221,
+      "step": 7416
+    },
+    {
+      "epoch": 0.632040903280784,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.942523018597374e-06,
+      "loss": 0.5319,
+      "step": 7417
+    },
+    {
+      "epoch": 0.632126118449084,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.942383122003276e-06,
+      "loss": 0.6062,
+      "step": 7418
+    },
+    {
+      "epoch": 0.6322113336173839,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.9422432128971187e-06,
+      "loss": 0.6049,
+      "step": 7419
+    },
+    {
+      "epoch": 0.6322965487856839,
+      "grad_norm": 13.75,
+      "learning_rate": 1.94210329128143e-06,
+      "loss": 0.8464,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6323817639539838,
+      "grad_norm": 12.0,
+      "learning_rate": 1.9419633571587392e-06,
+      "loss": 0.4576,
+      "step": 7421
+    },
+    {
+      "epoch": 0.6324669791222838,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9418234105315744e-06,
+      "loss": 0.7391,
+      "step": 7422
+    },
+    {
+      "epoch": 0.6325521942905837,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9416834514024654e-06,
+      "loss": 0.5661,
+      "step": 7423
+    },
+    {
+      "epoch": 0.6326374094588837,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.94154347977394e-06,
+      "loss": 0.706,
+      "step": 7424
+    },
+    {
+      "epoch": 0.6327226246271836,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9414034956485293e-06,
+      "loss": 0.8299,
+      "step": 7425
+    },
+    {
+      "epoch": 0.6328078397954836,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9412634990287616e-06,
+      "loss": 0.9554,
+      "step": 7426
+    },
+    {
+      "epoch": 0.6328930549637836,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.9411234899171685e-06,
+      "loss": 0.7038,
+      "step": 7427
+    },
+    {
+      "epoch": 0.6329782701320835,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9409834683162777e-06,
+      "loss": 0.8385,
+      "step": 7428
+    },
+    {
+      "epoch": 0.6330634853003835,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9408434342286213e-06,
+      "loss": 0.9287,
+      "step": 7429
+    },
+    {
+      "epoch": 0.6331487004686834,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9407033876567292e-06,
+      "loss": 0.7782,
+      "step": 7430
+    },
+    {
+      "epoch": 0.6332339156369834,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9405633286031324e-06,
+      "loss": 0.566,
+      "step": 7431
+    },
+    {
+      "epoch": 0.6333191308052833,
+      "grad_norm": 12.125,
+      "learning_rate": 1.940423257070362e-06,
+      "loss": 0.53,
+      "step": 7432
+    },
+    {
+      "epoch": 0.6334043459735833,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.940283173060949e-06,
+      "loss": 0.8393,
+      "step": 7433
+    },
+    {
+      "epoch": 0.6334895611418833,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9401430765774247e-06,
+      "loss": 0.957,
+      "step": 7434
+    },
+    {
+      "epoch": 0.6335747763101832,
+      "grad_norm": 13.5,
+      "learning_rate": 1.940002967622321e-06,
+      "loss": 0.6762,
+      "step": 7435
+    },
+    {
+      "epoch": 0.6336599914784832,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.93986284619817e-06,
+      "loss": 0.965,
+      "step": 7436
+    },
+    {
+      "epoch": 0.6337452066467831,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9397227123075035e-06,
+      "loss": 0.8258,
+      "step": 7437
+    },
+    {
+      "epoch": 0.6338304218150831,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.9395825659528543e-06,
+      "loss": 0.5571,
+      "step": 7438
+    },
+    {
+      "epoch": 0.633915636983383,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.9394424071367545e-06,
+      "loss": 0.4198,
+      "step": 7439
+    },
+    {
+      "epoch": 0.634000852151683,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9393022358617374e-06,
+      "loss": 0.6657,
+      "step": 7440
+    },
+    {
+      "epoch": 0.634086067319983,
+      "grad_norm": 18.25,
+      "learning_rate": 1.9391620521303355e-06,
+      "loss": 0.9386,
+      "step": 7441
+    },
+    {
+      "epoch": 0.6341712824882829,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9390218559450823e-06,
+      "loss": 0.3725,
+      "step": 7442
+    },
+    {
+      "epoch": 0.6342564976565829,
+      "grad_norm": 31.5,
+      "learning_rate": 1.9388816473085116e-06,
+      "loss": 1.049,
+      "step": 7443
+    },
+    {
+      "epoch": 0.6343417128248828,
+      "grad_norm": 11.125,
+      "learning_rate": 1.9387414262231562e-06,
+      "loss": 0.4952,
+      "step": 7444
+    },
+    {
+      "epoch": 0.6344269279931828,
+      "grad_norm": 17.625,
+      "learning_rate": 1.938601192691551e-06,
+      "loss": 0.9978,
+      "step": 7445
+    },
+    {
+      "epoch": 0.6345121431614827,
+      "grad_norm": 16.0,
+      "learning_rate": 1.93846094671623e-06,
+      "loss": 0.7123,
+      "step": 7446
+    },
+    {
+      "epoch": 0.6345973583297827,
+      "grad_norm": 20.75,
+      "learning_rate": 1.9383206882997275e-06,
+      "loss": 1.2092,
+      "step": 7447
+    },
+    {
+      "epoch": 0.6346825734980827,
+      "grad_norm": 11.625,
+      "learning_rate": 1.9381804174445775e-06,
+      "loss": 0.5536,
+      "step": 7448
+    },
+    {
+      "epoch": 0.6347677886663826,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.9380401341533155e-06,
+      "loss": 0.5701,
+      "step": 7449
+    },
+    {
+      "epoch": 0.6348530038346826,
+      "grad_norm": 21.5,
+      "learning_rate": 1.9378998384284764e-06,
+      "loss": 0.8794,
+      "step": 7450
+    },
+    {
+      "epoch": 0.6349382190029825,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9377595302725954e-06,
+      "loss": 0.8057,
+      "step": 7451
+    },
+    {
+      "epoch": 0.6350234341712825,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9376192096882083e-06,
+      "loss": 0.8704,
+      "step": 7452
+    },
+    {
+      "epoch": 0.6351086493395824,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9374788766778505e-06,
+      "loss": 0.8706,
+      "step": 7453
+    },
+    {
+      "epoch": 0.6351938645078824,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9373385312440583e-06,
+      "loss": 0.9047,
+      "step": 7454
+    },
+    {
+      "epoch": 0.6352790796761824,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9371981733893674e-06,
+      "loss": 0.9157,
+      "step": 7455
+    },
+    {
+      "epoch": 0.6353642948444823,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9370578031163145e-06,
+      "loss": 0.6883,
+      "step": 7456
+    },
+    {
+      "epoch": 0.6354495100127823,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9369174204274362e-06,
+      "loss": 1.1392,
+      "step": 7457
+    },
+    {
+      "epoch": 0.6355347251810822,
+      "grad_norm": 10.375,
+      "learning_rate": 1.9367770253252694e-06,
+      "loss": 0.4064,
+      "step": 7458
+    },
+    {
+      "epoch": 0.6356199403493822,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.936636617812351e-06,
+      "loss": 0.5448,
+      "step": 7459
+    },
+    {
+      "epoch": 0.6357051555176821,
+      "grad_norm": 18.5,
+      "learning_rate": 1.936496197891219e-06,
+      "loss": 0.9048,
+      "step": 7460
+    },
+    {
+      "epoch": 0.6357903706859821,
+      "grad_norm": 14.625,
+      "learning_rate": 1.93635576556441e-06,
+      "loss": 0.9042,
+      "step": 7461
+    },
+    {
+      "epoch": 0.635875585854282,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.936215320834462e-06,
+      "loss": 0.6657,
+      "step": 7462
+    },
+    {
+      "epoch": 0.635960801022582,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9360748637039136e-06,
+      "loss": 0.5172,
+      "step": 7463
+    },
+    {
+      "epoch": 0.636046016190882,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9359343941753024e-06,
+      "loss": 0.629,
+      "step": 7464
+    },
+    {
+      "epoch": 0.6361312313591819,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.9357939122511666e-06,
+      "loss": 0.7695,
+      "step": 7465
+    },
+    {
+      "epoch": 0.6362164465274819,
+      "grad_norm": 19.875,
+      "learning_rate": 1.935653417934045e-06,
+      "loss": 0.6998,
+      "step": 7466
+    },
+    {
+      "epoch": 0.6363016616957818,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.935512911226477e-06,
+      "loss": 0.5618,
+      "step": 7467
+    },
+    {
+      "epoch": 0.6363868768640818,
+      "grad_norm": 22.25,
+      "learning_rate": 1.935372392131001e-06,
+      "loss": 0.9994,
+      "step": 7468
+    },
+    {
+      "epoch": 0.6364720920323818,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.9352318606501576e-06,
+      "loss": 0.5832,
+      "step": 7469
+    },
+    {
+      "epoch": 0.6365573072006817,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9350913167864842e-06,
+      "loss": 1.0666,
+      "step": 7470
+    },
+    {
+      "epoch": 0.6366425223689817,
+      "grad_norm": 19.875,
+      "learning_rate": 1.9349507605425227e-06,
+      "loss": 1.0364,
+      "step": 7471
+    },
+    {
+      "epoch": 0.6367277375372816,
+      "grad_norm": 14.625,
+      "learning_rate": 1.934810191920812e-06,
+      "loss": 0.9384,
+      "step": 7472
+    },
+    {
+      "epoch": 0.6368129527055816,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9346696109238923e-06,
+      "loss": 0.7831,
+      "step": 7473
+    },
+    {
+      "epoch": 0.6368981678738815,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9345290175543043e-06,
+      "loss": 0.8197,
+      "step": 7474
+    },
+    {
+      "epoch": 0.6369833830421815,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9343884118145883e-06,
+      "loss": 0.4731,
+      "step": 7475
+    },
+    {
+      "epoch": 0.6370685982104815,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.934247793707286e-06,
+      "loss": 0.6042,
+      "step": 7476
+    },
+    {
+      "epoch": 0.6371538133787814,
+      "grad_norm": 22.375,
+      "learning_rate": 1.934107163234938e-06,
+      "loss": 1.2094,
+      "step": 7477
+    },
+    {
+      "epoch": 0.6372390285470814,
+      "grad_norm": 11.625,
+      "learning_rate": 1.9339665204000856e-06,
+      "loss": 0.5361,
+      "step": 7478
+    },
+    {
+      "epoch": 0.6373242437153813,
+      "grad_norm": 13.75,
+      "learning_rate": 1.93382586520527e-06,
+      "loss": 0.7193,
+      "step": 7479
+    },
+    {
+      "epoch": 0.6374094588836813,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9336851976530342e-06,
+      "loss": 1.0151,
+      "step": 7480
+    },
+    {
+      "epoch": 0.6374946740519812,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.933544517745918e-06,
+      "loss": 0.7259,
+      "step": 7481
+    },
+    {
+      "epoch": 0.6375798892202812,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.933403825486466e-06,
+      "loss": 0.2779,
+      "step": 7482
+    },
+    {
+      "epoch": 0.6376651043885811,
+      "grad_norm": 19.0,
+      "learning_rate": 1.9332631208772197e-06,
+      "loss": 0.8455,
+      "step": 7483
+    },
+    {
+      "epoch": 0.6377503195568811,
+      "grad_norm": 13.5,
+      "learning_rate": 1.9331224039207217e-06,
+      "loss": 0.6364,
+      "step": 7484
+    },
+    {
+      "epoch": 0.6378355347251811,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9329816746195146e-06,
+      "loss": 0.7591,
+      "step": 7485
+    },
+    {
+      "epoch": 0.637920749893481,
+      "grad_norm": 12.25,
+      "learning_rate": 1.932840932976142e-06,
+      "loss": 0.6894,
+      "step": 7486
+    },
+    {
+      "epoch": 0.638005965061781,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.932700178993147e-06,
+      "loss": 0.5843,
+      "step": 7487
+    },
+    {
+      "epoch": 0.6380911802300809,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9325594126730736e-06,
+      "loss": 0.5927,
+      "step": 7488
+    },
+    {
+      "epoch": 0.6381763953983809,
+      "grad_norm": 17.0,
+      "learning_rate": 1.932418634018465e-06,
+      "loss": 0.5767,
+      "step": 7489
+    },
+    {
+      "epoch": 0.6382616105666808,
+      "grad_norm": 11.75,
+      "learning_rate": 1.932277843031865e-06,
+      "loss": 0.4747,
+      "step": 7490
+    },
+    {
+      "epoch": 0.6383468257349808,
+      "grad_norm": 15.375,
+      "learning_rate": 1.932137039715819e-06,
+      "loss": 0.6273,
+      "step": 7491
+    },
+    {
+      "epoch": 0.6384320409032808,
+      "grad_norm": 22.25,
+      "learning_rate": 1.931996224072871e-06,
+      "loss": 0.851,
+      "step": 7492
+    },
+    {
+      "epoch": 0.6385172560715807,
+      "grad_norm": 11.25,
+      "learning_rate": 1.9318553961055645e-06,
+      "loss": 0.5278,
+      "step": 7493
+    },
+    {
+      "epoch": 0.6386024712398807,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9317145558164458e-06,
+      "loss": 0.989,
+      "step": 7494
+    },
+    {
+      "epoch": 0.6386876864081806,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.9315737032080595e-06,
+      "loss": 0.7262,
+      "step": 7495
+    },
+    {
+      "epoch": 0.6387729015764806,
+      "grad_norm": 18.5,
+      "learning_rate": 1.931432838282951e-06,
+      "loss": 1.0054,
+      "step": 7496
+    },
+    {
+      "epoch": 0.6388581167447805,
+      "grad_norm": 17.875,
+      "learning_rate": 1.931291961043666e-06,
+      "loss": 0.7548,
+      "step": 7497
+    },
+    {
+      "epoch": 0.6389433319130805,
+      "grad_norm": 20.0,
+      "learning_rate": 1.93115107149275e-06,
+      "loss": 0.9563,
+      "step": 7498
+    },
+    {
+      "epoch": 0.6390285470813805,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9310101696327494e-06,
+      "loss": 0.7491,
+      "step": 7499
+    },
+    {
+      "epoch": 0.6391137622496804,
+      "grad_norm": 22.25,
+      "learning_rate": 1.93086925546621e-06,
+      "loss": 0.6004,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6391989774179804,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9307283289956786e-06,
+      "loss": 0.5954,
+      "step": 7501
+    },
+    {
+      "epoch": 0.6392841925862803,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9305873902237014e-06,
+      "loss": 0.3651,
+      "step": 7502
+    },
+    {
+      "epoch": 0.6393694077545803,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.930446439152826e-06,
+      "loss": 0.6012,
+      "step": 7503
+    },
+    {
+      "epoch": 0.6394546229228802,
+      "grad_norm": 22.375,
+      "learning_rate": 1.9303054757855994e-06,
+      "loss": 0.9328,
+      "step": 7504
+    },
+    {
+      "epoch": 0.6395398380911802,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9301645001245686e-06,
+      "loss": 0.5892,
+      "step": 7505
+    },
+    {
+      "epoch": 0.6396250532594802,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9300235121722813e-06,
+      "loss": 0.7726,
+      "step": 7506
+    },
+    {
+      "epoch": 0.6397102684277801,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9298825119312853e-06,
+      "loss": 0.8592,
+      "step": 7507
+    },
+    {
+      "epoch": 0.6397954835960801,
+      "grad_norm": 17.625,
+      "learning_rate": 1.929741499404129e-06,
+      "loss": 0.5337,
+      "step": 7508
+    },
+    {
+      "epoch": 0.63988069876438,
+      "grad_norm": 12.75,
+      "learning_rate": 1.9296004745933596e-06,
+      "loss": 0.4694,
+      "step": 7509
+    },
+    {
+      "epoch": 0.63996591393268,
+      "grad_norm": 31.875,
+      "learning_rate": 1.9294594375015266e-06,
+      "loss": 1.1662,
+      "step": 7510
+    },
+    {
+      "epoch": 0.6400511291009799,
+      "grad_norm": 29.125,
+      "learning_rate": 1.9293183881311784e-06,
+      "loss": 1.1891,
+      "step": 7511
+    },
+    {
+      "epoch": 0.6401363442692799,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.929177326484864e-06,
+      "loss": 0.5254,
+      "step": 7512
+    },
+    {
+      "epoch": 0.6402215594375799,
+      "grad_norm": 12.125,
+      "learning_rate": 1.9290362525651318e-06,
+      "loss": 0.9344,
+      "step": 7513
+    },
+    {
+      "epoch": 0.6403067746058798,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9288951663745316e-06,
+      "loss": 0.7495,
+      "step": 7514
+    },
+    {
+      "epoch": 0.6403919897741798,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9287540679156134e-06,
+      "loss": 0.5234,
+      "step": 7515
+    },
+    {
+      "epoch": 0.6404772049424797,
+      "grad_norm": 12.125,
+      "learning_rate": 1.9286129571909267e-06,
+      "loss": 0.7819,
+      "step": 7516
+    },
+    {
+      "epoch": 0.6405624201107797,
+      "grad_norm": 13.5,
+      "learning_rate": 1.928471834203022e-06,
+      "loss": 0.424,
+      "step": 7517
+    },
+    {
+      "epoch": 0.6406476352790796,
+      "grad_norm": 24.125,
+      "learning_rate": 1.928330698954448e-06,
+      "loss": 0.8901,
+      "step": 7518
+    },
+    {
+      "epoch": 0.6407328504473796,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9281895514477568e-06,
+      "loss": 0.6623,
+      "step": 7519
+    },
+    {
+      "epoch": 0.6408180656156796,
+      "grad_norm": 17.375,
+      "learning_rate": 1.928048391685498e-06,
+      "loss": 0.6539,
+      "step": 7520
+    },
+    {
+      "epoch": 0.6409032807839795,
+      "grad_norm": 19.0,
+      "learning_rate": 1.927907219670223e-06,
+      "loss": 0.9933,
+      "step": 7521
+    },
+    {
+      "epoch": 0.6409884959522795,
+      "grad_norm": 16.625,
+      "learning_rate": 1.927766035404483e-06,
+      "loss": 0.7251,
+      "step": 7522
+    },
+    {
+      "epoch": 0.6410737111205794,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9276248388908293e-06,
+      "loss": 0.6019,
+      "step": 7523
+    },
+    {
+      "epoch": 0.6411589262888794,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9274836301318135e-06,
+      "loss": 0.5628,
+      "step": 7524
+    },
+    {
+      "epoch": 0.6412441414571793,
+      "grad_norm": 11.25,
+      "learning_rate": 1.9273424091299867e-06,
+      "loss": 0.5809,
+      "step": 7525
+    },
+    {
+      "epoch": 0.6413293566254793,
+      "grad_norm": 21.25,
+      "learning_rate": 1.9272011758879015e-06,
+      "loss": 0.9849,
+      "step": 7526
+    },
+    {
+      "epoch": 0.6414145717937793,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9270599304081105e-06,
+      "loss": 0.5735,
+      "step": 7527
+    },
+    {
+      "epoch": 0.6414997869620792,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9269186726931653e-06,
+      "loss": 0.5916,
+      "step": 7528
+    },
+    {
+      "epoch": 0.6415850021303792,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9267774027456193e-06,
+      "loss": 0.8224,
+      "step": 7529
+    },
+    {
+      "epoch": 0.6416702172986791,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.926636120568025e-06,
+      "loss": 0.6435,
+      "step": 7530
+    },
+    {
+      "epoch": 0.6417554324669791,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.926494826162936e-06,
+      "loss": 0.5791,
+      "step": 7531
+    },
+    {
+      "epoch": 0.641840647635279,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9263535195329044e-06,
+      "loss": 0.741,
+      "step": 7532
+    },
+    {
+      "epoch": 0.641925862803579,
+      "grad_norm": 16.75,
+      "learning_rate": 1.926212200680485e-06,
+      "loss": 1.0166,
+      "step": 7533
+    },
+    {
+      "epoch": 0.642011077971879,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9260708696082312e-06,
+      "loss": 0.9037,
+      "step": 7534
+    },
+    {
+      "epoch": 0.6420962931401789,
+      "grad_norm": 18.5,
+      "learning_rate": 1.925929526318697e-06,
+      "loss": 0.8912,
+      "step": 7535
+    },
+    {
+      "epoch": 0.6421815083084789,
+      "grad_norm": 17.75,
+      "learning_rate": 1.925788170814436e-06,
+      "loss": 1.0491,
+      "step": 7536
+    },
+    {
+      "epoch": 0.6422667234767788,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9256468030980037e-06,
+      "loss": 0.7845,
+      "step": 7537
+    },
+    {
+      "epoch": 0.6423519386450788,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9255054231719544e-06,
+      "loss": 0.9729,
+      "step": 7538
+    },
+    {
+      "epoch": 0.6424371538133787,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9253640310388427e-06,
+      "loss": 1.1976,
+      "step": 7539
+    },
+    {
+      "epoch": 0.6425223689816787,
+      "grad_norm": 21.5,
+      "learning_rate": 1.9252226267012238e-06,
+      "loss": 0.7589,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6426075841499787,
+      "grad_norm": 22.25,
+      "learning_rate": 1.9250812101616527e-06,
+      "loss": 0.6298,
+      "step": 7541
+    },
+    {
+      "epoch": 0.6426927993182786,
+      "grad_norm": 30.625,
+      "learning_rate": 1.9249397814226857e-06,
+      "loss": 1.174,
+      "step": 7542
+    },
+    {
+      "epoch": 0.6427780144865786,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.924798340486878e-06,
+      "loss": 0.4541,
+      "step": 7543
+    },
+    {
+      "epoch": 0.6428632296548785,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9246568873567855e-06,
+      "loss": 0.4666,
+      "step": 7544
+    },
+    {
+      "epoch": 0.6429484448231785,
+      "grad_norm": 16.625,
+      "learning_rate": 1.924515422034965e-06,
+      "loss": 0.871,
+      "step": 7545
+    },
+    {
+      "epoch": 0.6430336599914784,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9243739445239725e-06,
+      "loss": 0.6814,
+      "step": 7546
+    },
+    {
+      "epoch": 0.6431188751597784,
+      "grad_norm": 13.5,
+      "learning_rate": 1.924232454826365e-06,
+      "loss": 0.7081,
+      "step": 7547
+    },
+    {
+      "epoch": 0.6432040903280783,
+      "grad_norm": 19.125,
+      "learning_rate": 1.924090952944698e-06,
+      "loss": 1.1679,
+      "step": 7548
+    },
+    {
+      "epoch": 0.6432893054963783,
+      "grad_norm": 20.375,
+      "learning_rate": 1.9239494388815304e-06,
+      "loss": 0.8666,
+      "step": 7549
+    },
+    {
+      "epoch": 0.6433745206646783,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.923807912639419e-06,
+      "loss": 0.5498,
+      "step": 7550
+    },
+    {
+      "epoch": 0.6434597358329782,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9236663742209203e-06,
+      "loss": 0.8145,
+      "step": 7551
+    },
+    {
+      "epoch": 0.6435449510012782,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.923524823628593e-06,
+      "loss": 0.4805,
+      "step": 7552
+    },
+    {
+      "epoch": 0.6436301661695781,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9233832608649955e-06,
+      "loss": 0.7224,
+      "step": 7553
+    },
+    {
+      "epoch": 0.6437153813378781,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9232416859326845e-06,
+      "loss": 0.7638,
+      "step": 7554
+    },
+    {
+      "epoch": 0.643800596506178,
+      "grad_norm": 15.125,
+      "learning_rate": 1.92310009883422e-06,
+      "loss": 0.7241,
+      "step": 7555
+    },
+    {
+      "epoch": 0.643885811674478,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.922958499572159e-06,
+      "loss": 0.5145,
+      "step": 7556
+    },
+    {
+      "epoch": 0.643971026842778,
+      "grad_norm": 39.5,
+      "learning_rate": 1.9228168881490616e-06,
+      "loss": 0.9358,
+      "step": 7557
+    },
+    {
+      "epoch": 0.6440562420110779,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.922675264567487e-06,
+      "loss": 0.7633,
+      "step": 7558
+    },
+    {
+      "epoch": 0.6441414571793779,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.922533628829993e-06,
+      "loss": 0.9419,
+      "step": 7559
+    },
+    {
+      "epoch": 0.6442266723476778,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9223919809391407e-06,
+      "loss": 0.7895,
+      "step": 7560
+    },
+    {
+      "epoch": 0.6443118875159779,
+      "grad_norm": 14.125,
+      "learning_rate": 1.922250320897489e-06,
+      "loss": 0.7973,
+      "step": 7561
+    },
+    {
+      "epoch": 0.6443971026842779,
+      "grad_norm": 12.75,
+      "learning_rate": 1.922108648707598e-06,
+      "loss": 0.7334,
+      "step": 7562
+    },
+    {
+      "epoch": 0.6444823178525778,
+      "grad_norm": 13.0,
+      "learning_rate": 1.9219669643720275e-06,
+      "loss": 0.5542,
+      "step": 7563
+    },
+    {
+      "epoch": 0.6445675330208778,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.921825267893339e-06,
+      "loss": 0.6766,
+      "step": 7564
+    },
+    {
+      "epoch": 0.6446527481891777,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9216835592740917e-06,
+      "loss": 0.7353,
+      "step": 7565
+    },
+    {
+      "epoch": 0.6447379633574777,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9215418385168473e-06,
+      "loss": 1.0316,
+      "step": 7566
+    },
+    {
+      "epoch": 0.6448231785257776,
+      "grad_norm": 15.125,
+      "learning_rate": 1.9214001056241665e-06,
+      "loss": 0.5764,
+      "step": 7567
+    },
+    {
+      "epoch": 0.6449083936940776,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9212583605986103e-06,
+      "loss": 0.8525,
+      "step": 7568
+    },
+    {
+      "epoch": 0.6449936088623776,
+      "grad_norm": 11.125,
+      "learning_rate": 1.9211166034427415e-06,
+      "loss": 0.3902,
+      "step": 7569
+    },
+    {
+      "epoch": 0.6450788240306775,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.92097483415912e-06,
+      "loss": 0.6017,
+      "step": 7570
+    },
+    {
+      "epoch": 0.6451640391989775,
+      "grad_norm": 17.75,
+      "learning_rate": 1.920833052750309e-06,
+      "loss": 1.0817,
+      "step": 7571
+    },
+    {
+      "epoch": 0.6452492543672774,
+      "grad_norm": 18.25,
+      "learning_rate": 1.9206912592188693e-06,
+      "loss": 0.5975,
+      "step": 7572
+    },
+    {
+      "epoch": 0.6453344695355774,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9205494535673653e-06,
+      "loss": 0.5062,
+      "step": 7573
+    },
+    {
+      "epoch": 0.6454196847038773,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9204076357983575e-06,
+      "loss": 0.51,
+      "step": 7574
+    },
+    {
+      "epoch": 0.6455048998721773,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.9202658059144098e-06,
+      "loss": 0.3104,
+      "step": 7575
+    },
+    {
+      "epoch": 0.6455901150404773,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9201239639180853e-06,
+      "loss": 0.5224,
+      "step": 7576
+    },
+    {
+      "epoch": 0.6456753302087772,
+      "grad_norm": 21.75,
+      "learning_rate": 1.9199821098119468e-06,
+      "loss": 0.8429,
+      "step": 7577
+    },
+    {
+      "epoch": 0.6457605453770772,
+      "grad_norm": 20.875,
+      "learning_rate": 1.9198402435985573e-06,
+      "loss": 0.6916,
+      "step": 7578
+    },
+    {
+      "epoch": 0.6458457605453771,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9196983652804817e-06,
+      "loss": 0.5824,
+      "step": 7579
+    },
+    {
+      "epoch": 0.6459309757136771,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9195564748602828e-06,
+      "loss": 0.6706,
+      "step": 7580
+    },
+    {
+      "epoch": 0.646016190881977,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9194145723405256e-06,
+      "loss": 0.9729,
+      "step": 7581
+    },
+    {
+      "epoch": 0.646101406050277,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9192726577237738e-06,
+      "loss": 1.0382,
+      "step": 7582
+    },
+    {
+      "epoch": 0.646186621218577,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9191307310125913e-06,
+      "loss": 1.0176,
+      "step": 7583
+    },
+    {
+      "epoch": 0.6462718363868769,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9189887922095443e-06,
+      "loss": 0.7483,
+      "step": 7584
+    },
+    {
+      "epoch": 0.6463570515551769,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.918846841317197e-06,
+      "loss": 0.4302,
+      "step": 7585
+    },
+    {
+      "epoch": 0.6464422667234768,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9187048783381144e-06,
+      "loss": 0.3955,
+      "step": 7586
+    },
+    {
+      "epoch": 0.6465274818917768,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9185629032748626e-06,
+      "loss": 0.904,
+      "step": 7587
+    },
+    {
+      "epoch": 0.6466126970600767,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.918420916130007e-06,
+      "loss": 0.6171,
+      "step": 7588
+    },
+    {
+      "epoch": 0.6466979122283767,
+      "grad_norm": 17.75,
+      "learning_rate": 1.918278916906113e-06,
+      "loss": 1.0285,
+      "step": 7589
+    },
+    {
+      "epoch": 0.6467831273966766,
+      "grad_norm": 16.25,
+      "learning_rate": 1.9181369056057474e-06,
+      "loss": 0.7404,
+      "step": 7590
+    },
+    {
+      "epoch": 0.6468683425649766,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9179948822314754e-06,
+      "loss": 0.4816,
+      "step": 7591
+    },
+    {
+      "epoch": 0.6469535577332766,
+      "grad_norm": 11.75,
+      "learning_rate": 1.917852846785864e-06,
+      "loss": 0.5135,
+      "step": 7592
+    },
+    {
+      "epoch": 0.6470387729015765,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.917710799271481e-06,
+      "loss": 0.462,
+      "step": 7593
+    },
+    {
+      "epoch": 0.6471239880698765,
+      "grad_norm": 17.375,
+      "learning_rate": 1.917568739690892e-06,
+      "loss": 0.7491,
+      "step": 7594
+    },
+    {
+      "epoch": 0.6472092032381764,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.917426668046664e-06,
+      "loss": 0.848,
+      "step": 7595
+    },
+    {
+      "epoch": 0.6472944184064764,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9172845843413655e-06,
+      "loss": 0.4944,
+      "step": 7596
+    },
+    {
+      "epoch": 0.6473796335747763,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.917142488577564e-06,
+      "loss": 0.4256,
+      "step": 7597
+    },
+    {
+      "epoch": 0.6474648487430763,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.917000380757826e-06,
+      "loss": 1.0364,
+      "step": 7598
+    },
+    {
+      "epoch": 0.6475500639113763,
+      "grad_norm": 14.0,
+      "learning_rate": 1.916858260884721e-06,
+      "loss": 0.7619,
+      "step": 7599
+    },
+    {
+      "epoch": 0.6476352790796762,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9167161289608164e-06,
+      "loss": 0.8506,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6477204942479762,
+      "grad_norm": 24.375,
+      "learning_rate": 1.916573984988681e-06,
+      "loss": 0.9152,
+      "step": 7601
+    },
+    {
+      "epoch": 0.6478057094162761,
+      "grad_norm": 11.625,
+      "learning_rate": 1.9164318289708834e-06,
+      "loss": 0.5738,
+      "step": 7602
+    },
+    {
+      "epoch": 0.6478909245845761,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9162896609099924e-06,
+      "loss": 1.0527,
+      "step": 7603
+    },
+    {
+      "epoch": 0.647976139752876,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9161474808085773e-06,
+      "loss": 0.7811,
+      "step": 7604
+    },
+    {
+      "epoch": 0.648061354921176,
+      "grad_norm": 15.375,
+      "learning_rate": 1.916005288669207e-06,
+      "loss": 1.0938,
+      "step": 7605
+    },
+    {
+      "epoch": 0.648146570089476,
+      "grad_norm": 10.875,
+      "learning_rate": 1.915863084494452e-06,
+      "loss": 0.3323,
+      "step": 7606
+    },
+    {
+      "epoch": 0.6482317852577759,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.915720868286881e-06,
+      "loss": 0.4926,
+      "step": 7607
+    },
+    {
+      "epoch": 0.6483170004260759,
+      "grad_norm": 13.75,
+      "learning_rate": 1.915578640049065e-06,
+      "loss": 0.8232,
+      "step": 7608
+    },
+    {
+      "epoch": 0.6484022155943758,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9154363997835736e-06,
+      "loss": 0.8178,
+      "step": 7609
+    },
+    {
+      "epoch": 0.6484874307626758,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.915294147492977e-06,
+      "loss": 1.0978,
+      "step": 7610
+    },
+    {
+      "epoch": 0.6485726459309757,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9151518831798462e-06,
+      "loss": 0.8506,
+      "step": 7611
+    },
+    {
+      "epoch": 0.6486578610992757,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.915009606846752e-06,
+      "loss": 0.9594,
+      "step": 7612
+    },
+    {
+      "epoch": 0.6487430762675757,
+      "grad_norm": 16.0,
+      "learning_rate": 1.914867318496266e-06,
+      "loss": 0.8196,
+      "step": 7613
+    },
+    {
+      "epoch": 0.6488282914358756,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.914725018130958e-06,
+      "loss": 0.6289,
+      "step": 7614
+    },
+    {
+      "epoch": 0.6489135066041756,
+      "grad_norm": 19.0,
+      "learning_rate": 1.914582705753401e-06,
+      "loss": 0.9905,
+      "step": 7615
+    },
+    {
+      "epoch": 0.6489987217724755,
+      "grad_norm": 11.625,
+      "learning_rate": 1.9144403813661665e-06,
+      "loss": 0.5533,
+      "step": 7616
+    },
+    {
+      "epoch": 0.6490839369407755,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9142980449718256e-06,
+      "loss": 0.7281,
+      "step": 7617
+    },
+    {
+      "epoch": 0.6491691521090754,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9141556965729518e-06,
+      "loss": 0.4316,
+      "step": 7618
+    },
+    {
+      "epoch": 0.6492543672773754,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9140133361721157e-06,
+      "loss": 0.8764,
+      "step": 7619
+    },
+    {
+      "epoch": 0.6493395824456754,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9138709637718916e-06,
+      "loss": 1.0729,
+      "step": 7620
+    },
+    {
+      "epoch": 0.6494247976139753,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.913728579374851e-06,
+      "loss": 0.4932,
+      "step": 7621
+    },
+    {
+      "epoch": 0.6495100127822753,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9135861829835686e-06,
+      "loss": 0.6855,
+      "step": 7622
+    },
+    {
+      "epoch": 0.6495952279505752,
+      "grad_norm": 18.625,
+      "learning_rate": 1.9134437746006158e-06,
+      "loss": 0.854,
+      "step": 7623
+    },
+    {
+      "epoch": 0.6496804431188752,
+      "grad_norm": 23.0,
+      "learning_rate": 1.9133013542285668e-06,
+      "loss": 1.3407,
+      "step": 7624
+    },
+    {
+      "epoch": 0.6497656582871751,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9131589218699957e-06,
+      "loss": 0.8345,
+      "step": 7625
+    },
+    {
+      "epoch": 0.6498508734554751,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9130164775274763e-06,
+      "loss": 0.7151,
+      "step": 7626
+    },
+    {
+      "epoch": 0.649936088623775,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9128740212035814e-06,
+      "loss": 1.013,
+      "step": 7627
+    },
+    {
+      "epoch": 0.650021303792075,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9127315529008873e-06,
+      "loss": 0.8041,
+      "step": 7628
+    },
+    {
+      "epoch": 0.650106518960375,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.912589072621967e-06,
+      "loss": 0.5704,
+      "step": 7629
+    },
+    {
+      "epoch": 0.6501917341286749,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9124465803693962e-06,
+      "loss": 0.5162,
+      "step": 7630
+    },
+    {
+      "epoch": 0.6502769492969749,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9123040761457497e-06,
+      "loss": 0.6625,
+      "step": 7631
+    },
+    {
+      "epoch": 0.6503621644652748,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.912161559953602e-06,
+      "loss": 0.407,
+      "step": 7632
+    },
+    {
+      "epoch": 0.6504473796335748,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.9120190317955293e-06,
+      "loss": 0.5644,
+      "step": 7633
+    },
+    {
+      "epoch": 0.6505325948018748,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.911876491674107e-06,
+      "loss": 0.6873,
+      "step": 7634
+    },
+    {
+      "epoch": 0.6506178099701747,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9117339395919115e-06,
+      "loss": 0.7525,
+      "step": 7635
+    },
+    {
+      "epoch": 0.6507030251384747,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9115913755515173e-06,
+      "loss": 0.9009,
+      "step": 7636
+    },
+    {
+      "epoch": 0.6507882403067746,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9114487995555025e-06,
+      "loss": 0.749,
+      "step": 7637
+    },
+    {
+      "epoch": 0.6508734554750746,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9113062116064426e-06,
+      "loss": 1.1835,
+      "step": 7638
+    },
+    {
+      "epoch": 0.6509586706433745,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9111636117069145e-06,
+      "loss": 0.9973,
+      "step": 7639
+    },
+    {
+      "epoch": 0.6510438858116745,
+      "grad_norm": 14.125,
+      "learning_rate": 1.911020999859494e-06,
+      "loss": 0.7477,
+      "step": 7640
+    },
+    {
+      "epoch": 0.6511291009799745,
+      "grad_norm": 27.25,
+      "learning_rate": 1.9108783760667607e-06,
+      "loss": 0.8946,
+      "step": 7641
+    },
+    {
+      "epoch": 0.6512143161482744,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.91073574033129e-06,
+      "loss": 0.8122,
+      "step": 7642
+    },
+    {
+      "epoch": 0.6512995313165744,
+      "grad_norm": 11.0,
+      "learning_rate": 1.9105930926556606e-06,
+      "loss": 0.4939,
+      "step": 7643
+    },
+    {
+      "epoch": 0.6513847464848743,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.910450433042449e-06,
+      "loss": 0.6006,
+      "step": 7644
+    },
+    {
+      "epoch": 0.6514699616531743,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.910307761494234e-06,
+      "loss": 0.9568,
+      "step": 7645
+    },
+    {
+      "epoch": 0.6515551768214742,
+      "grad_norm": 69.5,
+      "learning_rate": 1.9101650780135937e-06,
+      "loss": 0.9464,
+      "step": 7646
+    },
+    {
+      "epoch": 0.6516403919897742,
+      "grad_norm": 34.0,
+      "learning_rate": 1.9100223826031074e-06,
+      "loss": 0.715,
+      "step": 7647
+    },
+    {
+      "epoch": 0.6517256071580741,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.909879675265352e-06,
+      "loss": 0.4619,
+      "step": 7648
+    },
+    {
+      "epoch": 0.6518108223263741,
+      "grad_norm": 24.5,
+      "learning_rate": 1.9097369560029078e-06,
+      "loss": 1.0209,
+      "step": 7649
+    },
+    {
+      "epoch": 0.6518960374946741,
+      "grad_norm": 14.375,
+      "learning_rate": 1.9095942248183534e-06,
+      "loss": 0.5722,
+      "step": 7650
+    },
+    {
+      "epoch": 0.651981252662974,
+      "grad_norm": 18.25,
+      "learning_rate": 1.909451481714268e-06,
+      "loss": 0.7237,
+      "step": 7651
+    },
+    {
+      "epoch": 0.652066467831274,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.909308726693231e-06,
+      "loss": 0.5568,
+      "step": 7652
+    },
+    {
+      "epoch": 0.6521516829995739,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9091659597578223e-06,
+      "loss": 0.4262,
+      "step": 7653
+    },
+    {
+      "epoch": 0.6522368981678739,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9090231809106223e-06,
+      "loss": 0.6109,
+      "step": 7654
+    },
+    {
+      "epoch": 0.6523221133361738,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9088803901542103e-06,
+      "loss": 0.5379,
+      "step": 7655
+    },
+    {
+      "epoch": 0.6524073285044738,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.9087375874911674e-06,
+      "loss": 0.6077,
+      "step": 7656
+    },
+    {
+      "epoch": 0.6524925436727738,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9085947729240735e-06,
+      "loss": 0.8948,
+      "step": 7657
+    },
+    {
+      "epoch": 0.6525777588410737,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9084519464555106e-06,
+      "loss": 0.8755,
+      "step": 7658
+    },
+    {
+      "epoch": 0.6526629740093737,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.908309108088058e-06,
+      "loss": 0.7233,
+      "step": 7659
+    },
+    {
+      "epoch": 0.6527481891776736,
+      "grad_norm": 27.0,
+      "learning_rate": 1.9081662578242983e-06,
+      "loss": 0.3556,
+      "step": 7660
+    },
+    {
+      "epoch": 0.6528334043459736,
+      "grad_norm": 11.5,
+      "learning_rate": 1.908023395666812e-06,
+      "loss": 0.4448,
+      "step": 7661
+    },
+    {
+      "epoch": 0.6529186195142735,
+      "grad_norm": 16.625,
+      "learning_rate": 1.907880521618182e-06,
+      "loss": 0.8879,
+      "step": 7662
+    },
+    {
+      "epoch": 0.6530038346825735,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.907737635680989e-06,
+      "loss": 0.5524,
+      "step": 7663
+    },
+    {
+      "epoch": 0.6530890498508735,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9075947378578157e-06,
+      "loss": 0.8419,
+      "step": 7664
+    },
+    {
+      "epoch": 0.6531742650191734,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9074518281512438e-06,
+      "loss": 0.6376,
+      "step": 7665
+    },
+    {
+      "epoch": 0.6532594801874734,
+      "grad_norm": 20.25,
+      "learning_rate": 1.9073089065638563e-06,
+      "loss": 0.8453,
+      "step": 7666
+    },
+    {
+      "epoch": 0.6533446953557733,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9071659730982362e-06,
+      "loss": 0.5867,
+      "step": 7667
+    },
+    {
+      "epoch": 0.6534299105240733,
+      "grad_norm": 19.125,
+      "learning_rate": 1.907023027756966e-06,
+      "loss": 0.5683,
+      "step": 7668
+    },
+    {
+      "epoch": 0.6535151256923732,
+      "grad_norm": 10.5,
+      "learning_rate": 1.906880070542629e-06,
+      "loss": 0.3983,
+      "step": 7669
+    },
+    {
+      "epoch": 0.6536003408606732,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9067371014578087e-06,
+      "loss": 1.0957,
+      "step": 7670
+    },
+    {
+      "epoch": 0.6536855560289732,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9065941205050886e-06,
+      "loss": 1.0778,
+      "step": 7671
+    },
+    {
+      "epoch": 0.6537707711972731,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9064511276870523e-06,
+      "loss": 0.6851,
+      "step": 7672
+    },
+    {
+      "epoch": 0.6538559863655731,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9063081230062841e-06,
+      "loss": 0.9018,
+      "step": 7673
+    },
+    {
+      "epoch": 0.653941201533873,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.906165106465368e-06,
+      "loss": 0.6788,
+      "step": 7674
+    },
+    {
+      "epoch": 0.654026416702173,
+      "grad_norm": 12.625,
+      "learning_rate": 1.906022078066889e-06,
+      "loss": 0.6593,
+      "step": 7675
+    },
+    {
+      "epoch": 0.6541116318704729,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9058790378134309e-06,
+      "loss": 0.5763,
+      "step": 7676
+    },
+    {
+      "epoch": 0.6541968470387729,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9057359857075794e-06,
+      "loss": 0.8212,
+      "step": 7677
+    },
+    {
+      "epoch": 0.6542820622070729,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9055929217519186e-06,
+      "loss": 0.5392,
+      "step": 7678
+    },
+    {
+      "epoch": 0.6543672773753728,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.9054498459490357e-06,
+      "loss": 0.1814,
+      "step": 7679
+    },
+    {
+      "epoch": 0.6544524925436728,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.9053067583015138e-06,
+      "loss": 0.4883,
+      "step": 7680
+    },
+    {
+      "epoch": 0.6545377077119727,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9051636588119405e-06,
+      "loss": 1.0578,
+      "step": 7681
+    },
+    {
+      "epoch": 0.6546229228802727,
+      "grad_norm": 11.875,
+      "learning_rate": 1.9050205474829005e-06,
+      "loss": 0.7196,
+      "step": 7682
+    },
+    {
+      "epoch": 0.6547081380485726,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.904877424316981e-06,
+      "loss": 0.6645,
+      "step": 7683
+    },
+    {
+      "epoch": 0.6547933532168726,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9047342893167678e-06,
+      "loss": 0.5131,
+      "step": 7684
+    },
+    {
+      "epoch": 0.6548785683851726,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9045911424848476e-06,
+      "loss": 0.8631,
+      "step": 7685
+    },
+    {
+      "epoch": 0.6549637835534725,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.904447983823807e-06,
+      "loss": 0.4195,
+      "step": 7686
+    },
+    {
+      "epoch": 0.6550489987217725,
+      "grad_norm": 21.0,
+      "learning_rate": 1.9043048133362334e-06,
+      "loss": 1.0715,
+      "step": 7687
+    },
+    {
+      "epoch": 0.6551342138900724,
+      "grad_norm": 17.75,
+      "learning_rate": 1.904161631024714e-06,
+      "loss": 1.0429,
+      "step": 7688
+    },
+    {
+      "epoch": 0.6552194290583724,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9040184368918358e-06,
+      "loss": 0.9046,
+      "step": 7689
+    },
+    {
+      "epoch": 0.6553046442266723,
+      "grad_norm": 15.375,
+      "learning_rate": 1.903875230940187e-06,
+      "loss": 0.9655,
+      "step": 7690
+    },
+    {
+      "epoch": 0.6553898593949723,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9037320131723552e-06,
+      "loss": 1.1764,
+      "step": 7691
+    },
+    {
+      "epoch": 0.6554750745632723,
+      "grad_norm": 18.125,
+      "learning_rate": 1.9035887835909289e-06,
+      "loss": 1.071,
+      "step": 7692
+    },
+    {
+      "epoch": 0.6555602897315722,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9034455421984955e-06,
+      "loss": 0.63,
+      "step": 7693
+    },
+    {
+      "epoch": 0.6556455048998722,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9033022889976444e-06,
+      "loss": 0.5354,
+      "step": 7694
+    },
+    {
+      "epoch": 0.6557307200681721,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.903159023990964e-06,
+      "loss": 1.0363,
+      "step": 7695
+    },
+    {
+      "epoch": 0.6558159352364721,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.9030157471810428e-06,
+      "loss": 0.6402,
+      "step": 7696
+    },
+    {
+      "epoch": 0.655901150404772,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9028724585704708e-06,
+      "loss": 1.0597,
+      "step": 7697
+    },
+    {
+      "epoch": 0.655986365573072,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9027291581618368e-06,
+      "loss": 0.698,
+      "step": 7698
+    },
+    {
+      "epoch": 0.656071580741372,
+      "grad_norm": 12.25,
+      "learning_rate": 1.9025858459577306e-06,
+      "loss": 0.4694,
+      "step": 7699
+    },
+    {
+      "epoch": 0.6561567959096719,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9024425219607422e-06,
+      "loss": 0.6339,
+      "step": 7700
+    },
+    {
+      "epoch": 0.6562420110779719,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9022991861734607e-06,
+      "loss": 0.2362,
+      "step": 7701
+    },
+    {
+      "epoch": 0.6563272262462718,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.9021558385984776e-06,
+      "loss": 0.4518,
+      "step": 7702
+    },
+    {
+      "epoch": 0.6564124414145718,
+      "grad_norm": 16.25,
+      "learning_rate": 1.9020124792383826e-06,
+      "loss": 0.5414,
+      "step": 7703
+    },
+    {
+      "epoch": 0.6564976565828717,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9018691080957663e-06,
+      "loss": 0.8466,
+      "step": 7704
+    },
+    {
+      "epoch": 0.6565828717511717,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.9017257251732197e-06,
+      "loss": 0.4506,
+      "step": 7705
+    },
+    {
+      "epoch": 0.6566680869194716,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9015823304733337e-06,
+      "loss": 0.9776,
+      "step": 7706
+    },
+    {
+      "epoch": 0.6567533020877716,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.9014389239987003e-06,
+      "loss": 0.5014,
+      "step": 7707
+    },
+    {
+      "epoch": 0.6568385172560716,
+      "grad_norm": 19.75,
+      "learning_rate": 1.90129550575191e-06,
+      "loss": 1.0714,
+      "step": 7708
+    },
+    {
+      "epoch": 0.6569237324243715,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.901152075735555e-06,
+      "loss": 0.876,
+      "step": 7709
+    },
+    {
+      "epoch": 0.6570089475926715,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9010086339522272e-06,
+      "loss": 0.8053,
+      "step": 7710
+    },
+    {
+      "epoch": 0.6570941627609714,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9008651804045192e-06,
+      "loss": 0.6776,
+      "step": 7711
+    },
+    {
+      "epoch": 0.6571793779292714,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9007217150950224e-06,
+      "loss": 0.8731,
+      "step": 7712
+    },
+    {
+      "epoch": 0.6572645930975713,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9005782380263301e-06,
+      "loss": 0.6487,
+      "step": 7713
+    },
+    {
+      "epoch": 0.6573498082658713,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9004347492010344e-06,
+      "loss": 0.5321,
+      "step": 7714
+    },
+    {
+      "epoch": 0.6574350234341713,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9002912486217294e-06,
+      "loss": 0.7253,
+      "step": 7715
+    },
+    {
+      "epoch": 0.6575202386024712,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9001477362910072e-06,
+      "loss": 0.9307,
+      "step": 7716
+    },
+    {
+      "epoch": 0.6576054537707712,
+      "grad_norm": 24.125,
+      "learning_rate": 1.900004212211462e-06,
+      "loss": 0.8846,
+      "step": 7717
+    },
+    {
+      "epoch": 0.6576906689390711,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8998606763856864e-06,
+      "loss": 0.375,
+      "step": 7718
+    },
+    {
+      "epoch": 0.6577758841073711,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8997171288162756e-06,
+      "loss": 0.5148,
+      "step": 7719
+    },
+    {
+      "epoch": 0.657861099275671,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.8995735695058226e-06,
+      "loss": 0.4347,
+      "step": 7720
+    },
+    {
+      "epoch": 0.657946314443971,
+      "grad_norm": 10.625,
+      "learning_rate": 1.899429998456922e-06,
+      "loss": 0.3172,
+      "step": 7721
+    },
+    {
+      "epoch": 0.658031529612271,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8992864156721682e-06,
+      "loss": 0.7518,
+      "step": 7722
+    },
+    {
+      "epoch": 0.6581167447805709,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8991428211541564e-06,
+      "loss": 0.7375,
+      "step": 7723
+    },
+    {
+      "epoch": 0.6582019599488709,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8989992149054806e-06,
+      "loss": 0.7669,
+      "step": 7724
+    },
+    {
+      "epoch": 0.6582871751171708,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8988555969287365e-06,
+      "loss": 0.6209,
+      "step": 7725
+    },
+    {
+      "epoch": 0.6583723902854708,
+      "grad_norm": 16.75,
+      "learning_rate": 1.898711967226519e-06,
+      "loss": 0.8698,
+      "step": 7726
+    },
+    {
+      "epoch": 0.6584576054537707,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8985683258014244e-06,
+      "loss": 0.7797,
+      "step": 7727
+    },
+    {
+      "epoch": 0.6585428206220707,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.8984246726560478e-06,
+      "loss": 0.7394,
+      "step": 7728
+    },
+    {
+      "epoch": 0.6586280357903707,
+      "grad_norm": 17.875,
+      "learning_rate": 1.8982810077929854e-06,
+      "loss": 0.6197,
+      "step": 7729
+    },
+    {
+      "epoch": 0.6587132509586706,
+      "grad_norm": 15.75,
+      "learning_rate": 1.898137331214833e-06,
+      "loss": 0.4921,
+      "step": 7730
+    },
+    {
+      "epoch": 0.6587984661269706,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8979936429241874e-06,
+      "loss": 0.9318,
+      "step": 7731
+    },
+    {
+      "epoch": 0.6588836812952705,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8978499429236451e-06,
+      "loss": 0.6064,
+      "step": 7732
+    },
+    {
+      "epoch": 0.6589688964635705,
+      "grad_norm": 15.75,
+      "learning_rate": 1.897706231215803e-06,
+      "loss": 0.3024,
+      "step": 7733
+    },
+    {
+      "epoch": 0.6590541116318704,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8975625078032577e-06,
+      "loss": 0.7487,
+      "step": 7734
+    },
+    {
+      "epoch": 0.6591393268001704,
+      "grad_norm": 14.125,
+      "learning_rate": 1.8974187726886068e-06,
+      "loss": 0.7034,
+      "step": 7735
+    },
+    {
+      "epoch": 0.6592245419684704,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8972750258744478e-06,
+      "loss": 0.7132,
+      "step": 7736
+    },
+    {
+      "epoch": 0.6593097571367703,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8971312673633782e-06,
+      "loss": 0.8852,
+      "step": 7737
+    },
+    {
+      "epoch": 0.6593949723050703,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8969874971579954e-06,
+      "loss": 0.8059,
+      "step": 7738
+    },
+    {
+      "epoch": 0.6594801874733702,
+      "grad_norm": 9.875,
+      "learning_rate": 1.8968437152608987e-06,
+      "loss": 0.3816,
+      "step": 7739
+    },
+    {
+      "epoch": 0.6595654026416702,
+      "grad_norm": 23.75,
+      "learning_rate": 1.896699921674685e-06,
+      "loss": 0.9941,
+      "step": 7740
+    },
+    {
+      "epoch": 0.6596506178099701,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8965561164019538e-06,
+      "loss": 1.0777,
+      "step": 7741
+    },
+    {
+      "epoch": 0.6597358329782701,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8964122994453032e-06,
+      "loss": 0.8716,
+      "step": 7742
+    },
+    {
+      "epoch": 0.65982104814657,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.8962684708073327e-06,
+      "loss": 0.4156,
+      "step": 7743
+    },
+    {
+      "epoch": 0.65990626331487,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8961246304906406e-06,
+      "loss": 0.84,
+      "step": 7744
+    },
+    {
+      "epoch": 0.65999147848317,
+      "grad_norm": 22.25,
+      "learning_rate": 1.8959807784978268e-06,
+      "loss": 0.9273,
+      "step": 7745
+    },
+    {
+      "epoch": 0.6600766936514699,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.8958369148314907e-06,
+      "loss": 0.7878,
+      "step": 7746
+    },
+    {
+      "epoch": 0.6601619088197699,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8956930394942323e-06,
+      "loss": 1.1043,
+      "step": 7747
+    },
+    {
+      "epoch": 0.6602471239880698,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8955491524886513e-06,
+      "loss": 0.6509,
+      "step": 7748
+    },
+    {
+      "epoch": 0.6603323391563698,
+      "grad_norm": 10.625,
+      "learning_rate": 1.8954052538173478e-06,
+      "loss": 0.3943,
+      "step": 7749
+    },
+    {
+      "epoch": 0.6604175543246698,
+      "grad_norm": 26.125,
+      "learning_rate": 1.8952613434829226e-06,
+      "loss": 1.2321,
+      "step": 7750
+    },
+    {
+      "epoch": 0.6605027694929697,
+      "grad_norm": 17.875,
+      "learning_rate": 1.8951174214879757e-06,
+      "loss": 1.2233,
+      "step": 7751
+    },
+    {
+      "epoch": 0.6605879846612697,
+      "grad_norm": 18.75,
+      "learning_rate": 1.894973487835109e-06,
+      "loss": 0.7206,
+      "step": 7752
+    },
+    {
+      "epoch": 0.6606731998295696,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.894829542526922e-06,
+      "loss": 0.4861,
+      "step": 7753
+    },
+    {
+      "epoch": 0.6607584149978696,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8946855855660174e-06,
+      "loss": 0.6899,
+      "step": 7754
+    },
+    {
+      "epoch": 0.6608436301661695,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8945416169549954e-06,
+      "loss": 1.0371,
+      "step": 7755
+    },
+    {
+      "epoch": 0.6609288453344695,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8943976366964586e-06,
+      "loss": 0.8959,
+      "step": 7756
+    },
+    {
+      "epoch": 0.6610140605027695,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8942536447930081e-06,
+      "loss": 0.8161,
+      "step": 7757
+    },
+    {
+      "epoch": 0.6610992756710694,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8941096412472472e-06,
+      "loss": 0.4195,
+      "step": 7758
+    },
+    {
+      "epoch": 0.6611844908393694,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8939656260617767e-06,
+      "loss": 1.0856,
+      "step": 7759
+    },
+    {
+      "epoch": 0.6612697060076693,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8938215992392e-06,
+      "loss": 0.5116,
+      "step": 7760
+    },
+    {
+      "epoch": 0.6613549211759693,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8936775607821194e-06,
+      "loss": 0.8593,
+      "step": 7761
+    },
+    {
+      "epoch": 0.6614401363442692,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8935335106931385e-06,
+      "loss": 0.5395,
+      "step": 7762
+    },
+    {
+      "epoch": 0.6615253515125692,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8933894489748595e-06,
+      "loss": 0.649,
+      "step": 7763
+    },
+    {
+      "epoch": 0.6616105666808691,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8932453756298864e-06,
+      "loss": 0.8281,
+      "step": 7764
+    },
+    {
+      "epoch": 0.6616957818491691,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8931012906608222e-06,
+      "loss": 0.918,
+      "step": 7765
+    },
+    {
+      "epoch": 0.6617809970174691,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.8929571940702712e-06,
+      "loss": 0.5456,
+      "step": 7766
+    },
+    {
+      "epoch": 0.661866212185769,
+      "grad_norm": 25.5,
+      "learning_rate": 1.8928130858608373e-06,
+      "loss": 1.1243,
+      "step": 7767
+    },
+    {
+      "epoch": 0.661951427354069,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.8926689660351243e-06,
+      "loss": 0.2324,
+      "step": 7768
+    },
+    {
+      "epoch": 0.6620366425223689,
+      "grad_norm": 14.125,
+      "learning_rate": 1.892524834595737e-06,
+      "loss": 0.7314,
+      "step": 7769
+    },
+    {
+      "epoch": 0.6621218576906689,
+      "grad_norm": 15.5,
+      "learning_rate": 1.89238069154528e-06,
+      "loss": 0.8729,
+      "step": 7770
+    },
+    {
+      "epoch": 0.6622070728589688,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8922365368863578e-06,
+      "loss": 0.6991,
+      "step": 7771
+    },
+    {
+      "epoch": 0.6622922880272688,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.8920923706215751e-06,
+      "loss": 0.5239,
+      "step": 7772
+    },
+    {
+      "epoch": 0.6623775031955688,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8919481927535382e-06,
+      "loss": 0.6511,
+      "step": 7773
+    },
+    {
+      "epoch": 0.6624627183638687,
+      "grad_norm": 20.75,
+      "learning_rate": 1.8918040032848515e-06,
+      "loss": 0.684,
+      "step": 7774
+    },
+    {
+      "epoch": 0.6625479335321687,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8916598022181213e-06,
+      "loss": 1.0745,
+      "step": 7775
+    },
+    {
+      "epoch": 0.6626331487004686,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8915155895559534e-06,
+      "loss": 0.7396,
+      "step": 7776
+    },
+    {
+      "epoch": 0.6627183638687686,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8913713653009536e-06,
+      "loss": 0.5338,
+      "step": 7777
+    },
+    {
+      "epoch": 0.6628035790370685,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.8912271294557277e-06,
+      "loss": 0.5244,
+      "step": 7778
+    },
+    {
+      "epoch": 0.6628887942053685,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8910828820228836e-06,
+      "loss": 0.4698,
+      "step": 7779
+    },
+    {
+      "epoch": 0.6629740093736685,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.890938623005027e-06,
+      "loss": 0.7289,
+      "step": 7780
+    },
+    {
+      "epoch": 0.6630592245419684,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8907943524047645e-06,
+      "loss": 0.9359,
+      "step": 7781
+    },
+    {
+      "epoch": 0.6631444397102684,
+      "grad_norm": 19.75,
+      "learning_rate": 1.8906500702247038e-06,
+      "loss": 1.0226,
+      "step": 7782
+    },
+    {
+      "epoch": 0.6632296548785683,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8905057764674522e-06,
+      "loss": 0.416,
+      "step": 7783
+    },
+    {
+      "epoch": 0.6633148700468683,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8903614711356178e-06,
+      "loss": 0.4751,
+      "step": 7784
+    },
+    {
+      "epoch": 0.6634000852151682,
+      "grad_norm": 13.5,
+      "learning_rate": 1.890217154231807e-06,
+      "loss": 0.7512,
+      "step": 7785
+    },
+    {
+      "epoch": 0.6634853003834683,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8900728257586282e-06,
+      "loss": 0.7259,
+      "step": 7786
+    },
+    {
+      "epoch": 0.6635705155517683,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8899284857186903e-06,
+      "loss": 0.9254,
+      "step": 7787
+    },
+    {
+      "epoch": 0.6636557307200682,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8897841341146012e-06,
+      "loss": 0.849,
+      "step": 7788
+    },
+    {
+      "epoch": 0.6637409458883682,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8896397709489688e-06,
+      "loss": 1.0384,
+      "step": 7789
+    },
+    {
+      "epoch": 0.6638261610566681,
+      "grad_norm": 13.625,
+      "learning_rate": 1.889495396224403e-06,
+      "loss": 0.7918,
+      "step": 7790
+    },
+    {
+      "epoch": 0.6639113762249681,
+      "grad_norm": 25.0,
+      "learning_rate": 1.8893510099435122e-06,
+      "loss": 1.2188,
+      "step": 7791
+    },
+    {
+      "epoch": 0.663996591393268,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.889206612108906e-06,
+      "loss": 0.2705,
+      "step": 7792
+    },
+    {
+      "epoch": 0.664081806561568,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8890622027231937e-06,
+      "loss": 0.7039,
+      "step": 7793
+    },
+    {
+      "epoch": 0.664167021729868,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8889177817889841e-06,
+      "loss": 0.9733,
+      "step": 7794
+    },
+    {
+      "epoch": 0.6642522368981679,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8887733493088882e-06,
+      "loss": 0.621,
+      "step": 7795
+    },
+    {
+      "epoch": 0.6643374520664679,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.8886289052855155e-06,
+      "loss": 0.4317,
+      "step": 7796
+    },
+    {
+      "epoch": 0.6644226672347678,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.888484449721476e-06,
+      "loss": 0.7371,
+      "step": 7797
+    },
+    {
+      "epoch": 0.6645078824030678,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8883399826193807e-06,
+      "loss": 0.553,
+      "step": 7798
+    },
+    {
+      "epoch": 0.6645930975713678,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8881955039818398e-06,
+      "loss": 0.3941,
+      "step": 7799
+    },
+    {
+      "epoch": 0.6646783127396677,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8880510138114646e-06,
+      "loss": 0.4722,
+      "step": 7800
+    },
+    {
+      "epoch": 0.6647635279079677,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.8879065121108659e-06,
+      "loss": 0.4653,
+      "step": 7801
+    },
+    {
+      "epoch": 0.6648487430762676,
+      "grad_norm": 24.625,
+      "learning_rate": 1.887761998882655e-06,
+      "loss": 1.0067,
+      "step": 7802
+    },
+    {
+      "epoch": 0.6649339582445676,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8876174741294434e-06,
+      "loss": 0.991,
+      "step": 7803
+    },
+    {
+      "epoch": 0.6650191734128675,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8874729378538432e-06,
+      "loss": 0.396,
+      "step": 7804
+    },
+    {
+      "epoch": 0.6651043885811675,
+      "grad_norm": 19.75,
+      "learning_rate": 1.887328390058466e-06,
+      "loss": 1.0518,
+      "step": 7805
+    },
+    {
+      "epoch": 0.6651896037494675,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8871838307459234e-06,
+      "loss": 0.8238,
+      "step": 7806
+    },
+    {
+      "epoch": 0.6652748189177674,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.8870392599188286e-06,
+      "loss": 0.3528,
+      "step": 7807
+    },
+    {
+      "epoch": 0.6653600340860674,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8868946775797936e-06,
+      "loss": 0.4562,
+      "step": 7808
+    },
+    {
+      "epoch": 0.6654452492543673,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8867500837314318e-06,
+      "loss": 0.941,
+      "step": 7809
+    },
+    {
+      "epoch": 0.6655304644226673,
+      "grad_norm": 17.75,
+      "learning_rate": 1.886605478376355e-06,
+      "loss": 0.6784,
+      "step": 7810
+    },
+    {
+      "epoch": 0.6656156795909672,
+      "grad_norm": 15.25,
+      "learning_rate": 1.8864608615171776e-06,
+      "loss": 0.7678,
+      "step": 7811
+    },
+    {
+      "epoch": 0.6657008947592672,
+      "grad_norm": 17.375,
+      "learning_rate": 1.886316233156512e-06,
+      "loss": 0.7316,
+      "step": 7812
+    },
+    {
+      "epoch": 0.6657861099275671,
+      "grad_norm": 29.625,
+      "learning_rate": 1.8861715932969727e-06,
+      "loss": 0.8083,
+      "step": 7813
+    },
+    {
+      "epoch": 0.6658713250958671,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8860269419411724e-06,
+      "loss": 0.4648,
+      "step": 7814
+    },
+    {
+      "epoch": 0.6659565402641671,
+      "grad_norm": 26.125,
+      "learning_rate": 1.885882279091726e-06,
+      "loss": 0.6458,
+      "step": 7815
+    },
+    {
+      "epoch": 0.666041755432467,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8857376047512471e-06,
+      "loss": 0.6986,
+      "step": 7816
+    },
+    {
+      "epoch": 0.666126970600767,
+      "grad_norm": 20.625,
+      "learning_rate": 1.885592918922351e-06,
+      "loss": 1.1474,
+      "step": 7817
+    },
+    {
+      "epoch": 0.6662121857690669,
+      "grad_norm": 23.875,
+      "learning_rate": 1.8854482216076514e-06,
+      "loss": 0.8777,
+      "step": 7818
+    },
+    {
+      "epoch": 0.6662974009373669,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8853035128097635e-06,
+      "loss": 0.6345,
+      "step": 7819
+    },
+    {
+      "epoch": 0.6663826161056668,
+      "grad_norm": 19.0,
+      "learning_rate": 1.885158792531302e-06,
+      "loss": 0.8317,
+      "step": 7820
+    },
+    {
+      "epoch": 0.6664678312739668,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8850140607748826e-06,
+      "loss": 0.9223,
+      "step": 7821
+    },
+    {
+      "epoch": 0.6665530464422668,
+      "grad_norm": 11.375,
+      "learning_rate": 1.8848693175431206e-06,
+      "loss": 0.5993,
+      "step": 7822
+    },
+    {
+      "epoch": 0.6666382616105667,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8847245628386313e-06,
+      "loss": 0.5059,
+      "step": 7823
+    },
+    {
+      "epoch": 0.6667234767788667,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.8845797966640314e-06,
+      "loss": 0.4261,
+      "step": 7824
+    },
+    {
+      "epoch": 0.6668086919471666,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8844350190219362e-06,
+      "loss": 0.5137,
+      "step": 7825
+    },
+    {
+      "epoch": 0.6668939071154666,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8842902299149624e-06,
+      "loss": 1.0134,
+      "step": 7826
+    },
+    {
+      "epoch": 0.6669791222837665,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8841454293457263e-06,
+      "loss": 0.5634,
+      "step": 7827
+    },
+    {
+      "epoch": 0.6670643374520665,
+      "grad_norm": 23.0,
+      "learning_rate": 1.8840006173168442e-06,
+      "loss": 0.4067,
+      "step": 7828
+    },
+    {
+      "epoch": 0.6671495526203665,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8838557938309337e-06,
+      "loss": 0.4343,
+      "step": 7829
+    },
+    {
+      "epoch": 0.6672347677886664,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.883710958890612e-06,
+      "loss": 0.817,
+      "step": 7830
+    },
+    {
+      "epoch": 0.6673199829569664,
+      "grad_norm": 19.5,
+      "learning_rate": 1.8835661124984956e-06,
+      "loss": 0.8981,
+      "step": 7831
+    },
+    {
+      "epoch": 0.6674051981252663,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8834212546572023e-06,
+      "loss": 0.698,
+      "step": 7832
+    },
+    {
+      "epoch": 0.6674904132935663,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8832763853693502e-06,
+      "loss": 0.7266,
+      "step": 7833
+    },
+    {
+      "epoch": 0.6675756284618662,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.883131504637557e-06,
+      "loss": 0.715,
+      "step": 7834
+    },
+    {
+      "epoch": 0.6676608436301662,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8829866124644408e-06,
+      "loss": 0.6582,
+      "step": 7835
+    },
+    {
+      "epoch": 0.6677460587984662,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8828417088526203e-06,
+      "loss": 0.7126,
+      "step": 7836
+    },
+    {
+      "epoch": 0.6678312739667661,
+      "grad_norm": 19.0,
+      "learning_rate": 1.8826967938047134e-06,
+      "loss": 1.1873,
+      "step": 7837
+    },
+    {
+      "epoch": 0.6679164891350661,
+      "grad_norm": 18.0,
+      "learning_rate": 1.882551867323339e-06,
+      "loss": 1.1855,
+      "step": 7838
+    },
+    {
+      "epoch": 0.668001704303366,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8824069294111166e-06,
+      "loss": 0.9735,
+      "step": 7839
+    },
+    {
+      "epoch": 0.668086919471666,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8822619800706649e-06,
+      "loss": 0.815,
+      "step": 7840
+    },
+    {
+      "epoch": 0.6681721346399659,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8821170193046038e-06,
+      "loss": 0.6461,
+      "step": 7841
+    },
+    {
+      "epoch": 0.6682573498082659,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8819720471155519e-06,
+      "loss": 0.8103,
+      "step": 7842
+    },
+    {
+      "epoch": 0.6683425649765659,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.8818270635061303e-06,
+      "loss": 0.5115,
+      "step": 7843
+    },
+    {
+      "epoch": 0.6684277801448658,
+      "grad_norm": 21.5,
+      "learning_rate": 1.8816820684789578e-06,
+      "loss": 1.2185,
+      "step": 7844
+    },
+    {
+      "epoch": 0.6685129953131658,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8815370620366554e-06,
+      "loss": 1.0734,
+      "step": 7845
+    },
+    {
+      "epoch": 0.6685982104814657,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8813920441818428e-06,
+      "loss": 0.7405,
+      "step": 7846
+    },
+    {
+      "epoch": 0.6686834256497657,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8812470149171416e-06,
+      "loss": 0.3884,
+      "step": 7847
+    },
+    {
+      "epoch": 0.6687686408180656,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.881101974245172e-06,
+      "loss": 0.4499,
+      "step": 7848
+    },
+    {
+      "epoch": 0.6688538559863656,
+      "grad_norm": 26.625,
+      "learning_rate": 1.880956922168555e-06,
+      "loss": 0.769,
+      "step": 7849
+    },
+    {
+      "epoch": 0.6689390711546656,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8808118586899116e-06,
+      "loss": 0.709,
+      "step": 7850
+    },
+    {
+      "epoch": 0.6690242863229655,
+      "grad_norm": 12.25,
+      "learning_rate": 1.880666783811864e-06,
+      "loss": 0.5738,
+      "step": 7851
+    },
+    {
+      "epoch": 0.6691095014912655,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8805216975370334e-06,
+      "loss": 0.7661,
+      "step": 7852
+    },
+    {
+      "epoch": 0.6691947166595654,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.8803765998680414e-06,
+      "loss": 0.7259,
+      "step": 7853
+    },
+    {
+      "epoch": 0.6692799318278654,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8802314908075107e-06,
+      "loss": 0.7106,
+      "step": 7854
+    },
+    {
+      "epoch": 0.6693651469961653,
+      "grad_norm": 18.25,
+      "learning_rate": 1.880086370358063e-06,
+      "loss": 0.5793,
+      "step": 7855
+    },
+    {
+      "epoch": 0.6694503621644653,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8799412385223208e-06,
+      "loss": 0.4874,
+      "step": 7856
+    },
+    {
+      "epoch": 0.6695355773327653,
+      "grad_norm": 10.75,
+      "learning_rate": 1.8797960953029076e-06,
+      "loss": 0.5157,
+      "step": 7857
+    },
+    {
+      "epoch": 0.6696207925010652,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8796509407024452e-06,
+      "loss": 0.7985,
+      "step": 7858
+    },
+    {
+      "epoch": 0.6697060076693652,
+      "grad_norm": 18.75,
+      "learning_rate": 1.8795057747235569e-06,
+      "loss": 0.6469,
+      "step": 7859
+    },
+    {
+      "epoch": 0.6697912228376651,
+      "grad_norm": 19.375,
+      "learning_rate": 1.8793605973688666e-06,
+      "loss": 0.8221,
+      "step": 7860
+    },
+    {
+      "epoch": 0.6698764380059651,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8792154086409974e-06,
+      "loss": 1.0967,
+      "step": 7861
+    },
+    {
+      "epoch": 0.669961653174265,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8790702085425732e-06,
+      "loss": 0.5723,
+      "step": 7862
+    },
+    {
+      "epoch": 0.670046868342565,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8789249970762174e-06,
+      "loss": 0.5327,
+      "step": 7863
+    },
+    {
+      "epoch": 0.670132083510865,
+      "grad_norm": 15.25,
+      "learning_rate": 1.8787797742445545e-06,
+      "loss": 0.567,
+      "step": 7864
+    },
+    {
+      "epoch": 0.6702172986791649,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8786345400502092e-06,
+      "loss": 0.5532,
+      "step": 7865
+    },
+    {
+      "epoch": 0.6703025138474649,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8784892944958055e-06,
+      "loss": 1.0269,
+      "step": 7866
+    },
+    {
+      "epoch": 0.6703877290157648,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8783440375839679e-06,
+      "loss": 1.0999,
+      "step": 7867
+    },
+    {
+      "epoch": 0.6704729441840648,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8781987693173218e-06,
+      "loss": 0.7037,
+      "step": 7868
+    },
+    {
+      "epoch": 0.6705581593523647,
+      "grad_norm": 11.125,
+      "learning_rate": 1.878053489698492e-06,
+      "loss": 0.364,
+      "step": 7869
+    },
+    {
+      "epoch": 0.6706433745206647,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.8779081987301049e-06,
+      "loss": 0.4882,
+      "step": 7870
+    },
+    {
+      "epoch": 0.6707285896889646,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8777628964147845e-06,
+      "loss": 0.6133,
+      "step": 7871
+    },
+    {
+      "epoch": 0.6708138048572646,
+      "grad_norm": 31.75,
+      "learning_rate": 1.8776175827551577e-06,
+      "loss": 1.3091,
+      "step": 7872
+    },
+    {
+      "epoch": 0.6708990200255646,
+      "grad_norm": 10.875,
+      "learning_rate": 1.8774722577538497e-06,
+      "loss": 0.534,
+      "step": 7873
+    },
+    {
+      "epoch": 0.6709842351938645,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8773269214134871e-06,
+      "loss": 0.8544,
+      "step": 7874
+    },
+    {
+      "epoch": 0.6710694503621645,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8771815737366967e-06,
+      "loss": 0.8822,
+      "step": 7875
+    },
+    {
+      "epoch": 0.6711546655304644,
+      "grad_norm": 16.625,
+      "learning_rate": 1.877036214726104e-06,
+      "loss": 0.6763,
+      "step": 7876
+    },
+    {
+      "epoch": 0.6712398806987644,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8768908443843363e-06,
+      "loss": 0.7085,
+      "step": 7877
+    },
+    {
+      "epoch": 0.6713250958670643,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.8767454627140214e-06,
+      "loss": 0.2832,
+      "step": 7878
+    },
+    {
+      "epoch": 0.6714103110353643,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8766000697177853e-06,
+      "loss": 0.8317,
+      "step": 7879
+    },
+    {
+      "epoch": 0.6714955262036643,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8764546653982558e-06,
+      "loss": 0.4419,
+      "step": 7880
+    },
+    {
+      "epoch": 0.6715807413719642,
+      "grad_norm": 12.375,
+      "learning_rate": 1.8763092497580603e-06,
+      "loss": 0.7417,
+      "step": 7881
+    },
+    {
+      "epoch": 0.6716659565402642,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8761638227998274e-06,
+      "loss": 0.6472,
+      "step": 7882
+    },
+    {
+      "epoch": 0.6717511717085641,
+      "grad_norm": 26.25,
+      "learning_rate": 1.8760183845261842e-06,
+      "loss": 0.9428,
+      "step": 7883
+    },
+    {
+      "epoch": 0.6718363868768641,
+      "grad_norm": 13.875,
+      "learning_rate": 1.875872934939759e-06,
+      "loss": 0.6219,
+      "step": 7884
+    },
+    {
+      "epoch": 0.671921602045164,
+      "grad_norm": 18.5,
+      "learning_rate": 1.8757274740431813e-06,
+      "loss": 0.9919,
+      "step": 7885
+    },
+    {
+      "epoch": 0.672006817213464,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.8755820018390786e-06,
+      "loss": 0.5577,
+      "step": 7886
+    },
+    {
+      "epoch": 0.672092032381764,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8754365183300798e-06,
+      "loss": 0.4373,
+      "step": 7887
+    },
+    {
+      "epoch": 0.6721772475500639,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8752910235188143e-06,
+      "loss": 0.652,
+      "step": 7888
+    },
+    {
+      "epoch": 0.6722624627183639,
+      "grad_norm": 10.25,
+      "learning_rate": 1.8751455174079115e-06,
+      "loss": 0.3525,
+      "step": 7889
+    },
+    {
+      "epoch": 0.6723476778866638,
+      "grad_norm": 19.5,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 0.8254,
+      "step": 7890
+    },
+    {
+      "epoch": 0.6724328930549638,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8748544712977104e-06,
+      "loss": 0.5942,
+      "step": 7891
+    },
+    {
+      "epoch": 0.6725181082232637,
+      "grad_norm": 24.625,
+      "learning_rate": 1.8747089313036722e-06,
+      "loss": 1.1972,
+      "step": 7892
+    },
+    {
+      "epoch": 0.6726033233915637,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.874563380020515e-06,
+      "loss": 0.7444,
+      "step": 7893
+    },
+    {
+      "epoch": 0.6726885385598637,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8744178174508703e-06,
+      "loss": 0.4562,
+      "step": 7894
+    },
+    {
+      "epoch": 0.6727737537281636,
+      "grad_norm": 19.625,
+      "learning_rate": 1.874272243597367e-06,
+      "loss": 1.2864,
+      "step": 7895
+    },
+    {
+      "epoch": 0.6728589688964636,
+      "grad_norm": 26.0,
+      "learning_rate": 1.874126658462637e-06,
+      "loss": 0.9159,
+      "step": 7896
+    },
+    {
+      "epoch": 0.6729441840647635,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8739810620493101e-06,
+      "loss": 0.3869,
+      "step": 7897
+    },
+    {
+      "epoch": 0.6730293992330635,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.8738354543600187e-06,
+      "loss": 0.7865,
+      "step": 7898
+    },
+    {
+      "epoch": 0.6731146144013634,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8736898353973925e-06,
+      "loss": 0.4406,
+      "step": 7899
+    },
+    {
+      "epoch": 0.6731998295696634,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.8735442051640644e-06,
+      "loss": 0.4261,
+      "step": 7900
+    },
+    {
+      "epoch": 0.6732850447379634,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8733985636626653e-06,
+      "loss": 0.6808,
+      "step": 7901
+    },
+    {
+      "epoch": 0.6733702599062633,
+      "grad_norm": 20.125,
+      "learning_rate": 1.8732529108958275e-06,
+      "loss": 0.9224,
+      "step": 7902
+    },
+    {
+      "epoch": 0.6734554750745633,
+      "grad_norm": 21.875,
+      "learning_rate": 1.8731072468661826e-06,
+      "loss": 1.3805,
+      "step": 7903
+    },
+    {
+      "epoch": 0.6735406902428632,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8729615715763633e-06,
+      "loss": 0.4637,
+      "step": 7904
+    },
+    {
+      "epoch": 0.6736259054111632,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8728158850290016e-06,
+      "loss": 0.4815,
+      "step": 7905
+    },
+    {
+      "epoch": 0.6737111205794631,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8726701872267309e-06,
+      "loss": 0.5942,
+      "step": 7906
+    },
+    {
+      "epoch": 0.6737963357477631,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.8725244781721837e-06,
+      "loss": 0.5458,
+      "step": 7907
+    },
+    {
+      "epoch": 0.673881550916063,
+      "grad_norm": 14.875,
+      "learning_rate": 1.872378757867993e-06,
+      "loss": 0.7109,
+      "step": 7908
+    },
+    {
+      "epoch": 0.673966766084363,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8722330263167926e-06,
+      "loss": 0.9752,
+      "step": 7909
+    },
+    {
+      "epoch": 0.674051981252663,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.872087283521216e-06,
+      "loss": 0.6463,
+      "step": 7910
+    },
+    {
+      "epoch": 0.6741371964209629,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8719415294838962e-06,
+      "loss": 0.5688,
+      "step": 7911
+    },
+    {
+      "epoch": 0.6742224115892629,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8717957642074675e-06,
+      "loss": 0.5499,
+      "step": 7912
+    },
+    {
+      "epoch": 0.6743076267575628,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.871649987694564e-06,
+      "loss": 0.7316,
+      "step": 7913
+    },
+    {
+      "epoch": 0.6743928419258628,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8715041999478204e-06,
+      "loss": 0.7385,
+      "step": 7914
+    },
+    {
+      "epoch": 0.6744780570941628,
+      "grad_norm": 14.75,
+      "learning_rate": 1.871358400969871e-06,
+      "loss": 0.6208,
+      "step": 7915
+    },
+    {
+      "epoch": 0.6745632722624627,
+      "grad_norm": 16.75,
+      "learning_rate": 1.87121259076335e-06,
+      "loss": 1.0545,
+      "step": 7916
+    },
+    {
+      "epoch": 0.6746484874307627,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8710667693308932e-06,
+      "loss": 0.6094,
+      "step": 7917
+    },
+    {
+      "epoch": 0.6747337025990626,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8709209366751351e-06,
+      "loss": 0.787,
+      "step": 7918
+    },
+    {
+      "epoch": 0.6748189177673626,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8707750927987116e-06,
+      "loss": 0.5271,
+      "step": 7919
+    },
+    {
+      "epoch": 0.6749041329356625,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8706292377042575e-06,
+      "loss": 0.4676,
+      "step": 7920
+    },
+    {
+      "epoch": 0.6749893481039625,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8704833713944095e-06,
+      "loss": 0.5728,
+      "step": 7921
+    },
+    {
+      "epoch": 0.6750745632722625,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8703374938718028e-06,
+      "loss": 0.779,
+      "step": 7922
+    },
+    {
+      "epoch": 0.6751597784405624,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8701916051390737e-06,
+      "loss": 0.7196,
+      "step": 7923
+    },
+    {
+      "epoch": 0.6752449936088624,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8700457051988585e-06,
+      "loss": 0.6841,
+      "step": 7924
+    },
+    {
+      "epoch": 0.6753302087771623,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8698997940537944e-06,
+      "loss": 0.8734,
+      "step": 7925
+    },
+    {
+      "epoch": 0.6754154239454623,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8697538717065175e-06,
+      "loss": 0.8064,
+      "step": 7926
+    },
+    {
+      "epoch": 0.6755006391137622,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.8696079381596647e-06,
+      "loss": 0.5771,
+      "step": 7927
+    },
+    {
+      "epoch": 0.6755858542820622,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.8694619934158734e-06,
+      "loss": 0.3223,
+      "step": 7928
+    },
+    {
+      "epoch": 0.6756710694503621,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.8693160374777809e-06,
+      "loss": 0.7083,
+      "step": 7929
+    },
+    {
+      "epoch": 0.6757562846186621,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8691700703480254e-06,
+      "loss": 0.7299,
+      "step": 7930
+    },
+    {
+      "epoch": 0.6758414997869621,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8690240920292436e-06,
+      "loss": 0.7834,
+      "step": 7931
+    },
+    {
+      "epoch": 0.675926714955262,
+      "grad_norm": 15.75,
+      "learning_rate": 1.868878102524074e-06,
+      "loss": 1.0039,
+      "step": 7932
+    },
+    {
+      "epoch": 0.676011930123562,
+      "grad_norm": 12.25,
+      "learning_rate": 1.8687321018351545e-06,
+      "loss": 0.8225,
+      "step": 7933
+    },
+    {
+      "epoch": 0.6760971452918619,
+      "grad_norm": 12.5,
+      "learning_rate": 1.8685860899651246e-06,
+      "loss": 0.423,
+      "step": 7934
+    },
+    {
+      "epoch": 0.6761823604601619,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8684400669166214e-06,
+      "loss": 0.9114,
+      "step": 7935
+    },
+    {
+      "epoch": 0.6762675756284618,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.8682940326922844e-06,
+      "loss": 0.8907,
+      "step": 7936
+    },
+    {
+      "epoch": 0.6763527907967618,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8681479872947523e-06,
+      "loss": 0.629,
+      "step": 7937
+    },
+    {
+      "epoch": 0.6764380059650618,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8680019307266647e-06,
+      "loss": 0.7546,
+      "step": 7938
+    },
+    {
+      "epoch": 0.6765232211333617,
+      "grad_norm": 24.0,
+      "learning_rate": 1.8678558629906606e-06,
+      "loss": 1.1357,
+      "step": 7939
+    },
+    {
+      "epoch": 0.6766084363016617,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8677097840893802e-06,
+      "loss": 1.2446,
+      "step": 7940
+    },
+    {
+      "epoch": 0.6766936514699616,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8675636940254622e-06,
+      "loss": 1.1777,
+      "step": 7941
+    },
+    {
+      "epoch": 0.6767788666382616,
+      "grad_norm": 9.75,
+      "learning_rate": 1.8674175928015475e-06,
+      "loss": 0.4478,
+      "step": 7942
+    },
+    {
+      "epoch": 0.6768640818065615,
+      "grad_norm": 10.375,
+      "learning_rate": 1.8672714804202763e-06,
+      "loss": 0.2777,
+      "step": 7943
+    },
+    {
+      "epoch": 0.6769492969748615,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8671253568842884e-06,
+      "loss": 0.498,
+      "step": 7944
+    },
+    {
+      "epoch": 0.6770345121431615,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8669792221962248e-06,
+      "loss": 0.4724,
+      "step": 7945
+    },
+    {
+      "epoch": 0.6771197273114614,
+      "grad_norm": 15.625,
+      "learning_rate": 1.866833076358726e-06,
+      "loss": 0.5586,
+      "step": 7946
+    },
+    {
+      "epoch": 0.6772049424797614,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8666869193744342e-06,
+      "loss": 0.6965,
+      "step": 7947
+    },
+    {
+      "epoch": 0.6772901576480613,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8665407512459887e-06,
+      "loss": 0.6138,
+      "step": 7948
+    },
+    {
+      "epoch": 0.6773753728163613,
+      "grad_norm": 13.125,
+      "learning_rate": 1.866394571976032e-06,
+      "loss": 0.6222,
+      "step": 7949
+    },
+    {
+      "epoch": 0.6774605879846612,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8662483815672055e-06,
+      "loss": 0.4369,
+      "step": 7950
+    },
+    {
+      "epoch": 0.6775458031529612,
+      "grad_norm": 15.875,
+      "learning_rate": 1.866102180022152e-06,
+      "loss": 0.5167,
+      "step": 7951
+    },
+    {
+      "epoch": 0.6776310183212612,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8659559673435113e-06,
+      "loss": 0.9183,
+      "step": 7952
+    },
+    {
+      "epoch": 0.6777162334895611,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8658097435339273e-06,
+      "loss": 0.808,
+      "step": 7953
+    },
+    {
+      "epoch": 0.6778014486578611,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.865663508596042e-06,
+      "loss": 0.5715,
+      "step": 7954
+    },
+    {
+      "epoch": 0.677886663826161,
+      "grad_norm": 19.125,
+      "learning_rate": 1.8655172625324986e-06,
+      "loss": 1.043,
+      "step": 7955
+    },
+    {
+      "epoch": 0.677971878994461,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8653710053459383e-06,
+      "loss": 0.6524,
+      "step": 7956
+    },
+    {
+      "epoch": 0.6780570941627609,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8652247370390059e-06,
+      "loss": 0.853,
+      "step": 7957
+    },
+    {
+      "epoch": 0.6781423093310609,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8650784576143433e-06,
+      "loss": 0.8039,
+      "step": 7958
+    },
+    {
+      "epoch": 0.6782275244993609,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8649321670745948e-06,
+      "loss": 0.5593,
+      "step": 7959
+    },
+    {
+      "epoch": 0.6783127396676608,
+      "grad_norm": 14.0,
+      "learning_rate": 1.8647858654224039e-06,
+      "loss": 0.6878,
+      "step": 7960
+    },
+    {
+      "epoch": 0.6783979548359608,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8646395526604134e-06,
+      "loss": 0.69,
+      "step": 7961
+    },
+    {
+      "epoch": 0.6784831700042607,
+      "grad_norm": 29.25,
+      "learning_rate": 1.8644932287912686e-06,
+      "loss": 0.9136,
+      "step": 7962
+    },
+    {
+      "epoch": 0.6785683851725607,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8643468938176134e-06,
+      "loss": 0.5083,
+      "step": 7963
+    },
+    {
+      "epoch": 0.6786536003408606,
+      "grad_norm": 18.5,
+      "learning_rate": 1.8642005477420922e-06,
+      "loss": 0.4323,
+      "step": 7964
+    },
+    {
+      "epoch": 0.6787388155091606,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8640541905673487e-06,
+      "loss": 0.7278,
+      "step": 7965
+    },
+    {
+      "epoch": 0.6788240306774606,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8639078222960292e-06,
+      "loss": 0.7939,
+      "step": 7966
+    },
+    {
+      "epoch": 0.6789092458457605,
+      "grad_norm": 13.625,
+      "learning_rate": 1.863761442930778e-06,
+      "loss": 0.6886,
+      "step": 7967
+    },
+    {
+      "epoch": 0.6789944610140605,
+      "grad_norm": 14.875,
+      "learning_rate": 1.86361505247424e-06,
+      "loss": 0.7742,
+      "step": 7968
+    },
+    {
+      "epoch": 0.6790796761823604,
+      "grad_norm": 18.75,
+      "learning_rate": 1.863468650929061e-06,
+      "loss": 0.9031,
+      "step": 7969
+    },
+    {
+      "epoch": 0.6791648913506604,
+      "grad_norm": 14.125,
+      "learning_rate": 1.8633222382978867e-06,
+      "loss": 0.8332,
+      "step": 7970
+    },
+    {
+      "epoch": 0.6792501065189603,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8631758145833628e-06,
+      "loss": 0.5905,
+      "step": 7971
+    },
+    {
+      "epoch": 0.6793353216872603,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8630293797881355e-06,
+      "loss": 1.1168,
+      "step": 7972
+    },
+    {
+      "epoch": 0.6794205368555603,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8628829339148504e-06,
+      "loss": 0.6075,
+      "step": 7973
+    },
+    {
+      "epoch": 0.6795057520238602,
+      "grad_norm": 19.0,
+      "learning_rate": 1.862736476966155e-06,
+      "loss": 0.9521,
+      "step": 7974
+    },
+    {
+      "epoch": 0.6795909671921602,
+      "grad_norm": 19.0,
+      "learning_rate": 1.862590008944695e-06,
+      "loss": 0.9583,
+      "step": 7975
+    },
+    {
+      "epoch": 0.6796761823604601,
+      "grad_norm": 33.75,
+      "learning_rate": 1.8624435298531174e-06,
+      "loss": 1.3099,
+      "step": 7976
+    },
+    {
+      "epoch": 0.6797613975287601,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8622970396940693e-06,
+      "loss": 0.5251,
+      "step": 7977
+    },
+    {
+      "epoch": 0.67984661269706,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8621505384701982e-06,
+      "loss": 0.9311,
+      "step": 7978
+    },
+    {
+      "epoch": 0.67993182786536,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8620040261841516e-06,
+      "loss": 0.9038,
+      "step": 7979
+    },
+    {
+      "epoch": 0.68001704303366,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.861857502838576e-06,
+      "loss": 0.2994,
+      "step": 7980
+    },
+    {
+      "epoch": 0.6801022582019599,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.8617109684361206e-06,
+      "loss": 0.3945,
+      "step": 7981
+    },
+    {
+      "epoch": 0.6801874733702599,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8615644229794327e-06,
+      "loss": 1.1678,
+      "step": 7982
+    },
+    {
+      "epoch": 0.6802726885385598,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8614178664711613e-06,
+      "loss": 0.8588,
+      "step": 7983
+    },
+    {
+      "epoch": 0.6803579037068598,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8612712989139534e-06,
+      "loss": 0.8733,
+      "step": 7984
+    },
+    {
+      "epoch": 0.6804431188751597,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8611247203104587e-06,
+      "loss": 0.6981,
+      "step": 7985
+    },
+    {
+      "epoch": 0.6805283340434597,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.860978130663326e-06,
+      "loss": 0.4357,
+      "step": 7986
+    },
+    {
+      "epoch": 0.6806135492117596,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8608315299752042e-06,
+      "loss": 0.6975,
+      "step": 7987
+    },
+    {
+      "epoch": 0.6806987643800596,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8606849182487418e-06,
+      "loss": 0.8316,
+      "step": 7988
+    },
+    {
+      "epoch": 0.6807839795483596,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8605382954865894e-06,
+      "loss": 0.7927,
+      "step": 7989
+    },
+    {
+      "epoch": 0.6808691947166595,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8603916616913959e-06,
+      "loss": 0.797,
+      "step": 7990
+    },
+    {
+      "epoch": 0.6809544098849595,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8602450168658113e-06,
+      "loss": 0.904,
+      "step": 7991
+    },
+    {
+      "epoch": 0.6810396250532594,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8600983610124853e-06,
+      "loss": 0.5518,
+      "step": 7992
+    },
+    {
+      "epoch": 0.6811248402215594,
+      "grad_norm": 17.875,
+      "learning_rate": 1.8599516941340689e-06,
+      "loss": 0.8319,
+      "step": 7993
+    },
+    {
+      "epoch": 0.6812100553898593,
+      "grad_norm": 18.75,
+      "learning_rate": 1.8598050162332116e-06,
+      "loss": 1.0023,
+      "step": 7994
+    },
+    {
+      "epoch": 0.6812952705581593,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.8596583273125646e-06,
+      "loss": 0.4096,
+      "step": 7995
+    },
+    {
+      "epoch": 0.6813804857264593,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8595116273747787e-06,
+      "loss": 0.3719,
+      "step": 7996
+    },
+    {
+      "epoch": 0.6814657008947592,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8593649164225044e-06,
+      "loss": 0.8419,
+      "step": 7997
+    },
+    {
+      "epoch": 0.6815509160630592,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8592181944583938e-06,
+      "loss": 0.532,
+      "step": 7998
+    },
+    {
+      "epoch": 0.6816361312313591,
+      "grad_norm": 15.375,
+      "learning_rate": 1.8590714614850975e-06,
+      "loss": 0.5619,
+      "step": 7999
+    },
+    {
+      "epoch": 0.6817213463996591,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8589247175052677e-06,
+      "loss": 0.9156,
+      "step": 8000
+    },
+    {
+      "epoch": 0.681806561567959,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8587779625215554e-06,
+      "loss": 0.6071,
+      "step": 8001
+    },
+    {
+      "epoch": 0.681891776736259,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8586311965366138e-06,
+      "loss": 0.9373,
+      "step": 8002
+    },
+    {
+      "epoch": 0.681976991904559,
+      "grad_norm": 21.5,
+      "learning_rate": 1.8584844195530939e-06,
+      "loss": 0.9696,
+      "step": 8003
+    },
+    {
+      "epoch": 0.6820622070728589,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8583376315736493e-06,
+      "loss": 0.7568,
+      "step": 8004
+    },
+    {
+      "epoch": 0.6821474222411589,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8581908326009312e-06,
+      "loss": 0.4698,
+      "step": 8005
+    },
+    {
+      "epoch": 0.6822326374094588,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.8580440226375937e-06,
+      "loss": 0.5944,
+      "step": 8006
+    },
+    {
+      "epoch": 0.6823178525777588,
+      "grad_norm": 17.75,
+      "learning_rate": 1.857897201686289e-06,
+      "loss": 1.0195,
+      "step": 8007
+    },
+    {
+      "epoch": 0.6824030677460587,
+      "grad_norm": 17.125,
+      "learning_rate": 1.857750369749671e-06,
+      "loss": 0.8105,
+      "step": 8008
+    },
+    {
+      "epoch": 0.6824882829143587,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.8576035268303922e-06,
+      "loss": 0.4069,
+      "step": 8009
+    },
+    {
+      "epoch": 0.6825734980826587,
+      "grad_norm": 11.5,
+      "learning_rate": 1.8574566729311066e-06,
+      "loss": 0.4938,
+      "step": 8010
+    },
+    {
+      "epoch": 0.6826587132509586,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8573098080544682e-06,
+      "loss": 0.8314,
+      "step": 8011
+    },
+    {
+      "epoch": 0.6827439284192587,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8571629322031312e-06,
+      "loss": 0.809,
+      "step": 8012
+    },
+    {
+      "epoch": 0.6828291435875586,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8570160453797493e-06,
+      "loss": 0.9014,
+      "step": 8013
+    },
+    {
+      "epoch": 0.6829143587558586,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8568691475869768e-06,
+      "loss": 0.3936,
+      "step": 8014
+    },
+    {
+      "epoch": 0.6829995739241586,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.856722238827469e-06,
+      "loss": 0.472,
+      "step": 8015
+    },
+    {
+      "epoch": 0.6830847890924585,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.85657531910388e-06,
+      "loss": 0.6809,
+      "step": 8016
+    },
+    {
+      "epoch": 0.6831700042607585,
+      "grad_norm": 19.0,
+      "learning_rate": 1.856428388418865e-06,
+      "loss": 0.6864,
+      "step": 8017
+    },
+    {
+      "epoch": 0.6832552194290584,
+      "grad_norm": 25.625,
+      "learning_rate": 1.8562814467750791e-06,
+      "loss": 1.1295,
+      "step": 8018
+    },
+    {
+      "epoch": 0.6833404345973584,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.856134494175178e-06,
+      "loss": 0.8235,
+      "step": 8019
+    },
+    {
+      "epoch": 0.6834256497656583,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8559875306218172e-06,
+      "loss": 0.8916,
+      "step": 8020
+    },
+    {
+      "epoch": 0.6835108649339583,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.855840556117652e-06,
+      "loss": 0.7067,
+      "step": 8021
+    },
+    {
+      "epoch": 0.6835960801022583,
+      "grad_norm": 11.0,
+      "learning_rate": 1.855693570665339e-06,
+      "loss": 0.5367,
+      "step": 8022
+    },
+    {
+      "epoch": 0.6836812952705582,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.855546574267534e-06,
+      "loss": 0.8688,
+      "step": 8023
+    },
+    {
+      "epoch": 0.6837665104388582,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.8553995669268935e-06,
+      "loss": 0.5116,
+      "step": 8024
+    },
+    {
+      "epoch": 0.6838517256071581,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.855252548646074e-06,
+      "loss": 0.7576,
+      "step": 8025
+    },
+    {
+      "epoch": 0.6839369407754581,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8551055194277324e-06,
+      "loss": 0.7792,
+      "step": 8026
+    },
+    {
+      "epoch": 0.684022155943758,
+      "grad_norm": 19.0,
+      "learning_rate": 1.8549584792745254e-06,
+      "loss": 0.968,
+      "step": 8027
+    },
+    {
+      "epoch": 0.684107371112058,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8548114281891104e-06,
+      "loss": 0.5374,
+      "step": 8028
+    },
+    {
+      "epoch": 0.684192586280358,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8546643661741448e-06,
+      "loss": 0.9379,
+      "step": 8029
+    },
+    {
+      "epoch": 0.6842778014486579,
+      "grad_norm": 16.0,
+      "learning_rate": 1.854517293232286e-06,
+      "loss": 0.7617,
+      "step": 8030
+    },
+    {
+      "epoch": 0.6843630166169579,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8543702093661918e-06,
+      "loss": 0.7383,
+      "step": 8031
+    },
+    {
+      "epoch": 0.6844482317852578,
+      "grad_norm": 14.0,
+      "learning_rate": 1.8542231145785205e-06,
+      "loss": 0.6043,
+      "step": 8032
+    },
+    {
+      "epoch": 0.6845334469535578,
+      "grad_norm": 14.0,
+      "learning_rate": 1.8540760088719295e-06,
+      "loss": 0.7912,
+      "step": 8033
+    },
+    {
+      "epoch": 0.6846186621218577,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8539288922490778e-06,
+      "loss": 0.9933,
+      "step": 8034
+    },
+    {
+      "epoch": 0.6847038772901577,
+      "grad_norm": 12.25,
+      "learning_rate": 1.8537817647126235e-06,
+      "loss": 0.6488,
+      "step": 8035
+    },
+    {
+      "epoch": 0.6847890924584576,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.853634626265226e-06,
+      "loss": 0.6229,
+      "step": 8036
+    },
+    {
+      "epoch": 0.6848743076267576,
+      "grad_norm": 10.25,
+      "learning_rate": 1.8534874769095434e-06,
+      "loss": 0.4627,
+      "step": 8037
+    },
+    {
+      "epoch": 0.6849595227950576,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8533403166482356e-06,
+      "loss": 1.0666,
+      "step": 8038
+    },
+    {
+      "epoch": 0.6850447379633575,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.853193145483961e-06,
+      "loss": 0.4895,
+      "step": 8039
+    },
+    {
+      "epoch": 0.6851299531316575,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8530459634193805e-06,
+      "loss": 0.5945,
+      "step": 8040
+    },
+    {
+      "epoch": 0.6852151682999574,
+      "grad_norm": 30.75,
+      "learning_rate": 1.8528987704571522e-06,
+      "loss": 1.6107,
+      "step": 8041
+    },
+    {
+      "epoch": 0.6853003834682574,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.8527515665999375e-06,
+      "loss": 0.6774,
+      "step": 8042
+    },
+    {
+      "epoch": 0.6853855986365573,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8526043518503953e-06,
+      "loss": 0.6746,
+      "step": 8043
+    },
+    {
+      "epoch": 0.6854708138048573,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8524571262111869e-06,
+      "loss": 0.5955,
+      "step": 8044
+    },
+    {
+      "epoch": 0.6855560289731573,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8523098896849723e-06,
+      "loss": 1.0742,
+      "step": 8045
+    },
+    {
+      "epoch": 0.6856412441414572,
+      "grad_norm": 17.125,
+      "learning_rate": 1.852162642274412e-06,
+      "loss": 0.1923,
+      "step": 8046
+    },
+    {
+      "epoch": 0.6857264593097572,
+      "grad_norm": 13.375,
+      "learning_rate": 1.852015383982168e-06,
+      "loss": 0.41,
+      "step": 8047
+    },
+    {
+      "epoch": 0.6858116744780571,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8518681148109002e-06,
+      "loss": 0.6981,
+      "step": 8048
+    },
+    {
+      "epoch": 0.6858968896463571,
+      "grad_norm": 23.0,
+      "learning_rate": 1.8517208347632705e-06,
+      "loss": 1.0914,
+      "step": 8049
+    },
+    {
+      "epoch": 0.685982104814657,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8515735438419396e-06,
+      "loss": 0.7991,
+      "step": 8050
+    },
+    {
+      "epoch": 0.686067319982957,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8514262420495704e-06,
+      "loss": 0.8384,
+      "step": 8051
+    },
+    {
+      "epoch": 0.686152535151257,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.8512789293888244e-06,
+      "loss": 0.4543,
+      "step": 8052
+    },
+    {
+      "epoch": 0.6862377503195569,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8511316058623635e-06,
+      "loss": 0.8889,
+      "step": 8053
+    },
+    {
+      "epoch": 0.6863229654878569,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8509842714728493e-06,
+      "loss": 0.6106,
+      "step": 8054
+    },
+    {
+      "epoch": 0.6864081806561568,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8508369262229457e-06,
+      "loss": 0.8317,
+      "step": 8055
+    },
+    {
+      "epoch": 0.6864933958244568,
+      "grad_norm": 17.75,
+      "learning_rate": 1.8506895701153144e-06,
+      "loss": 0.6371,
+      "step": 8056
+    },
+    {
+      "epoch": 0.6865786109927567,
+      "grad_norm": 10.875,
+      "learning_rate": 1.8505422031526186e-06,
+      "loss": 0.4233,
+      "step": 8057
+    },
+    {
+      "epoch": 0.6866638261610567,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8503948253375209e-06,
+      "loss": 0.4715,
+      "step": 8058
+    },
+    {
+      "epoch": 0.6867490413293567,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8502474366726856e-06,
+      "loss": 0.7788,
+      "step": 8059
+    },
+    {
+      "epoch": 0.6868342564976566,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8501000371607752e-06,
+      "loss": 0.7568,
+      "step": 8060
+    },
+    {
+      "epoch": 0.6869194716659566,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8499526268044534e-06,
+      "loss": 0.7253,
+      "step": 8061
+    },
+    {
+      "epoch": 0.6870046868342565,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8498052056063847e-06,
+      "loss": 0.7275,
+      "step": 8062
+    },
+    {
+      "epoch": 0.6870899020025565,
+      "grad_norm": 28.625,
+      "learning_rate": 1.849657773569233e-06,
+      "loss": 1.1963,
+      "step": 8063
+    },
+    {
+      "epoch": 0.6871751171708564,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8495103306956618e-06,
+      "loss": 0.7997,
+      "step": 8064
+    },
+    {
+      "epoch": 0.6872603323391564,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8493628769883363e-06,
+      "loss": 0.7399,
+      "step": 8065
+    },
+    {
+      "epoch": 0.6873455475074564,
+      "grad_norm": 21.125,
+      "learning_rate": 1.8492154124499208e-06,
+      "loss": 0.8936,
+      "step": 8066
+    },
+    {
+      "epoch": 0.6874307626757563,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8490679370830804e-06,
+      "loss": 0.7956,
+      "step": 8067
+    },
+    {
+      "epoch": 0.6875159778440563,
+      "grad_norm": 12.0,
+      "learning_rate": 1.84892045089048e-06,
+      "loss": 0.4879,
+      "step": 8068
+    },
+    {
+      "epoch": 0.6876011930123562,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8487729538747847e-06,
+      "loss": 0.7588,
+      "step": 8069
+    },
+    {
+      "epoch": 0.6876864081806562,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.8486254460386602e-06,
+      "loss": 0.4672,
+      "step": 8070
+    },
+    {
+      "epoch": 0.6877716233489561,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8484779273847713e-06,
+      "loss": 1.003,
+      "step": 8071
+    },
+    {
+      "epoch": 0.6878568385172561,
+      "grad_norm": 18.625,
+      "learning_rate": 1.8483303979157851e-06,
+      "loss": 0.7874,
+      "step": 8072
+    },
+    {
+      "epoch": 0.687942053685556,
+      "grad_norm": 46.0,
+      "learning_rate": 1.8481828576343668e-06,
+      "loss": 1.5661,
+      "step": 8073
+    },
+    {
+      "epoch": 0.688027268853856,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8480353065431828e-06,
+      "loss": 0.6623,
+      "step": 8074
+    },
+    {
+      "epoch": 0.688112484022156,
+      "grad_norm": 17.375,
+      "learning_rate": 1.847887744644899e-06,
+      "loss": 0.987,
+      "step": 8075
+    },
+    {
+      "epoch": 0.6881976991904559,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8477401719421827e-06,
+      "loss": 0.8627,
+      "step": 8076
+    },
+    {
+      "epoch": 0.6882829143587559,
+      "grad_norm": 18.0,
+      "learning_rate": 1.8475925884377008e-06,
+      "loss": 0.9467,
+      "step": 8077
+    },
+    {
+      "epoch": 0.6883681295270558,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8474449941341197e-06,
+      "loss": 0.7599,
+      "step": 8078
+    },
+    {
+      "epoch": 0.6884533446953558,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8472973890341062e-06,
+      "loss": 0.5263,
+      "step": 8079
+    },
+    {
+      "epoch": 0.6885385598636558,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8471497731403287e-06,
+      "loss": 1.1068,
+      "step": 8080
+    },
+    {
+      "epoch": 0.6886237750319557,
+      "grad_norm": 11.75,
+      "learning_rate": 1.8470021464554545e-06,
+      "loss": 0.403,
+      "step": 8081
+    },
+    {
+      "epoch": 0.6887089902002557,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.846854508982151e-06,
+      "loss": 0.5711,
+      "step": 8082
+    },
+    {
+      "epoch": 0.6887942053685556,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8467068607230861e-06,
+      "loss": 0.3717,
+      "step": 8083
+    },
+    {
+      "epoch": 0.6888794205368556,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.8465592016809283e-06,
+      "loss": 0.3918,
+      "step": 8084
+    },
+    {
+      "epoch": 0.6889646357051555,
+      "grad_norm": 23.0,
+      "learning_rate": 1.8464115318583464e-06,
+      "loss": 0.3721,
+      "step": 8085
+    },
+    {
+      "epoch": 0.6890498508734555,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8462638512580078e-06,
+      "loss": 0.575,
+      "step": 8086
+    },
+    {
+      "epoch": 0.6891350660417555,
+      "grad_norm": 16.125,
+      "learning_rate": 1.846116159882582e-06,
+      "loss": 0.8211,
+      "step": 8087
+    },
+    {
+      "epoch": 0.6892202812100554,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8459684577347376e-06,
+      "loss": 0.7709,
+      "step": 8088
+    },
+    {
+      "epoch": 0.6893054963783554,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8458207448171442e-06,
+      "loss": 0.5794,
+      "step": 8089
+    },
+    {
+      "epoch": 0.6893907115466553,
+      "grad_norm": 22.125,
+      "learning_rate": 1.8456730211324703e-06,
+      "loss": 1.0998,
+      "step": 8090
+    },
+    {
+      "epoch": 0.6894759267149553,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8455252866833864e-06,
+      "loss": 0.7661,
+      "step": 8091
+    },
+    {
+      "epoch": 0.6895611418832552,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8453775414725613e-06,
+      "loss": 0.6825,
+      "step": 8092
+    },
+    {
+      "epoch": 0.6896463570515552,
+      "grad_norm": 30.5,
+      "learning_rate": 1.8452297855026656e-06,
+      "loss": 0.9084,
+      "step": 8093
+    },
+    {
+      "epoch": 0.6897315722198551,
+      "grad_norm": 14.0,
+      "learning_rate": 1.8450820187763691e-06,
+      "loss": 0.6193,
+      "step": 8094
+    },
+    {
+      "epoch": 0.6898167873881551,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8449342412963422e-06,
+      "loss": 0.6635,
+      "step": 8095
+    },
+    {
+      "epoch": 0.6899020025564551,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8447864530652552e-06,
+      "loss": 1.0097,
+      "step": 8096
+    },
+    {
+      "epoch": 0.689987217724755,
+      "grad_norm": 16.75,
+      "learning_rate": 1.844638654085779e-06,
+      "loss": 1.0979,
+      "step": 8097
+    },
+    {
+      "epoch": 0.690072432893055,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8444908443605842e-06,
+      "loss": 0.6398,
+      "step": 8098
+    },
+    {
+      "epoch": 0.6901576480613549,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.8443430238923418e-06,
+      "loss": 0.6056,
+      "step": 8099
+    },
+    {
+      "epoch": 0.6902428632296549,
+      "grad_norm": 30.5,
+      "learning_rate": 1.8441951926837237e-06,
+      "loss": 1.3652,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6903280783979548,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8440473507374007e-06,
+      "loss": 1.1108,
+      "step": 8101
+    },
+    {
+      "epoch": 0.6904132935662548,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8438994980560451e-06,
+      "loss": 0.9355,
+      "step": 8102
+    },
+    {
+      "epoch": 0.6904985087345548,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8437516346423279e-06,
+      "loss": 0.6826,
+      "step": 8103
+    },
+    {
+      "epoch": 0.6905837239028547,
+      "grad_norm": 26.875,
+      "learning_rate": 1.843603760498922e-06,
+      "loss": 0.7131,
+      "step": 8104
+    },
+    {
+      "epoch": 0.6906689390711547,
+      "grad_norm": 17.375,
+      "learning_rate": 1.843455875628499e-06,
+      "loss": 0.6287,
+      "step": 8105
+    },
+    {
+      "epoch": 0.6907541542394546,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8433079800337317e-06,
+      "loss": 0.8101,
+      "step": 8106
+    },
+    {
+      "epoch": 0.6908393694077546,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8431600737172922e-06,
+      "loss": 1.0646,
+      "step": 8107
+    },
+    {
+      "epoch": 0.6909245845760545,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8430121566818543e-06,
+      "loss": 0.7475,
+      "step": 8108
+    },
+    {
+      "epoch": 0.6910097997443545,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8428642289300903e-06,
+      "loss": 0.9519,
+      "step": 8109
+    },
+    {
+      "epoch": 0.6910950149126545,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8427162904646731e-06,
+      "loss": 0.7998,
+      "step": 8110
+    },
+    {
+      "epoch": 0.6911802300809544,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8425683412882766e-06,
+      "loss": 1.1156,
+      "step": 8111
+    },
+    {
+      "epoch": 0.6912654452492544,
+      "grad_norm": 12.5,
+      "learning_rate": 1.8424203814035746e-06,
+      "loss": 0.569,
+      "step": 8112
+    },
+    {
+      "epoch": 0.6913506604175543,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8422724108132404e-06,
+      "loss": 0.7167,
+      "step": 8113
+    },
+    {
+      "epoch": 0.6914358755858543,
+      "grad_norm": 26.25,
+      "learning_rate": 1.8421244295199481e-06,
+      "loss": 0.7344,
+      "step": 8114
+    },
+    {
+      "epoch": 0.6915210907541542,
+      "grad_norm": 12.25,
+      "learning_rate": 1.8419764375263718e-06,
+      "loss": 0.6383,
+      "step": 8115
+    },
+    {
+      "epoch": 0.6916063059224542,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8418284348351862e-06,
+      "loss": 0.7088,
+      "step": 8116
+    },
+    {
+      "epoch": 0.6916915210907542,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8416804214490658e-06,
+      "loss": 0.537,
+      "step": 8117
+    },
+    {
+      "epoch": 0.6917767362590541,
+      "grad_norm": 16.375,
+      "learning_rate": 1.841532397370685e-06,
+      "loss": 0.789,
+      "step": 8118
+    },
+    {
+      "epoch": 0.6918619514273541,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8413843626027189e-06,
+      "loss": 0.4099,
+      "step": 8119
+    },
+    {
+      "epoch": 0.691947166595654,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8412363171478426e-06,
+      "loss": 0.6251,
+      "step": 8120
+    },
+    {
+      "epoch": 0.692032381763954,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8410882610087322e-06,
+      "loss": 0.8444,
+      "step": 8121
+    },
+    {
+      "epoch": 0.6921175969322539,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.8409401941880618e-06,
+      "loss": 0.5268,
+      "step": 8122
+    },
+    {
+      "epoch": 0.6922028121005539,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.8407921166885082e-06,
+      "loss": 0.3121,
+      "step": 8123
+    },
+    {
+      "epoch": 0.6922880272688539,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8406440285127466e-06,
+      "loss": 0.897,
+      "step": 8124
+    },
+    {
+      "epoch": 0.6923732424371538,
+      "grad_norm": 27.0,
+      "learning_rate": 1.840495929663454e-06,
+      "loss": 0.8406,
+      "step": 8125
+    },
+    {
+      "epoch": 0.6924584576054538,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8403478201433055e-06,
+      "loss": 0.6323,
+      "step": 8126
+    },
+    {
+      "epoch": 0.6925436727737537,
+      "grad_norm": 18.375,
+      "learning_rate": 1.840199699954979e-06,
+      "loss": 0.4347,
+      "step": 8127
+    },
+    {
+      "epoch": 0.6926288879420537,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8400515691011496e-06,
+      "loss": 0.8205,
+      "step": 8128
+    },
+    {
+      "epoch": 0.6927141031103536,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8399034275844957e-06,
+      "loss": 0.6021,
+      "step": 8129
+    },
+    {
+      "epoch": 0.6927993182786536,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8397552754076937e-06,
+      "loss": 0.8508,
+      "step": 8130
+    },
+    {
+      "epoch": 0.6928845334469536,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8396071125734202e-06,
+      "loss": 0.7914,
+      "step": 8131
+    },
+    {
+      "epoch": 0.6929697486152535,
+      "grad_norm": 33.25,
+      "learning_rate": 1.8394589390843536e-06,
+      "loss": 1.1633,
+      "step": 8132
+    },
+    {
+      "epoch": 0.6930549637835535,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8393107549431716e-06,
+      "loss": 0.692,
+      "step": 8133
+    },
+    {
+      "epoch": 0.6931401789518534,
+      "grad_norm": 12.375,
+      "learning_rate": 1.839162560152551e-06,
+      "loss": 0.4968,
+      "step": 8134
+    },
+    {
+      "epoch": 0.6932253941201534,
+      "grad_norm": 16.125,
+      "learning_rate": 1.839014354715171e-06,
+      "loss": 0.813,
+      "step": 8135
+    },
+    {
+      "epoch": 0.6933106092884533,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8388661386337093e-06,
+      "loss": 0.8119,
+      "step": 8136
+    },
+    {
+      "epoch": 0.6933958244567533,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8387179119108439e-06,
+      "loss": 0.6631,
+      "step": 8137
+    },
+    {
+      "epoch": 0.6934810396250533,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.8385696745492542e-06,
+      "loss": 0.4312,
+      "step": 8138
+    },
+    {
+      "epoch": 0.6935662547933532,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.8384214265516184e-06,
+      "loss": 0.3797,
+      "step": 8139
+    },
+    {
+      "epoch": 0.6936514699616532,
+      "grad_norm": 9.25,
+      "learning_rate": 1.838273167920616e-06,
+      "loss": 0.3417,
+      "step": 8140
+    },
+    {
+      "epoch": 0.6937366851299531,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8381248986589253e-06,
+      "loss": 1.0847,
+      "step": 8141
+    },
+    {
+      "epoch": 0.6938219002982531,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8379766187692272e-06,
+      "loss": 0.7426,
+      "step": 8142
+    },
+    {
+      "epoch": 0.693907115466553,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8378283282541997e-06,
+      "loss": 0.8764,
+      "step": 8143
+    },
+    {
+      "epoch": 0.693992330634853,
+      "grad_norm": 23.25,
+      "learning_rate": 1.8376800271165234e-06,
+      "loss": 0.6341,
+      "step": 8144
+    },
+    {
+      "epoch": 0.694077545803153,
+      "grad_norm": 9.625,
+      "learning_rate": 1.8375317153588778e-06,
+      "loss": 0.335,
+      "step": 8145
+    },
+    {
+      "epoch": 0.6941627609714529,
+      "grad_norm": 19.125,
+      "learning_rate": 1.8373833929839436e-06,
+      "loss": 0.8884,
+      "step": 8146
+    },
+    {
+      "epoch": 0.6942479761397529,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8372350599944008e-06,
+      "loss": 0.4556,
+      "step": 8147
+    },
+    {
+      "epoch": 0.6943331913080528,
+      "grad_norm": 30.5,
+      "learning_rate": 1.83708671639293e-06,
+      "loss": 1.385,
+      "step": 8148
+    },
+    {
+      "epoch": 0.6944184064763528,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8369383621822116e-06,
+      "loss": 0.4308,
+      "step": 8149
+    },
+    {
+      "epoch": 0.6945036216446527,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8367899973649267e-06,
+      "loss": 0.8422,
+      "step": 8150
+    },
+    {
+      "epoch": 0.6945888368129527,
+      "grad_norm": 18.375,
+      "learning_rate": 1.836641621943757e-06,
+      "loss": 0.8391,
+      "step": 8151
+    },
+    {
+      "epoch": 0.6946740519812526,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8364932359213828e-06,
+      "loss": 0.5521,
+      "step": 8152
+    },
+    {
+      "epoch": 0.6947592671495526,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8363448393004862e-06,
+      "loss": 0.9841,
+      "step": 8153
+    },
+    {
+      "epoch": 0.6948444823178526,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.836196432083749e-06,
+      "loss": 0.5429,
+      "step": 8154
+    },
+    {
+      "epoch": 0.6949296974861525,
+      "grad_norm": 10.875,
+      "learning_rate": 1.8360480142738527e-06,
+      "loss": 0.4257,
+      "step": 8155
+    },
+    {
+      "epoch": 0.6950149126544525,
+      "grad_norm": 14.625,
+      "learning_rate": 1.835899585873479e-06,
+      "loss": 0.9718,
+      "step": 8156
+    },
+    {
+      "epoch": 0.6951001278227524,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.835751146885311e-06,
+      "loss": 0.8682,
+      "step": 8157
+    },
+    {
+      "epoch": 0.6951853429910524,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.835602697312031e-06,
+      "loss": 0.6956,
+      "step": 8158
+    },
+    {
+      "epoch": 0.6952705581593523,
+      "grad_norm": 12.625,
+      "learning_rate": 1.8354542371563212e-06,
+      "loss": 0.5683,
+      "step": 8159
+    },
+    {
+      "epoch": 0.6953557733276523,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8353057664208646e-06,
+      "loss": 0.7104,
+      "step": 8160
+    },
+    {
+      "epoch": 0.6954409884959523,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8351572851083442e-06,
+      "loss": 0.5315,
+      "step": 8161
+    },
+    {
+      "epoch": 0.6955262036642522,
+      "grad_norm": 26.625,
+      "learning_rate": 1.835008793221443e-06,
+      "loss": 0.9861,
+      "step": 8162
+    },
+    {
+      "epoch": 0.6956114188325522,
+      "grad_norm": 18.125,
+      "learning_rate": 1.8348602907628448e-06,
+      "loss": 1.0019,
+      "step": 8163
+    },
+    {
+      "epoch": 0.6956966340008521,
+      "grad_norm": 17.375,
+      "learning_rate": 1.834711777735233e-06,
+      "loss": 1.0653,
+      "step": 8164
+    },
+    {
+      "epoch": 0.6957818491691521,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8345632541412917e-06,
+      "loss": 0.6701,
+      "step": 8165
+    },
+    {
+      "epoch": 0.695867064337452,
+      "grad_norm": 9.375,
+      "learning_rate": 1.8344147199837041e-06,
+      "loss": 0.281,
+      "step": 8166
+    },
+    {
+      "epoch": 0.695952279505752,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.8342661752651554e-06,
+      "loss": 0.9143,
+      "step": 8167
+    },
+    {
+      "epoch": 0.696037494674052,
+      "grad_norm": 13.25,
+      "learning_rate": 1.834117619988329e-06,
+      "loss": 0.7012,
+      "step": 8168
+    },
+    {
+      "epoch": 0.6961227098423519,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8339690541559097e-06,
+      "loss": 0.7927,
+      "step": 8169
+    },
+    {
+      "epoch": 0.6962079250106519,
+      "grad_norm": 12.5,
+      "learning_rate": 1.833820477770583e-06,
+      "loss": 0.6952,
+      "step": 8170
+    },
+    {
+      "epoch": 0.6962931401789518,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8336718908350326e-06,
+      "loss": 0.5869,
+      "step": 8171
+    },
+    {
+      "epoch": 0.6963783553472518,
+      "grad_norm": 9.5,
+      "learning_rate": 1.8335232933519443e-06,
+      "loss": 0.2891,
+      "step": 8172
+    },
+    {
+      "epoch": 0.6964635705155517,
+      "grad_norm": 17.125,
+      "learning_rate": 1.833374685324003e-06,
+      "loss": 0.9662,
+      "step": 8173
+    },
+    {
+      "epoch": 0.6965487856838517,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8332260667538953e-06,
+      "loss": 0.6379,
+      "step": 8174
+    },
+    {
+      "epoch": 0.6966340008521517,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8330774376443056e-06,
+      "loss": 0.9416,
+      "step": 8175
+    },
+    {
+      "epoch": 0.6967192160204516,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8329287979979205e-06,
+      "loss": 0.9773,
+      "step": 8176
+    },
+    {
+      "epoch": 0.6968044311887516,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8327801478174253e-06,
+      "loss": 0.9329,
+      "step": 8177
+    },
+    {
+      "epoch": 0.6968896463570515,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8326314871055075e-06,
+      "loss": 0.9319,
+      "step": 8178
+    },
+    {
+      "epoch": 0.6969748615253515,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8324828158648525e-06,
+      "loss": 0.5919,
+      "step": 8179
+    },
+    {
+      "epoch": 0.6970600766936514,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.8323341340981474e-06,
+      "loss": 0.6949,
+      "step": 8180
+    },
+    {
+      "epoch": 0.6971452918619514,
+      "grad_norm": 20.125,
+      "learning_rate": 1.8321854418080786e-06,
+      "loss": 1.0211,
+      "step": 8181
+    },
+    {
+      "epoch": 0.6972305070302514,
+      "grad_norm": 22.25,
+      "learning_rate": 1.8320367389973337e-06,
+      "loss": 1.0448,
+      "step": 8182
+    },
+    {
+      "epoch": 0.6973157221985513,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8318880256685998e-06,
+      "loss": 0.7677,
+      "step": 8183
+    },
+    {
+      "epoch": 0.6974009373668513,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8317393018245638e-06,
+      "loss": 0.7157,
+      "step": 8184
+    },
+    {
+      "epoch": 0.6974861525351512,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8315905674679139e-06,
+      "loss": 0.8377,
+      "step": 8185
+    },
+    {
+      "epoch": 0.6975713677034512,
+      "grad_norm": 11.125,
+      "learning_rate": 1.8314418226013375e-06,
+      "loss": 0.3535,
+      "step": 8186
+    },
+    {
+      "epoch": 0.6976565828717511,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8312930672275229e-06,
+      "loss": 0.7206,
+      "step": 8187
+    },
+    {
+      "epoch": 0.6977417980400511,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8311443013491576e-06,
+      "loss": 0.5121,
+      "step": 8188
+    },
+    {
+      "epoch": 0.697827013208351,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8309955249689305e-06,
+      "loss": 0.8198,
+      "step": 8189
+    },
+    {
+      "epoch": 0.697912228376651,
+      "grad_norm": 23.25,
+      "learning_rate": 1.8308467380895304e-06,
+      "loss": 0.5556,
+      "step": 8190
+    },
+    {
+      "epoch": 0.697997443544951,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8306979407136452e-06,
+      "loss": 0.5531,
+      "step": 8191
+    },
+    {
+      "epoch": 0.6980826587132509,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8305491328439646e-06,
+      "loss": 0.6348,
+      "step": 8192
+    },
+    {
+      "epoch": 0.6981678738815509,
+      "grad_norm": 19.25,
+      "learning_rate": 1.8304003144831772e-06,
+      "loss": 0.8518,
+      "step": 8193
+    },
+    {
+      "epoch": 0.6982530890498508,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8302514856339726e-06,
+      "loss": 0.7462,
+      "step": 8194
+    },
+    {
+      "epoch": 0.6983383042181508,
+      "grad_norm": 17.875,
+      "learning_rate": 1.83010264629904e-06,
+      "loss": 1.0599,
+      "step": 8195
+    },
+    {
+      "epoch": 0.6984235193864508,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8299537964810693e-06,
+      "loss": 0.9519,
+      "step": 8196
+    },
+    {
+      "epoch": 0.6985087345547507,
+      "grad_norm": 13.625,
+      "learning_rate": 1.8298049361827503e-06,
+      "loss": 0.7515,
+      "step": 8197
+    },
+    {
+      "epoch": 0.6985939497230507,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8296560654067732e-06,
+      "loss": 0.8538,
+      "step": 8198
+    },
+    {
+      "epoch": 0.6986791648913506,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8295071841558278e-06,
+      "loss": 0.8301,
+      "step": 8199
+    },
+    {
+      "epoch": 0.6987643800596506,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8293582924326053e-06,
+      "loss": 0.8123,
+      "step": 8200
+    },
+    {
+      "epoch": 0.6988495952279505,
+      "grad_norm": 15.25,
+      "learning_rate": 1.8292093902397954e-06,
+      "loss": 0.6887,
+      "step": 8201
+    },
+    {
+      "epoch": 0.6989348103962505,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.82906047758009e-06,
+      "loss": 0.6365,
+      "step": 8202
+    },
+    {
+      "epoch": 0.6990200255645505,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.8289115544561787e-06,
+      "loss": 0.7295,
+      "step": 8203
+    },
+    {
+      "epoch": 0.6991052407328504,
+      "grad_norm": 14.0,
+      "learning_rate": 1.828762620870754e-06,
+      "loss": 0.7369,
+      "step": 8204
+    },
+    {
+      "epoch": 0.6991904559011504,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8286136768265064e-06,
+      "loss": 0.8952,
+      "step": 8205
+    },
+    {
+      "epoch": 0.6992756710694503,
+      "grad_norm": 13.625,
+      "learning_rate": 1.828464722326128e-06,
+      "loss": 0.7115,
+      "step": 8206
+    },
+    {
+      "epoch": 0.6993608862377503,
+      "grad_norm": 19.375,
+      "learning_rate": 1.8283157573723106e-06,
+      "loss": 0.9904,
+      "step": 8207
+    },
+    {
+      "epoch": 0.6994461014060502,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8281667819677457e-06,
+      "loss": 1.1654,
+      "step": 8208
+    },
+    {
+      "epoch": 0.6995313165743502,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8280177961151258e-06,
+      "loss": 0.6762,
+      "step": 8209
+    },
+    {
+      "epoch": 0.6996165317426501,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.827868799817143e-06,
+      "loss": 0.3427,
+      "step": 8210
+    },
+    {
+      "epoch": 0.6997017469109501,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8277197930764898e-06,
+      "loss": 0.599,
+      "step": 8211
+    },
+    {
+      "epoch": 0.6997869620792501,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8275707758958592e-06,
+      "loss": 0.3797,
+      "step": 8212
+    },
+    {
+      "epoch": 0.69987217724755,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8274217482779435e-06,
+      "loss": 0.6451,
+      "step": 8213
+    },
+    {
+      "epoch": 0.69995739241585,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8272727102254365e-06,
+      "loss": 0.7015,
+      "step": 8214
+    },
+    {
+      "epoch": 0.7000426075841499,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.827123661741031e-06,
+      "loss": 0.3925,
+      "step": 8215
+    },
+    {
+      "epoch": 0.7001278227524499,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8269746028274203e-06,
+      "loss": 0.5469,
+      "step": 8216
+    },
+    {
+      "epoch": 0.7002130379207498,
+      "grad_norm": 12.625,
+      "learning_rate": 1.826825533487299e-06,
+      "loss": 0.6324,
+      "step": 8217
+    },
+    {
+      "epoch": 0.7002982530890498,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.82667645372336e-06,
+      "loss": 0.4119,
+      "step": 8218
+    },
+    {
+      "epoch": 0.7003834682573498,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.8265273635382973e-06,
+      "loss": 1.025,
+      "step": 8219
+    },
+    {
+      "epoch": 0.7004686834256497,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8263782629348056e-06,
+      "loss": 0.8116,
+      "step": 8220
+    },
+    {
+      "epoch": 0.7005538985939497,
+      "grad_norm": 15.375,
+      "learning_rate": 1.8262291519155793e-06,
+      "loss": 0.7729,
+      "step": 8221
+    },
+    {
+      "epoch": 0.7006391137622496,
+      "grad_norm": 14.875,
+      "learning_rate": 1.8260800304833123e-06,
+      "loss": 0.7839,
+      "step": 8222
+    },
+    {
+      "epoch": 0.7007243289305496,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8259308986407006e-06,
+      "loss": 0.4612,
+      "step": 8223
+    },
+    {
+      "epoch": 0.7008095440988495,
+      "grad_norm": 18.0,
+      "learning_rate": 1.8257817563904378e-06,
+      "loss": 0.6319,
+      "step": 8224
+    },
+    {
+      "epoch": 0.7008947592671495,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8256326037352199e-06,
+      "loss": 0.9335,
+      "step": 8225
+    },
+    {
+      "epoch": 0.7009799744354495,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8254834406777416e-06,
+      "loss": 1.0858,
+      "step": 8226
+    },
+    {
+      "epoch": 0.7010651896037494,
+      "grad_norm": 12.375,
+      "learning_rate": 1.8253342672206997e-06,
+      "loss": 0.5482,
+      "step": 8227
+    },
+    {
+      "epoch": 0.7011504047720494,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8251850833667882e-06,
+      "loss": 0.8393,
+      "step": 8228
+    },
+    {
+      "epoch": 0.7012356199403493,
+      "grad_norm": 14.125,
+      "learning_rate": 1.8250358891187042e-06,
+      "loss": 0.5882,
+      "step": 8229
+    },
+    {
+      "epoch": 0.7013208351086493,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8248866844791435e-06,
+      "loss": 0.7213,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7014060502769492,
+      "grad_norm": 12.625,
+      "learning_rate": 1.824737469450802e-06,
+      "loss": 0.5338,
+      "step": 8231
+    },
+    {
+      "epoch": 0.7014912654452492,
+      "grad_norm": 13.0,
+      "learning_rate": 1.824588244036377e-06,
+      "loss": 0.6128,
+      "step": 8232
+    },
+    {
+      "epoch": 0.7015764806135492,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8244390082385644e-06,
+      "loss": 0.6144,
+      "step": 8233
+    },
+    {
+      "epoch": 0.7016616957818491,
+      "grad_norm": 10.875,
+      "learning_rate": 1.8242897620600612e-06,
+      "loss": 0.4473,
+      "step": 8234
+    },
+    {
+      "epoch": 0.7017469109501491,
+      "grad_norm": 15.625,
+      "learning_rate": 1.8241405055035644e-06,
+      "loss": 1.1708,
+      "step": 8235
+    },
+    {
+      "epoch": 0.701832126118449,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.8239912385717718e-06,
+      "loss": 0.3221,
+      "step": 8236
+    },
+    {
+      "epoch": 0.701917341286749,
+      "grad_norm": 12.25,
+      "learning_rate": 1.8238419612673796e-06,
+      "loss": 0.516,
+      "step": 8237
+    },
+    {
+      "epoch": 0.702002556455049,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8236926735930865e-06,
+      "loss": 0.5556,
+      "step": 8238
+    },
+    {
+      "epoch": 0.702087771623349,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8235433755515903e-06,
+      "loss": 0.8166,
+      "step": 8239
+    },
+    {
+      "epoch": 0.702172986791649,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.823394067145588e-06,
+      "loss": 0.453,
+      "step": 8240
+    },
+    {
+      "epoch": 0.7022582019599489,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8232447483777783e-06,
+      "loss": 1.1542,
+      "step": 8241
+    },
+    {
+      "epoch": 0.7023434171282489,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8230954192508598e-06,
+      "loss": 0.5877,
+      "step": 8242
+    },
+    {
+      "epoch": 0.7024286322965488,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8229460797675308e-06,
+      "loss": 0.8381,
+      "step": 8243
+    },
+    {
+      "epoch": 0.7025138474648488,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8227967299304899e-06,
+      "loss": 1.1072,
+      "step": 8244
+    },
+    {
+      "epoch": 0.7025990626331488,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.822647369742436e-06,
+      "loss": 1.1704,
+      "step": 8245
+    },
+    {
+      "epoch": 0.7026842778014487,
+      "grad_norm": 12.375,
+      "learning_rate": 1.8224979992060685e-06,
+      "loss": 0.469,
+      "step": 8246
+    },
+    {
+      "epoch": 0.7027694929697487,
+      "grad_norm": 29.5,
+      "learning_rate": 1.8223486183240865e-06,
+      "loss": 1.0302,
+      "step": 8247
+    },
+    {
+      "epoch": 0.7028547081380486,
+      "grad_norm": 20.125,
+      "learning_rate": 1.8221992270991894e-06,
+      "loss": 1.0763,
+      "step": 8248
+    },
+    {
+      "epoch": 0.7029399233063486,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8220498255340766e-06,
+      "loss": 0.8155,
+      "step": 8249
+    },
+    {
+      "epoch": 0.7030251384746485,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8219004136314482e-06,
+      "loss": 0.5337,
+      "step": 8250
+    },
+    {
+      "epoch": 0.7031103536429485,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8217509913940046e-06,
+      "loss": 0.909,
+      "step": 8251
+    },
+    {
+      "epoch": 0.7031955688112485,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.8216015588244453e-06,
+      "loss": 0.6935,
+      "step": 8252
+    },
+    {
+      "epoch": 0.7032807839795484,
+      "grad_norm": 24.875,
+      "learning_rate": 1.8214521159254715e-06,
+      "loss": 1.37,
+      "step": 8253
+    },
+    {
+      "epoch": 0.7033659991478484,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8213026626997826e-06,
+      "loss": 0.8778,
+      "step": 8254
+    },
+    {
+      "epoch": 0.7034512143161483,
+      "grad_norm": 16.25,
+      "learning_rate": 1.8211531991500809e-06,
+      "loss": 0.467,
+      "step": 8255
+    },
+    {
+      "epoch": 0.7035364294844483,
+      "grad_norm": 14.25,
+      "learning_rate": 1.821003725279066e-06,
+      "loss": 0.8079,
+      "step": 8256
+    },
+    {
+      "epoch": 0.7036216446527482,
+      "grad_norm": 23.0,
+      "learning_rate": 1.8208542410894398e-06,
+      "loss": 0.7194,
+      "step": 8257
+    },
+    {
+      "epoch": 0.7037068598210482,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8207047465839035e-06,
+      "loss": 0.723,
+      "step": 8258
+    },
+    {
+      "epoch": 0.7037920749893481,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8205552417651585e-06,
+      "loss": 0.6498,
+      "step": 8259
+    },
+    {
+      "epoch": 0.7038772901576481,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8204057266359062e-06,
+      "loss": 0.7109,
+      "step": 8260
+    },
+    {
+      "epoch": 0.7039625053259481,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8202562011988492e-06,
+      "loss": 0.6741,
+      "step": 8261
+    },
+    {
+      "epoch": 0.704047720494248,
+      "grad_norm": 19.0,
+      "learning_rate": 1.820106665456689e-06,
+      "loss": 0.6911,
+      "step": 8262
+    },
+    {
+      "epoch": 0.704132935662548,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.8199571194121285e-06,
+      "loss": 0.3597,
+      "step": 8263
+    },
+    {
+      "epoch": 0.7042181508308479,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.819807563067869e-06,
+      "loss": 0.483,
+      "step": 8264
+    },
+    {
+      "epoch": 0.7043033659991479,
+      "grad_norm": 24.125,
+      "learning_rate": 1.8196579964266145e-06,
+      "loss": 1.0289,
+      "step": 8265
+    },
+    {
+      "epoch": 0.7043885811674478,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8195084194910668e-06,
+      "loss": 0.8748,
+      "step": 8266
+    },
+    {
+      "epoch": 0.7044737963357478,
+      "grad_norm": 22.625,
+      "learning_rate": 1.8193588322639294e-06,
+      "loss": 1.3506,
+      "step": 8267
+    },
+    {
+      "epoch": 0.7045590115040478,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8192092347479057e-06,
+      "loss": 0.693,
+      "step": 8268
+    },
+    {
+      "epoch": 0.7046442266723477,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.819059626945698e-06,
+      "loss": 0.977,
+      "step": 8269
+    },
+    {
+      "epoch": 0.7047294418406477,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8189100088600111e-06,
+      "loss": 0.8157,
+      "step": 8270
+    },
+    {
+      "epoch": 0.7048146570089476,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.8187603804935482e-06,
+      "loss": 0.6772,
+      "step": 8271
+    },
+    {
+      "epoch": 0.7048998721772476,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8186107418490135e-06,
+      "loss": 0.9156,
+      "step": 8272
+    },
+    {
+      "epoch": 0.7049850873455475,
+      "grad_norm": 26.625,
+      "learning_rate": 1.8184610929291106e-06,
+      "loss": 1.1045,
+      "step": 8273
+    },
+    {
+      "epoch": 0.7050703025138475,
+      "grad_norm": 15.125,
+      "learning_rate": 1.818311433736544e-06,
+      "loss": 0.7433,
+      "step": 8274
+    },
+    {
+      "epoch": 0.7051555176821475,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8181617642740186e-06,
+      "loss": 0.8027,
+      "step": 8275
+    },
+    {
+      "epoch": 0.7052407328504474,
+      "grad_norm": 17.125,
+      "learning_rate": 1.818012084544239e-06,
+      "loss": 0.5,
+      "step": 8276
+    },
+    {
+      "epoch": 0.7053259480187474,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8178623945499092e-06,
+      "loss": 0.8944,
+      "step": 8277
+    },
+    {
+      "epoch": 0.7054111631870473,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8177126942937354e-06,
+      "loss": 0.704,
+      "step": 8278
+    },
+    {
+      "epoch": 0.7054963783553473,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.817562983778422e-06,
+      "loss": 0.4785,
+      "step": 8279
+    },
+    {
+      "epoch": 0.7055815935236472,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8174132630066751e-06,
+      "loss": 0.8266,
+      "step": 8280
+    },
+    {
+      "epoch": 0.7056668086919472,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.8172635319812e-06,
+      "loss": 0.5889,
+      "step": 8281
+    },
+    {
+      "epoch": 0.7057520238602472,
+      "grad_norm": 15.625,
+      "learning_rate": 1.817113790704702e-06,
+      "loss": 0.7331,
+      "step": 8282
+    },
+    {
+      "epoch": 0.7058372390285471,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8169640391798876e-06,
+      "loss": 0.5778,
+      "step": 8283
+    },
+    {
+      "epoch": 0.7059224541968471,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8168142774094633e-06,
+      "loss": 0.906,
+      "step": 8284
+    },
+    {
+      "epoch": 0.706007669365147,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8166645053961347e-06,
+      "loss": 0.6478,
+      "step": 8285
+    },
+    {
+      "epoch": 0.706092884533447,
+      "grad_norm": 12.25,
+      "learning_rate": 1.8165147231426088e-06,
+      "loss": 0.5336,
+      "step": 8286
+    },
+    {
+      "epoch": 0.7061780997017469,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.8163649306515919e-06,
+      "loss": 0.3632,
+      "step": 8287
+    },
+    {
+      "epoch": 0.7062633148700469,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8162151279257918e-06,
+      "loss": 0.9746,
+      "step": 8288
+    },
+    {
+      "epoch": 0.7063485300383469,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8160653149679145e-06,
+      "loss": 1.0229,
+      "step": 8289
+    },
+    {
+      "epoch": 0.7064337452066468,
+      "grad_norm": 25.0,
+      "learning_rate": 1.8159154917806675e-06,
+      "loss": 0.9772,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7065189603749468,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8157656583667587e-06,
+      "loss": 0.7552,
+      "step": 8291
+    },
+    {
+      "epoch": 0.7066041755432467,
+      "grad_norm": 14.875,
+      "learning_rate": 1.815615814728896e-06,
+      "loss": 0.8258,
+      "step": 8292
+    },
+    {
+      "epoch": 0.7066893907115467,
+      "grad_norm": 23.375,
+      "learning_rate": 1.8154659608697864e-06,
+      "loss": 0.8631,
+      "step": 8293
+    },
+    {
+      "epoch": 0.7067746058798466,
+      "grad_norm": 13.25,
+      "learning_rate": 1.815316096792138e-06,
+      "loss": 0.7681,
+      "step": 8294
+    },
+    {
+      "epoch": 0.7068598210481466,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.8151662224986596e-06,
+      "loss": 0.9717,
+      "step": 8295
+    },
+    {
+      "epoch": 0.7069450362164466,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8150163379920592e-06,
+      "loss": 0.8529,
+      "step": 8296
+    },
+    {
+      "epoch": 0.7070302513847465,
+      "grad_norm": 13.75,
+      "learning_rate": 1.8148664432750455e-06,
+      "loss": 0.393,
+      "step": 8297
+    },
+    {
+      "epoch": 0.7071154665530465,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8147165383503267e-06,
+      "loss": 0.6728,
+      "step": 8298
+    },
+    {
+      "epoch": 0.7072006817213464,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.8145666232206128e-06,
+      "loss": 0.7041,
+      "step": 8299
+    },
+    {
+      "epoch": 0.7072858968896464,
+      "grad_norm": 13.25,
+      "learning_rate": 1.814416697888612e-06,
+      "loss": 0.5002,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7073711120579463,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.814266762357034e-06,
+      "loss": 0.4458,
+      "step": 8301
+    },
+    {
+      "epoch": 0.7074563272262463,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8141168166285877e-06,
+      "loss": 0.8011,
+      "step": 8302
+    },
+    {
+      "epoch": 0.7075415423945463,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.8139668607059841e-06,
+      "loss": 0.3866,
+      "step": 8303
+    },
+    {
+      "epoch": 0.7076267575628462,
+      "grad_norm": 18.75,
+      "learning_rate": 1.8138168945919316e-06,
+      "loss": 0.8114,
+      "step": 8304
+    },
+    {
+      "epoch": 0.7077119727311462,
+      "grad_norm": 22.0,
+      "learning_rate": 1.8136669182891408e-06,
+      "loss": 1.0237,
+      "step": 8305
+    },
+    {
+      "epoch": 0.7077971878994461,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8135169318003222e-06,
+      "loss": 0.4133,
+      "step": 8306
+    },
+    {
+      "epoch": 0.7078824030677461,
+      "grad_norm": 13.0,
+      "learning_rate": 1.813366935128186e-06,
+      "loss": 0.642,
+      "step": 8307
+    },
+    {
+      "epoch": 0.707967618236046,
+      "grad_norm": 11.25,
+      "learning_rate": 1.813216928275443e-06,
+      "loss": 0.2526,
+      "step": 8308
+    },
+    {
+      "epoch": 0.708052833404346,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8130669112448035e-06,
+      "loss": 0.8389,
+      "step": 8309
+    },
+    {
+      "epoch": 0.708138048572646,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8129168840389789e-06,
+      "loss": 0.9668,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7082232637409459,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8127668466606796e-06,
+      "loss": 0.8989,
+      "step": 8311
+    },
+    {
+      "epoch": 0.7083084789092459,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8126167991126182e-06,
+      "loss": 0.7233,
+      "step": 8312
+    },
+    {
+      "epoch": 0.7083936940775458,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.8124667413975047e-06,
+      "loss": 0.5178,
+      "step": 8313
+    },
+    {
+      "epoch": 0.7084789092458458,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8123166735180521e-06,
+      "loss": 0.7463,
+      "step": 8314
+    },
+    {
+      "epoch": 0.7085641244141457,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8121665954769715e-06,
+      "loss": 0.8885,
+      "step": 8315
+    },
+    {
+      "epoch": 0.7086493395824457,
+      "grad_norm": 22.125,
+      "learning_rate": 1.8120165072769756e-06,
+      "loss": 1.2882,
+      "step": 8316
+    },
+    {
+      "epoch": 0.7087345547507456,
+      "grad_norm": 10.25,
+      "learning_rate": 1.811866408920776e-06,
+      "loss": 0.6725,
+      "step": 8317
+    },
+    {
+      "epoch": 0.7088197699190456,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8117163004110854e-06,
+      "loss": 1.0656,
+      "step": 8318
+    },
+    {
+      "epoch": 0.7089049850873456,
+      "grad_norm": 20.0,
+      "learning_rate": 1.8115661817506164e-06,
+      "loss": 1.1641,
+      "step": 8319
+    },
+    {
+      "epoch": 0.7089902002556455,
+      "grad_norm": 24.375,
+      "learning_rate": 1.8114160529420818e-06,
+      "loss": 0.7925,
+      "step": 8320
+    },
+    {
+      "epoch": 0.7090754154239455,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8112659139881946e-06,
+      "loss": 1.0362,
+      "step": 8321
+    },
+    {
+      "epoch": 0.7091606305922454,
+      "grad_norm": 15.75,
+      "learning_rate": 1.8111157648916677e-06,
+      "loss": 0.7768,
+      "step": 8322
+    },
+    {
+      "epoch": 0.7092458457605454,
+      "grad_norm": 17.75,
+      "learning_rate": 1.8109656056552153e-06,
+      "loss": 1.0236,
+      "step": 8323
+    },
+    {
+      "epoch": 0.7093310609288453,
+      "grad_norm": 10.625,
+      "learning_rate": 1.81081543628155e-06,
+      "loss": 0.5315,
+      "step": 8324
+    },
+    {
+      "epoch": 0.7094162760971453,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8106652567733856e-06,
+      "loss": 0.7208,
+      "step": 8325
+    },
+    {
+      "epoch": 0.7095014912654453,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.8105150671334363e-06,
+      "loss": 0.4416,
+      "step": 8326
+    },
+    {
+      "epoch": 0.7095867064337452,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8103648673644164e-06,
+      "loss": 0.5388,
+      "step": 8327
+    },
+    {
+      "epoch": 0.7096719216020452,
+      "grad_norm": 24.0,
+      "learning_rate": 1.8102146574690396e-06,
+      "loss": 1.357,
+      "step": 8328
+    },
+    {
+      "epoch": 0.7097571367703451,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.8100644374500207e-06,
+      "loss": 0.8613,
+      "step": 8329
+    },
+    {
+      "epoch": 0.7098423519386451,
+      "grad_norm": 17.875,
+      "learning_rate": 1.809914207310074e-06,
+      "loss": 0.7143,
+      "step": 8330
+    },
+    {
+      "epoch": 0.709927567106945,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.8097639670519145e-06,
+      "loss": 0.7625,
+      "step": 8331
+    },
+    {
+      "epoch": 0.710012782275245,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8096137166782576e-06,
+      "loss": 0.7302,
+      "step": 8332
+    },
+    {
+      "epoch": 0.710097997443545,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8094634561918182e-06,
+      "loss": 0.981,
+      "step": 8333
+    },
+    {
+      "epoch": 0.7101832126118449,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.8093131855953111e-06,
+      "loss": 0.3318,
+      "step": 8334
+    },
+    {
+      "epoch": 0.7102684277801449,
+      "grad_norm": 19.5,
+      "learning_rate": 1.8091629048914525e-06,
+      "loss": 0.983,
+      "step": 8335
+    },
+    {
+      "epoch": 0.7103536429484448,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8090126140829577e-06,
+      "loss": 0.9012,
+      "step": 8336
+    },
+    {
+      "epoch": 0.7104388581167448,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.808862313172543e-06,
+      "loss": 0.6816,
+      "step": 8337
+    },
+    {
+      "epoch": 0.7105240732850447,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8087120021629245e-06,
+      "loss": 0.9793,
+      "step": 8338
+    },
+    {
+      "epoch": 0.7106092884533447,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8085616810568179e-06,
+      "loss": 0.8137,
+      "step": 8339
+    },
+    {
+      "epoch": 0.7106945036216447,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8084113498569406e-06,
+      "loss": 0.4541,
+      "step": 8340
+    },
+    {
+      "epoch": 0.7107797187899446,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8082610085660085e-06,
+      "loss": 0.8404,
+      "step": 8341
+    },
+    {
+      "epoch": 0.7108649339582446,
+      "grad_norm": 13.25,
+      "learning_rate": 1.8081106571867385e-06,
+      "loss": 0.8269,
+      "step": 8342
+    },
+    {
+      "epoch": 0.7109501491265445,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8079602957218474e-06,
+      "loss": 0.729,
+      "step": 8343
+    },
+    {
+      "epoch": 0.7110353642948445,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8078099241740533e-06,
+      "loss": 0.7775,
+      "step": 8344
+    },
+    {
+      "epoch": 0.7111205794631444,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8076595425460725e-06,
+      "loss": 0.7163,
+      "step": 8345
+    },
+    {
+      "epoch": 0.7112057946314444,
+      "grad_norm": 14.5,
+      "learning_rate": 1.807509150840623e-06,
+      "loss": 0.7853,
+      "step": 8346
+    },
+    {
+      "epoch": 0.7112910097997444,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8073587490604227e-06,
+      "loss": 0.5224,
+      "step": 8347
+    },
+    {
+      "epoch": 0.7113762249680443,
+      "grad_norm": 14.25,
+      "learning_rate": 1.8072083372081892e-06,
+      "loss": 0.9063,
+      "step": 8348
+    },
+    {
+      "epoch": 0.7114614401363443,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.807057915286641e-06,
+      "loss": 0.581,
+      "step": 8349
+    },
+    {
+      "epoch": 0.7115466553046442,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8069074832984958e-06,
+      "loss": 0.6699,
+      "step": 8350
+    },
+    {
+      "epoch": 0.7116318704729442,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8067570412464725e-06,
+      "loss": 0.5923,
+      "step": 8351
+    },
+    {
+      "epoch": 0.7117170856412441,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.8066065891332896e-06,
+      "loss": 0.3533,
+      "step": 8352
+    },
+    {
+      "epoch": 0.7118023008095441,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.806456126961666e-06,
+      "loss": 0.2979,
+      "step": 8353
+    },
+    {
+      "epoch": 0.711887515977844,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.8063056547343203e-06,
+      "loss": 0.6823,
+      "step": 8354
+    },
+    {
+      "epoch": 0.711972731146144,
+      "grad_norm": 12.375,
+      "learning_rate": 1.8061551724539722e-06,
+      "loss": 0.5469,
+      "step": 8355
+    },
+    {
+      "epoch": 0.712057946314444,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8060046801233411e-06,
+      "loss": 0.4253,
+      "step": 8356
+    },
+    {
+      "epoch": 0.7121431614827439,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8058541777451464e-06,
+      "loss": 0.659,
+      "step": 8357
+    },
+    {
+      "epoch": 0.7122283766510439,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8057036653221071e-06,
+      "loss": 0.667,
+      "step": 8358
+    },
+    {
+      "epoch": 0.7123135918193438,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8055531428569443e-06,
+      "loss": 0.4437,
+      "step": 8359
+    },
+    {
+      "epoch": 0.7123988069876438,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8054026103523772e-06,
+      "loss": 0.3802,
+      "step": 8360
+    },
+    {
+      "epoch": 0.7124840221559438,
+      "grad_norm": 15.875,
+      "learning_rate": 1.8052520678111272e-06,
+      "loss": 0.4102,
+      "step": 8361
+    },
+    {
+      "epoch": 0.7125692373242437,
+      "grad_norm": 15.5,
+      "learning_rate": 1.805101515235913e-06,
+      "loss": 0.8885,
+      "step": 8362
+    },
+    {
+      "epoch": 0.7126544524925437,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8049509526294567e-06,
+      "loss": 0.8602,
+      "step": 8363
+    },
+    {
+      "epoch": 0.7127396676608436,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8048003799944786e-06,
+      "loss": 0.5404,
+      "step": 8364
+    },
+    {
+      "epoch": 0.7128248828291436,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.8046497973336998e-06,
+      "loss": 0.834,
+      "step": 8365
+    },
+    {
+      "epoch": 0.7129100979974435,
+      "grad_norm": 12.375,
+      "learning_rate": 1.8044992046498413e-06,
+      "loss": 0.5426,
+      "step": 8366
+    },
+    {
+      "epoch": 0.7129953131657435,
+      "grad_norm": 15.125,
+      "learning_rate": 1.8043486019456247e-06,
+      "loss": 0.6189,
+      "step": 8367
+    },
+    {
+      "epoch": 0.7130805283340435,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8041979892237713e-06,
+      "loss": 0.6886,
+      "step": 8368
+    },
+    {
+      "epoch": 0.7131657435023434,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.804047366487003e-06,
+      "loss": 0.5865,
+      "step": 8369
+    },
+    {
+      "epoch": 0.7132509586706434,
+      "grad_norm": 24.25,
+      "learning_rate": 1.8038967337380417e-06,
+      "loss": 1.3272,
+      "step": 8370
+    },
+    {
+      "epoch": 0.7133361738389433,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8037460909796096e-06,
+      "loss": 0.9567,
+      "step": 8371
+    },
+    {
+      "epoch": 0.7134213890072433,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8035954382144282e-06,
+      "loss": 0.8599,
+      "step": 8372
+    },
+    {
+      "epoch": 0.7135066041755432,
+      "grad_norm": 12.0,
+      "learning_rate": 1.8034447754452211e-06,
+      "loss": 0.6604,
+      "step": 8373
+    },
+    {
+      "epoch": 0.7135918193438432,
+      "grad_norm": 30.375,
+      "learning_rate": 1.8032941026747102e-06,
+      "loss": 0.9519,
+      "step": 8374
+    },
+    {
+      "epoch": 0.7136770345121431,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8031434199056184e-06,
+      "loss": 0.9951,
+      "step": 8375
+    },
+    {
+      "epoch": 0.7137622496804431,
+      "grad_norm": 23.125,
+      "learning_rate": 1.8029927271406688e-06,
+      "loss": 0.8564,
+      "step": 8376
+    },
+    {
+      "epoch": 0.7138474648487431,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8028420243825848e-06,
+      "loss": 0.5792,
+      "step": 8377
+    },
+    {
+      "epoch": 0.713932680017043,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.8026913116340893e-06,
+      "loss": 0.8336,
+      "step": 8378
+    },
+    {
+      "epoch": 0.714017895185343,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.802540588897906e-06,
+      "loss": 0.6323,
+      "step": 8379
+    },
+    {
+      "epoch": 0.7141031103536429,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8023898561767586e-06,
+      "loss": 0.7817,
+      "step": 8380
+    },
+    {
+      "epoch": 0.7141883255219429,
+      "grad_norm": 20.625,
+      "learning_rate": 1.8022391134733713e-06,
+      "loss": 0.9023,
+      "step": 8381
+    },
+    {
+      "epoch": 0.7142735406902428,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.8020883607904677e-06,
+      "loss": 0.3712,
+      "step": 8382
+    },
+    {
+      "epoch": 0.7143587558585428,
+      "grad_norm": 12.5,
+      "learning_rate": 1.8019375981307722e-06,
+      "loss": 0.6155,
+      "step": 8383
+    },
+    {
+      "epoch": 0.7144439710268428,
+      "grad_norm": 10.125,
+      "learning_rate": 1.8017868254970095e-06,
+      "loss": 0.4282,
+      "step": 8384
+    },
+    {
+      "epoch": 0.7145291861951427,
+      "grad_norm": 14.0,
+      "learning_rate": 1.8016360428919038e-06,
+      "loss": 0.5352,
+      "step": 8385
+    },
+    {
+      "epoch": 0.7146144013634427,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.8014852503181802e-06,
+      "loss": 0.9668,
+      "step": 8386
+    },
+    {
+      "epoch": 0.7146996165317426,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8013344477785632e-06,
+      "loss": 0.8365,
+      "step": 8387
+    },
+    {
+      "epoch": 0.7147848317000426,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.8011836352757785e-06,
+      "loss": 0.5402,
+      "step": 8388
+    },
+    {
+      "epoch": 0.7148700468683425,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8010328128125515e-06,
+      "loss": 0.7868,
+      "step": 8389
+    },
+    {
+      "epoch": 0.7149552620366425,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8008819803916073e-06,
+      "loss": 0.5965,
+      "step": 8390
+    },
+    {
+      "epoch": 0.7150404772049425,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8007311380156716e-06,
+      "loss": 0.4634,
+      "step": 8391
+    },
+    {
+      "epoch": 0.7151256923732424,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8005802856874707e-06,
+      "loss": 0.8445,
+      "step": 8392
+    },
+    {
+      "epoch": 0.7152109075415424,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.8004294234097304e-06,
+      "loss": 0.8743,
+      "step": 8393
+    },
+    {
+      "epoch": 0.7152961227098423,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8002785511851766e-06,
+      "loss": 0.8426,
+      "step": 8394
+    },
+    {
+      "epoch": 0.7153813378781423,
+      "grad_norm": 21.625,
+      "learning_rate": 1.8001276690165364e-06,
+      "loss": 0.9245,
+      "step": 8395
+    },
+    {
+      "epoch": 0.7154665530464422,
+      "grad_norm": 15.25,
+      "learning_rate": 1.7999767769065357e-06,
+      "loss": 0.384,
+      "step": 8396
+    },
+    {
+      "epoch": 0.7155517682147422,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7998258748579022e-06,
+      "loss": 0.5564,
+      "step": 8397
+    },
+    {
+      "epoch": 0.7156369833830422,
+      "grad_norm": 15.875,
+      "learning_rate": 1.7996749628733618e-06,
+      "loss": 0.7272,
+      "step": 8398
+    },
+    {
+      "epoch": 0.7157221985513421,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7995240409556421e-06,
+      "loss": 1.0192,
+      "step": 8399
+    },
+    {
+      "epoch": 0.7158074137196421,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7993731091074701e-06,
+      "loss": 0.9317,
+      "step": 8400
+    },
+    {
+      "epoch": 0.715892628887942,
+      "grad_norm": 21.5,
+      "learning_rate": 1.7992221673315743e-06,
+      "loss": 0.7983,
+      "step": 8401
+    },
+    {
+      "epoch": 0.715977844056242,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7990712156306812e-06,
+      "loss": 1.0339,
+      "step": 8402
+    },
+    {
+      "epoch": 0.7160630592245419,
+      "grad_norm": 14.75,
+      "learning_rate": 1.798920254007519e-06,
+      "loss": 0.8863,
+      "step": 8403
+    },
+    {
+      "epoch": 0.7161482743928419,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.7987692824648159e-06,
+      "loss": 0.4364,
+      "step": 8404
+    },
+    {
+      "epoch": 0.7162334895611419,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.7986183010053001e-06,
+      "loss": 0.6309,
+      "step": 8405
+    },
+    {
+      "epoch": 0.7163187047294418,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.7984673096317001e-06,
+      "loss": 0.4635,
+      "step": 8406
+    },
+    {
+      "epoch": 0.7164039198977418,
+      "grad_norm": 11.125,
+      "learning_rate": 1.798316308346744e-06,
+      "loss": 0.4399,
+      "step": 8407
+    },
+    {
+      "epoch": 0.7164891350660417,
+      "grad_norm": 16.375,
+      "learning_rate": 1.798165297153161e-06,
+      "loss": 0.9663,
+      "step": 8408
+    },
+    {
+      "epoch": 0.7165743502343417,
+      "grad_norm": 15.25,
+      "learning_rate": 1.7980142760536795e-06,
+      "loss": 0.8537,
+      "step": 8409
+    },
+    {
+      "epoch": 0.7166595654026416,
+      "grad_norm": 18.375,
+      "learning_rate": 1.7978632450510297e-06,
+      "loss": 0.8599,
+      "step": 8410
+    },
+    {
+      "epoch": 0.7167447805709416,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7977122041479394e-06,
+      "loss": 0.7619,
+      "step": 8411
+    },
+    {
+      "epoch": 0.7168299957392416,
+      "grad_norm": 13.625,
+      "learning_rate": 1.7975611533471393e-06,
+      "loss": 0.682,
+      "step": 8412
+    },
+    {
+      "epoch": 0.7169152109075415,
+      "grad_norm": 10.375,
+      "learning_rate": 1.7974100926513579e-06,
+      "loss": 0.7359,
+      "step": 8413
+    },
+    {
+      "epoch": 0.7170004260758415,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7972590220633264e-06,
+      "loss": 0.6809,
+      "step": 8414
+    },
+    {
+      "epoch": 0.7170856412441414,
+      "grad_norm": 27.625,
+      "learning_rate": 1.7971079415857737e-06,
+      "loss": 1.25,
+      "step": 8415
+    },
+    {
+      "epoch": 0.7171708564124414,
+      "grad_norm": 15.375,
+      "learning_rate": 1.7969568512214305e-06,
+      "loss": 0.7435,
+      "step": 8416
+    },
+    {
+      "epoch": 0.7172560715807413,
+      "grad_norm": 20.875,
+      "learning_rate": 1.796805750973027e-06,
+      "loss": 1.1183,
+      "step": 8417
+    },
+    {
+      "epoch": 0.7173412867490413,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.7966546408432936e-06,
+      "loss": 0.4151,
+      "step": 8418
+    },
+    {
+      "epoch": 0.7174265019173413,
+      "grad_norm": 23.5,
+      "learning_rate": 1.796503520834961e-06,
+      "loss": 0.8503,
+      "step": 8419
+    },
+    {
+      "epoch": 0.7175117170856412,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7963523909507606e-06,
+      "loss": 0.9599,
+      "step": 8420
+    },
+    {
+      "epoch": 0.7175969322539412,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.7962012511934226e-06,
+      "loss": 0.6236,
+      "step": 8421
+    },
+    {
+      "epoch": 0.7176821474222411,
+      "grad_norm": 18.125,
+      "learning_rate": 1.7960501015656793e-06,
+      "loss": 0.8873,
+      "step": 8422
+    },
+    {
+      "epoch": 0.7177673625905411,
+      "grad_norm": 11.875,
+      "learning_rate": 1.795898942070261e-06,
+      "loss": 0.3641,
+      "step": 8423
+    },
+    {
+      "epoch": 0.717852577758841,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.7957477727099e-06,
+      "loss": 0.7945,
+      "step": 8424
+    },
+    {
+      "epoch": 0.717937792927141,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7955965934873284e-06,
+      "loss": 0.6206,
+      "step": 8425
+    },
+    {
+      "epoch": 0.718023008095441,
+      "grad_norm": 17.625,
+      "learning_rate": 1.7954454044052775e-06,
+      "loss": 0.9856,
+      "step": 8426
+    },
+    {
+      "epoch": 0.7181082232637409,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.7952942054664795e-06,
+      "loss": 1.1367,
+      "step": 8427
+    },
+    {
+      "epoch": 0.7181934384320409,
+      "grad_norm": 13.625,
+      "learning_rate": 1.795142996673667e-06,
+      "loss": 0.6583,
+      "step": 8428
+    },
+    {
+      "epoch": 0.7182786536003408,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7949917780295722e-06,
+      "loss": 0.8533,
+      "step": 8429
+    },
+    {
+      "epoch": 0.7183638687686408,
+      "grad_norm": 19.5,
+      "learning_rate": 1.794840549536928e-06,
+      "loss": 1.093,
+      "step": 8430
+    },
+    {
+      "epoch": 0.7184490839369407,
+      "grad_norm": 21.625,
+      "learning_rate": 1.7946893111984675e-06,
+      "loss": 0.7753,
+      "step": 8431
+    },
+    {
+      "epoch": 0.7185342991052407,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7945380630169231e-06,
+      "loss": 1.0423,
+      "step": 8432
+    },
+    {
+      "epoch": 0.7186195142735406,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7943868049950286e-06,
+      "loss": 0.76,
+      "step": 8433
+    },
+    {
+      "epoch": 0.7187047294418406,
+      "grad_norm": 21.5,
+      "learning_rate": 1.794235537135517e-06,
+      "loss": 0.9214,
+      "step": 8434
+    },
+    {
+      "epoch": 0.7187899446101406,
+      "grad_norm": 10.375,
+      "learning_rate": 1.7940842594411217e-06,
+      "loss": 0.4538,
+      "step": 8435
+    },
+    {
+      "epoch": 0.7188751597784405,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7939329719145768e-06,
+      "loss": 0.7646,
+      "step": 8436
+    },
+    {
+      "epoch": 0.7189603749467405,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7937816745586162e-06,
+      "loss": 0.8822,
+      "step": 8437
+    },
+    {
+      "epoch": 0.7190455901150404,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7936303673759742e-06,
+      "loss": 0.8798,
+      "step": 8438
+    },
+    {
+      "epoch": 0.7191308052833404,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.7934790503693844e-06,
+      "loss": 0.4781,
+      "step": 8439
+    },
+    {
+      "epoch": 0.7192160204516403,
+      "grad_norm": 15.125,
+      "learning_rate": 1.793327723541582e-06,
+      "loss": 0.8587,
+      "step": 8440
+    },
+    {
+      "epoch": 0.7193012356199403,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.7931763868953008e-06,
+      "loss": 0.695,
+      "step": 8441
+    },
+    {
+      "epoch": 0.7193864507882403,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7930250404332766e-06,
+      "loss": 0.6579,
+      "step": 8442
+    },
+    {
+      "epoch": 0.7194716659565402,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.7928736841582434e-06,
+      "loss": 0.5921,
+      "step": 8443
+    },
+    {
+      "epoch": 0.7195568811248402,
+      "grad_norm": 12.125,
+      "learning_rate": 1.7927223180729372e-06,
+      "loss": 0.5393,
+      "step": 8444
+    },
+    {
+      "epoch": 0.7196420962931401,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.7925709421800928e-06,
+      "loss": 0.7309,
+      "step": 8445
+    },
+    {
+      "epoch": 0.7197273114614401,
+      "grad_norm": 13.5,
+      "learning_rate": 1.792419556482446e-06,
+      "loss": 0.3641,
+      "step": 8446
+    },
+    {
+      "epoch": 0.71981252662974,
+      "grad_norm": 14.75,
+      "learning_rate": 1.7922681609827322e-06,
+      "loss": 0.671,
+      "step": 8447
+    },
+    {
+      "epoch": 0.71989774179804,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.7921167556836874e-06,
+      "loss": 0.8593,
+      "step": 8448
+    },
+    {
+      "epoch": 0.71998295696634,
+      "grad_norm": 19.875,
+      "learning_rate": 1.7919653405880478e-06,
+      "loss": 0.7353,
+      "step": 8449
+    },
+    {
+      "epoch": 0.7200681721346399,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7918139156985499e-06,
+      "loss": 0.6623,
+      "step": 8450
+    },
+    {
+      "epoch": 0.7201533873029399,
+      "grad_norm": 16.5,
+      "learning_rate": 1.791662481017929e-06,
+      "loss": 1.1,
+      "step": 8451
+    },
+    {
+      "epoch": 0.7202386024712398,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.7915110365489228e-06,
+      "loss": 0.324,
+      "step": 8452
+    },
+    {
+      "epoch": 0.7203238176395398,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7913595822942673e-06,
+      "loss": 0.5537,
+      "step": 8453
+    },
+    {
+      "epoch": 0.7204090328078397,
+      "grad_norm": 16.0,
+      "learning_rate": 1.7912081182567e-06,
+      "loss": 0.9242,
+      "step": 8454
+    },
+    {
+      "epoch": 0.7204942479761397,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.7910566444389578e-06,
+      "loss": 0.7414,
+      "step": 8455
+    },
+    {
+      "epoch": 0.7205794631444397,
+      "grad_norm": 22.375,
+      "learning_rate": 1.790905160843778e-06,
+      "loss": 1.2834,
+      "step": 8456
+    },
+    {
+      "epoch": 0.7206646783127396,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7907536674738978e-06,
+      "loss": 0.9382,
+      "step": 8457
+    },
+    {
+      "epoch": 0.7207498934810396,
+      "grad_norm": 23.5,
+      "learning_rate": 1.7906021643320554e-06,
+      "loss": 0.6758,
+      "step": 8458
+    },
+    {
+      "epoch": 0.7208351086493395,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.790450651420988e-06,
+      "loss": 0.8636,
+      "step": 8459
+    },
+    {
+      "epoch": 0.7209203238176395,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7902991287434337e-06,
+      "loss": 1.0673,
+      "step": 8460
+    },
+    {
+      "epoch": 0.7210055389859394,
+      "grad_norm": 15.375,
+      "learning_rate": 1.7901475963021315e-06,
+      "loss": 1.0023,
+      "step": 8461
+    },
+    {
+      "epoch": 0.7210907541542394,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7899960540998185e-06,
+      "loss": 0.9036,
+      "step": 8462
+    },
+    {
+      "epoch": 0.7211759693225395,
+      "grad_norm": 12.875,
+      "learning_rate": 1.789844502139234e-06,
+      "loss": 0.8175,
+      "step": 8463
+    },
+    {
+      "epoch": 0.7212611844908394,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.789692940423116e-06,
+      "loss": 0.325,
+      "step": 8464
+    },
+    {
+      "epoch": 0.7213463996591394,
+      "grad_norm": 20.875,
+      "learning_rate": 1.7895413689542045e-06,
+      "loss": 0.7589,
+      "step": 8465
+    },
+    {
+      "epoch": 0.7214316148274393,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7893897877352378e-06,
+      "loss": 0.7837,
+      "step": 8466
+    },
+    {
+      "epoch": 0.7215168299957393,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.7892381967689554e-06,
+      "loss": 0.3613,
+      "step": 8467
+    },
+    {
+      "epoch": 0.7216020451640393,
+      "grad_norm": 11.625,
+      "learning_rate": 1.7890865960580961e-06,
+      "loss": 0.5492,
+      "step": 8468
+    },
+    {
+      "epoch": 0.7216872603323392,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7889349856054e-06,
+      "loss": 0.7181,
+      "step": 8469
+    },
+    {
+      "epoch": 0.7217724755006392,
+      "grad_norm": 13.5,
+      "learning_rate": 1.7887833654136073e-06,
+      "loss": 0.4474,
+      "step": 8470
+    },
+    {
+      "epoch": 0.7218576906689391,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7886317354854568e-06,
+      "loss": 0.5507,
+      "step": 8471
+    },
+    {
+      "epoch": 0.7219429058372391,
+      "grad_norm": 14.75,
+      "learning_rate": 1.7884800958236892e-06,
+      "loss": 0.9146,
+      "step": 8472
+    },
+    {
+      "epoch": 0.722028121005539,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7883284464310453e-06,
+      "loss": 0.7204,
+      "step": 8473
+    },
+    {
+      "epoch": 0.722113336173839,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7881767873102646e-06,
+      "loss": 0.8883,
+      "step": 8474
+    },
+    {
+      "epoch": 0.722198551342139,
+      "grad_norm": 22.0,
+      "learning_rate": 1.7880251184640878e-06,
+      "loss": 0.965,
+      "step": 8475
+    },
+    {
+      "epoch": 0.7222837665104389,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7878734398952568e-06,
+      "loss": 0.8425,
+      "step": 8476
+    },
+    {
+      "epoch": 0.7223689816787389,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7877217516065115e-06,
+      "loss": 0.7847,
+      "step": 8477
+    },
+    {
+      "epoch": 0.7224541968470388,
+      "grad_norm": 24.25,
+      "learning_rate": 1.7875700536005935e-06,
+      "loss": 1.0565,
+      "step": 8478
+    },
+    {
+      "epoch": 0.7225394120153388,
+      "grad_norm": 22.25,
+      "learning_rate": 1.7874183458802445e-06,
+      "loss": 1.0721,
+      "step": 8479
+    },
+    {
+      "epoch": 0.7226246271836387,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.7872666284482052e-06,
+      "loss": 0.7625,
+      "step": 8480
+    },
+    {
+      "epoch": 0.7227098423519387,
+      "grad_norm": 9.875,
+      "learning_rate": 1.7871149013072175e-06,
+      "loss": 0.2838,
+      "step": 8481
+    },
+    {
+      "epoch": 0.7227950575202386,
+      "grad_norm": 14.375,
+      "learning_rate": 1.7869631644600236e-06,
+      "loss": 0.5975,
+      "step": 8482
+    },
+    {
+      "epoch": 0.7228802726885386,
+      "grad_norm": 10.5,
+      "learning_rate": 1.7868114179093655e-06,
+      "loss": 0.4271,
+      "step": 8483
+    },
+    {
+      "epoch": 0.7229654878568386,
+      "grad_norm": 17.0,
+      "learning_rate": 1.786659661657985e-06,
+      "loss": 0.7355,
+      "step": 8484
+    },
+    {
+      "epoch": 0.7230507030251385,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.7865078957086245e-06,
+      "loss": 0.4958,
+      "step": 8485
+    },
+    {
+      "epoch": 0.7231359181934385,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7863561200640275e-06,
+      "loss": 1.3118,
+      "step": 8486
+    },
+    {
+      "epoch": 0.7232211333617384,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.786204334726936e-06,
+      "loss": 0.6518,
+      "step": 8487
+    },
+    {
+      "epoch": 0.7233063485300384,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.7860525397000927e-06,
+      "loss": 0.7232,
+      "step": 8488
+    },
+    {
+      "epoch": 0.7233915636983383,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7859007349862405e-06,
+      "loss": 0.8012,
+      "step": 8489
+    },
+    {
+      "epoch": 0.7234767788666383,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7857489205881239e-06,
+      "loss": 0.8907,
+      "step": 8490
+    },
+    {
+      "epoch": 0.7235619940349383,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7855970965084855e-06,
+      "loss": 0.5191,
+      "step": 8491
+    },
+    {
+      "epoch": 0.7236472092032382,
+      "grad_norm": 11.875,
+      "learning_rate": 1.7854452627500684e-06,
+      "loss": 0.5629,
+      "step": 8492
+    },
+    {
+      "epoch": 0.7237324243715382,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7852934193156174e-06,
+      "loss": 0.7152,
+      "step": 8493
+    },
+    {
+      "epoch": 0.7238176395398381,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7851415662078757e-06,
+      "loss": 0.6132,
+      "step": 8494
+    },
+    {
+      "epoch": 0.7239028547081381,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7849897034295885e-06,
+      "loss": 0.8485,
+      "step": 8495
+    },
+    {
+      "epoch": 0.723988069876438,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.7848378309834987e-06,
+      "loss": 0.5086,
+      "step": 8496
+    },
+    {
+      "epoch": 0.724073285044738,
+      "grad_norm": 13.0,
+      "learning_rate": 1.784685948872352e-06,
+      "loss": 0.7946,
+      "step": 8497
+    },
+    {
+      "epoch": 0.724158500213038,
+      "grad_norm": 16.125,
+      "learning_rate": 1.784534057098892e-06,
+      "loss": 0.6018,
+      "step": 8498
+    },
+    {
+      "epoch": 0.7242437153813379,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.7843821556658646e-06,
+      "loss": 0.496,
+      "step": 8499
+    },
+    {
+      "epoch": 0.7243289305496379,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.784230244576014e-06,
+      "loss": 0.5848,
+      "step": 8500
+    },
+    {
+      "epoch": 0.7244141457179378,
+      "grad_norm": 19.625,
+      "learning_rate": 1.7840783238320856e-06,
+      "loss": 0.7122,
+      "step": 8501
+    },
+    {
+      "epoch": 0.7244993608862378,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7839263934368247e-06,
+      "loss": 0.7983,
+      "step": 8502
+    },
+    {
+      "epoch": 0.7245845760545377,
+      "grad_norm": 28.125,
+      "learning_rate": 1.7837744533929775e-06,
+      "loss": 0.7489,
+      "step": 8503
+    },
+    {
+      "epoch": 0.7246697912228377,
+      "grad_norm": 21.625,
+      "learning_rate": 1.7836225037032887e-06,
+      "loss": 0.7452,
+      "step": 8504
+    },
+    {
+      "epoch": 0.7247550063911377,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7834705443705047e-06,
+      "loss": 0.7934,
+      "step": 8505
+    },
+    {
+      "epoch": 0.7248402215594376,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7833185753973716e-06,
+      "loss": 0.8602,
+      "step": 8506
+    },
+    {
+      "epoch": 0.7249254367277376,
+      "grad_norm": 19.875,
+      "learning_rate": 1.7831665967866358e-06,
+      "loss": 0.9688,
+      "step": 8507
+    },
+    {
+      "epoch": 0.7250106518960375,
+      "grad_norm": 15.375,
+      "learning_rate": 1.7830146085410433e-06,
+      "loss": 0.7071,
+      "step": 8508
+    },
+    {
+      "epoch": 0.7250958670643375,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7828626106633403e-06,
+      "loss": 0.4915,
+      "step": 8509
+    },
+    {
+      "epoch": 0.7251810822326374,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.7827106031562747e-06,
+      "loss": 0.5039,
+      "step": 8510
+    },
+    {
+      "epoch": 0.7252662974009374,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.7825585860225927e-06,
+      "loss": 0.4425,
+      "step": 8511
+    },
+    {
+      "epoch": 0.7253515125692374,
+      "grad_norm": 21.5,
+      "learning_rate": 1.7824065592650417e-06,
+      "loss": 0.7248,
+      "step": 8512
+    },
+    {
+      "epoch": 0.7254367277375373,
+      "grad_norm": 10.875,
+      "learning_rate": 1.7822545228863684e-06,
+      "loss": 0.4146,
+      "step": 8513
+    },
+    {
+      "epoch": 0.7255219429058373,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.7821024768893208e-06,
+      "loss": 0.801,
+      "step": 8514
+    },
+    {
+      "epoch": 0.7256071580741372,
+      "grad_norm": 18.875,
+      "learning_rate": 1.7819504212766463e-06,
+      "loss": 0.8424,
+      "step": 8515
+    },
+    {
+      "epoch": 0.7256923732424372,
+      "grad_norm": 13.125,
+      "learning_rate": 1.7817983560510932e-06,
+      "loss": 0.7092,
+      "step": 8516
+    },
+    {
+      "epoch": 0.7257775884107371,
+      "grad_norm": 10.875,
+      "learning_rate": 1.7816462812154086e-06,
+      "loss": 0.4323,
+      "step": 8517
+    },
+    {
+      "epoch": 0.7258628035790371,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7814941967723413e-06,
+      "loss": 0.8228,
+      "step": 8518
+    },
+    {
+      "epoch": 0.725948018747337,
+      "grad_norm": 14.625,
+      "learning_rate": 1.7813421027246395e-06,
+      "loss": 0.4931,
+      "step": 8519
+    },
+    {
+      "epoch": 0.726033233915637,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7811899990750514e-06,
+      "loss": 0.6817,
+      "step": 8520
+    },
+    {
+      "epoch": 0.726118449083937,
+      "grad_norm": 17.5,
+      "learning_rate": 1.7810378858263256e-06,
+      "loss": 0.7145,
+      "step": 8521
+    },
+    {
+      "epoch": 0.7262036642522369,
+      "grad_norm": 9.5,
+      "learning_rate": 1.7808857629812114e-06,
+      "loss": 0.283,
+      "step": 8522
+    },
+    {
+      "epoch": 0.7262888794205369,
+      "grad_norm": 14.5,
+      "learning_rate": 1.780733630542458e-06,
+      "loss": 0.6382,
+      "step": 8523
+    },
+    {
+      "epoch": 0.7263740945888368,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7805814885128136e-06,
+      "loss": 0.6895,
+      "step": 8524
+    },
+    {
+      "epoch": 0.7264593097571368,
+      "grad_norm": 21.5,
+      "learning_rate": 1.7804293368950286e-06,
+      "loss": 0.9316,
+      "step": 8525
+    },
+    {
+      "epoch": 0.7265445249254368,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7802771756918519e-06,
+      "loss": 0.6778,
+      "step": 8526
+    },
+    {
+      "epoch": 0.7266297400937367,
+      "grad_norm": 13.875,
+      "learning_rate": 1.7801250049060336e-06,
+      "loss": 0.7596,
+      "step": 8527
+    },
+    {
+      "epoch": 0.7267149552620367,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.779972824540323e-06,
+      "loss": 0.7848,
+      "step": 8528
+    },
+    {
+      "epoch": 0.7268001704303366,
+      "grad_norm": 26.625,
+      "learning_rate": 1.7798206345974712e-06,
+      "loss": 1.4553,
+      "step": 8529
+    },
+    {
+      "epoch": 0.7268853855986366,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7796684350802268e-06,
+      "loss": 0.4869,
+      "step": 8530
+    },
+    {
+      "epoch": 0.7269706007669365,
+      "grad_norm": 12.875,
+      "learning_rate": 1.7795162259913421e-06,
+      "loss": 0.5979,
+      "step": 8531
+    },
+    {
+      "epoch": 0.7270558159352365,
+      "grad_norm": 19.5,
+      "learning_rate": 1.779364007333566e-06,
+      "loss": 0.6755,
+      "step": 8532
+    },
+    {
+      "epoch": 0.7271410311035365,
+      "grad_norm": 10.25,
+      "learning_rate": 1.7792117791096506e-06,
+      "loss": 0.4563,
+      "step": 8533
+    },
+    {
+      "epoch": 0.7272262462718364,
+      "grad_norm": 13.875,
+      "learning_rate": 1.7790595413223455e-06,
+      "loss": 0.5445,
+      "step": 8534
+    },
+    {
+      "epoch": 0.7273114614401364,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.778907293974403e-06,
+      "loss": 0.606,
+      "step": 8535
+    },
+    {
+      "epoch": 0.7273966766084363,
+      "grad_norm": 17.875,
+      "learning_rate": 1.7787550370685735e-06,
+      "loss": 0.9824,
+      "step": 8536
+    },
+    {
+      "epoch": 0.7274818917767363,
+      "grad_norm": 14.25,
+      "learning_rate": 1.7786027706076092e-06,
+      "loss": 0.7652,
+      "step": 8537
+    },
+    {
+      "epoch": 0.7275671069450362,
+      "grad_norm": 13.625,
+      "learning_rate": 1.778450494594261e-06,
+      "loss": 0.5231,
+      "step": 8538
+    },
+    {
+      "epoch": 0.7276523221133362,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.778298209031281e-06,
+      "loss": 0.7698,
+      "step": 8539
+    },
+    {
+      "epoch": 0.7277375372816361,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7781459139214213e-06,
+      "loss": 1.0011,
+      "step": 8540
+    },
+    {
+      "epoch": 0.7278227524499361,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7779936092674339e-06,
+      "loss": 0.876,
+      "step": 8541
+    },
+    {
+      "epoch": 0.7279079676182361,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7778412950720704e-06,
+      "loss": 0.54,
+      "step": 8542
+    },
+    {
+      "epoch": 0.727993182786536,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7776889713380844e-06,
+      "loss": 0.5822,
+      "step": 8543
+    },
+    {
+      "epoch": 0.728078397954836,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.777536638068228e-06,
+      "loss": 0.5462,
+      "step": 8544
+    },
+    {
+      "epoch": 0.7281636131231359,
+      "grad_norm": 20.5,
+      "learning_rate": 1.7773842952652539e-06,
+      "loss": 1.0861,
+      "step": 8545
+    },
+    {
+      "epoch": 0.7282488282914359,
+      "grad_norm": 19.375,
+      "learning_rate": 1.7772319429319153e-06,
+      "loss": 1.0924,
+      "step": 8546
+    },
+    {
+      "epoch": 0.7283340434597358,
+      "grad_norm": 9.625,
+      "learning_rate": 1.7770795810709653e-06,
+      "loss": 0.2995,
+      "step": 8547
+    },
+    {
+      "epoch": 0.7284192586280358,
+      "grad_norm": 13.5,
+      "learning_rate": 1.7769272096851578e-06,
+      "loss": 0.7393,
+      "step": 8548
+    },
+    {
+      "epoch": 0.7285044737963358,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.776774828777245e-06,
+      "loss": 0.453,
+      "step": 8549
+    },
+    {
+      "epoch": 0.7285896889646357,
+      "grad_norm": 17.625,
+      "learning_rate": 1.7766224383499819e-06,
+      "loss": 0.7386,
+      "step": 8550
+    },
+    {
+      "epoch": 0.7286749041329357,
+      "grad_norm": 19.0,
+      "learning_rate": 1.7764700384061212e-06,
+      "loss": 0.9498,
+      "step": 8551
+    },
+    {
+      "epoch": 0.7287601193012356,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.7763176289484182e-06,
+      "loss": 0.5819,
+      "step": 8552
+    },
+    {
+      "epoch": 0.7288453344695356,
+      "grad_norm": 12.5,
+      "learning_rate": 1.776165209979626e-06,
+      "loss": 0.6056,
+      "step": 8553
+    },
+    {
+      "epoch": 0.7289305496378355,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.7760127815024994e-06,
+      "loss": 0.3769,
+      "step": 8554
+    },
+    {
+      "epoch": 0.7290157648061355,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7758603435197926e-06,
+      "loss": 0.6686,
+      "step": 8555
+    },
+    {
+      "epoch": 0.7291009799744355,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7757078960342613e-06,
+      "loss": 0.8655,
+      "step": 8556
+    },
+    {
+      "epoch": 0.7291861951427354,
+      "grad_norm": 20.25,
+      "learning_rate": 1.775555439048659e-06,
+      "loss": 0.6658,
+      "step": 8557
+    },
+    {
+      "epoch": 0.7292714103110354,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7754029725657417e-06,
+      "loss": 0.8874,
+      "step": 8558
+    },
+    {
+      "epoch": 0.7293566254793353,
+      "grad_norm": 14.625,
+      "learning_rate": 1.7752504965882643e-06,
+      "loss": 0.6466,
+      "step": 8559
+    },
+    {
+      "epoch": 0.7294418406476353,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7750980111189826e-06,
+      "loss": 0.5399,
+      "step": 8560
+    },
+    {
+      "epoch": 0.7295270558159352,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.7749455161606513e-06,
+      "loss": 0.4239,
+      "step": 8561
+    },
+    {
+      "epoch": 0.7296122709842352,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7747930117160267e-06,
+      "loss": 0.2861,
+      "step": 8562
+    },
+    {
+      "epoch": 0.7296974861525352,
+      "grad_norm": 11.625,
+      "learning_rate": 1.7746404977878651e-06,
+      "loss": 0.419,
+      "step": 8563
+    },
+    {
+      "epoch": 0.7297827013208351,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7744879743789218e-06,
+      "loss": 0.7734,
+      "step": 8564
+    },
+    {
+      "epoch": 0.7298679164891351,
+      "grad_norm": 14.375,
+      "learning_rate": 1.7743354414919534e-06,
+      "loss": 0.8239,
+      "step": 8565
+    },
+    {
+      "epoch": 0.729953131657435,
+      "grad_norm": 18.375,
+      "learning_rate": 1.7741828991297161e-06,
+      "loss": 0.8964,
+      "step": 8566
+    },
+    {
+      "epoch": 0.730038346825735,
+      "grad_norm": 12.75,
+      "learning_rate": 1.774030347294967e-06,
+      "loss": 0.5365,
+      "step": 8567
+    },
+    {
+      "epoch": 0.7301235619940349,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.7738777859904626e-06,
+      "loss": 0.5052,
+      "step": 8568
+    },
+    {
+      "epoch": 0.7302087771623349,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7737252152189599e-06,
+      "loss": 0.4295,
+      "step": 8569
+    },
+    {
+      "epoch": 0.7302939923306349,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7735726349832156e-06,
+      "loss": 0.4678,
+      "step": 8570
+    },
+    {
+      "epoch": 0.7303792074989348,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.7734200452859876e-06,
+      "loss": 0.4154,
+      "step": 8571
+    },
+    {
+      "epoch": 0.7304644226672348,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7732674461300328e-06,
+      "loss": 0.5816,
+      "step": 8572
+    },
+    {
+      "epoch": 0.7305496378355347,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.7731148375181092e-06,
+      "loss": 0.4047,
+      "step": 8573
+    },
+    {
+      "epoch": 0.7306348530038347,
+      "grad_norm": 18.875,
+      "learning_rate": 1.7729622194529744e-06,
+      "loss": 1.0405,
+      "step": 8574
+    },
+    {
+      "epoch": 0.7307200681721346,
+      "grad_norm": 21.875,
+      "learning_rate": 1.7728095919373863e-06,
+      "loss": 0.8328,
+      "step": 8575
+    },
+    {
+      "epoch": 0.7308052833404346,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7726569549741035e-06,
+      "loss": 0.5895,
+      "step": 8576
+    },
+    {
+      "epoch": 0.7308904985087346,
+      "grad_norm": 24.0,
+      "learning_rate": 1.7725043085658834e-06,
+      "loss": 0.8665,
+      "step": 8577
+    },
+    {
+      "epoch": 0.7309757136770345,
+      "grad_norm": 11.625,
+      "learning_rate": 1.7723516527154855e-06,
+      "loss": 0.6448,
+      "step": 8578
+    },
+    {
+      "epoch": 0.7310609288453345,
+      "grad_norm": 18.5,
+      "learning_rate": 1.772198987425668e-06,
+      "loss": 0.6823,
+      "step": 8579
+    },
+    {
+      "epoch": 0.7311461440136344,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.7720463126991893e-06,
+      "loss": 0.8641,
+      "step": 8580
+    },
+    {
+      "epoch": 0.7312313591819344,
+      "grad_norm": 29.125,
+      "learning_rate": 1.7718936285388088e-06,
+      "loss": 0.8062,
+      "step": 8581
+    },
+    {
+      "epoch": 0.7313165743502343,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.7717409349472858e-06,
+      "loss": 0.6037,
+      "step": 8582
+    },
+    {
+      "epoch": 0.7314017895185343,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7715882319273794e-06,
+      "loss": 1.1658,
+      "step": 8583
+    },
+    {
+      "epoch": 0.7314870046868343,
+      "grad_norm": 20.75,
+      "learning_rate": 1.7714355194818494e-06,
+      "loss": 1.082,
+      "step": 8584
+    },
+    {
+      "epoch": 0.7315722198551342,
+      "grad_norm": 16.0,
+      "learning_rate": 1.771282797613455e-06,
+      "loss": 0.9383,
+      "step": 8585
+    },
+    {
+      "epoch": 0.7316574350234342,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7711300663249564e-06,
+      "loss": 0.78,
+      "step": 8586
+    },
+    {
+      "epoch": 0.7317426501917341,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7709773256191132e-06,
+      "loss": 0.757,
+      "step": 8587
+    },
+    {
+      "epoch": 0.7318278653600341,
+      "grad_norm": 17.75,
+      "learning_rate": 1.7708245754986861e-06,
+      "loss": 0.9254,
+      "step": 8588
+    },
+    {
+      "epoch": 0.731913080528334,
+      "grad_norm": 17.875,
+      "learning_rate": 1.770671815966435e-06,
+      "loss": 0.835,
+      "step": 8589
+    },
+    {
+      "epoch": 0.731998295696634,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.7705190470251207e-06,
+      "loss": 0.7231,
+      "step": 8590
+    },
+    {
+      "epoch": 0.732083510864934,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7703662686775036e-06,
+      "loss": 1.0814,
+      "step": 8591
+    },
+    {
+      "epoch": 0.7321687260332339,
+      "grad_norm": 21.75,
+      "learning_rate": 1.7702134809263449e-06,
+      "loss": 0.9058,
+      "step": 8592
+    },
+    {
+      "epoch": 0.7322539412015339,
+      "grad_norm": 19.875,
+      "learning_rate": 1.7700606837744056e-06,
+      "loss": 0.6174,
+      "step": 8593
+    },
+    {
+      "epoch": 0.7323391563698338,
+      "grad_norm": 14.75,
+      "learning_rate": 1.7699078772244467e-06,
+      "loss": 0.9005,
+      "step": 8594
+    },
+    {
+      "epoch": 0.7324243715381338,
+      "grad_norm": 11.25,
+      "learning_rate": 1.7697550612792296e-06,
+      "loss": 0.4177,
+      "step": 8595
+    },
+    {
+      "epoch": 0.7325095867064337,
+      "grad_norm": 11.125,
+      "learning_rate": 1.769602235941516e-06,
+      "loss": 0.4524,
+      "step": 8596
+    },
+    {
+      "epoch": 0.7325948018747337,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.7694494012140676e-06,
+      "loss": 0.7261,
+      "step": 8597
+    },
+    {
+      "epoch": 0.7326800170430336,
+      "grad_norm": 13.0,
+      "learning_rate": 1.7692965570996456e-06,
+      "loss": 0.5186,
+      "step": 8598
+    },
+    {
+      "epoch": 0.7327652322113336,
+      "grad_norm": 18.625,
+      "learning_rate": 1.7691437036010135e-06,
+      "loss": 0.4464,
+      "step": 8599
+    },
+    {
+      "epoch": 0.7328504473796336,
+      "grad_norm": 22.125,
+      "learning_rate": 1.7689908407209321e-06,
+      "loss": 0.9657,
+      "step": 8600
+    },
+    {
+      "epoch": 0.7329356625479335,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7688379684621642e-06,
+      "loss": 0.6473,
+      "step": 8601
+    },
+    {
+      "epoch": 0.7330208777162335,
+      "grad_norm": 19.625,
+      "learning_rate": 1.7686850868274726e-06,
+      "loss": 0.6331,
+      "step": 8602
+    },
+    {
+      "epoch": 0.7331060928845334,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.76853219581962e-06,
+      "loss": 0.5246,
+      "step": 8603
+    },
+    {
+      "epoch": 0.7331913080528334,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7683792954413694e-06,
+      "loss": 0.5925,
+      "step": 8604
+    },
+    {
+      "epoch": 0.7332765232211333,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7682263856954835e-06,
+      "loss": 0.9027,
+      "step": 8605
+    },
+    {
+      "epoch": 0.7333617383894333,
+      "grad_norm": 21.875,
+      "learning_rate": 1.7680734665847254e-06,
+      "loss": 0.8009,
+      "step": 8606
+    },
+    {
+      "epoch": 0.7334469535577333,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7679205381118592e-06,
+      "loss": 0.5548,
+      "step": 8607
+    },
+    {
+      "epoch": 0.7335321687260332,
+      "grad_norm": 17.75,
+      "learning_rate": 1.767767600279648e-06,
+      "loss": 1.0133,
+      "step": 8608
+    },
+    {
+      "epoch": 0.7336173838943332,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7676146530908552e-06,
+      "loss": 0.9443,
+      "step": 8609
+    },
+    {
+      "epoch": 0.7337025990626331,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7674616965482454e-06,
+      "loss": 0.7443,
+      "step": 8610
+    },
+    {
+      "epoch": 0.7337878142309331,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.7673087306545825e-06,
+      "loss": 0.5754,
+      "step": 8611
+    },
+    {
+      "epoch": 0.733873029399233,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.7671557554126303e-06,
+      "loss": 0.6166,
+      "step": 8612
+    },
+    {
+      "epoch": 0.733958244567533,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7670027708251542e-06,
+      "loss": 0.6266,
+      "step": 8613
+    },
+    {
+      "epoch": 0.734043459735833,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7668497768949177e-06,
+      "loss": 0.7187,
+      "step": 8614
+    },
+    {
+      "epoch": 0.7341286749041329,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.766696773624686e-06,
+      "loss": 0.4585,
+      "step": 8615
+    },
+    {
+      "epoch": 0.7342138900724329,
+      "grad_norm": 19.375,
+      "learning_rate": 1.766543761017224e-06,
+      "loss": 0.441,
+      "step": 8616
+    },
+    {
+      "epoch": 0.7342991052407328,
+      "grad_norm": 18.375,
+      "learning_rate": 1.766390739075297e-06,
+      "loss": 0.8743,
+      "step": 8617
+    },
+    {
+      "epoch": 0.7343843204090328,
+      "grad_norm": 16.75,
+      "learning_rate": 1.76623770780167e-06,
+      "loss": 0.5088,
+      "step": 8618
+    },
+    {
+      "epoch": 0.7344695355773327,
+      "grad_norm": 23.25,
+      "learning_rate": 1.7660846671991083e-06,
+      "loss": 1.2049,
+      "step": 8619
+    },
+    {
+      "epoch": 0.7345547507456327,
+      "grad_norm": 21.0,
+      "learning_rate": 1.7659316172703777e-06,
+      "loss": 0.7994,
+      "step": 8620
+    },
+    {
+      "epoch": 0.7346399659139327,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.7657785580182443e-06,
+      "loss": 0.7801,
+      "step": 8621
+    },
+    {
+      "epoch": 0.7347251810822326,
+      "grad_norm": 18.5,
+      "learning_rate": 1.7656254894454736e-06,
+      "loss": 0.7233,
+      "step": 8622
+    },
+    {
+      "epoch": 0.7348103962505326,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7654724115548316e-06,
+      "loss": 0.3936,
+      "step": 8623
+    },
+    {
+      "epoch": 0.7348956114188325,
+      "grad_norm": 14.625,
+      "learning_rate": 1.7653193243490852e-06,
+      "loss": 1.1214,
+      "step": 8624
+    },
+    {
+      "epoch": 0.7349808265871325,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.765166227831e-06,
+      "loss": 0.5206,
+      "step": 8625
+    },
+    {
+      "epoch": 0.7350660417554324,
+      "grad_norm": 13.625,
+      "learning_rate": 1.765013122003343e-06,
+      "loss": 0.733,
+      "step": 8626
+    },
+    {
+      "epoch": 0.7351512569237324,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.7648600068688811e-06,
+      "loss": 0.7921,
+      "step": 8627
+    },
+    {
+      "epoch": 0.7352364720920324,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7647068824303815e-06,
+      "loss": 0.8531,
+      "step": 8628
+    },
+    {
+      "epoch": 0.7353216872603323,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7645537486906107e-06,
+      "loss": 0.6251,
+      "step": 8629
+    },
+    {
+      "epoch": 0.7354069024286323,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7644006056523358e-06,
+      "loss": 0.9287,
+      "step": 8630
+    },
+    {
+      "epoch": 0.7354921175969322,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7642474533183252e-06,
+      "loss": 0.6049,
+      "step": 8631
+    },
+    {
+      "epoch": 0.7355773327652322,
+      "grad_norm": 19.375,
+      "learning_rate": 1.7640942916913457e-06,
+      "loss": 0.8854,
+      "step": 8632
+    },
+    {
+      "epoch": 0.7356625479335321,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.7639411207741657e-06,
+      "loss": 0.3173,
+      "step": 8633
+    },
+    {
+      "epoch": 0.7357477631018321,
+      "grad_norm": 12.5,
+      "learning_rate": 1.7637879405695526e-06,
+      "loss": 0.4365,
+      "step": 8634
+    },
+    {
+      "epoch": 0.735832978270132,
+      "grad_norm": 20.625,
+      "learning_rate": 1.7636347510802751e-06,
+      "loss": 0.729,
+      "step": 8635
+    },
+    {
+      "epoch": 0.735918193438432,
+      "grad_norm": 13.625,
+      "learning_rate": 1.7634815523091007e-06,
+      "loss": 0.9007,
+      "step": 8636
+    },
+    {
+      "epoch": 0.736003408606732,
+      "grad_norm": 21.375,
+      "learning_rate": 1.7633283442587989e-06,
+      "loss": 0.8828,
+      "step": 8637
+    },
+    {
+      "epoch": 0.7360886237750319,
+      "grad_norm": 13.375,
+      "learning_rate": 1.763175126932137e-06,
+      "loss": 0.5454,
+      "step": 8638
+    },
+    {
+      "epoch": 0.7361738389433319,
+      "grad_norm": 16.625,
+      "learning_rate": 1.763021900331885e-06,
+      "loss": 0.8719,
+      "step": 8639
+    },
+    {
+      "epoch": 0.7362590541116318,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7628686644608113e-06,
+      "loss": 0.5728,
+      "step": 8640
+    },
+    {
+      "epoch": 0.7363442692799318,
+      "grad_norm": 17.375,
+      "learning_rate": 1.762715419321685e-06,
+      "loss": 0.9567,
+      "step": 8641
+    },
+    {
+      "epoch": 0.7364294844482318,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7625621649172755e-06,
+      "loss": 0.7863,
+      "step": 8642
+    },
+    {
+      "epoch": 0.7365146996165317,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7624089012503523e-06,
+      "loss": 0.7025,
+      "step": 8643
+    },
+    {
+      "epoch": 0.7365999147848317,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.762255628323685e-06,
+      "loss": 0.7628,
+      "step": 8644
+    },
+    {
+      "epoch": 0.7366851299531316,
+      "grad_norm": 10.625,
+      "learning_rate": 1.7621023461400432e-06,
+      "loss": 0.3108,
+      "step": 8645
+    },
+    {
+      "epoch": 0.7367703451214316,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7619490547021976e-06,
+      "loss": 0.6916,
+      "step": 8646
+    },
+    {
+      "epoch": 0.7368555602897315,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.7617957540129171e-06,
+      "loss": 0.4654,
+      "step": 8647
+    },
+    {
+      "epoch": 0.7369407754580315,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.761642444074973e-06,
+      "loss": 0.7203,
+      "step": 8648
+    },
+    {
+      "epoch": 0.7370259906263315,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.7614891248911353e-06,
+      "loss": 0.4617,
+      "step": 8649
+    },
+    {
+      "epoch": 0.7371112057946314,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7613357964641753e-06,
+      "loss": 0.5827,
+      "step": 8650
+    },
+    {
+      "epoch": 0.7371964209629314,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7611824587968624e-06,
+      "loss": 0.7434,
+      "step": 8651
+    },
+    {
+      "epoch": 0.7372816361312313,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.7610291118919689e-06,
+      "loss": 0.3914,
+      "step": 8652
+    },
+    {
+      "epoch": 0.7373668512995313,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7608757557522656e-06,
+      "loss": 0.6868,
+      "step": 8653
+    },
+    {
+      "epoch": 0.7374520664678312,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7607223903805237e-06,
+      "loss": 0.6665,
+      "step": 8654
+    },
+    {
+      "epoch": 0.7375372816361312,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7605690157795141e-06,
+      "loss": 0.657,
+      "step": 8655
+    },
+    {
+      "epoch": 0.7376224968044311,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7604156319520097e-06,
+      "loss": 1.1056,
+      "step": 8656
+    },
+    {
+      "epoch": 0.7377077119727311,
+      "grad_norm": 19.375,
+      "learning_rate": 1.7602622389007811e-06,
+      "loss": 0.8459,
+      "step": 8657
+    },
+    {
+      "epoch": 0.7377929271410311,
+      "grad_norm": 16.875,
+      "learning_rate": 1.760108836628601e-06,
+      "loss": 0.756,
+      "step": 8658
+    },
+    {
+      "epoch": 0.737878142309331,
+      "grad_norm": 12.125,
+      "learning_rate": 1.759955425138241e-06,
+      "loss": 0.5786,
+      "step": 8659
+    },
+    {
+      "epoch": 0.737963357477631,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7598020044324738e-06,
+      "loss": 1.063,
+      "step": 8660
+    },
+    {
+      "epoch": 0.7380485726459309,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.7596485745140718e-06,
+      "loss": 0.6755,
+      "step": 8661
+    },
+    {
+      "epoch": 0.7381337878142309,
+      "grad_norm": 15.375,
+      "learning_rate": 1.7594951353858069e-06,
+      "loss": 0.6532,
+      "step": 8662
+    },
+    {
+      "epoch": 0.7382190029825308,
+      "grad_norm": 15.25,
+      "learning_rate": 1.7593416870504535e-06,
+      "loss": 0.7708,
+      "step": 8663
+    },
+    {
+      "epoch": 0.7383042181508308,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.759188229510783e-06,
+      "loss": 0.7735,
+      "step": 8664
+    },
+    {
+      "epoch": 0.7383894333191308,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7590347627695698e-06,
+      "loss": 0.6911,
+      "step": 8665
+    },
+    {
+      "epoch": 0.7384746484874307,
+      "grad_norm": 12.875,
+      "learning_rate": 1.758881286829586e-06,
+      "loss": 0.6542,
+      "step": 8666
+    },
+    {
+      "epoch": 0.7385598636557307,
+      "grad_norm": 16.875,
+      "learning_rate": 1.758727801693606e-06,
+      "loss": 0.9329,
+      "step": 8667
+    },
+    {
+      "epoch": 0.7386450788240306,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7585743073644027e-06,
+      "loss": 1.1321,
+      "step": 8668
+    },
+    {
+      "epoch": 0.7387302939923306,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7584208038447506e-06,
+      "loss": 0.5965,
+      "step": 8669
+    },
+    {
+      "epoch": 0.7388155091606305,
+      "grad_norm": 17.75,
+      "learning_rate": 1.758267291137423e-06,
+      "loss": 1.1661,
+      "step": 8670
+    },
+    {
+      "epoch": 0.7389007243289305,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7581137692451945e-06,
+      "loss": 0.5061,
+      "step": 8671
+    },
+    {
+      "epoch": 0.7389859394972305,
+      "grad_norm": 16.125,
+      "learning_rate": 1.757960238170839e-06,
+      "loss": 0.3777,
+      "step": 8672
+    },
+    {
+      "epoch": 0.7390711546655304,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7578066979171319e-06,
+      "loss": 1.1345,
+      "step": 8673
+    },
+    {
+      "epoch": 0.7391563698338304,
+      "grad_norm": 22.875,
+      "learning_rate": 1.7576531484868464e-06,
+      "loss": 0.7862,
+      "step": 8674
+    },
+    {
+      "epoch": 0.7392415850021303,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7574995898827585e-06,
+      "loss": 0.734,
+      "step": 8675
+    },
+    {
+      "epoch": 0.7393268001704303,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7573460221076422e-06,
+      "loss": 0.7169,
+      "step": 8676
+    },
+    {
+      "epoch": 0.7394120153387302,
+      "grad_norm": 17.75,
+      "learning_rate": 1.7571924451642735e-06,
+      "loss": 0.903,
+      "step": 8677
+    },
+    {
+      "epoch": 0.7394972305070302,
+      "grad_norm": 19.625,
+      "learning_rate": 1.7570388590554272e-06,
+      "loss": 1.1479,
+      "step": 8678
+    },
+    {
+      "epoch": 0.7395824456753302,
+      "grad_norm": 32.5,
+      "learning_rate": 1.756885263783879e-06,
+      "loss": 1.3808,
+      "step": 8679
+    },
+    {
+      "epoch": 0.7396676608436301,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7567316593524038e-06,
+      "loss": 0.6421,
+      "step": 8680
+    },
+    {
+      "epoch": 0.7397528760119301,
+      "grad_norm": 10.375,
+      "learning_rate": 1.7565780457637784e-06,
+      "loss": 0.4352,
+      "step": 8681
+    },
+    {
+      "epoch": 0.73983809118023,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7564244230207786e-06,
+      "loss": 0.9484,
+      "step": 8682
+    },
+    {
+      "epoch": 0.73992330634853,
+      "grad_norm": 18.125,
+      "learning_rate": 1.7562707911261796e-06,
+      "loss": 0.742,
+      "step": 8683
+    },
+    {
+      "epoch": 0.7400085215168299,
+      "grad_norm": 20.25,
+      "learning_rate": 1.7561171500827587e-06,
+      "loss": 1.0956,
+      "step": 8684
+    },
+    {
+      "epoch": 0.7400937366851299,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7559634998932918e-06,
+      "loss": 0.6639,
+      "step": 8685
+    },
+    {
+      "epoch": 0.7401789518534299,
+      "grad_norm": 13.25,
+      "learning_rate": 1.755809840560556e-06,
+      "loss": 0.6339,
+      "step": 8686
+    },
+    {
+      "epoch": 0.7402641670217298,
+      "grad_norm": 11.0,
+      "learning_rate": 1.7556561720873272e-06,
+      "loss": 0.3729,
+      "step": 8687
+    },
+    {
+      "epoch": 0.7403493821900298,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.7555024944763833e-06,
+      "loss": 0.6727,
+      "step": 8688
+    },
+    {
+      "epoch": 0.7404345973583298,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7553488077305006e-06,
+      "loss": 0.7158,
+      "step": 8689
+    },
+    {
+      "epoch": 0.7405198125266298,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.7551951118524574e-06,
+      "loss": 0.6316,
+      "step": 8690
+    },
+    {
+      "epoch": 0.7406050276949298,
+      "grad_norm": 15.875,
+      "learning_rate": 1.75504140684503e-06,
+      "loss": 1.0444,
+      "step": 8691
+    },
+    {
+      "epoch": 0.7406902428632297,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7548876927109967e-06,
+      "loss": 0.5795,
+      "step": 8692
+    },
+    {
+      "epoch": 0.7407754580315297,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7547339694531349e-06,
+      "loss": 0.7421,
+      "step": 8693
+    },
+    {
+      "epoch": 0.7408606731998296,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7545802370742233e-06,
+      "loss": 0.7057,
+      "step": 8694
+    },
+    {
+      "epoch": 0.7409458883681296,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.7544264955770388e-06,
+      "loss": 0.491,
+      "step": 8695
+    },
+    {
+      "epoch": 0.7410311035364295,
+      "grad_norm": 41.75,
+      "learning_rate": 1.7542727449643605e-06,
+      "loss": 1.3983,
+      "step": 8696
+    },
+    {
+      "epoch": 0.7411163187047295,
+      "grad_norm": 33.5,
+      "learning_rate": 1.7541189852389664e-06,
+      "loss": 1.3734,
+      "step": 8697
+    },
+    {
+      "epoch": 0.7412015338730295,
+      "grad_norm": 17.5,
+      "learning_rate": 1.7539652164036359e-06,
+      "loss": 0.895,
+      "step": 8698
+    },
+    {
+      "epoch": 0.7412867490413294,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7538114384611466e-06,
+      "loss": 0.7135,
+      "step": 8699
+    },
+    {
+      "epoch": 0.7413719642096294,
+      "grad_norm": 10.0,
+      "learning_rate": 1.7536576514142781e-06,
+      "loss": 0.6409,
+      "step": 8700
+    },
+    {
+      "epoch": 0.7414571793779293,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7535038552658097e-06,
+      "loss": 0.7891,
+      "step": 8701
+    },
+    {
+      "epoch": 0.7415423945462293,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7533500500185198e-06,
+      "loss": 0.5417,
+      "step": 8702
+    },
+    {
+      "epoch": 0.7416276097145292,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7531962356751886e-06,
+      "loss": 0.7711,
+      "step": 8703
+    },
+    {
+      "epoch": 0.7417128248828292,
+      "grad_norm": 18.625,
+      "learning_rate": 1.7530424122385953e-06,
+      "loss": 0.7421,
+      "step": 8704
+    },
+    {
+      "epoch": 0.7417980400511291,
+      "grad_norm": 24.75,
+      "learning_rate": 1.75288857971152e-06,
+      "loss": 0.8814,
+      "step": 8705
+    },
+    {
+      "epoch": 0.7418832552194291,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.7527347380967424e-06,
+      "loss": 0.5617,
+      "step": 8706
+    },
+    {
+      "epoch": 0.7419684703877291,
+      "grad_norm": 14.625,
+      "learning_rate": 1.7525808873970424e-06,
+      "loss": 0.8583,
+      "step": 8707
+    },
+    {
+      "epoch": 0.742053685556029,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.7524270276152003e-06,
+      "loss": 0.6855,
+      "step": 8708
+    },
+    {
+      "epoch": 0.742138900724329,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.7522731587539967e-06,
+      "loss": 0.2737,
+      "step": 8709
+    },
+    {
+      "epoch": 0.7422241158926289,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7521192808162121e-06,
+      "loss": 0.4955,
+      "step": 8710
+    },
+    {
+      "epoch": 0.7423093310609289,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7519653938046271e-06,
+      "loss": 0.8397,
+      "step": 8711
+    },
+    {
+      "epoch": 0.7423945462292288,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7518114977220223e-06,
+      "loss": 1.0708,
+      "step": 8712
+    },
+    {
+      "epoch": 0.7424797613975288,
+      "grad_norm": 14.5,
+      "learning_rate": 1.75165759257118e-06,
+      "loss": 0.9256,
+      "step": 8713
+    },
+    {
+      "epoch": 0.7425649765658288,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.75150367835488e-06,
+      "loss": 0.8408,
+      "step": 8714
+    },
+    {
+      "epoch": 0.7426501917341287,
+      "grad_norm": 20.0,
+      "learning_rate": 1.751349755075904e-06,
+      "loss": 1.176,
+      "step": 8715
+    },
+    {
+      "epoch": 0.7427354069024287,
+      "grad_norm": 19.875,
+      "learning_rate": 1.751195822737034e-06,
+      "loss": 0.818,
+      "step": 8716
+    },
+    {
+      "epoch": 0.7428206220707286,
+      "grad_norm": 24.125,
+      "learning_rate": 1.7510418813410513e-06,
+      "loss": 1.343,
+      "step": 8717
+    },
+    {
+      "epoch": 0.7429058372390286,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7508879308907387e-06,
+      "loss": 0.3106,
+      "step": 8718
+    },
+    {
+      "epoch": 0.7429910524073285,
+      "grad_norm": 14.125,
+      "learning_rate": 1.7507339713888765e-06,
+      "loss": 0.6693,
+      "step": 8719
+    },
+    {
+      "epoch": 0.7430762675756285,
+      "grad_norm": 23.75,
+      "learning_rate": 1.7505800028382485e-06,
+      "loss": 1.0575,
+      "step": 8720
+    },
+    {
+      "epoch": 0.7431614827439285,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7504260252416363e-06,
+      "loss": 0.7606,
+      "step": 8721
+    },
+    {
+      "epoch": 0.7432466979122284,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7502720386018233e-06,
+      "loss": 0.8169,
+      "step": 8722
+    },
+    {
+      "epoch": 0.7433319130805284,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7501180429215905e-06,
+      "loss": 0.9029,
+      "step": 8723
+    },
+    {
+      "epoch": 0.7434171282488283,
+      "grad_norm": 17.875,
+      "learning_rate": 1.7499640382037222e-06,
+      "loss": 0.7621,
+      "step": 8724
+    },
+    {
+      "epoch": 0.7435023434171283,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.7498100244510007e-06,
+      "loss": 0.9402,
+      "step": 8725
+    },
+    {
+      "epoch": 0.7435875585854282,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.7496560016662102e-06,
+      "loss": 1.0549,
+      "step": 8726
+    },
+    {
+      "epoch": 0.7436727737537282,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7495019698521327e-06,
+      "loss": 1.1288,
+      "step": 8727
+    },
+    {
+      "epoch": 0.7437579889220282,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7493479290115523e-06,
+      "loss": 0.5801,
+      "step": 8728
+    },
+    {
+      "epoch": 0.7438432040903281,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7491938791472527e-06,
+      "loss": 0.3154,
+      "step": 8729
+    },
+    {
+      "epoch": 0.7439284192586281,
+      "grad_norm": 11.25,
+      "learning_rate": 1.749039820262018e-06,
+      "loss": 0.5015,
+      "step": 8730
+    },
+    {
+      "epoch": 0.744013634426928,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7488857523586317e-06,
+      "loss": 0.6858,
+      "step": 8731
+    },
+    {
+      "epoch": 0.744098849595228,
+      "grad_norm": 21.375,
+      "learning_rate": 1.748731675439878e-06,
+      "loss": 0.5715,
+      "step": 8732
+    },
+    {
+      "epoch": 0.7441840647635279,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7485775895085416e-06,
+      "loss": 0.4485,
+      "step": 8733
+    },
+    {
+      "epoch": 0.7442692799318279,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.7484234945674072e-06,
+      "loss": 0.5973,
+      "step": 8734
+    },
+    {
+      "epoch": 0.7443544951001279,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7482693906192588e-06,
+      "loss": 0.495,
+      "step": 8735
+    },
+    {
+      "epoch": 0.7444397102684278,
+      "grad_norm": 11.625,
+      "learning_rate": 1.7481152776668812e-06,
+      "loss": 1.3128,
+      "step": 8736
+    },
+    {
+      "epoch": 0.7445249254367278,
+      "grad_norm": 14.75,
+      "learning_rate": 1.74796115571306e-06,
+      "loss": 0.7134,
+      "step": 8737
+    },
+    {
+      "epoch": 0.7446101406050277,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7478070247605798e-06,
+      "loss": 0.6541,
+      "step": 8738
+    },
+    {
+      "epoch": 0.7446953557733277,
+      "grad_norm": 21.125,
+      "learning_rate": 1.747652884812226e-06,
+      "loss": 0.7929,
+      "step": 8739
+    },
+    {
+      "epoch": 0.7447805709416276,
+      "grad_norm": 35.25,
+      "learning_rate": 1.747498735870784e-06,
+      "loss": 1.3412,
+      "step": 8740
+    },
+    {
+      "epoch": 0.7448657861099276,
+      "grad_norm": 20.875,
+      "learning_rate": 1.7473445779390396e-06,
+      "loss": 1.2407,
+      "step": 8741
+    },
+    {
+      "epoch": 0.7449510012782276,
+      "grad_norm": 16.75,
+      "learning_rate": 1.747190411019779e-06,
+      "loss": 0.9036,
+      "step": 8742
+    },
+    {
+      "epoch": 0.7450362164465275,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7470362351157875e-06,
+      "loss": 0.8357,
+      "step": 8743
+    },
+    {
+      "epoch": 0.7451214316148275,
+      "grad_norm": 18.625,
+      "learning_rate": 1.7468820502298511e-06,
+      "loss": 0.9005,
+      "step": 8744
+    },
+    {
+      "epoch": 0.7452066467831274,
+      "grad_norm": 12.625,
+      "learning_rate": 1.746727856364757e-06,
+      "loss": 0.5742,
+      "step": 8745
+    },
+    {
+      "epoch": 0.7452918619514274,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7465736535232901e-06,
+      "loss": 0.7629,
+      "step": 8746
+    },
+    {
+      "epoch": 0.7453770771197273,
+      "grad_norm": 16.75,
+      "learning_rate": 1.746419441708239e-06,
+      "loss": 0.8716,
+      "step": 8747
+    },
+    {
+      "epoch": 0.7454622922880273,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7462652209223888e-06,
+      "loss": 0.5686,
+      "step": 8748
+    },
+    {
+      "epoch": 0.7455475074563273,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7461109911685272e-06,
+      "loss": 0.6032,
+      "step": 8749
+    },
+    {
+      "epoch": 0.7456327226246272,
+      "grad_norm": 11.0,
+      "learning_rate": 1.7459567524494409e-06,
+      "loss": 0.3629,
+      "step": 8750
+    },
+    {
+      "epoch": 0.7457179377929272,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7458025047679178e-06,
+      "loss": 0.6784,
+      "step": 8751
+    },
+    {
+      "epoch": 0.7458031529612271,
+      "grad_norm": 26.25,
+      "learning_rate": 1.7456482481267445e-06,
+      "loss": 0.576,
+      "step": 8752
+    },
+    {
+      "epoch": 0.7458883681295271,
+      "grad_norm": 25.0,
+      "learning_rate": 1.7454939825287087e-06,
+      "loss": 0.7948,
+      "step": 8753
+    },
+    {
+      "epoch": 0.745973583297827,
+      "grad_norm": 18.625,
+      "learning_rate": 1.7453397079765991e-06,
+      "loss": 0.8094,
+      "step": 8754
+    },
+    {
+      "epoch": 0.746058798466127,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7451854244732025e-06,
+      "loss": 0.8838,
+      "step": 8755
+    },
+    {
+      "epoch": 0.746144013634427,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7450311320213074e-06,
+      "loss": 0.779,
+      "step": 8756
+    },
+    {
+      "epoch": 0.7462292288027269,
+      "grad_norm": 19.25,
+      "learning_rate": 1.7448768306237015e-06,
+      "loss": 0.7891,
+      "step": 8757
+    },
+    {
+      "epoch": 0.7463144439710269,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.744722520283174e-06,
+      "loss": 0.9333,
+      "step": 8758
+    },
+    {
+      "epoch": 0.7463996591393268,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7445682010025133e-06,
+      "loss": 0.6183,
+      "step": 8759
+    },
+    {
+      "epoch": 0.7464848743076268,
+      "grad_norm": 12.125,
+      "learning_rate": 1.7444138727845078e-06,
+      "loss": 0.3199,
+      "step": 8760
+    },
+    {
+      "epoch": 0.7465700894759267,
+      "grad_norm": 13.125,
+      "learning_rate": 1.7442595356319461e-06,
+      "loss": 0.4685,
+      "step": 8761
+    },
+    {
+      "epoch": 0.7466553046442267,
+      "grad_norm": 18.25,
+      "learning_rate": 1.744105189547618e-06,
+      "loss": 0.9534,
+      "step": 8762
+    },
+    {
+      "epoch": 0.7467405198125266,
+      "grad_norm": 14.5,
+      "learning_rate": 1.743950834534312e-06,
+      "loss": 0.7155,
+      "step": 8763
+    },
+    {
+      "epoch": 0.7468257349808266,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.743796470594818e-06,
+      "loss": 0.6915,
+      "step": 8764
+    },
+    {
+      "epoch": 0.7469109501491266,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7436420977319246e-06,
+      "loss": 0.9821,
+      "step": 8765
+    },
+    {
+      "epoch": 0.7469961653174265,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7434877159484225e-06,
+      "loss": 0.763,
+      "step": 8766
+    },
+    {
+      "epoch": 0.7470813804857265,
+      "grad_norm": 17.625,
+      "learning_rate": 1.7433333252471013e-06,
+      "loss": 0.6521,
+      "step": 8767
+    },
+    {
+      "epoch": 0.7471665956540264,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7431789256307505e-06,
+      "loss": 0.7515,
+      "step": 8768
+    },
+    {
+      "epoch": 0.7472518108223264,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7430245171021608e-06,
+      "loss": 0.3938,
+      "step": 8769
+    },
+    {
+      "epoch": 0.7473370259906263,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.742870099664122e-06,
+      "loss": 0.9272,
+      "step": 8770
+    },
+    {
+      "epoch": 0.7474222411589263,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.7427156733194255e-06,
+      "loss": 0.4833,
+      "step": 8771
+    },
+    {
+      "epoch": 0.7475074563272263,
+      "grad_norm": 15.5,
+      "learning_rate": 1.742561238070861e-06,
+      "loss": 0.8576,
+      "step": 8772
+    },
+    {
+      "epoch": 0.7475926714955262,
+      "grad_norm": 18.375,
+      "learning_rate": 1.7424067939212195e-06,
+      "loss": 0.9216,
+      "step": 8773
+    },
+    {
+      "epoch": 0.7476778866638262,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.742252340873292e-06,
+      "loss": 0.6485,
+      "step": 8774
+    },
+    {
+      "epoch": 0.7477631018321261,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7420978789298702e-06,
+      "loss": 0.8847,
+      "step": 8775
+    },
+    {
+      "epoch": 0.7478483170004261,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7419434080937442e-06,
+      "loss": 0.508,
+      "step": 8776
+    },
+    {
+      "epoch": 0.747933532168726,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7417889283677066e-06,
+      "loss": 1.007,
+      "step": 8777
+    },
+    {
+      "epoch": 0.748018747337026,
+      "grad_norm": 10.25,
+      "learning_rate": 1.7416344397545482e-06,
+      "loss": 0.291,
+      "step": 8778
+    },
+    {
+      "epoch": 0.748103962505326,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7414799422570616e-06,
+      "loss": 0.864,
+      "step": 8779
+    },
+    {
+      "epoch": 0.7481891776736259,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7413254358780377e-06,
+      "loss": 0.9047,
+      "step": 8780
+    },
+    {
+      "epoch": 0.7482743928419259,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.7411709206202692e-06,
+      "loss": 0.646,
+      "step": 8781
+    },
+    {
+      "epoch": 0.7483596080102258,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7410163964865483e-06,
+      "loss": 0.7902,
+      "step": 8782
+    },
+    {
+      "epoch": 0.7484448231785258,
+      "grad_norm": 23.5,
+      "learning_rate": 1.7408618634796675e-06,
+      "loss": 0.8568,
+      "step": 8783
+    },
+    {
+      "epoch": 0.7485300383468257,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7407073216024195e-06,
+      "loss": 0.8488,
+      "step": 8784
+    },
+    {
+      "epoch": 0.7486152535151257,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7405527708575962e-06,
+      "loss": 0.7358,
+      "step": 8785
+    },
+    {
+      "epoch": 0.7487004686834257,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7403982112479912e-06,
+      "loss": 0.7834,
+      "step": 8786
+    },
+    {
+      "epoch": 0.7487856838517256,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7402436427763974e-06,
+      "loss": 0.8336,
+      "step": 8787
+    },
+    {
+      "epoch": 0.7488708990200256,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7400890654456083e-06,
+      "loss": 0.5317,
+      "step": 8788
+    },
+    {
+      "epoch": 0.7489561141883255,
+      "grad_norm": 10.0,
+      "learning_rate": 1.7399344792584163e-06,
+      "loss": 0.3257,
+      "step": 8789
+    },
+    {
+      "epoch": 0.7490413293566255,
+      "grad_norm": 14.375,
+      "learning_rate": 1.739779884217616e-06,
+      "loss": 0.3037,
+      "step": 8790
+    },
+    {
+      "epoch": 0.7491265445249254,
+      "grad_norm": 13.375,
+      "learning_rate": 1.739625280326001e-06,
+      "loss": 0.78,
+      "step": 8791
+    },
+    {
+      "epoch": 0.7492117596932254,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7394706675863645e-06,
+      "loss": 0.5396,
+      "step": 8792
+    },
+    {
+      "epoch": 0.7492969748615254,
+      "grad_norm": 13.0,
+      "learning_rate": 1.7393160460015007e-06,
+      "loss": 0.6034,
+      "step": 8793
+    },
+    {
+      "epoch": 0.7493821900298253,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7391614155742042e-06,
+      "loss": 0.5717,
+      "step": 8794
+    },
+    {
+      "epoch": 0.7494674051981253,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7390067763072688e-06,
+      "loss": 0.7804,
+      "step": 8795
+    },
+    {
+      "epoch": 0.7495526203664252,
+      "grad_norm": 25.0,
+      "learning_rate": 1.7388521282034895e-06,
+      "loss": 0.6993,
+      "step": 8796
+    },
+    {
+      "epoch": 0.7496378355347252,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.73869747126566e-06,
+      "loss": 0.3596,
+      "step": 8797
+    },
+    {
+      "epoch": 0.7497230507030251,
+      "grad_norm": 13.5,
+      "learning_rate": 1.7385428054965764e-06,
+      "loss": 0.8027,
+      "step": 8798
+    },
+    {
+      "epoch": 0.7498082658713251,
+      "grad_norm": 16.25,
+      "learning_rate": 1.738388130899033e-06,
+      "loss": 0.7779,
+      "step": 8799
+    },
+    {
+      "epoch": 0.749893481039625,
+      "grad_norm": 33.25,
+      "learning_rate": 1.7382334474758248e-06,
+      "loss": 0.7611,
+      "step": 8800
+    },
+    {
+      "epoch": 0.749978696207925,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.7380787552297474e-06,
+      "loss": 0.8858,
+      "step": 8801
+    },
+    {
+      "epoch": 0.750063911376225,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7379240541635957e-06,
+      "loss": 0.8558,
+      "step": 8802
+    },
+    {
+      "epoch": 0.7501491265445249,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.7377693442801664e-06,
+      "loss": 0.4349,
+      "step": 8803
+    },
+    {
+      "epoch": 0.7502343417128249,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.7376146255822537e-06,
+      "loss": 0.4543,
+      "step": 8804
+    },
+    {
+      "epoch": 0.7503195568811248,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7374598980726548e-06,
+      "loss": 0.6571,
+      "step": 8805
+    },
+    {
+      "epoch": 0.7504047720494248,
+      "grad_norm": 21.5,
+      "learning_rate": 1.7373051617541654e-06,
+      "loss": 1.2828,
+      "step": 8806
+    },
+    {
+      "epoch": 0.7504899872177248,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7371504166295818e-06,
+      "loss": 0.549,
+      "step": 8807
+    },
+    {
+      "epoch": 0.7505752023860247,
+      "grad_norm": 11.25,
+      "learning_rate": 1.7369956627017e-06,
+      "loss": 0.721,
+      "step": 8808
+    },
+    {
+      "epoch": 0.7506604175543247,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7368408999733172e-06,
+      "loss": 0.8061,
+      "step": 8809
+    },
+    {
+      "epoch": 0.7507456327226246,
+      "grad_norm": 16.0,
+      "learning_rate": 1.7366861284472295e-06,
+      "loss": 0.3442,
+      "step": 8810
+    },
+    {
+      "epoch": 0.7508308478909246,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7365313481262344e-06,
+      "loss": 0.8165,
+      "step": 8811
+    },
+    {
+      "epoch": 0.7509160630592245,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7363765590131282e-06,
+      "loss": 0.8678,
+      "step": 8812
+    },
+    {
+      "epoch": 0.7510012782275245,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7362217611107087e-06,
+      "loss": 0.7003,
+      "step": 8813
+    },
+    {
+      "epoch": 0.7510864933958245,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7360669544217728e-06,
+      "loss": 0.6057,
+      "step": 8814
+    },
+    {
+      "epoch": 0.7511717085641244,
+      "grad_norm": 23.625,
+      "learning_rate": 1.7359121389491184e-06,
+      "loss": 1.0507,
+      "step": 8815
+    },
+    {
+      "epoch": 0.7512569237324244,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7357573146955434e-06,
+      "loss": 0.7692,
+      "step": 8816
+    },
+    {
+      "epoch": 0.7513421389007243,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.7356024816638447e-06,
+      "loss": 0.6305,
+      "step": 8817
+    },
+    {
+      "epoch": 0.7514273540690243,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.7354476398568213e-06,
+      "loss": 0.4532,
+      "step": 8818
+    },
+    {
+      "epoch": 0.7515125692373242,
+      "grad_norm": 22.5,
+      "learning_rate": 1.7352927892772707e-06,
+      "loss": 1.1847,
+      "step": 8819
+    },
+    {
+      "epoch": 0.7515977844056242,
+      "grad_norm": 15.25,
+      "learning_rate": 1.7351379299279913e-06,
+      "loss": 0.539,
+      "step": 8820
+    },
+    {
+      "epoch": 0.7516829995739241,
+      "grad_norm": 10.75,
+      "learning_rate": 1.7349830618117815e-06,
+      "loss": 0.3418,
+      "step": 8821
+    },
+    {
+      "epoch": 0.7517682147422241,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.7348281849314405e-06,
+      "loss": 0.3841,
+      "step": 8822
+    },
+    {
+      "epoch": 0.7518534299105241,
+      "grad_norm": 20.375,
+      "learning_rate": 1.7346732992897663e-06,
+      "loss": 1.0008,
+      "step": 8823
+    },
+    {
+      "epoch": 0.751938645078824,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7345184048895588e-06,
+      "loss": 0.8183,
+      "step": 8824
+    },
+    {
+      "epoch": 0.752023860247124,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.7343635017336158e-06,
+      "loss": 0.8955,
+      "step": 8825
+    },
+    {
+      "epoch": 0.7521090754154239,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7342085898247377e-06,
+      "loss": 0.8762,
+      "step": 8826
+    },
+    {
+      "epoch": 0.7521942905837239,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.734053669165723e-06,
+      "loss": 0.2592,
+      "step": 8827
+    },
+    {
+      "epoch": 0.7522795057520238,
+      "grad_norm": 14.125,
+      "learning_rate": 1.7338987397593727e-06,
+      "loss": 0.9215,
+      "step": 8828
+    },
+    {
+      "epoch": 0.7523647209203238,
+      "grad_norm": 20.125,
+      "learning_rate": 1.7337438016084845e-06,
+      "loss": 1.1696,
+      "step": 8829
+    },
+    {
+      "epoch": 0.7524499360886238,
+      "grad_norm": 23.125,
+      "learning_rate": 1.73358885471586e-06,
+      "loss": 1.0865,
+      "step": 8830
+    },
+    {
+      "epoch": 0.7525351512569237,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7334338990842982e-06,
+      "loss": 1.0081,
+      "step": 8831
+    },
+    {
+      "epoch": 0.7526203664252237,
+      "grad_norm": 13.0,
+      "learning_rate": 1.7332789347166e-06,
+      "loss": 0.5814,
+      "step": 8832
+    },
+    {
+      "epoch": 0.7527055815935236,
+      "grad_norm": 20.5,
+      "learning_rate": 1.7331239616155655e-06,
+      "loss": 1.1821,
+      "step": 8833
+    },
+    {
+      "epoch": 0.7527907967618236,
+      "grad_norm": 21.0,
+      "learning_rate": 1.7329689797839953e-06,
+      "loss": 0.8885,
+      "step": 8834
+    },
+    {
+      "epoch": 0.7528760119301235,
+      "grad_norm": 49.75,
+      "learning_rate": 1.7328139892246894e-06,
+      "loss": 0.995,
+      "step": 8835
+    },
+    {
+      "epoch": 0.7529612270984235,
+      "grad_norm": 13.0,
+      "learning_rate": 1.7326589899404498e-06,
+      "loss": 0.4504,
+      "step": 8836
+    },
+    {
+      "epoch": 0.7530464422667235,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7325039819340767e-06,
+      "loss": 0.9001,
+      "step": 8837
+    },
+    {
+      "epoch": 0.7531316574350234,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7323489652083714e-06,
+      "loss": 0.6195,
+      "step": 8838
+    },
+    {
+      "epoch": 0.7532168726033234,
+      "grad_norm": 16.0,
+      "learning_rate": 1.7321939397661352e-06,
+      "loss": 0.7212,
+      "step": 8839
+    },
+    {
+      "epoch": 0.7533020877716233,
+      "grad_norm": 17.875,
+      "learning_rate": 1.73203890561017e-06,
+      "loss": 0.9277,
+      "step": 8840
+    },
+    {
+      "epoch": 0.7533873029399233,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7318838627432765e-06,
+      "loss": 0.7808,
+      "step": 8841
+    },
+    {
+      "epoch": 0.7534725181082232,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7317288111682573e-06,
+      "loss": 0.7741,
+      "step": 8842
+    },
+    {
+      "epoch": 0.7535577332765232,
+      "grad_norm": 14.5,
+      "learning_rate": 1.731573750887914e-06,
+      "loss": 0.8483,
+      "step": 8843
+    },
+    {
+      "epoch": 0.7536429484448232,
+      "grad_norm": 11.75,
+      "learning_rate": 1.7314186819050492e-06,
+      "loss": 0.43,
+      "step": 8844
+    },
+    {
+      "epoch": 0.7537281636131231,
+      "grad_norm": 22.875,
+      "learning_rate": 1.7312636042224643e-06,
+      "loss": 1.0391,
+      "step": 8845
+    },
+    {
+      "epoch": 0.7538133787814231,
+      "grad_norm": 14.625,
+      "learning_rate": 1.731108517842962e-06,
+      "loss": 1.134,
+      "step": 8846
+    },
+    {
+      "epoch": 0.753898593949723,
+      "grad_norm": 10.625,
+      "learning_rate": 1.7309534227693453e-06,
+      "loss": 0.3056,
+      "step": 8847
+    },
+    {
+      "epoch": 0.753983809118023,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7307983190044164e-06,
+      "loss": 0.4403,
+      "step": 8848
+    },
+    {
+      "epoch": 0.7540690242863229,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7306432065509784e-06,
+      "loss": 0.4834,
+      "step": 8849
+    },
+    {
+      "epoch": 0.7541542394546229,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7304880854118345e-06,
+      "loss": 0.8416,
+      "step": 8850
+    },
+    {
+      "epoch": 0.7542394546229229,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7303329555897873e-06,
+      "loss": 0.9184,
+      "step": 8851
+    },
+    {
+      "epoch": 0.7543246697912228,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7301778170876413e-06,
+      "loss": 0.7101,
+      "step": 8852
+    },
+    {
+      "epoch": 0.7544098849595228,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7300226699081984e-06,
+      "loss": 0.5973,
+      "step": 8853
+    },
+    {
+      "epoch": 0.7544951001278227,
+      "grad_norm": 16.5,
+      "learning_rate": 1.729867514054264e-06,
+      "loss": 1.0795,
+      "step": 8854
+    },
+    {
+      "epoch": 0.7545803152961227,
+      "grad_norm": 15.25,
+      "learning_rate": 1.7297123495286406e-06,
+      "loss": 0.6148,
+      "step": 8855
+    },
+    {
+      "epoch": 0.7546655304644226,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.729557176334133e-06,
+      "loss": 0.5185,
+      "step": 8856
+    },
+    {
+      "epoch": 0.7547507456327226,
+      "grad_norm": 14.25,
+      "learning_rate": 1.7294019944735447e-06,
+      "loss": 0.7482,
+      "step": 8857
+    },
+    {
+      "epoch": 0.7548359608010226,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.7292468039496804e-06,
+      "loss": 0.4201,
+      "step": 8858
+    },
+    {
+      "epoch": 0.7549211759693225,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7290916047653442e-06,
+      "loss": 0.8541,
+      "step": 8859
+    },
+    {
+      "epoch": 0.7550063911376225,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7289363969233413e-06,
+      "loss": 0.8045,
+      "step": 8860
+    },
+    {
+      "epoch": 0.7550916063059224,
+      "grad_norm": 20.5,
+      "learning_rate": 1.728781180426476e-06,
+      "loss": 0.9231,
+      "step": 8861
+    },
+    {
+      "epoch": 0.7551768214742224,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.728625955277553e-06,
+      "loss": 0.7687,
+      "step": 8862
+    },
+    {
+      "epoch": 0.7552620366425223,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7284707214793777e-06,
+      "loss": 1.1376,
+      "step": 8863
+    },
+    {
+      "epoch": 0.7553472518108223,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.728315479034756e-06,
+      "loss": 0.4929,
+      "step": 8864
+    },
+    {
+      "epoch": 0.7554324669791223,
+      "grad_norm": 23.875,
+      "learning_rate": 1.728160227946492e-06,
+      "loss": 0.9484,
+      "step": 8865
+    },
+    {
+      "epoch": 0.7555176821474222,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.728004968217392e-06,
+      "loss": 0.5762,
+      "step": 8866
+    },
+    {
+      "epoch": 0.7556028973157222,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7278496998502614e-06,
+      "loss": 0.6943,
+      "step": 8867
+    },
+    {
+      "epoch": 0.7556881124840221,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7276944228479067e-06,
+      "loss": 0.4078,
+      "step": 8868
+    },
+    {
+      "epoch": 0.7557733276523221,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.7275391372131331e-06,
+      "loss": 0.6442,
+      "step": 8869
+    },
+    {
+      "epoch": 0.755858542820622,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7273838429487468e-06,
+      "loss": 1.0687,
+      "step": 8870
+    },
+    {
+      "epoch": 0.755943757988922,
+      "grad_norm": 11.75,
+      "learning_rate": 1.727228540057555e-06,
+      "loss": 0.6581,
+      "step": 8871
+    },
+    {
+      "epoch": 0.756028973157222,
+      "grad_norm": 11.875,
+      "learning_rate": 1.7270732285423633e-06,
+      "loss": 0.4837,
+      "step": 8872
+    },
+    {
+      "epoch": 0.7561141883255219,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7269179084059788e-06,
+      "loss": 0.6691,
+      "step": 8873
+    },
+    {
+      "epoch": 0.7561994034938219,
+      "grad_norm": 29.25,
+      "learning_rate": 1.7267625796512078e-06,
+      "loss": 0.9593,
+      "step": 8874
+    },
+    {
+      "epoch": 0.7562846186621218,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7266072422808578e-06,
+      "loss": 0.6315,
+      "step": 8875
+    },
+    {
+      "epoch": 0.7563698338304218,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7264518962977356e-06,
+      "loss": 0.7552,
+      "step": 8876
+    },
+    {
+      "epoch": 0.7564550489987217,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7262965417046483e-06,
+      "loss": 0.6384,
+      "step": 8877
+    },
+    {
+      "epoch": 0.7565402641670217,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7261411785044036e-06,
+      "loss": 0.5805,
+      "step": 8878
+    },
+    {
+      "epoch": 0.7566254793353216,
+      "grad_norm": 17.75,
+      "learning_rate": 1.7259858066998091e-06,
+      "loss": 1.0247,
+      "step": 8879
+    },
+    {
+      "epoch": 0.7567106945036216,
+      "grad_norm": 11.25,
+      "learning_rate": 1.7258304262936726e-06,
+      "loss": 0.3864,
+      "step": 8880
+    },
+    {
+      "epoch": 0.7567959096719216,
+      "grad_norm": 10.375,
+      "learning_rate": 1.725675037288802e-06,
+      "loss": 0.4187,
+      "step": 8881
+    },
+    {
+      "epoch": 0.7568811248402215,
+      "grad_norm": 11.875,
+      "learning_rate": 1.7255196396880044e-06,
+      "loss": 0.5016,
+      "step": 8882
+    },
+    {
+      "epoch": 0.7569663400085215,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.7253642334940896e-06,
+      "loss": 0.7801,
+      "step": 8883
+    },
+    {
+      "epoch": 0.7570515551768214,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.7252088187098642e-06,
+      "loss": 0.5035,
+      "step": 8884
+    },
+    {
+      "epoch": 0.7571367703451214,
+      "grad_norm": 18.0,
+      "learning_rate": 1.7250533953381381e-06,
+      "loss": 0.6229,
+      "step": 8885
+    },
+    {
+      "epoch": 0.7572219855134213,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7248979633817195e-06,
+      "loss": 0.6052,
+      "step": 8886
+    },
+    {
+      "epoch": 0.7573072006817213,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.724742522843417e-06,
+      "loss": 0.7106,
+      "step": 8887
+    },
+    {
+      "epoch": 0.7573924158500213,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7245870737260398e-06,
+      "loss": 0.612,
+      "step": 8888
+    },
+    {
+      "epoch": 0.7574776310183212,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.724431616032397e-06,
+      "loss": 0.842,
+      "step": 8889
+    },
+    {
+      "epoch": 0.7575628461866212,
+      "grad_norm": 21.75,
+      "learning_rate": 1.7242761497652976e-06,
+      "loss": 1.211,
+      "step": 8890
+    },
+    {
+      "epoch": 0.7576480613549211,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7241206749275512e-06,
+      "loss": 0.6871,
+      "step": 8891
+    },
+    {
+      "epoch": 0.7577332765232211,
+      "grad_norm": 18.875,
+      "learning_rate": 1.723965191521968e-06,
+      "loss": 0.998,
+      "step": 8892
+    },
+    {
+      "epoch": 0.757818491691521,
+      "grad_norm": 11.75,
+      "learning_rate": 1.7238096995513567e-06,
+      "loss": 0.4797,
+      "step": 8893
+    },
+    {
+      "epoch": 0.757903706859821,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.7236541990185279e-06,
+      "loss": 0.7879,
+      "step": 8894
+    },
+    {
+      "epoch": 0.757988922028121,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7234986899262912e-06,
+      "loss": 0.8003,
+      "step": 8895
+    },
+    {
+      "epoch": 0.7580741371964209,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.723343172277457e-06,
+      "loss": 0.606,
+      "step": 8896
+    },
+    {
+      "epoch": 0.7581593523647209,
+      "grad_norm": 18.625,
+      "learning_rate": 1.723187646074836e-06,
+      "loss": 1.0115,
+      "step": 8897
+    },
+    {
+      "epoch": 0.7582445675330208,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.7230321113212383e-06,
+      "loss": 0.5035,
+      "step": 8898
+    },
+    {
+      "epoch": 0.7583297827013208,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7228765680194745e-06,
+      "loss": 0.8066,
+      "step": 8899
+    },
+    {
+      "epoch": 0.7584149978696207,
+      "grad_norm": 12.25,
+      "learning_rate": 1.7227210161723556e-06,
+      "loss": 0.6151,
+      "step": 8900
+    },
+    {
+      "epoch": 0.7585002130379207,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.722565455782693e-06,
+      "loss": 0.8543,
+      "step": 8901
+    },
+    {
+      "epoch": 0.7585854282062207,
+      "grad_norm": 19.25,
+      "learning_rate": 1.7224098868532968e-06,
+      "loss": 0.8868,
+      "step": 8902
+    },
+    {
+      "epoch": 0.7586706433745206,
+      "grad_norm": 15.875,
+      "learning_rate": 1.7222543093869793e-06,
+      "loss": 0.8081,
+      "step": 8903
+    },
+    {
+      "epoch": 0.7587558585428206,
+      "grad_norm": 13.0,
+      "learning_rate": 1.7220987233865517e-06,
+      "loss": 0.3198,
+      "step": 8904
+    },
+    {
+      "epoch": 0.7588410737111205,
+      "grad_norm": 19.75,
+      "learning_rate": 1.721943128854825e-06,
+      "loss": 0.8929,
+      "step": 8905
+    },
+    {
+      "epoch": 0.7589262888794205,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.7217875257946115e-06,
+      "loss": 0.3741,
+      "step": 8906
+    },
+    {
+      "epoch": 0.7590115040477204,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7216319142087232e-06,
+      "loss": 0.8314,
+      "step": 8907
+    },
+    {
+      "epoch": 0.7590967192160204,
+      "grad_norm": 31.5,
+      "learning_rate": 1.7214762940999716e-06,
+      "loss": 1.2375,
+      "step": 8908
+    },
+    {
+      "epoch": 0.7591819343843204,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.7213206654711697e-06,
+      "loss": 0.9969,
+      "step": 8909
+    },
+    {
+      "epoch": 0.7592671495526203,
+      "grad_norm": 50.0,
+      "learning_rate": 1.721165028325129e-06,
+      "loss": 0.6129,
+      "step": 8910
+    },
+    {
+      "epoch": 0.7593523647209203,
+      "grad_norm": 14.75,
+      "learning_rate": 1.7210093826646628e-06,
+      "loss": 0.8879,
+      "step": 8911
+    },
+    {
+      "epoch": 0.7594375798892202,
+      "grad_norm": 16.75,
+      "learning_rate": 1.720853728492583e-06,
+      "loss": 0.8162,
+      "step": 8912
+    },
+    {
+      "epoch": 0.7595227950575202,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7206980658117034e-06,
+      "loss": 0.967,
+      "step": 8913
+    },
+    {
+      "epoch": 0.7596080102258201,
+      "grad_norm": 16.875,
+      "learning_rate": 1.720542394624836e-06,
+      "loss": 0.6018,
+      "step": 8914
+    },
+    {
+      "epoch": 0.7596932253941202,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7203867149347946e-06,
+      "loss": 0.3263,
+      "step": 8915
+    },
+    {
+      "epoch": 0.7597784405624202,
+      "grad_norm": 9.75,
+      "learning_rate": 1.720231026744392e-06,
+      "loss": 0.203,
+      "step": 8916
+    },
+    {
+      "epoch": 0.7598636557307201,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7200753300564422e-06,
+      "loss": 0.576,
+      "step": 8917
+    },
+    {
+      "epoch": 0.7599488708990201,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7199196248737582e-06,
+      "loss": 0.7405,
+      "step": 8918
+    },
+    {
+      "epoch": 0.76003408606732,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7197639111991543e-06,
+      "loss": 0.9682,
+      "step": 8919
+    },
+    {
+      "epoch": 0.76011930123562,
+      "grad_norm": 17.625,
+      "learning_rate": 1.7196081890354439e-06,
+      "loss": 0.8688,
+      "step": 8920
+    },
+    {
+      "epoch": 0.76020451640392,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.7194524583854414e-06,
+      "loss": 0.5735,
+      "step": 8921
+    },
+    {
+      "epoch": 0.7602897315722199,
+      "grad_norm": 14.25,
+      "learning_rate": 1.719296719251961e-06,
+      "loss": 0.5425,
+      "step": 8922
+    },
+    {
+      "epoch": 0.7603749467405199,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7191409716378165e-06,
+      "loss": 0.4827,
+      "step": 8923
+    },
+    {
+      "epoch": 0.7604601619088198,
+      "grad_norm": 15.875,
+      "learning_rate": 1.7189852155458236e-06,
+      "loss": 0.9541,
+      "step": 8924
+    },
+    {
+      "epoch": 0.7605453770771198,
+      "grad_norm": 18.625,
+      "learning_rate": 1.718829450978796e-06,
+      "loss": 0.9378,
+      "step": 8925
+    },
+    {
+      "epoch": 0.7606305922454197,
+      "grad_norm": 12.25,
+      "learning_rate": 1.7186736779395485e-06,
+      "loss": 0.5963,
+      "step": 8926
+    },
+    {
+      "epoch": 0.7607158074137197,
+      "grad_norm": 12.875,
+      "learning_rate": 1.7185178964308965e-06,
+      "loss": 0.5026,
+      "step": 8927
+    },
+    {
+      "epoch": 0.7608010225820196,
+      "grad_norm": 18.875,
+      "learning_rate": 1.7183621064556551e-06,
+      "loss": 0.6259,
+      "step": 8928
+    },
+    {
+      "epoch": 0.7608862377503196,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7182063080166395e-06,
+      "loss": 0.9694,
+      "step": 8929
+    },
+    {
+      "epoch": 0.7609714529186196,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.718050501116665e-06,
+      "loss": 0.4604,
+      "step": 8930
+    },
+    {
+      "epoch": 0.7610566680869195,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.7178946857585472e-06,
+      "loss": 0.4042,
+      "step": 8931
+    },
+    {
+      "epoch": 0.7611418832552195,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.717738861945102e-06,
+      "loss": 0.2951,
+      "step": 8932
+    },
+    {
+      "epoch": 0.7612270984235194,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7175830296791451e-06,
+      "loss": 1.141,
+      "step": 8933
+    },
+    {
+      "epoch": 0.7613123135918194,
+      "grad_norm": 14.0,
+      "learning_rate": 1.717427188963493e-06,
+      "loss": 0.809,
+      "step": 8934
+    },
+    {
+      "epoch": 0.7613975287601193,
+      "grad_norm": 14.25,
+      "learning_rate": 1.717271339800961e-06,
+      "loss": 0.8487,
+      "step": 8935
+    },
+    {
+      "epoch": 0.7614827439284193,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7171154821943664e-06,
+      "loss": 0.6533,
+      "step": 8936
+    },
+    {
+      "epoch": 0.7615679590967193,
+      "grad_norm": 13.875,
+      "learning_rate": 1.7169596161465256e-06,
+      "loss": 0.6976,
+      "step": 8937
+    },
+    {
+      "epoch": 0.7616531742650192,
+      "grad_norm": 17.875,
+      "learning_rate": 1.7168037416602543e-06,
+      "loss": 0.9174,
+      "step": 8938
+    },
+    {
+      "epoch": 0.7617383894333192,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7166478587383707e-06,
+      "loss": 0.784,
+      "step": 8939
+    },
+    {
+      "epoch": 0.7618236046016191,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.7164919673836902e-06,
+      "loss": 0.57,
+      "step": 8940
+    },
+    {
+      "epoch": 0.7619088197699191,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7163360675990315e-06,
+      "loss": 0.8436,
+      "step": 8941
+    },
+    {
+      "epoch": 0.761994034938219,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7161801593872106e-06,
+      "loss": 0.4477,
+      "step": 8942
+    },
+    {
+      "epoch": 0.762079250106519,
+      "grad_norm": 15.375,
+      "learning_rate": 1.716024242751046e-06,
+      "loss": 0.8258,
+      "step": 8943
+    },
+    {
+      "epoch": 0.762164465274819,
+      "grad_norm": 14.375,
+      "learning_rate": 1.7158683176933542e-06,
+      "loss": 0.8524,
+      "step": 8944
+    },
+    {
+      "epoch": 0.7622496804431189,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7157123842169537e-06,
+      "loss": 0.8037,
+      "step": 8945
+    },
+    {
+      "epoch": 0.7623348956114189,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.7155564423246617e-06,
+      "loss": 0.4825,
+      "step": 8946
+    },
+    {
+      "epoch": 0.7624201107797188,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.7154004920192973e-06,
+      "loss": 0.6762,
+      "step": 8947
+    },
+    {
+      "epoch": 0.7625053259480188,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7152445333036774e-06,
+      "loss": 0.4607,
+      "step": 8948
+    },
+    {
+      "epoch": 0.7625905411163187,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.7150885661806212e-06,
+      "loss": 0.6954,
+      "step": 8949
+    },
+    {
+      "epoch": 0.7626757562846187,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.7149325906529471e-06,
+      "loss": 0.5826,
+      "step": 8950
+    },
+    {
+      "epoch": 0.7627609714529187,
+      "grad_norm": 20.5,
+      "learning_rate": 1.7147766067234734e-06,
+      "loss": 1.1907,
+      "step": 8951
+    },
+    {
+      "epoch": 0.7628461866212186,
+      "grad_norm": 17.75,
+      "learning_rate": 1.714620614395019e-06,
+      "loss": 1.0698,
+      "step": 8952
+    },
+    {
+      "epoch": 0.7629314017895186,
+      "grad_norm": 21.125,
+      "learning_rate": 1.714464613670403e-06,
+      "loss": 0.8887,
+      "step": 8953
+    },
+    {
+      "epoch": 0.7630166169578185,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7143086045524443e-06,
+      "loss": 0.73,
+      "step": 8954
+    },
+    {
+      "epoch": 0.7631018321261185,
+      "grad_norm": 11.0,
+      "learning_rate": 1.7141525870439622e-06,
+      "loss": 0.4569,
+      "step": 8955
+    },
+    {
+      "epoch": 0.7631870472944184,
+      "grad_norm": 36.25,
+      "learning_rate": 1.713996561147776e-06,
+      "loss": 1.3506,
+      "step": 8956
+    },
+    {
+      "epoch": 0.7632722624627184,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.7138405268667058e-06,
+      "loss": 0.5579,
+      "step": 8957
+    },
+    {
+      "epoch": 0.7633574776310184,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7136844842035704e-06,
+      "loss": 0.8242,
+      "step": 8958
+    },
+    {
+      "epoch": 0.7634426927993183,
+      "grad_norm": 10.375,
+      "learning_rate": 1.71352843316119e-06,
+      "loss": 0.5989,
+      "step": 8959
+    },
+    {
+      "epoch": 0.7635279079676183,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7133723737423852e-06,
+      "loss": 1.0877,
+      "step": 8960
+    },
+    {
+      "epoch": 0.7636131231359182,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7132163059499751e-06,
+      "loss": 1.0761,
+      "step": 8961
+    },
+    {
+      "epoch": 0.7636983383042182,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7130602297867814e-06,
+      "loss": 0.4392,
+      "step": 8962
+    },
+    {
+      "epoch": 0.7637835534725181,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7129041452556227e-06,
+      "loss": 0.837,
+      "step": 8963
+    },
+    {
+      "epoch": 0.7638687686408181,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.7127480523593213e-06,
+      "loss": 0.6768,
+      "step": 8964
+    },
+    {
+      "epoch": 0.763953983809118,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7125919511006966e-06,
+      "loss": 0.6845,
+      "step": 8965
+    },
+    {
+      "epoch": 0.764039198977418,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.712435841482571e-06,
+      "loss": 0.3896,
+      "step": 8966
+    },
+    {
+      "epoch": 0.764124414145718,
+      "grad_norm": 30.25,
+      "learning_rate": 1.7122797235077642e-06,
+      "loss": 1.0668,
+      "step": 8967
+    },
+    {
+      "epoch": 0.7642096293140179,
+      "grad_norm": 10.625,
+      "learning_rate": 1.7121235971790982e-06,
+      "loss": 0.2162,
+      "step": 8968
+    },
+    {
+      "epoch": 0.7642948444823179,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.7119674624993937e-06,
+      "loss": 0.6982,
+      "step": 8969
+    },
+    {
+      "epoch": 0.7643800596506178,
+      "grad_norm": 12.25,
+      "learning_rate": 1.711811319471473e-06,
+      "loss": 0.4853,
+      "step": 8970
+    },
+    {
+      "epoch": 0.7644652748189178,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.7116551680981572e-06,
+      "loss": 0.339,
+      "step": 8971
+    },
+    {
+      "epoch": 0.7645504899872178,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.7114990083822686e-06,
+      "loss": 0.7008,
+      "step": 8972
+    },
+    {
+      "epoch": 0.7646357051555177,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7113428403266282e-06,
+      "loss": 0.7619,
+      "step": 8973
+    },
+    {
+      "epoch": 0.7647209203238177,
+      "grad_norm": 16.0,
+      "learning_rate": 1.711186663934059e-06,
+      "loss": 1.0166,
+      "step": 8974
+    },
+    {
+      "epoch": 0.7648061354921176,
+      "grad_norm": 13.625,
+      "learning_rate": 1.7110304792073835e-06,
+      "loss": 0.5381,
+      "step": 8975
+    },
+    {
+      "epoch": 0.7648913506604176,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7108742861494233e-06,
+      "loss": 0.8544,
+      "step": 8976
+    },
+    {
+      "epoch": 0.7649765658287175,
+      "grad_norm": 20.375,
+      "learning_rate": 1.710718084763001e-06,
+      "loss": 0.9204,
+      "step": 8977
+    },
+    {
+      "epoch": 0.7650617809970175,
+      "grad_norm": 19.125,
+      "learning_rate": 1.71056187505094e-06,
+      "loss": 0.9626,
+      "step": 8978
+    },
+    {
+      "epoch": 0.7651469961653175,
+      "grad_norm": 16.0,
+      "learning_rate": 1.710405657016063e-06,
+      "loss": 0.7943,
+      "step": 8979
+    },
+    {
+      "epoch": 0.7652322113336174,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7102494306611921e-06,
+      "loss": 0.9519,
+      "step": 8980
+    },
+    {
+      "epoch": 0.7653174265019174,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.7100931959891517e-06,
+      "loss": 1.0013,
+      "step": 8981
+    },
+    {
+      "epoch": 0.7654026416702173,
+      "grad_norm": 19.375,
+      "learning_rate": 1.7099369530027645e-06,
+      "loss": 0.9115,
+      "step": 8982
+    },
+    {
+      "epoch": 0.7654878568385173,
+      "grad_norm": 13.625,
+      "learning_rate": 1.709780701704854e-06,
+      "loss": 0.7726,
+      "step": 8983
+    },
+    {
+      "epoch": 0.7655730720068172,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7096244420982438e-06,
+      "loss": 0.8881,
+      "step": 8984
+    },
+    {
+      "epoch": 0.7656582871751172,
+      "grad_norm": 9.875,
+      "learning_rate": 1.709468174185758e-06,
+      "loss": 0.3125,
+      "step": 8985
+    },
+    {
+      "epoch": 0.7657435023434171,
+      "grad_norm": 16.25,
+      "learning_rate": 1.70931189797022e-06,
+      "loss": 0.721,
+      "step": 8986
+    },
+    {
+      "epoch": 0.7658287175117171,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7091556134544547e-06,
+      "loss": 0.8042,
+      "step": 8987
+    },
+    {
+      "epoch": 0.7659139326800171,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7089993206412848e-06,
+      "loss": 0.8803,
+      "step": 8988
+    },
+    {
+      "epoch": 0.765999147848317,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7088430195335362e-06,
+      "loss": 0.9869,
+      "step": 8989
+    },
+    {
+      "epoch": 0.766084363016617,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7086867101340327e-06,
+      "loss": 0.6307,
+      "step": 8990
+    },
+    {
+      "epoch": 0.7661695781849169,
+      "grad_norm": 14.25,
+      "learning_rate": 1.708530392445599e-06,
+      "loss": 0.7213,
+      "step": 8991
+    },
+    {
+      "epoch": 0.7662547933532169,
+      "grad_norm": 19.75,
+      "learning_rate": 1.7083740664710602e-06,
+      "loss": 0.7864,
+      "step": 8992
+    },
+    {
+      "epoch": 0.7663400085215168,
+      "grad_norm": 12.875,
+      "learning_rate": 1.708217732213241e-06,
+      "loss": 0.5595,
+      "step": 8993
+    },
+    {
+      "epoch": 0.7664252236898168,
+      "grad_norm": 24.875,
+      "learning_rate": 1.7080613896749667e-06,
+      "loss": 0.6869,
+      "step": 8994
+    },
+    {
+      "epoch": 0.7665104388581168,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7079050388590623e-06,
+      "loss": 0.8646,
+      "step": 8995
+    },
+    {
+      "epoch": 0.7665956540264167,
+      "grad_norm": 13.875,
+      "learning_rate": 1.7077486797683538e-06,
+      "loss": 0.6568,
+      "step": 8996
+    },
+    {
+      "epoch": 0.7666808691947167,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.7075923124056657e-06,
+      "loss": 0.5769,
+      "step": 8997
+    },
+    {
+      "epoch": 0.7667660843630166,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7074359367738252e-06,
+      "loss": 0.6569,
+      "step": 8998
+    },
+    {
+      "epoch": 0.7668512995313166,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.7072795528756566e-06,
+      "loss": 0.5376,
+      "step": 8999
+    },
+    {
+      "epoch": 0.7669365146996165,
+      "grad_norm": 12.0,
+      "learning_rate": 1.707123160713987e-06,
+      "loss": 0.5588,
+      "step": 9000
+    },
+    {
+      "epoch": 0.7670217298679165,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7069667602916417e-06,
+      "loss": 0.5436,
+      "step": 9001
+    },
+    {
+      "epoch": 0.7671069450362165,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7068103516114484e-06,
+      "loss": 0.6801,
+      "step": 9002
+    },
+    {
+      "epoch": 0.7671921602045164,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.7066539346762322e-06,
+      "loss": 0.6156,
+      "step": 9003
+    },
+    {
+      "epoch": 0.7672773753728164,
+      "grad_norm": 17.875,
+      "learning_rate": 1.7064975094888202e-06,
+      "loss": 0.7216,
+      "step": 9004
+    },
+    {
+      "epoch": 0.7673625905411163,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7063410760520392e-06,
+      "loss": 0.8245,
+      "step": 9005
+    },
+    {
+      "epoch": 0.7674478057094163,
+      "grad_norm": 10.875,
+      "learning_rate": 1.706184634368716e-06,
+      "loss": 0.3248,
+      "step": 9006
+    },
+    {
+      "epoch": 0.7675330208777162,
+      "grad_norm": 12.75,
+      "learning_rate": 1.706028184441678e-06,
+      "loss": 0.4611,
+      "step": 9007
+    },
+    {
+      "epoch": 0.7676182360460162,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7058717262737517e-06,
+      "loss": 0.2809,
+      "step": 9008
+    },
+    {
+      "epoch": 0.7677034512143162,
+      "grad_norm": 24.125,
+      "learning_rate": 1.7057152598677651e-06,
+      "loss": 0.917,
+      "step": 9009
+    },
+    {
+      "epoch": 0.7677886663826161,
+      "grad_norm": 14.75,
+      "learning_rate": 1.7055587852265457e-06,
+      "loss": 0.7854,
+      "step": 9010
+    },
+    {
+      "epoch": 0.7678738815509161,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7054023023529208e-06,
+      "loss": 0.875,
+      "step": 9011
+    },
+    {
+      "epoch": 0.767959096719216,
+      "grad_norm": 18.75,
+      "learning_rate": 1.705245811249718e-06,
+      "loss": 0.7601,
+      "step": 9012
+    },
+    {
+      "epoch": 0.768044311887516,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.7050893119197663e-06,
+      "loss": 0.4686,
+      "step": 9013
+    },
+    {
+      "epoch": 0.7681295270558159,
+      "grad_norm": 13.125,
+      "learning_rate": 1.7049328043658926e-06,
+      "loss": 0.5051,
+      "step": 9014
+    },
+    {
+      "epoch": 0.7682147422241159,
+      "grad_norm": 14.625,
+      "learning_rate": 1.704776288590926e-06,
+      "loss": 0.76,
+      "step": 9015
+    },
+    {
+      "epoch": 0.7682999573924159,
+      "grad_norm": 12.5,
+      "learning_rate": 1.7046197645976943e-06,
+      "loss": 0.3944,
+      "step": 9016
+    },
+    {
+      "epoch": 0.7683851725607158,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7044632323890261e-06,
+      "loss": 0.8577,
+      "step": 9017
+    },
+    {
+      "epoch": 0.7684703877290158,
+      "grad_norm": 23.75,
+      "learning_rate": 1.7043066919677503e-06,
+      "loss": 0.5575,
+      "step": 9018
+    },
+    {
+      "epoch": 0.7685556028973157,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7041501433366965e-06,
+      "loss": 0.6031,
+      "step": 9019
+    },
+    {
+      "epoch": 0.7686408180656157,
+      "grad_norm": 13.25,
+      "learning_rate": 1.7039935864986923e-06,
+      "loss": 0.6532,
+      "step": 9020
+    },
+    {
+      "epoch": 0.7687260332339156,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.7038370214565675e-06,
+      "loss": 0.6905,
+      "step": 9021
+    },
+    {
+      "epoch": 0.7688112484022156,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.703680448213151e-06,
+      "loss": 0.698,
+      "step": 9022
+    },
+    {
+      "epoch": 0.7688964635705156,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7035238667712733e-06,
+      "loss": 0.548,
+      "step": 9023
+    },
+    {
+      "epoch": 0.7689816787388155,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.703367277133763e-06,
+      "loss": 0.5977,
+      "step": 9024
+    },
+    {
+      "epoch": 0.7690668939071155,
+      "grad_norm": 14.375,
+      "learning_rate": 1.7032106793034501e-06,
+      "loss": 0.5938,
+      "step": 9025
+    },
+    {
+      "epoch": 0.7691521090754154,
+      "grad_norm": 12.625,
+      "learning_rate": 1.7030540732831644e-06,
+      "loss": 0.4421,
+      "step": 9026
+    },
+    {
+      "epoch": 0.7692373242437154,
+      "grad_norm": 14.375,
+      "learning_rate": 1.702897459075736e-06,
+      "loss": 0.9577,
+      "step": 9027
+    },
+    {
+      "epoch": 0.7693225394120153,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.7027408366839953e-06,
+      "loss": 0.443,
+      "step": 9028
+    },
+    {
+      "epoch": 0.7694077545803153,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.7025842061107723e-06,
+      "loss": 0.8565,
+      "step": 9029
+    },
+    {
+      "epoch": 0.7694929697486153,
+      "grad_norm": 15.125,
+      "learning_rate": 1.7024275673588975e-06,
+      "loss": 0.6184,
+      "step": 9030
+    },
+    {
+      "epoch": 0.7695781849169152,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7022709204312018e-06,
+      "loss": 1.1135,
+      "step": 9031
+    },
+    {
+      "epoch": 0.7696634000852152,
+      "grad_norm": 14.125,
+      "learning_rate": 1.7021142653305156e-06,
+      "loss": 0.5914,
+      "step": 9032
+    },
+    {
+      "epoch": 0.7697486152535151,
+      "grad_norm": 16.25,
+      "learning_rate": 1.70195760205967e-06,
+      "loss": 0.6726,
+      "step": 9033
+    },
+    {
+      "epoch": 0.7698338304218151,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7018009306214962e-06,
+      "loss": 0.8695,
+      "step": 9034
+    },
+    {
+      "epoch": 0.769919045590115,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7016442510188251e-06,
+      "loss": 0.4837,
+      "step": 9035
+    },
+    {
+      "epoch": 0.770004260758415,
+      "grad_norm": 9.875,
+      "learning_rate": 1.7014875632544885e-06,
+      "loss": 0.1759,
+      "step": 9036
+    },
+    {
+      "epoch": 0.770089475926715,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7013308673313173e-06,
+      "loss": 0.7079,
+      "step": 9037
+    },
+    {
+      "epoch": 0.7701746910950149,
+      "grad_norm": 17.625,
+      "learning_rate": 1.7011741632521436e-06,
+      "loss": 1.1155,
+      "step": 9038
+    },
+    {
+      "epoch": 0.7702599062633149,
+      "grad_norm": 17.0,
+      "learning_rate": 1.7010174510197992e-06,
+      "loss": 0.9273,
+      "step": 9039
+    },
+    {
+      "epoch": 0.7703451214316148,
+      "grad_norm": 14.25,
+      "learning_rate": 1.7008607306371155e-06,
+      "loss": 0.7976,
+      "step": 9040
+    },
+    {
+      "epoch": 0.7704303365999148,
+      "grad_norm": 14.0,
+      "learning_rate": 1.7007040021069254e-06,
+      "loss": 0.5912,
+      "step": 9041
+    },
+    {
+      "epoch": 0.7705155517682147,
+      "grad_norm": 16.75,
+      "learning_rate": 1.700547265432061e-06,
+      "loss": 0.5672,
+      "step": 9042
+    },
+    {
+      "epoch": 0.7706007669365147,
+      "grad_norm": 11.625,
+      "learning_rate": 1.700390520615354e-06,
+      "loss": 0.5232,
+      "step": 9043
+    },
+    {
+      "epoch": 0.7706859821048146,
+      "grad_norm": 19.0,
+      "learning_rate": 1.7002337676596373e-06,
+      "loss": 1.1698,
+      "step": 9044
+    },
+    {
+      "epoch": 0.7707711972731146,
+      "grad_norm": 18.75,
+      "learning_rate": 1.700077006567744e-06,
+      "loss": 0.8062,
+      "step": 9045
+    },
+    {
+      "epoch": 0.7708564124414146,
+      "grad_norm": 13.625,
+      "learning_rate": 1.6999202373425063e-06,
+      "loss": 0.5261,
+      "step": 9046
+    },
+    {
+      "epoch": 0.7709416276097145,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.699763459986758e-06,
+      "loss": 0.5351,
+      "step": 9047
+    },
+    {
+      "epoch": 0.7710268427780145,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.6996066745033312e-06,
+      "loss": 0.7916,
+      "step": 9048
+    },
+    {
+      "epoch": 0.7711120579463144,
+      "grad_norm": 16.0,
+      "learning_rate": 1.69944988089506e-06,
+      "loss": 0.8742,
+      "step": 9049
+    },
+    {
+      "epoch": 0.7711972731146144,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.699293079164777e-06,
+      "loss": 0.4032,
+      "step": 9050
+    },
+    {
+      "epoch": 0.7712824882829143,
+      "grad_norm": 16.0,
+      "learning_rate": 1.699136269315317e-06,
+      "loss": 0.7968,
+      "step": 9051
+    },
+    {
+      "epoch": 0.7713677034512143,
+      "grad_norm": 11.0,
+      "learning_rate": 1.6989794513495125e-06,
+      "loss": 0.4347,
+      "step": 9052
+    },
+    {
+      "epoch": 0.7714529186195143,
+      "grad_norm": 11.875,
+      "learning_rate": 1.6988226252701982e-06,
+      "loss": 0.5018,
+      "step": 9053
+    },
+    {
+      "epoch": 0.7715381337878142,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.698665791080207e-06,
+      "loss": 0.3569,
+      "step": 9054
+    },
+    {
+      "epoch": 0.7716233489561142,
+      "grad_norm": 17.75,
+      "learning_rate": 1.6985089487823745e-06,
+      "loss": 0.6398,
+      "step": 9055
+    },
+    {
+      "epoch": 0.7717085641244141,
+      "grad_norm": 16.5,
+      "learning_rate": 1.698352098379534e-06,
+      "loss": 0.7814,
+      "step": 9056
+    },
+    {
+      "epoch": 0.7717937792927141,
+      "grad_norm": 26.0,
+      "learning_rate": 1.6981952398745205e-06,
+      "loss": 1.3219,
+      "step": 9057
+    },
+    {
+      "epoch": 0.771878994461014,
+      "grad_norm": 17.875,
+      "learning_rate": 1.6980383732701678e-06,
+      "loss": 0.7834,
+      "step": 9058
+    },
+    {
+      "epoch": 0.771964209629314,
+      "grad_norm": 17.625,
+      "learning_rate": 1.6978814985693115e-06,
+      "loss": 0.7673,
+      "step": 9059
+    },
+    {
+      "epoch": 0.772049424797614,
+      "grad_norm": 24.5,
+      "learning_rate": 1.6977246157747862e-06,
+      "loss": 0.9748,
+      "step": 9060
+    },
+    {
+      "epoch": 0.7721346399659139,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.6975677248894266e-06,
+      "loss": 0.4087,
+      "step": 9061
+    },
+    {
+      "epoch": 0.7722198551342139,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.6974108259160683e-06,
+      "loss": 0.6819,
+      "step": 9062
+    },
+    {
+      "epoch": 0.7723050703025138,
+      "grad_norm": 27.625,
+      "learning_rate": 1.6972539188575465e-06,
+      "loss": 0.9034,
+      "step": 9063
+    },
+    {
+      "epoch": 0.7723902854708138,
+      "grad_norm": 23.25,
+      "learning_rate": 1.6970970037166967e-06,
+      "loss": 0.8258,
+      "step": 9064
+    },
+    {
+      "epoch": 0.7724755006391137,
+      "grad_norm": 18.875,
+      "learning_rate": 1.6969400804963542e-06,
+      "loss": 0.7505,
+      "step": 9065
+    },
+    {
+      "epoch": 0.7725607158074137,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.6967831491993553e-06,
+      "loss": 0.5268,
+      "step": 9066
+    },
+    {
+      "epoch": 0.7726459309757137,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6966262098285355e-06,
+      "loss": 0.3593,
+      "step": 9067
+    },
+    {
+      "epoch": 0.7727311461440136,
+      "grad_norm": 16.875,
+      "learning_rate": 1.696469262386731e-06,
+      "loss": 0.776,
+      "step": 9068
+    },
+    {
+      "epoch": 0.7728163613123136,
+      "grad_norm": 17.125,
+      "learning_rate": 1.696312306876778e-06,
+      "loss": 1.0498,
+      "step": 9069
+    },
+    {
+      "epoch": 0.7729015764806135,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6961553433015127e-06,
+      "loss": 0.6401,
+      "step": 9070
+    },
+    {
+      "epoch": 0.7729867916489135,
+      "grad_norm": 11.125,
+      "learning_rate": 1.6959983716637717e-06,
+      "loss": 0.5161,
+      "step": 9071
+    },
+    {
+      "epoch": 0.7730720068172134,
+      "grad_norm": 13.125,
+      "learning_rate": 1.6958413919663918e-06,
+      "loss": 0.4393,
+      "step": 9072
+    },
+    {
+      "epoch": 0.7731572219855134,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6956844042122094e-06,
+      "loss": 0.3946,
+      "step": 9073
+    },
+    {
+      "epoch": 0.7732424371538134,
+      "grad_norm": 19.25,
+      "learning_rate": 1.6955274084040618e-06,
+      "loss": 1.0158,
+      "step": 9074
+    },
+    {
+      "epoch": 0.7733276523221133,
+      "grad_norm": 15.0,
+      "learning_rate": 1.695370404544786e-06,
+      "loss": 0.787,
+      "step": 9075
+    },
+    {
+      "epoch": 0.7734128674904133,
+      "grad_norm": 17.0,
+      "learning_rate": 1.6952133926372188e-06,
+      "loss": 0.8103,
+      "step": 9076
+    },
+    {
+      "epoch": 0.7734980826587132,
+      "grad_norm": 30.0,
+      "learning_rate": 1.695056372684198e-06,
+      "loss": 0.9871,
+      "step": 9077
+    },
+    {
+      "epoch": 0.7735832978270132,
+      "grad_norm": 18.875,
+      "learning_rate": 1.694899344688561e-06,
+      "loss": 1.0234,
+      "step": 9078
+    },
+    {
+      "epoch": 0.7736685129953131,
+      "grad_norm": 29.875,
+      "learning_rate": 1.6947423086531458e-06,
+      "loss": 0.5055,
+      "step": 9079
+    },
+    {
+      "epoch": 0.7737537281636131,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6945852645807895e-06,
+      "loss": 0.9303,
+      "step": 9080
+    },
+    {
+      "epoch": 0.773838943331913,
+      "grad_norm": 24.625,
+      "learning_rate": 1.6944282124743305e-06,
+      "loss": 0.8925,
+      "step": 9081
+    },
+    {
+      "epoch": 0.773924158500213,
+      "grad_norm": 24.0,
+      "learning_rate": 1.6942711523366069e-06,
+      "loss": 0.8611,
+      "step": 9082
+    },
+    {
+      "epoch": 0.774009373668513,
+      "grad_norm": 16.75,
+      "learning_rate": 1.694114084170457e-06,
+      "loss": 0.8173,
+      "step": 9083
+    },
+    {
+      "epoch": 0.7740945888368129,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6939570079787184e-06,
+      "loss": 0.9825,
+      "step": 9084
+    },
+    {
+      "epoch": 0.7741798040051129,
+      "grad_norm": 14.75,
+      "learning_rate": 1.6937999237642308e-06,
+      "loss": 0.8613,
+      "step": 9085
+    },
+    {
+      "epoch": 0.7742650191734128,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.693642831529832e-06,
+      "loss": 0.3232,
+      "step": 9086
+    },
+    {
+      "epoch": 0.7743502343417128,
+      "grad_norm": 20.375,
+      "learning_rate": 1.6934857312783613e-06,
+      "loss": 1.2607,
+      "step": 9087
+    },
+    {
+      "epoch": 0.7744354495100128,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6933286230126575e-06,
+      "loss": 0.5303,
+      "step": 9088
+    },
+    {
+      "epoch": 0.7745206646783127,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6931715067355595e-06,
+      "loss": 0.4752,
+      "step": 9089
+    },
+    {
+      "epoch": 0.7746058798466127,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.6930143824499068e-06,
+      "loss": 0.7418,
+      "step": 9090
+    },
+    {
+      "epoch": 0.7746910950149126,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6928572501585386e-06,
+      "loss": 0.8885,
+      "step": 9091
+    },
+    {
+      "epoch": 0.7747763101832126,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6927001098642948e-06,
+      "loss": 0.497,
+      "step": 9092
+    },
+    {
+      "epoch": 0.7748615253515125,
+      "grad_norm": 21.25,
+      "learning_rate": 1.6925429615700146e-06,
+      "loss": 1.0675,
+      "step": 9093
+    },
+    {
+      "epoch": 0.7749467405198125,
+      "grad_norm": 27.125,
+      "learning_rate": 1.6923858052785383e-06,
+      "loss": 1.0814,
+      "step": 9094
+    },
+    {
+      "epoch": 0.7750319556881125,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.6922286409927052e-06,
+      "loss": 0.6301,
+      "step": 9095
+    },
+    {
+      "epoch": 0.7751171708564124,
+      "grad_norm": 13.625,
+      "learning_rate": 1.6920714687153567e-06,
+      "loss": 0.6973,
+      "step": 9096
+    },
+    {
+      "epoch": 0.7752023860247124,
+      "grad_norm": 17.625,
+      "learning_rate": 1.6919142884493315e-06,
+      "loss": 0.8307,
+      "step": 9097
+    },
+    {
+      "epoch": 0.7752876011930123,
+      "grad_norm": 15.625,
+      "learning_rate": 1.6917571001974712e-06,
+      "loss": 0.5827,
+      "step": 9098
+    },
+    {
+      "epoch": 0.7753728163613123,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6915999039626153e-06,
+      "loss": 0.8677,
+      "step": 9099
+    },
+    {
+      "epoch": 0.7754580315296122,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6914426997476057e-06,
+      "loss": 0.968,
+      "step": 9100
+    },
+    {
+      "epoch": 0.7755432466979122,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6912854875552822e-06,
+      "loss": 0.6724,
+      "step": 9101
+    },
+    {
+      "epoch": 0.7756284618662121,
+      "grad_norm": 11.5,
+      "learning_rate": 1.6911282673884863e-06,
+      "loss": 0.521,
+      "step": 9102
+    },
+    {
+      "epoch": 0.7757136770345121,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6909710392500588e-06,
+      "loss": 0.7789,
+      "step": 9103
+    },
+    {
+      "epoch": 0.7757988922028121,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.690813803142842e-06,
+      "loss": 0.3545,
+      "step": 9104
+    },
+    {
+      "epoch": 0.775884107371112,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6906565590696755e-06,
+      "loss": 0.9103,
+      "step": 9105
+    },
+    {
+      "epoch": 0.775969322539412,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.6904993070334024e-06,
+      "loss": 0.5881,
+      "step": 9106
+    },
+    {
+      "epoch": 0.7760545377077119,
+      "grad_norm": 15.875,
+      "learning_rate": 1.6903420470368636e-06,
+      "loss": 0.8179,
+      "step": 9107
+    },
+    {
+      "epoch": 0.7761397528760119,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.6901847790829015e-06,
+      "loss": 0.2609,
+      "step": 9108
+    },
+    {
+      "epoch": 0.7762249680443118,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.690027503174358e-06,
+      "loss": 0.4712,
+      "step": 9109
+    },
+    {
+      "epoch": 0.7763101832126118,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6898702193140743e-06,
+      "loss": 0.7857,
+      "step": 9110
+    },
+    {
+      "epoch": 0.7763953983809118,
+      "grad_norm": 19.0,
+      "learning_rate": 1.689712927504894e-06,
+      "loss": 0.7153,
+      "step": 9111
+    },
+    {
+      "epoch": 0.7764806135492117,
+      "grad_norm": 14.5,
+      "learning_rate": 1.6895556277496588e-06,
+      "loss": 0.6683,
+      "step": 9112
+    },
+    {
+      "epoch": 0.7765658287175117,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6893983200512115e-06,
+      "loss": 1.2218,
+      "step": 9113
+    },
+    {
+      "epoch": 0.7766510438858116,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6892410044123942e-06,
+      "loss": 0.3115,
+      "step": 9114
+    },
+    {
+      "epoch": 0.7767362590541116,
+      "grad_norm": 18.125,
+      "learning_rate": 1.6890836808360509e-06,
+      "loss": 0.652,
+      "step": 9115
+    },
+    {
+      "epoch": 0.7768214742224115,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.6889263493250235e-06,
+      "loss": 0.5894,
+      "step": 9116
+    },
+    {
+      "epoch": 0.7769066893907115,
+      "grad_norm": 13.75,
+      "learning_rate": 1.688769009882156e-06,
+      "loss": 0.7257,
+      "step": 9117
+    },
+    {
+      "epoch": 0.7769919045590115,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.6886116625102907e-06,
+      "loss": 1.2324,
+      "step": 9118
+    },
+    {
+      "epoch": 0.7770771197273114,
+      "grad_norm": 14.25,
+      "learning_rate": 1.6884543072122721e-06,
+      "loss": 0.8572,
+      "step": 9119
+    },
+    {
+      "epoch": 0.7771623348956114,
+      "grad_norm": 13.375,
+      "learning_rate": 1.6882969439909432e-06,
+      "loss": 0.7263,
+      "step": 9120
+    },
+    {
+      "epoch": 0.7772475500639113,
+      "grad_norm": 15.5,
+      "learning_rate": 1.688139572849148e-06,
+      "loss": 0.6301,
+      "step": 9121
+    },
+    {
+      "epoch": 0.7773327652322113,
+      "grad_norm": 17.0,
+      "learning_rate": 1.6879821937897295e-06,
+      "loss": 0.6442,
+      "step": 9122
+    },
+    {
+      "epoch": 0.7774179804005112,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6878248068155328e-06,
+      "loss": 0.4774,
+      "step": 9123
+    },
+    {
+      "epoch": 0.7775031955688112,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6876674119294014e-06,
+      "loss": 0.9256,
+      "step": 9124
+    },
+    {
+      "epoch": 0.7775884107371112,
+      "grad_norm": 17.875,
+      "learning_rate": 1.6875100091341795e-06,
+      "loss": 0.9333,
+      "step": 9125
+    },
+    {
+      "epoch": 0.7776736259054111,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6873525984327122e-06,
+      "loss": 0.6986,
+      "step": 9126
+    },
+    {
+      "epoch": 0.7777588410737111,
+      "grad_norm": 12.25,
+      "learning_rate": 1.6871951798278435e-06,
+      "loss": 0.7764,
+      "step": 9127
+    },
+    {
+      "epoch": 0.777844056242011,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.6870377533224186e-06,
+      "loss": 0.6542,
+      "step": 9128
+    },
+    {
+      "epoch": 0.777929271410311,
+      "grad_norm": 11.25,
+      "learning_rate": 1.6868803189192815e-06,
+      "loss": 0.4523,
+      "step": 9129
+    },
+    {
+      "epoch": 0.7780144865786109,
+      "grad_norm": 18.875,
+      "learning_rate": 1.686722876621278e-06,
+      "loss": 0.6665,
+      "step": 9130
+    },
+    {
+      "epoch": 0.7780997017469109,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6865654264312529e-06,
+      "loss": 0.4918,
+      "step": 9131
+    },
+    {
+      "epoch": 0.7781849169152109,
+      "grad_norm": 12.875,
+      "learning_rate": 1.6864079683520518e-06,
+      "loss": 0.5856,
+      "step": 9132
+    },
+    {
+      "epoch": 0.7782701320835108,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.6862505023865194e-06,
+      "loss": 1.0167,
+      "step": 9133
+    },
+    {
+      "epoch": 0.7783553472518108,
+      "grad_norm": 10.875,
+      "learning_rate": 1.686093028537502e-06,
+      "loss": 0.3041,
+      "step": 9134
+    },
+    {
+      "epoch": 0.7784405624201107,
+      "grad_norm": 11.75,
+      "learning_rate": 1.6859355468078448e-06,
+      "loss": 0.5819,
+      "step": 9135
+    },
+    {
+      "epoch": 0.7785257775884107,
+      "grad_norm": 11.625,
+      "learning_rate": 1.6857780572003946e-06,
+      "loss": 0.4546,
+      "step": 9136
+    },
+    {
+      "epoch": 0.7786109927567106,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.6856205597179959e-06,
+      "loss": 0.8451,
+      "step": 9137
+    },
+    {
+      "epoch": 0.7786962079250106,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.685463054363496e-06,
+      "loss": 0.8406,
+      "step": 9138
+    },
+    {
+      "epoch": 0.7787814230933106,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6853055411397406e-06,
+      "loss": 1.0408,
+      "step": 9139
+    },
+    {
+      "epoch": 0.7788666382616106,
+      "grad_norm": 15.875,
+      "learning_rate": 1.6851480200495765e-06,
+      "loss": 0.721,
+      "step": 9140
+    },
+    {
+      "epoch": 0.7789518534299106,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6849904910958503e-06,
+      "loss": 0.603,
+      "step": 9141
+    },
+    {
+      "epoch": 0.7790370685982105,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.6848329542814081e-06,
+      "loss": 0.4813,
+      "step": 9142
+    },
+    {
+      "epoch": 0.7791222837665105,
+      "grad_norm": 13.5,
+      "learning_rate": 1.6846754096090974e-06,
+      "loss": 0.7694,
+      "step": 9143
+    },
+    {
+      "epoch": 0.7792074989348104,
+      "grad_norm": 16.0,
+      "learning_rate": 1.684517857081765e-06,
+      "loss": 1.0108,
+      "step": 9144
+    },
+    {
+      "epoch": 0.7792927141031104,
+      "grad_norm": 11.125,
+      "learning_rate": 1.684360296702258e-06,
+      "loss": 0.5011,
+      "step": 9145
+    },
+    {
+      "epoch": 0.7793779292714104,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6842027284734233e-06,
+      "loss": 0.801,
+      "step": 9146
+    },
+    {
+      "epoch": 0.7794631444397103,
+      "grad_norm": 18.25,
+      "learning_rate": 1.6840451523981089e-06,
+      "loss": 0.6285,
+      "step": 9147
+    },
+    {
+      "epoch": 0.7795483596080103,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6838875684791622e-06,
+      "loss": 0.5928,
+      "step": 9148
+    },
+    {
+      "epoch": 0.7796335747763102,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.683729976719431e-06,
+      "loss": 0.7728,
+      "step": 9149
+    },
+    {
+      "epoch": 0.7797187899446102,
+      "grad_norm": 14.5,
+      "learning_rate": 1.6835723771217624e-06,
+      "loss": 0.4603,
+      "step": 9150
+    },
+    {
+      "epoch": 0.7798040051129101,
+      "grad_norm": 16.0,
+      "learning_rate": 1.6834147696890056e-06,
+      "loss": 0.6017,
+      "step": 9151
+    },
+    {
+      "epoch": 0.7798892202812101,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6832571544240076e-06,
+      "loss": 1.0502,
+      "step": 9152
+    },
+    {
+      "epoch": 0.7799744354495101,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6830995313296175e-06,
+      "loss": 0.7913,
+      "step": 9153
+    },
+    {
+      "epoch": 0.78005965061781,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.682941900408683e-06,
+      "loss": 0.383,
+      "step": 9154
+    },
+    {
+      "epoch": 0.78014486578611,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.6827842616640533e-06,
+      "loss": 0.6808,
+      "step": 9155
+    },
+    {
+      "epoch": 0.7802300809544099,
+      "grad_norm": 17.75,
+      "learning_rate": 1.6826266150985765e-06,
+      "loss": 0.631,
+      "step": 9156
+    },
+    {
+      "epoch": 0.7803152961227099,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.6824689607151023e-06,
+      "loss": 0.8862,
+      "step": 9157
+    },
+    {
+      "epoch": 0.7804005112910098,
+      "grad_norm": 15.375,
+      "learning_rate": 1.6823112985164786e-06,
+      "loss": 0.7749,
+      "step": 9158
+    },
+    {
+      "epoch": 0.7804857264593098,
+      "grad_norm": 16.75,
+      "learning_rate": 1.682153628505555e-06,
+      "loss": 0.9177,
+      "step": 9159
+    },
+    {
+      "epoch": 0.7805709416276098,
+      "grad_norm": 16.125,
+      "learning_rate": 1.6819959506851807e-06,
+      "loss": 0.5646,
+      "step": 9160
+    },
+    {
+      "epoch": 0.7806561567959097,
+      "grad_norm": 15.25,
+      "learning_rate": 1.6818382650582055e-06,
+      "loss": 0.9146,
+      "step": 9161
+    },
+    {
+      "epoch": 0.7807413719642097,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6816805716274787e-06,
+      "loss": 0.6513,
+      "step": 9162
+    },
+    {
+      "epoch": 0.7808265871325096,
+      "grad_norm": 11.125,
+      "learning_rate": 1.681522870395849e-06,
+      "loss": 0.5193,
+      "step": 9163
+    },
+    {
+      "epoch": 0.7809118023008096,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.681365161366168e-06,
+      "loss": 0.7052,
+      "step": 9164
+    },
+    {
+      "epoch": 0.7809970174691095,
+      "grad_norm": 14.0,
+      "learning_rate": 1.6812074445412845e-06,
+      "loss": 0.4651,
+      "step": 9165
+    },
+    {
+      "epoch": 0.7810822326374095,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.6810497199240491e-06,
+      "loss": 0.5505,
+      "step": 9166
+    },
+    {
+      "epoch": 0.7811674478057095,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6808919875173113e-06,
+      "loss": 0.9399,
+      "step": 9167
+    },
+    {
+      "epoch": 0.7812526629740094,
+      "grad_norm": 19.875,
+      "learning_rate": 1.6807342473239224e-06,
+      "loss": 0.9514,
+      "step": 9168
+    },
+    {
+      "epoch": 0.7813378781423094,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.6805764993467326e-06,
+      "loss": 0.6986,
+      "step": 9169
+    },
+    {
+      "epoch": 0.7814230933106093,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.6804187435885921e-06,
+      "loss": 0.6275,
+      "step": 9170
+    },
+    {
+      "epoch": 0.7815083084789093,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6802609800523523e-06,
+      "loss": 0.8237,
+      "step": 9171
+    },
+    {
+      "epoch": 0.7815935236472092,
+      "grad_norm": 19.0,
+      "learning_rate": 1.680103208740864e-06,
+      "loss": 0.9538,
+      "step": 9172
+    },
+    {
+      "epoch": 0.7816787388155092,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6799454296569778e-06,
+      "loss": 0.7257,
+      "step": 9173
+    },
+    {
+      "epoch": 0.7817639539838092,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.679787642803546e-06,
+      "loss": 0.8362,
+      "step": 9174
+    },
+    {
+      "epoch": 0.7818491691521091,
+      "grad_norm": 14.25,
+      "learning_rate": 1.6796298481834187e-06,
+      "loss": 0.9333,
+      "step": 9175
+    },
+    {
+      "epoch": 0.7819343843204091,
+      "grad_norm": 15.875,
+      "learning_rate": 1.6794720457994484e-06,
+      "loss": 0.8981,
+      "step": 9176
+    },
+    {
+      "epoch": 0.782019599488709,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.679314235654486e-06,
+      "loss": 0.5872,
+      "step": 9177
+    },
+    {
+      "epoch": 0.782104814657009,
+      "grad_norm": 15.125,
+      "learning_rate": 1.6791564177513838e-06,
+      "loss": 0.7796,
+      "step": 9178
+    },
+    {
+      "epoch": 0.7821900298253089,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.6789985920929937e-06,
+      "loss": 1.0153,
+      "step": 9179
+    },
+    {
+      "epoch": 0.7822752449936089,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.6788407586821675e-06,
+      "loss": 0.58,
+      "step": 9180
+    },
+    {
+      "epoch": 0.7823604601619089,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6786829175217582e-06,
+      "loss": 0.4868,
+      "step": 9181
+    },
+    {
+      "epoch": 0.7824456753302088,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6785250686146164e-06,
+      "loss": 0.8926,
+      "step": 9182
+    },
+    {
+      "epoch": 0.7825308904985088,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.6783672119635966e-06,
+      "loss": 0.6253,
+      "step": 9183
+    },
+    {
+      "epoch": 0.7826161056668087,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6782093475715499e-06,
+      "loss": 0.8603,
+      "step": 9184
+    },
+    {
+      "epoch": 0.7827013208351087,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6780514754413306e-06,
+      "loss": 0.6022,
+      "step": 9185
+    },
+    {
+      "epoch": 0.7827865360034086,
+      "grad_norm": 16.375,
+      "learning_rate": 1.6778935955757897e-06,
+      "loss": 0.7856,
+      "step": 9186
+    },
+    {
+      "epoch": 0.7828717511717086,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6777357079777817e-06,
+      "loss": 0.552,
+      "step": 9187
+    },
+    {
+      "epoch": 0.7829569663400086,
+      "grad_norm": 17.375,
+      "learning_rate": 1.677577812650159e-06,
+      "loss": 0.562,
+      "step": 9188
+    },
+    {
+      "epoch": 0.7830421815083085,
+      "grad_norm": 13.0,
+      "learning_rate": 1.677419909595776e-06,
+      "loss": 0.5429,
+      "step": 9189
+    },
+    {
+      "epoch": 0.7831273966766085,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.6772619988174846e-06,
+      "loss": 0.7473,
+      "step": 9190
+    },
+    {
+      "epoch": 0.7832126118449084,
+      "grad_norm": 16.125,
+      "learning_rate": 1.6771040803181398e-06,
+      "loss": 0.8336,
+      "step": 9191
+    },
+    {
+      "epoch": 0.7832978270132084,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6769461541005944e-06,
+      "loss": 0.8144,
+      "step": 9192
+    },
+    {
+      "epoch": 0.7833830421815083,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.6767882201677028e-06,
+      "loss": 0.5138,
+      "step": 9193
+    },
+    {
+      "epoch": 0.7834682573498083,
+      "grad_norm": 14.75,
+      "learning_rate": 1.676630278522319e-06,
+      "loss": 0.8737,
+      "step": 9194
+    },
+    {
+      "epoch": 0.7835534725181083,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.676472329167297e-06,
+      "loss": 0.8462,
+      "step": 9195
+    },
+    {
+      "epoch": 0.7836386876864082,
+      "grad_norm": 14.25,
+      "learning_rate": 1.6763143721054912e-06,
+      "loss": 0.8616,
+      "step": 9196
+    },
+    {
+      "epoch": 0.7837239028547082,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.676156407339756e-06,
+      "loss": 0.6099,
+      "step": 9197
+    },
+    {
+      "epoch": 0.7838091180230081,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.6759984348729458e-06,
+      "loss": 0.9848,
+      "step": 9198
+    },
+    {
+      "epoch": 0.7838943331913081,
+      "grad_norm": 17.625,
+      "learning_rate": 1.6758404547079155e-06,
+      "loss": 0.7159,
+      "step": 9199
+    },
+    {
+      "epoch": 0.783979548359608,
+      "grad_norm": 14.875,
+      "learning_rate": 1.6756824668475202e-06,
+      "loss": 0.8403,
+      "step": 9200
+    },
+    {
+      "epoch": 0.784064763527908,
+      "grad_norm": 13.0,
+      "learning_rate": 1.6755244712946148e-06,
+      "loss": 0.7129,
+      "step": 9201
+    },
+    {
+      "epoch": 0.784149978696208,
+      "grad_norm": 15.125,
+      "learning_rate": 1.675366468052054e-06,
+      "loss": 0.5531,
+      "step": 9202
+    },
+    {
+      "epoch": 0.7842351938645079,
+      "grad_norm": 12.75,
+      "learning_rate": 1.6752084571226935e-06,
+      "loss": 0.6888,
+      "step": 9203
+    },
+    {
+      "epoch": 0.7843204090328079,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6750504385093885e-06,
+      "loss": 0.8672,
+      "step": 9204
+    },
+    {
+      "epoch": 0.7844056242011078,
+      "grad_norm": 12.0,
+      "learning_rate": 1.6748924122149948e-06,
+      "loss": 0.5386,
+      "step": 9205
+    },
+    {
+      "epoch": 0.7844908393694078,
+      "grad_norm": 24.125,
+      "learning_rate": 1.6747343782423681e-06,
+      "loss": 0.9947,
+      "step": 9206
+    },
+    {
+      "epoch": 0.7845760545377077,
+      "grad_norm": 18.0,
+      "learning_rate": 1.6745763365943634e-06,
+      "loss": 0.7725,
+      "step": 9207
+    },
+    {
+      "epoch": 0.7846612697060077,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.6744182872738379e-06,
+      "loss": 0.3866,
+      "step": 9208
+    },
+    {
+      "epoch": 0.7847464848743076,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.6742602302836471e-06,
+      "loss": 0.7012,
+      "step": 9209
+    },
+    {
+      "epoch": 0.7848317000426076,
+      "grad_norm": 60.25,
+      "learning_rate": 1.6741021656266474e-06,
+      "loss": 0.9639,
+      "step": 9210
+    },
+    {
+      "epoch": 0.7849169152109076,
+      "grad_norm": 17.0,
+      "learning_rate": 1.673944093305695e-06,
+      "loss": 0.9644,
+      "step": 9211
+    },
+    {
+      "epoch": 0.7850021303792075,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6737860133236466e-06,
+      "loss": 0.8649,
+      "step": 9212
+    },
+    {
+      "epoch": 0.7850873455475075,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6736279256833588e-06,
+      "loss": 0.6805,
+      "step": 9213
+    },
+    {
+      "epoch": 0.7851725607158074,
+      "grad_norm": 15.625,
+      "learning_rate": 1.673469830387688e-06,
+      "loss": 1.0532,
+      "step": 9214
+    },
+    {
+      "epoch": 0.7852577758841074,
+      "grad_norm": 31.125,
+      "learning_rate": 1.673311727439492e-06,
+      "loss": 0.4484,
+      "step": 9215
+    },
+    {
+      "epoch": 0.7853429910524073,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6731536168416268e-06,
+      "loss": 0.497,
+      "step": 9216
+    },
+    {
+      "epoch": 0.7854282062207073,
+      "grad_norm": 20.0,
+      "learning_rate": 1.6729954985969512e-06,
+      "loss": 1.251,
+      "step": 9217
+    },
+    {
+      "epoch": 0.7855134213890073,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6728373727083209e-06,
+      "loss": 0.6152,
+      "step": 9218
+    },
+    {
+      "epoch": 0.7855986365573072,
+      "grad_norm": 11.5,
+      "learning_rate": 1.6726792391785942e-06,
+      "loss": 0.4704,
+      "step": 9219
+    },
+    {
+      "epoch": 0.7856838517256072,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6725210980106285e-06,
+      "loss": 0.8961,
+      "step": 9220
+    },
+    {
+      "epoch": 0.7857690668939071,
+      "grad_norm": 15.0,
+      "learning_rate": 1.672362949207282e-06,
+      "loss": 0.497,
+      "step": 9221
+    },
+    {
+      "epoch": 0.7858542820622071,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6722047927714123e-06,
+      "loss": 1.0312,
+      "step": 9222
+    },
+    {
+      "epoch": 0.785939497230507,
+      "grad_norm": 21.375,
+      "learning_rate": 1.672046628705877e-06,
+      "loss": 1.1299,
+      "step": 9223
+    },
+    {
+      "epoch": 0.786024712398807,
+      "grad_norm": 21.375,
+      "learning_rate": 1.671888457013535e-06,
+      "loss": 0.9332,
+      "step": 9224
+    },
+    {
+      "epoch": 0.786109927567107,
+      "grad_norm": 13.375,
+      "learning_rate": 1.6717302776972443e-06,
+      "loss": 0.4465,
+      "step": 9225
+    },
+    {
+      "epoch": 0.7861951427354069,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6715720907598635e-06,
+      "loss": 0.6631,
+      "step": 9226
+    },
+    {
+      "epoch": 0.7862803579037069,
+      "grad_norm": 13.25,
+      "learning_rate": 1.6714138962042512e-06,
+      "loss": 0.6299,
+      "step": 9227
+    },
+    {
+      "epoch": 0.7863655730720068,
+      "grad_norm": 21.125,
+      "learning_rate": 1.6712556940332655e-06,
+      "loss": 0.7696,
+      "step": 9228
+    },
+    {
+      "epoch": 0.7864507882403068,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6710974842497663e-06,
+      "loss": 0.9325,
+      "step": 9229
+    },
+    {
+      "epoch": 0.7865360034086067,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.670939266856612e-06,
+      "loss": 0.4064,
+      "step": 9230
+    },
+    {
+      "epoch": 0.7866212185769067,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6707810418566617e-06,
+      "loss": 0.6951,
+      "step": 9231
+    },
+    {
+      "epoch": 0.7867064337452067,
+      "grad_norm": 28.875,
+      "learning_rate": 1.670622809252775e-06,
+      "loss": 1.1272,
+      "step": 9232
+    },
+    {
+      "epoch": 0.7867916489135066,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.6704645690478108e-06,
+      "loss": 0.8552,
+      "step": 9233
+    },
+    {
+      "epoch": 0.7868768640818066,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.6703063212446298e-06,
+      "loss": 0.8898,
+      "step": 9234
+    },
+    {
+      "epoch": 0.7869620792501065,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6701480658460903e-06,
+      "loss": 0.4441,
+      "step": 9235
+    },
+    {
+      "epoch": 0.7870472944184065,
+      "grad_norm": 15.125,
+      "learning_rate": 1.6699898028550528e-06,
+      "loss": 0.5961,
+      "step": 9236
+    },
+    {
+      "epoch": 0.7871325095867064,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6698315322743774e-06,
+      "loss": 0.6096,
+      "step": 9237
+    },
+    {
+      "epoch": 0.7872177247550064,
+      "grad_norm": 13.5,
+      "learning_rate": 1.6696732541069244e-06,
+      "loss": 0.794,
+      "step": 9238
+    },
+    {
+      "epoch": 0.7873029399233064,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.6695149683555528e-06,
+      "loss": 0.8106,
+      "step": 9239
+    },
+    {
+      "epoch": 0.7873881550916063,
+      "grad_norm": 12.5,
+      "learning_rate": 1.6693566750231246e-06,
+      "loss": 0.8082,
+      "step": 9240
+    },
+    {
+      "epoch": 0.7874733702599063,
+      "grad_norm": 11.5,
+      "learning_rate": 1.6691983741124992e-06,
+      "loss": 0.5338,
+      "step": 9241
+    },
+    {
+      "epoch": 0.7875585854282062,
+      "grad_norm": 16.0,
+      "learning_rate": 1.6690400656265382e-06,
+      "loss": 0.7448,
+      "step": 9242
+    },
+    {
+      "epoch": 0.7876438005965062,
+      "grad_norm": 20.5,
+      "learning_rate": 1.6688817495681012e-06,
+      "loss": 0.8265,
+      "step": 9243
+    },
+    {
+      "epoch": 0.7877290157648061,
+      "grad_norm": 16.125,
+      "learning_rate": 1.66872342594005e-06,
+      "loss": 0.8309,
+      "step": 9244
+    },
+    {
+      "epoch": 0.7878142309331061,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6685650947452452e-06,
+      "loss": 0.7554,
+      "step": 9245
+    },
+    {
+      "epoch": 0.787899446101406,
+      "grad_norm": 21.75,
+      "learning_rate": 1.6684067559865486e-06,
+      "loss": 1.0237,
+      "step": 9246
+    },
+    {
+      "epoch": 0.787984661269706,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.668248409666821e-06,
+      "loss": 0.4739,
+      "step": 9247
+    },
+    {
+      "epoch": 0.788069876438006,
+      "grad_norm": 18.0,
+      "learning_rate": 1.668090055788924e-06,
+      "loss": 0.8306,
+      "step": 9248
+    },
+    {
+      "epoch": 0.7881550916063059,
+      "grad_norm": 18.25,
+      "learning_rate": 1.6679316943557195e-06,
+      "loss": 0.801,
+      "step": 9249
+    },
+    {
+      "epoch": 0.7882403067746059,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.667773325370069e-06,
+      "loss": 0.8462,
+      "step": 9250
+    },
+    {
+      "epoch": 0.7883255219429058,
+      "grad_norm": 35.5,
+      "learning_rate": 1.6676149488348341e-06,
+      "loss": 1.1651,
+      "step": 9251
+    },
+    {
+      "epoch": 0.7884107371112058,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6674565647528773e-06,
+      "loss": 0.862,
+      "step": 9252
+    },
+    {
+      "epoch": 0.7884959522795058,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.6672981731270608e-06,
+      "loss": 0.4958,
+      "step": 9253
+    },
+    {
+      "epoch": 0.7885811674478057,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.6671397739602464e-06,
+      "loss": 0.927,
+      "step": 9254
+    },
+    {
+      "epoch": 0.7886663826161057,
+      "grad_norm": 29.0,
+      "learning_rate": 1.666981367255297e-06,
+      "loss": 0.8799,
+      "step": 9255
+    },
+    {
+      "epoch": 0.7887515977844056,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6668229530150749e-06,
+      "loss": 0.6446,
+      "step": 9256
+    },
+    {
+      "epoch": 0.7888368129527056,
+      "grad_norm": 19.75,
+      "learning_rate": 1.666664531242443e-06,
+      "loss": 0.8819,
+      "step": 9257
+    },
+    {
+      "epoch": 0.7889220281210055,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.666506101940264e-06,
+      "loss": 0.2891,
+      "step": 9258
+    },
+    {
+      "epoch": 0.7890072432893055,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.6663476651114008e-06,
+      "loss": 0.4347,
+      "step": 9259
+    },
+    {
+      "epoch": 0.7890924584576055,
+      "grad_norm": 2112.0,
+      "learning_rate": 1.6661892207587165e-06,
+      "loss": 0.4986,
+      "step": 9260
+    },
+    {
+      "epoch": 0.7891776736259054,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6660307688850747e-06,
+      "loss": 0.7337,
+      "step": 9261
+    },
+    {
+      "epoch": 0.7892628887942054,
+      "grad_norm": 16.5,
+      "learning_rate": 1.6658723094933385e-06,
+      "loss": 0.7602,
+      "step": 9262
+    },
+    {
+      "epoch": 0.7893481039625053,
+      "grad_norm": 15.75,
+      "learning_rate": 1.665713842586371e-06,
+      "loss": 0.9238,
+      "step": 9263
+    },
+    {
+      "epoch": 0.7894333191308053,
+      "grad_norm": 17.75,
+      "learning_rate": 1.6655553681670373e-06,
+      "loss": 0.8818,
+      "step": 9264
+    },
+    {
+      "epoch": 0.7895185342991052,
+      "grad_norm": 12.875,
+      "learning_rate": 1.6653968862381998e-06,
+      "loss": 0.6838,
+      "step": 9265
+    },
+    {
+      "epoch": 0.7896037494674052,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.6652383968027224e-06,
+      "loss": 0.9095,
+      "step": 9266
+    },
+    {
+      "epoch": 0.7896889646357051,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.66507989986347e-06,
+      "loss": 1.1284,
+      "step": 9267
+    },
+    {
+      "epoch": 0.7897741798040051,
+      "grad_norm": 25.875,
+      "learning_rate": 1.6649213954233066e-06,
+      "loss": 0.8976,
+      "step": 9268
+    },
+    {
+      "epoch": 0.7898593949723051,
+      "grad_norm": 15.5,
+      "learning_rate": 1.664762883485096e-06,
+      "loss": 0.5452,
+      "step": 9269
+    },
+    {
+      "epoch": 0.789944610140605,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6646043640517034e-06,
+      "loss": 0.4223,
+      "step": 9270
+    },
+    {
+      "epoch": 0.790029825308905,
+      "grad_norm": 15.125,
+      "learning_rate": 1.6644458371259926e-06,
+      "loss": 0.6444,
+      "step": 9271
+    },
+    {
+      "epoch": 0.7901150404772049,
+      "grad_norm": 19.75,
+      "learning_rate": 1.6642873027108292e-06,
+      "loss": 1.3016,
+      "step": 9272
+    },
+    {
+      "epoch": 0.7902002556455049,
+      "grad_norm": 17.75,
+      "learning_rate": 1.664128760809077e-06,
+      "loss": 0.6226,
+      "step": 9273
+    },
+    {
+      "epoch": 0.7902854708138048,
+      "grad_norm": 16.0,
+      "learning_rate": 1.6639702114236024e-06,
+      "loss": 0.859,
+      "step": 9274
+    },
+    {
+      "epoch": 0.7903706859821048,
+      "grad_norm": 25.625,
+      "learning_rate": 1.663811654557269e-06,
+      "loss": 0.9279,
+      "step": 9275
+    },
+    {
+      "epoch": 0.7904559011504048,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6636530902129436e-06,
+      "loss": 0.8733,
+      "step": 9276
+    },
+    {
+      "epoch": 0.7905411163187047,
+      "grad_norm": 15.5,
+      "learning_rate": 1.6634945183934905e-06,
+      "loss": 0.8732,
+      "step": 9277
+    },
+    {
+      "epoch": 0.7906263314870047,
+      "grad_norm": 17.375,
+      "learning_rate": 1.663335939101776e-06,
+      "loss": 0.9286,
+      "step": 9278
+    },
+    {
+      "epoch": 0.7907115466553046,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6631773523406652e-06,
+      "loss": 0.6232,
+      "step": 9279
+    },
+    {
+      "epoch": 0.7907967618236046,
+      "grad_norm": 15.625,
+      "learning_rate": 1.6630187581130241e-06,
+      "loss": 0.8644,
+      "step": 9280
+    },
+    {
+      "epoch": 0.7908819769919045,
+      "grad_norm": 15.5,
+      "learning_rate": 1.6628601564217187e-06,
+      "loss": 0.5991,
+      "step": 9281
+    },
+    {
+      "epoch": 0.7909671921602045,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6627015472696154e-06,
+      "loss": 0.6777,
+      "step": 9282
+    },
+    {
+      "epoch": 0.7910524073285045,
+      "grad_norm": 12.0,
+      "learning_rate": 1.66254293065958e-06,
+      "loss": 0.476,
+      "step": 9283
+    },
+    {
+      "epoch": 0.7911376224968044,
+      "grad_norm": 17.125,
+      "learning_rate": 1.6623843065944786e-06,
+      "loss": 0.8345,
+      "step": 9284
+    },
+    {
+      "epoch": 0.7912228376651044,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.6622256750771787e-06,
+      "loss": 0.9714,
+      "step": 9285
+    },
+    {
+      "epoch": 0.7913080528334043,
+      "grad_norm": 20.25,
+      "learning_rate": 1.6620670361105458e-06,
+      "loss": 0.796,
+      "step": 9286
+    },
+    {
+      "epoch": 0.7913932680017043,
+      "grad_norm": 14.875,
+      "learning_rate": 1.6619083896974477e-06,
+      "loss": 0.8358,
+      "step": 9287
+    },
+    {
+      "epoch": 0.7914784831700042,
+      "grad_norm": 15.375,
+      "learning_rate": 1.6617497358407505e-06,
+      "loss": 0.6862,
+      "step": 9288
+    },
+    {
+      "epoch": 0.7915636983383042,
+      "grad_norm": 14.75,
+      "learning_rate": 1.6615910745433218e-06,
+      "loss": 0.7191,
+      "step": 9289
+    },
+    {
+      "epoch": 0.7916489135066042,
+      "grad_norm": 13.25,
+      "learning_rate": 1.6614324058080277e-06,
+      "loss": 0.853,
+      "step": 9290
+    },
+    {
+      "epoch": 0.7917341286749041,
+      "grad_norm": 13.625,
+      "learning_rate": 1.6612737296377374e-06,
+      "loss": 0.6146,
+      "step": 9291
+    },
+    {
+      "epoch": 0.7918193438432041,
+      "grad_norm": 18.375,
+      "learning_rate": 1.6611150460353164e-06,
+      "loss": 0.9618,
+      "step": 9292
+    },
+    {
+      "epoch": 0.791904559011504,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6609563550036336e-06,
+      "loss": 0.5117,
+      "step": 9293
+    },
+    {
+      "epoch": 0.791989774179804,
+      "grad_norm": 11.25,
+      "learning_rate": 1.6607976565455557e-06,
+      "loss": 0.5,
+      "step": 9294
+    },
+    {
+      "epoch": 0.7920749893481039,
+      "grad_norm": 10.125,
+      "learning_rate": 1.6606389506639517e-06,
+      "loss": 0.2575,
+      "step": 9295
+    },
+    {
+      "epoch": 0.7921602045164039,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6604802373616888e-06,
+      "loss": 0.6682,
+      "step": 9296
+    },
+    {
+      "epoch": 0.7922454196847039,
+      "grad_norm": 13.25,
+      "learning_rate": 1.660321516641635e-06,
+      "loss": 0.8294,
+      "step": 9297
+    },
+    {
+      "epoch": 0.7923306348530038,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.6601627885066588e-06,
+      "loss": 0.4739,
+      "step": 9298
+    },
+    {
+      "epoch": 0.7924158500213038,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.6600040529596287e-06,
+      "loss": 0.4657,
+      "step": 9299
+    },
+    {
+      "epoch": 0.7925010651896037,
+      "grad_norm": 10.875,
+      "learning_rate": 1.659845310003413e-06,
+      "loss": 0.298,
+      "step": 9300
+    },
+    {
+      "epoch": 0.7925862803579037,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6596865596408802e-06,
+      "loss": 0.8404,
+      "step": 9301
+    },
+    {
+      "epoch": 0.7926714955262036,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.6595278018748994e-06,
+      "loss": 0.3716,
+      "step": 9302
+    },
+    {
+      "epoch": 0.7927567106945036,
+      "grad_norm": 13.125,
+      "learning_rate": 1.6593690367083398e-06,
+      "loss": 0.6689,
+      "step": 9303
+    },
+    {
+      "epoch": 0.7928419258628036,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.65921026414407e-06,
+      "loss": 0.8589,
+      "step": 9304
+    },
+    {
+      "epoch": 0.7929271410311035,
+      "grad_norm": 17.0,
+      "learning_rate": 1.6590514841849587e-06,
+      "loss": 0.8676,
+      "step": 9305
+    },
+    {
+      "epoch": 0.7930123561994035,
+      "grad_norm": 13.25,
+      "learning_rate": 1.658892696833876e-06,
+      "loss": 0.5505,
+      "step": 9306
+    },
+    {
+      "epoch": 0.7930975713677034,
+      "grad_norm": 19.75,
+      "learning_rate": 1.6587339020936912e-06,
+      "loss": 0.5966,
+      "step": 9307
+    },
+    {
+      "epoch": 0.7931827865360034,
+      "grad_norm": 14.875,
+      "learning_rate": 1.658575099967274e-06,
+      "loss": 0.7274,
+      "step": 9308
+    },
+    {
+      "epoch": 0.7932680017043033,
+      "grad_norm": 14.5,
+      "learning_rate": 1.6584162904574932e-06,
+      "loss": 0.8129,
+      "step": 9309
+    },
+    {
+      "epoch": 0.7933532168726033,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.65825747356722e-06,
+      "loss": 0.4015,
+      "step": 9310
+    },
+    {
+      "epoch": 0.7934384320409033,
+      "grad_norm": 11.0,
+      "learning_rate": 1.6580986492993234e-06,
+      "loss": 0.4577,
+      "step": 9311
+    },
+    {
+      "epoch": 0.7935236472092032,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.657939817656674e-06,
+      "loss": 0.677,
+      "step": 9312
+    },
+    {
+      "epoch": 0.7936088623775032,
+      "grad_norm": 14.5,
+      "learning_rate": 1.6577809786421417e-06,
+      "loss": 0.5482,
+      "step": 9313
+    },
+    {
+      "epoch": 0.7936940775458031,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6576221322585973e-06,
+      "loss": 1.1866,
+      "step": 9314
+    },
+    {
+      "epoch": 0.7937792927141031,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.6574632785089107e-06,
+      "loss": 0.5453,
+      "step": 9315
+    },
+    {
+      "epoch": 0.793864507882403,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.657304417395953e-06,
+      "loss": 0.4564,
+      "step": 9316
+    },
+    {
+      "epoch": 0.793949723050703,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.657145548922595e-06,
+      "loss": 0.3868,
+      "step": 9317
+    },
+    {
+      "epoch": 0.794034938219003,
+      "grad_norm": 11.625,
+      "learning_rate": 1.6569866730917074e-06,
+      "loss": 0.342,
+      "step": 9318
+    },
+    {
+      "epoch": 0.7941201533873029,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.6568277899061618e-06,
+      "loss": 0.4057,
+      "step": 9319
+    },
+    {
+      "epoch": 0.7942053685556029,
+      "grad_norm": 21.5,
+      "learning_rate": 1.6566688993688285e-06,
+      "loss": 1.0049,
+      "step": 9320
+    },
+    {
+      "epoch": 0.7942905837239028,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.6565100014825798e-06,
+      "loss": 0.5596,
+      "step": 9321
+    },
+    {
+      "epoch": 0.7943757988922028,
+      "grad_norm": 24.75,
+      "learning_rate": 1.6563510962502857e-06,
+      "loss": 0.9233,
+      "step": 9322
+    },
+    {
+      "epoch": 0.7944610140605027,
+      "grad_norm": 21.125,
+      "learning_rate": 1.6561921836748195e-06,
+      "loss": 0.673,
+      "step": 9323
+    },
+    {
+      "epoch": 0.7945462292288027,
+      "grad_norm": 22.375,
+      "learning_rate": 1.6560332637590514e-06,
+      "loss": 0.9202,
+      "step": 9324
+    },
+    {
+      "epoch": 0.7946314443971026,
+      "grad_norm": 23.0,
+      "learning_rate": 1.6558743365058546e-06,
+      "loss": 0.7403,
+      "step": 9325
+    },
+    {
+      "epoch": 0.7947166595654026,
+      "grad_norm": 15.125,
+      "learning_rate": 1.6557154019180998e-06,
+      "loss": 0.665,
+      "step": 9326
+    },
+    {
+      "epoch": 0.7948018747337026,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.6555564599986603e-06,
+      "loss": 0.8816,
+      "step": 9327
+    },
+    {
+      "epoch": 0.7948870899020025,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6553975107504072e-06,
+      "loss": 0.707,
+      "step": 9328
+    },
+    {
+      "epoch": 0.7949723050703025,
+      "grad_norm": 13.25,
+      "learning_rate": 1.655238554176214e-06,
+      "loss": 0.5945,
+      "step": 9329
+    },
+    {
+      "epoch": 0.7950575202386024,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.655079590278952e-06,
+      "loss": 0.6046,
+      "step": 9330
+    },
+    {
+      "epoch": 0.7951427354069024,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6549206190614948e-06,
+      "loss": 0.8768,
+      "step": 9331
+    },
+    {
+      "epoch": 0.7952279505752023,
+      "grad_norm": 17.0,
+      "learning_rate": 1.6547616405267147e-06,
+      "loss": 0.5921,
+      "step": 9332
+    },
+    {
+      "epoch": 0.7953131657435023,
+      "grad_norm": 13.375,
+      "learning_rate": 1.6546026546774848e-06,
+      "loss": 0.5038,
+      "step": 9333
+    },
+    {
+      "epoch": 0.7953983809118023,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6544436615166781e-06,
+      "loss": 0.6684,
+      "step": 9334
+    },
+    {
+      "epoch": 0.7954835960801022,
+      "grad_norm": 16.0,
+      "learning_rate": 1.654284661047168e-06,
+      "loss": 0.8307,
+      "step": 9335
+    },
+    {
+      "epoch": 0.7955688112484022,
+      "grad_norm": 13.375,
+      "learning_rate": 1.6541256532718272e-06,
+      "loss": 0.6982,
+      "step": 9336
+    },
+    {
+      "epoch": 0.7956540264167021,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6539666381935293e-06,
+      "loss": 0.7893,
+      "step": 9337
+    },
+    {
+      "epoch": 0.7957392415850021,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.6538076158151484e-06,
+      "loss": 0.6543,
+      "step": 9338
+    },
+    {
+      "epoch": 0.795824456753302,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.6536485861395576e-06,
+      "loss": 0.4526,
+      "step": 9339
+    },
+    {
+      "epoch": 0.795909671921602,
+      "grad_norm": 17.125,
+      "learning_rate": 1.6534895491696313e-06,
+      "loss": 0.9724,
+      "step": 9340
+    },
+    {
+      "epoch": 0.795994887089902,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.6533305049082427e-06,
+      "loss": 0.6917,
+      "step": 9341
+    },
+    {
+      "epoch": 0.7960801022582019,
+      "grad_norm": 10.875,
+      "learning_rate": 1.6531714533582664e-06,
+      "loss": 0.4756,
+      "step": 9342
+    },
+    {
+      "epoch": 0.7961653174265019,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6530123945225765e-06,
+      "loss": 0.6805,
+      "step": 9343
+    },
+    {
+      "epoch": 0.7962505325948018,
+      "grad_norm": 15.375,
+      "learning_rate": 1.6528533284040476e-06,
+      "loss": 0.8315,
+      "step": 9344
+    },
+    {
+      "epoch": 0.7963357477631018,
+      "grad_norm": 14.875,
+      "learning_rate": 1.6526942550055537e-06,
+      "loss": 0.7992,
+      "step": 9345
+    },
+    {
+      "epoch": 0.7964209629314017,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.6525351743299697e-06,
+      "loss": 0.607,
+      "step": 9346
+    },
+    {
+      "epoch": 0.7965061780997017,
+      "grad_norm": 25.125,
+      "learning_rate": 1.6523760863801705e-06,
+      "loss": 1.0917,
+      "step": 9347
+    },
+    {
+      "epoch": 0.7965913932680017,
+      "grad_norm": 17.125,
+      "learning_rate": 1.6522169911590306e-06,
+      "loss": 1.0436,
+      "step": 9348
+    },
+    {
+      "epoch": 0.7966766084363016,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6520578886694253e-06,
+      "loss": 0.8071,
+      "step": 9349
+    },
+    {
+      "epoch": 0.7967618236046016,
+      "grad_norm": 15.375,
+      "learning_rate": 1.6518987789142296e-06,
+      "loss": 0.7745,
+      "step": 9350
+    },
+    {
+      "epoch": 0.7968470387729015,
+      "grad_norm": 19.625,
+      "learning_rate": 1.6517396618963188e-06,
+      "loss": 0.8329,
+      "step": 9351
+    },
+    {
+      "epoch": 0.7969322539412015,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.6515805376185685e-06,
+      "loss": 0.5778,
+      "step": 9352
+    },
+    {
+      "epoch": 0.7970174691095014,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6514214060838538e-06,
+      "loss": 0.7726,
+      "step": 9353
+    },
+    {
+      "epoch": 0.7971026842778014,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.6512622672950506e-06,
+      "loss": 0.8389,
+      "step": 9354
+    },
+    {
+      "epoch": 0.7971878994461014,
+      "grad_norm": 19.0,
+      "learning_rate": 1.651103121255035e-06,
+      "loss": 1.2951,
+      "step": 9355
+    },
+    {
+      "epoch": 0.7972731146144013,
+      "grad_norm": 22.875,
+      "learning_rate": 1.6509439679666826e-06,
+      "loss": 1.0622,
+      "step": 9356
+    },
+    {
+      "epoch": 0.7973583297827013,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6507848074328694e-06,
+      "loss": 0.865,
+      "step": 9357
+    },
+    {
+      "epoch": 0.7974435449510012,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.6506256396564718e-06,
+      "loss": 0.5667,
+      "step": 9358
+    },
+    {
+      "epoch": 0.7975287601193012,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6504664646403662e-06,
+      "loss": 0.6396,
+      "step": 9359
+    },
+    {
+      "epoch": 0.7976139752876011,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6503072823874287e-06,
+      "loss": 0.7614,
+      "step": 9360
+    },
+    {
+      "epoch": 0.7976991904559011,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6501480929005358e-06,
+      "loss": 1.3279,
+      "step": 9361
+    },
+    {
+      "epoch": 0.797784405624201,
+      "grad_norm": 12.0,
+      "learning_rate": 1.649988896182565e-06,
+      "loss": 0.7284,
+      "step": 9362
+    },
+    {
+      "epoch": 0.797869620792501,
+      "grad_norm": 20.0,
+      "learning_rate": 1.649829692236392e-06,
+      "loss": 0.9211,
+      "step": 9363
+    },
+    {
+      "epoch": 0.797954835960801,
+      "grad_norm": 12.125,
+      "learning_rate": 1.649670481064895e-06,
+      "loss": 0.6325,
+      "step": 9364
+    },
+    {
+      "epoch": 0.7980400511291009,
+      "grad_norm": 13.875,
+      "learning_rate": 1.6495112626709505e-06,
+      "loss": 0.7835,
+      "step": 9365
+    },
+    {
+      "epoch": 0.798125266297401,
+      "grad_norm": 19.0,
+      "learning_rate": 1.6493520370574354e-06,
+      "loss": 0.8174,
+      "step": 9366
+    },
+    {
+      "epoch": 0.798210481465701,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.6491928042272276e-06,
+      "loss": 0.8003,
+      "step": 9367
+    },
+    {
+      "epoch": 0.7982956966340009,
+      "grad_norm": 20.5,
+      "learning_rate": 1.6490335641832045e-06,
+      "loss": 0.6849,
+      "step": 9368
+    },
+    {
+      "epoch": 0.7983809118023009,
+      "grad_norm": 14.125,
+      "learning_rate": 1.648874316928243e-06,
+      "loss": 0.8997,
+      "step": 9369
+    },
+    {
+      "epoch": 0.7984661269706008,
+      "grad_norm": 10.0,
+      "learning_rate": 1.648715062465222e-06,
+      "loss": 0.2884,
+      "step": 9370
+    },
+    {
+      "epoch": 0.7985513421389008,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.648555800797019e-06,
+      "loss": 0.6575,
+      "step": 9371
+    },
+    {
+      "epoch": 0.7986365573072007,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.6483965319265121e-06,
+      "loss": 0.7147,
+      "step": 9372
+    },
+    {
+      "epoch": 0.7987217724755007,
+      "grad_norm": 20.375,
+      "learning_rate": 1.6482372558565787e-06,
+      "loss": 1.119,
+      "step": 9373
+    },
+    {
+      "epoch": 0.7988069876438006,
+      "grad_norm": 12.125,
+      "learning_rate": 1.6480779725900979e-06,
+      "loss": 0.5244,
+      "step": 9374
+    },
+    {
+      "epoch": 0.7988922028121006,
+      "grad_norm": 16.0,
+      "learning_rate": 1.6479186821299475e-06,
+      "loss": 0.6104,
+      "step": 9375
+    },
+    {
+      "epoch": 0.7989774179804006,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.647759384479007e-06,
+      "loss": 0.5298,
+      "step": 9376
+    },
+    {
+      "epoch": 0.7990626331487005,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.647600079640154e-06,
+      "loss": 0.3543,
+      "step": 9377
+    },
+    {
+      "epoch": 0.7991478483170005,
+      "grad_norm": 10.5,
+      "learning_rate": 1.647440767616268e-06,
+      "loss": 0.4199,
+      "step": 9378
+    },
+    {
+      "epoch": 0.7992330634853004,
+      "grad_norm": 14.0,
+      "learning_rate": 1.6472814484102273e-06,
+      "loss": 0.7481,
+      "step": 9379
+    },
+    {
+      "epoch": 0.7993182786536004,
+      "grad_norm": 21.5,
+      "learning_rate": 1.6471221220249117e-06,
+      "loss": 0.8683,
+      "step": 9380
+    },
+    {
+      "epoch": 0.7994034938219003,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6469627884631997e-06,
+      "loss": 0.6093,
+      "step": 9381
+    },
+    {
+      "epoch": 0.7994887089902003,
+      "grad_norm": 13.5,
+      "learning_rate": 1.6468034477279712e-06,
+      "loss": 0.7172,
+      "step": 9382
+    },
+    {
+      "epoch": 0.7995739241585003,
+      "grad_norm": 19.375,
+      "learning_rate": 1.646644099822105e-06,
+      "loss": 0.8839,
+      "step": 9383
+    },
+    {
+      "epoch": 0.7996591393268002,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6464847447484817e-06,
+      "loss": 0.6429,
+      "step": 9384
+    },
+    {
+      "epoch": 0.7997443544951002,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6463253825099799e-06,
+      "loss": 0.5567,
+      "step": 9385
+    },
+    {
+      "epoch": 0.7998295696634001,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.6461660131094797e-06,
+      "loss": 0.5549,
+      "step": 9386
+    },
+    {
+      "epoch": 0.7999147848317001,
+      "grad_norm": 19.125,
+      "learning_rate": 1.6460066365498617e-06,
+      "loss": 1.2088,
+      "step": 9387
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6458472528340054e-06,
+      "loss": 0.9374,
+      "step": 9388
+    },
+    {
+      "epoch": 0.8000852151683,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6456878619647914e-06,
+      "loss": 0.5813,
+      "step": 9389
+    },
+    {
+      "epoch": 0.8001704303366,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.6455284639450996e-06,
+      "loss": 0.5815,
+      "step": 9390
+    },
+    {
+      "epoch": 0.8002556455048999,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.645369058777811e-06,
+      "loss": 0.6624,
+      "step": 9391
+    },
+    {
+      "epoch": 0.8003408606731999,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.6452096464658056e-06,
+      "loss": 0.5533,
+      "step": 9392
+    },
+    {
+      "epoch": 0.8004260758414998,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.6450502270119645e-06,
+      "loss": 0.4451,
+      "step": 9393
+    },
+    {
+      "epoch": 0.8005112910097998,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.6448908004191684e-06,
+      "loss": 0.5631,
+      "step": 9394
+    },
+    {
+      "epoch": 0.8005965061780997,
+      "grad_norm": 30.25,
+      "learning_rate": 1.644731366690299e-06,
+      "loss": 1.4518,
+      "step": 9395
+    },
+    {
+      "epoch": 0.8006817213463997,
+      "grad_norm": 12.0,
+      "learning_rate": 1.6445719258282367e-06,
+      "loss": 0.3933,
+      "step": 9396
+    },
+    {
+      "epoch": 0.8007669365146997,
+      "grad_norm": 18.125,
+      "learning_rate": 1.6444124778358628e-06,
+      "loss": 0.8528,
+      "step": 9397
+    },
+    {
+      "epoch": 0.8008521516829996,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6442530227160585e-06,
+      "loss": 0.6588,
+      "step": 9398
+    },
+    {
+      "epoch": 0.8009373668512996,
+      "grad_norm": 13.375,
+      "learning_rate": 1.6440935604717059e-06,
+      "loss": 0.9586,
+      "step": 9399
+    },
+    {
+      "epoch": 0.8010225820195995,
+      "grad_norm": 20.125,
+      "learning_rate": 1.6439340911056868e-06,
+      "loss": 1.258,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8011077971878995,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.6437746146208816e-06,
+      "loss": 0.712,
+      "step": 9401
+    },
+    {
+      "epoch": 0.8011930123561994,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.643615131020174e-06,
+      "loss": 0.7403,
+      "step": 9402
+    },
+    {
+      "epoch": 0.8012782275244994,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6434556403064445e-06,
+      "loss": 0.5882,
+      "step": 9403
+    },
+    {
+      "epoch": 0.8013634426927994,
+      "grad_norm": 15.0,
+      "learning_rate": 1.6432961424825767e-06,
+      "loss": 0.7012,
+      "step": 9404
+    },
+    {
+      "epoch": 0.8014486578610993,
+      "grad_norm": 27.5,
+      "learning_rate": 1.6431366375514515e-06,
+      "loss": 1.2114,
+      "step": 9405
+    },
+    {
+      "epoch": 0.8015338730293993,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.642977125515952e-06,
+      "loss": 0.8754,
+      "step": 9406
+    },
+    {
+      "epoch": 0.8016190881976992,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6428176063789607e-06,
+      "loss": 0.8336,
+      "step": 9407
+    },
+    {
+      "epoch": 0.8017043033659992,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6426580801433607e-06,
+      "loss": 0.7895,
+      "step": 9408
+    },
+    {
+      "epoch": 0.8017895185342991,
+      "grad_norm": 13.25,
+      "learning_rate": 1.642498546812034e-06,
+      "loss": 0.4936,
+      "step": 9409
+    },
+    {
+      "epoch": 0.8018747337025991,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6423390063878636e-06,
+      "loss": 0.5878,
+      "step": 9410
+    },
+    {
+      "epoch": 0.801959948870899,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.642179458873733e-06,
+      "loss": 0.6331,
+      "step": 9411
+    },
+    {
+      "epoch": 0.802045164039199,
+      "grad_norm": 18.75,
+      "learning_rate": 1.6420199042725258e-06,
+      "loss": 1.1841,
+      "step": 9412
+    },
+    {
+      "epoch": 0.802130379207499,
+      "grad_norm": 32.5,
+      "learning_rate": 1.6418603425871239e-06,
+      "loss": 1.244,
+      "step": 9413
+    },
+    {
+      "epoch": 0.8022155943757989,
+      "grad_norm": 11.25,
+      "learning_rate": 1.6417007738204118e-06,
+      "loss": 0.6814,
+      "step": 9414
+    },
+    {
+      "epoch": 0.8023008095440989,
+      "grad_norm": 9.25,
+      "learning_rate": 1.641541197975273e-06,
+      "loss": 0.2862,
+      "step": 9415
+    },
+    {
+      "epoch": 0.8023860247123988,
+      "grad_norm": 15.0,
+      "learning_rate": 1.6413816150545908e-06,
+      "loss": 0.767,
+      "step": 9416
+    },
+    {
+      "epoch": 0.8024712398806988,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.6412220250612496e-06,
+      "loss": 0.752,
+      "step": 9417
+    },
+    {
+      "epoch": 0.8025564550489988,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.6410624279981324e-06,
+      "loss": 0.7724,
+      "step": 9418
+    },
+    {
+      "epoch": 0.8026416702172987,
+      "grad_norm": 16.125,
+      "learning_rate": 1.6409028238681247e-06,
+      "loss": 0.9049,
+      "step": 9419
+    },
+    {
+      "epoch": 0.8027268853855987,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6407432126741093e-06,
+      "loss": 0.984,
+      "step": 9420
+    },
+    {
+      "epoch": 0.8028121005538986,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6405835944189713e-06,
+      "loss": 0.845,
+      "step": 9421
+    },
+    {
+      "epoch": 0.8028973157221986,
+      "grad_norm": 13.625,
+      "learning_rate": 1.6404239691055947e-06,
+      "loss": 0.8253,
+      "step": 9422
+    },
+    {
+      "epoch": 0.8029825308904985,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6402643367368649e-06,
+      "loss": 0.2041,
+      "step": 9423
+    },
+    {
+      "epoch": 0.8030677460587985,
+      "grad_norm": 13.0,
+      "learning_rate": 1.6401046973156656e-06,
+      "loss": 0.4541,
+      "step": 9424
+    },
+    {
+      "epoch": 0.8031529612270984,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6399450508448827e-06,
+      "loss": 0.6505,
+      "step": 9425
+    },
+    {
+      "epoch": 0.8032381763953984,
+      "grad_norm": 23.25,
+      "learning_rate": 1.6397853973274003e-06,
+      "loss": 1.4802,
+      "step": 9426
+    },
+    {
+      "epoch": 0.8033233915636984,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.6396257367661037e-06,
+      "loss": 0.6071,
+      "step": 9427
+    },
+    {
+      "epoch": 0.8034086067319983,
+      "grad_norm": 16.625,
+      "learning_rate": 1.639466069163878e-06,
+      "loss": 0.6639,
+      "step": 9428
+    },
+    {
+      "epoch": 0.8034938219002983,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.6393063945236095e-06,
+      "loss": 0.811,
+      "step": 9429
+    },
+    {
+      "epoch": 0.8035790370685982,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6391467128481825e-06,
+      "loss": 0.6467,
+      "step": 9430
+    },
+    {
+      "epoch": 0.8036642522368982,
+      "grad_norm": 13.125,
+      "learning_rate": 1.6389870241404834e-06,
+      "loss": 0.6829,
+      "step": 9431
+    },
+    {
+      "epoch": 0.8037494674051981,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6388273284033972e-06,
+      "loss": 0.8062,
+      "step": 9432
+    },
+    {
+      "epoch": 0.8038346825734981,
+      "grad_norm": 19.125,
+      "learning_rate": 1.6386676256398104e-06,
+      "loss": 0.8652,
+      "step": 9433
+    },
+    {
+      "epoch": 0.8039198977417981,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6385079158526088e-06,
+      "loss": 0.5597,
+      "step": 9434
+    },
+    {
+      "epoch": 0.804005112910098,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6383481990446787e-06,
+      "loss": 0.4233,
+      "step": 9435
+    },
+    {
+      "epoch": 0.804090328078398,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.6381884752189056e-06,
+      "loss": 0.252,
+      "step": 9436
+    },
+    {
+      "epoch": 0.8041755432466979,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6380287443781768e-06,
+      "loss": 0.8427,
+      "step": 9437
+    },
+    {
+      "epoch": 0.8042607584149979,
+      "grad_norm": 12.5,
+      "learning_rate": 1.6378690065253784e-06,
+      "loss": 0.5465,
+      "step": 9438
+    },
+    {
+      "epoch": 0.8043459735832978,
+      "grad_norm": 35.0,
+      "learning_rate": 1.6377092616633967e-06,
+      "loss": 0.5623,
+      "step": 9439
+    },
+    {
+      "epoch": 0.8044311887515978,
+      "grad_norm": 15.25,
+      "learning_rate": 1.6375495097951188e-06,
+      "loss": 0.7402,
+      "step": 9440
+    },
+    {
+      "epoch": 0.8045164039198978,
+      "grad_norm": 14.75,
+      "learning_rate": 1.6373897509234318e-06,
+      "loss": 0.7543,
+      "step": 9441
+    },
+    {
+      "epoch": 0.8046016190881977,
+      "grad_norm": 20.25,
+      "learning_rate": 1.6372299850512224e-06,
+      "loss": 1.0142,
+      "step": 9442
+    },
+    {
+      "epoch": 0.8046868342564977,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.6370702121813776e-06,
+      "loss": 1.0541,
+      "step": 9443
+    },
+    {
+      "epoch": 0.8047720494247976,
+      "grad_norm": 15.0,
+      "learning_rate": 1.6369104323167851e-06,
+      "loss": 0.525,
+      "step": 9444
+    },
+    {
+      "epoch": 0.8048572645930976,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.6367506454603319e-06,
+      "loss": 0.7357,
+      "step": 9445
+    },
+    {
+      "epoch": 0.8049424797613975,
+      "grad_norm": 15.875,
+      "learning_rate": 1.6365908516149056e-06,
+      "loss": 0.8136,
+      "step": 9446
+    },
+    {
+      "epoch": 0.8050276949296975,
+      "grad_norm": 19.0,
+      "learning_rate": 1.6364310507833938e-06,
+      "loss": 0.7438,
+      "step": 9447
+    },
+    {
+      "epoch": 0.8051129100979975,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6362712429686844e-06,
+      "loss": 0.3985,
+      "step": 9448
+    },
+    {
+      "epoch": 0.8051981252662974,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6361114281736653e-06,
+      "loss": 0.803,
+      "step": 9449
+    },
+    {
+      "epoch": 0.8052833404345974,
+      "grad_norm": 31.5,
+      "learning_rate": 1.6359516064012246e-06,
+      "loss": 0.6475,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8053685556028973,
+      "grad_norm": 20.125,
+      "learning_rate": 1.6357917776542497e-06,
+      "loss": 0.9331,
+      "step": 9451
+    },
+    {
+      "epoch": 0.8054537707711973,
+      "grad_norm": 20.125,
+      "learning_rate": 1.6356319419356298e-06,
+      "loss": 1.2658,
+      "step": 9452
+    },
+    {
+      "epoch": 0.8055389859394972,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.635472099248253e-06,
+      "loss": 0.6567,
+      "step": 9453
+    },
+    {
+      "epoch": 0.8056242011077972,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.6353122495950075e-06,
+      "loss": 0.5096,
+      "step": 9454
+    },
+    {
+      "epoch": 0.8057094162760972,
+      "grad_norm": 18.125,
+      "learning_rate": 1.635152392978782e-06,
+      "loss": 0.808,
+      "step": 9455
+    },
+    {
+      "epoch": 0.8057946314443971,
+      "grad_norm": 15.125,
+      "learning_rate": 1.634992529402466e-06,
+      "loss": 0.9815,
+      "step": 9456
+    },
+    {
+      "epoch": 0.8058798466126971,
+      "grad_norm": 30.25,
+      "learning_rate": 1.634832658868948e-06,
+      "loss": 0.9481,
+      "step": 9457
+    },
+    {
+      "epoch": 0.805965061780997,
+      "grad_norm": 23.75,
+      "learning_rate": 1.6346727813811162e-06,
+      "loss": 1.0077,
+      "step": 9458
+    },
+    {
+      "epoch": 0.806050276949297,
+      "grad_norm": 10.75,
+      "learning_rate": 1.6345128969418608e-06,
+      "loss": 0.439,
+      "step": 9459
+    },
+    {
+      "epoch": 0.8061354921175969,
+      "grad_norm": 15.5,
+      "learning_rate": 1.6343530055540704e-06,
+      "loss": 0.8602,
+      "step": 9460
+    },
+    {
+      "epoch": 0.8062207072858969,
+      "grad_norm": 22.125,
+      "learning_rate": 1.6341931072206352e-06,
+      "loss": 0.7514,
+      "step": 9461
+    },
+    {
+      "epoch": 0.8063059224541969,
+      "grad_norm": 21.375,
+      "learning_rate": 1.6340332019444438e-06,
+      "loss": 0.935,
+      "step": 9462
+    },
+    {
+      "epoch": 0.8063911376224968,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.6338732897283866e-06,
+      "loss": 1.0326,
+      "step": 9463
+    },
+    {
+      "epoch": 0.8064763527907968,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6337133705753523e-06,
+      "loss": 0.9387,
+      "step": 9464
+    },
+    {
+      "epoch": 0.8065615679590967,
+      "grad_norm": 10.375,
+      "learning_rate": 1.6335534444882326e-06,
+      "loss": 0.4275,
+      "step": 9465
+    },
+    {
+      "epoch": 0.8066467831273967,
+      "grad_norm": 20.0,
+      "learning_rate": 1.6333935114699156e-06,
+      "loss": 0.6612,
+      "step": 9466
+    },
+    {
+      "epoch": 0.8067319982956966,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6332335715232927e-06,
+      "loss": 0.5352,
+      "step": 9467
+    },
+    {
+      "epoch": 0.8068172134639966,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6330736246512535e-06,
+      "loss": 0.5707,
+      "step": 9468
+    },
+    {
+      "epoch": 0.8069024286322966,
+      "grad_norm": 24.5,
+      "learning_rate": 1.6329136708566892e-06,
+      "loss": 0.7378,
+      "step": 9469
+    },
+    {
+      "epoch": 0.8069876438005965,
+      "grad_norm": 11.75,
+      "learning_rate": 1.6327537101424894e-06,
+      "loss": 0.4521,
+      "step": 9470
+    },
+    {
+      "epoch": 0.8070728589688965,
+      "grad_norm": 13.25,
+      "learning_rate": 1.6325937425115454e-06,
+      "loss": 0.8363,
+      "step": 9471
+    },
+    {
+      "epoch": 0.8071580741371964,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6324337679667477e-06,
+      "loss": 0.4928,
+      "step": 9472
+    },
+    {
+      "epoch": 0.8072432893054964,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6322737865109872e-06,
+      "loss": 0.5882,
+      "step": 9473
+    },
+    {
+      "epoch": 0.8073285044737963,
+      "grad_norm": 11.875,
+      "learning_rate": 1.6321137981471551e-06,
+      "loss": 0.5267,
+      "step": 9474
+    },
+    {
+      "epoch": 0.8074137196420963,
+      "grad_norm": 14.25,
+      "learning_rate": 1.6319538028781423e-06,
+      "loss": 0.635,
+      "step": 9475
+    },
+    {
+      "epoch": 0.8074989348103963,
+      "grad_norm": 16.25,
+      "learning_rate": 1.63179380070684e-06,
+      "loss": 1.034,
+      "step": 9476
+    },
+    {
+      "epoch": 0.8075841499786962,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6316337916361402e-06,
+      "loss": 1.1004,
+      "step": 9477
+    },
+    {
+      "epoch": 0.8076693651469962,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.631473775668934e-06,
+      "loss": 0.5225,
+      "step": 9478
+    },
+    {
+      "epoch": 0.8077545803152961,
+      "grad_norm": 12.0,
+      "learning_rate": 1.6313137528081125e-06,
+      "loss": 0.6254,
+      "step": 9479
+    },
+    {
+      "epoch": 0.8078397954835961,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.6311537230565686e-06,
+      "loss": 0.5595,
+      "step": 9480
+    },
+    {
+      "epoch": 0.807925010651896,
+      "grad_norm": 15.0,
+      "learning_rate": 1.6309936864171935e-06,
+      "loss": 0.4604,
+      "step": 9481
+    },
+    {
+      "epoch": 0.808010225820196,
+      "grad_norm": 18.875,
+      "learning_rate": 1.6308336428928794e-06,
+      "loss": 0.9487,
+      "step": 9482
+    },
+    {
+      "epoch": 0.808095440988496,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6306735924865184e-06,
+      "loss": 0.5752,
+      "step": 9483
+    },
+    {
+      "epoch": 0.8081806561567959,
+      "grad_norm": 18.0,
+      "learning_rate": 1.6305135352010026e-06,
+      "loss": 0.9865,
+      "step": 9484
+    },
+    {
+      "epoch": 0.8082658713250959,
+      "grad_norm": 10.5,
+      "learning_rate": 1.630353471039225e-06,
+      "loss": 0.2656,
+      "step": 9485
+    },
+    {
+      "epoch": 0.8083510864933958,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6301934000040773e-06,
+      "loss": 0.8126,
+      "step": 9486
+    },
+    {
+      "epoch": 0.8084363016616958,
+      "grad_norm": 18.625,
+      "learning_rate": 1.6300333220984526e-06,
+      "loss": 1.0819,
+      "step": 9487
+    },
+    {
+      "epoch": 0.8085215168299957,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.6298732373252436e-06,
+      "loss": 0.8976,
+      "step": 9488
+    },
+    {
+      "epoch": 0.8086067319982957,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.6297131456873433e-06,
+      "loss": 0.6188,
+      "step": 9489
+    },
+    {
+      "epoch": 0.8086919471665956,
+      "grad_norm": 56.25,
+      "learning_rate": 1.6295530471876448e-06,
+      "loss": 0.7695,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8087771623348956,
+      "grad_norm": 15.25,
+      "learning_rate": 1.629392941829041e-06,
+      "loss": 0.7134,
+      "step": 9491
+    },
+    {
+      "epoch": 0.8088623775031956,
+      "grad_norm": 24.375,
+      "learning_rate": 1.629232829614425e-06,
+      "loss": 0.8882,
+      "step": 9492
+    },
+    {
+      "epoch": 0.8089475926714955,
+      "grad_norm": 14.75,
+      "learning_rate": 1.6290727105466906e-06,
+      "loss": 0.7056,
+      "step": 9493
+    },
+    {
+      "epoch": 0.8090328078397955,
+      "grad_norm": 12.0,
+      "learning_rate": 1.6289125846287313e-06,
+      "loss": 0.5221,
+      "step": 9494
+    },
+    {
+      "epoch": 0.8091180230080954,
+      "grad_norm": 16.375,
+      "learning_rate": 1.6287524518634402e-06,
+      "loss": 0.6398,
+      "step": 9495
+    },
+    {
+      "epoch": 0.8092032381763954,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6285923122537114e-06,
+      "loss": 0.8392,
+      "step": 9496
+    },
+    {
+      "epoch": 0.8092884533446953,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.628432165802439e-06,
+      "loss": 0.4834,
+      "step": 9497
+    },
+    {
+      "epoch": 0.8093736685129953,
+      "grad_norm": 12.75,
+      "learning_rate": 1.628272012512517e-06,
+      "loss": 0.7347,
+      "step": 9498
+    },
+    {
+      "epoch": 0.8094588836812953,
+      "grad_norm": 16.0,
+      "learning_rate": 1.6281118523868395e-06,
+      "loss": 0.9386,
+      "step": 9499
+    },
+    {
+      "epoch": 0.8095440988495952,
+      "grad_norm": 17.125,
+      "learning_rate": 1.6279516854283e-06,
+      "loss": 0.8599,
+      "step": 9500
+    },
+    {
+      "epoch": 0.8096293140178952,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.6277915116397938e-06,
+      "loss": 0.5091,
+      "step": 9501
+    },
+    {
+      "epoch": 0.8097145291861951,
+      "grad_norm": 18.75,
+      "learning_rate": 1.6276313310242151e-06,
+      "loss": 0.5912,
+      "step": 9502
+    },
+    {
+      "epoch": 0.8097997443544951,
+      "grad_norm": 15.0,
+      "learning_rate": 1.6274711435844582e-06,
+      "loss": 0.649,
+      "step": 9503
+    },
+    {
+      "epoch": 0.809884959522795,
+      "grad_norm": 10.625,
+      "learning_rate": 1.6273109493234184e-06,
+      "loss": 0.2716,
+      "step": 9504
+    },
+    {
+      "epoch": 0.809970174691095,
+      "grad_norm": 22.5,
+      "learning_rate": 1.6271507482439907e-06,
+      "loss": 0.3357,
+      "step": 9505
+    },
+    {
+      "epoch": 0.810055389859395,
+      "grad_norm": 10.75,
+      "learning_rate": 1.6269905403490693e-06,
+      "loss": 0.506,
+      "step": 9506
+    },
+    {
+      "epoch": 0.8101406050276949,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.6268303256415496e-06,
+      "loss": 1.1093,
+      "step": 9507
+    },
+    {
+      "epoch": 0.8102258201959949,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.6266701041243275e-06,
+      "loss": 0.4562,
+      "step": 9508
+    },
+    {
+      "epoch": 0.8103110353642948,
+      "grad_norm": 10.25,
+      "learning_rate": 1.6265098758002972e-06,
+      "loss": 0.2915,
+      "step": 9509
+    },
+    {
+      "epoch": 0.8103962505325948,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6263496406723556e-06,
+      "loss": 0.7986,
+      "step": 9510
+    },
+    {
+      "epoch": 0.8104814657008947,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.6261893987433968e-06,
+      "loss": 0.6524,
+      "step": 9511
+    },
+    {
+      "epoch": 0.8105666808691947,
+      "grad_norm": 10.75,
+      "learning_rate": 1.6260291500163177e-06,
+      "loss": 0.4812,
+      "step": 9512
+    },
+    {
+      "epoch": 0.8106518960374947,
+      "grad_norm": 25.875,
+      "learning_rate": 1.6258688944940138e-06,
+      "loss": 0.9877,
+      "step": 9513
+    },
+    {
+      "epoch": 0.8107371112057946,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6257086321793812e-06,
+      "loss": 1.069,
+      "step": 9514
+    },
+    {
+      "epoch": 0.8108223263740946,
+      "grad_norm": 22.625,
+      "learning_rate": 1.625548363075315e-06,
+      "loss": 0.8805,
+      "step": 9515
+    },
+    {
+      "epoch": 0.8109075415423945,
+      "grad_norm": 12.75,
+      "learning_rate": 1.6253880871847127e-06,
+      "loss": 0.493,
+      "step": 9516
+    },
+    {
+      "epoch": 0.8109927567106945,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.6252278045104703e-06,
+      "loss": 0.6482,
+      "step": 9517
+    },
+    {
+      "epoch": 0.8110779718789944,
+      "grad_norm": 14.125,
+      "learning_rate": 1.625067515055484e-06,
+      "loss": 0.3299,
+      "step": 9518
+    },
+    {
+      "epoch": 0.8111631870472944,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6249072188226505e-06,
+      "loss": 0.9892,
+      "step": 9519
+    },
+    {
+      "epoch": 0.8112484022155944,
+      "grad_norm": 18.625,
+      "learning_rate": 1.6247469158148668e-06,
+      "loss": 0.6329,
+      "step": 9520
+    },
+    {
+      "epoch": 0.8113336173838943,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.624586606035029e-06,
+      "loss": 0.6345,
+      "step": 9521
+    },
+    {
+      "epoch": 0.8114188325521943,
+      "grad_norm": 25.625,
+      "learning_rate": 1.6244262894860352e-06,
+      "loss": 0.9274,
+      "step": 9522
+    },
+    {
+      "epoch": 0.8115040477204942,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6242659661707816e-06,
+      "loss": 1.0147,
+      "step": 9523
+    },
+    {
+      "epoch": 0.8115892628887942,
+      "grad_norm": 22.375,
+      "learning_rate": 1.6241056360921654e-06,
+      "loss": 1.0602,
+      "step": 9524
+    },
+    {
+      "epoch": 0.8116744780570941,
+      "grad_norm": 23.375,
+      "learning_rate": 1.6239452992530843e-06,
+      "loss": 0.6488,
+      "step": 9525
+    },
+    {
+      "epoch": 0.8117596932253941,
+      "grad_norm": 23.0,
+      "learning_rate": 1.6237849556564355e-06,
+      "loss": 1.0389,
+      "step": 9526
+    },
+    {
+      "epoch": 0.811844908393694,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.6236246053051172e-06,
+      "loss": 0.6787,
+      "step": 9527
+    },
+    {
+      "epoch": 0.811930123561994,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6234642482020257e-06,
+      "loss": 0.5816,
+      "step": 9528
+    },
+    {
+      "epoch": 0.812015338730294,
+      "grad_norm": 19.0,
+      "learning_rate": 1.6233038843500604e-06,
+      "loss": 0.6062,
+      "step": 9529
+    },
+    {
+      "epoch": 0.8121005538985939,
+      "grad_norm": 30.25,
+      "learning_rate": 1.6231435137521184e-06,
+      "loss": 0.9707,
+      "step": 9530
+    },
+    {
+      "epoch": 0.8121857690668939,
+      "grad_norm": 14.75,
+      "learning_rate": 1.622983136411098e-06,
+      "loss": 0.7146,
+      "step": 9531
+    },
+    {
+      "epoch": 0.8122709842351938,
+      "grad_norm": 11.5,
+      "learning_rate": 1.6228227523298968e-06,
+      "loss": 0.4218,
+      "step": 9532
+    },
+    {
+      "epoch": 0.8123561994034938,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.622662361511414e-06,
+      "loss": 0.395,
+      "step": 9533
+    },
+    {
+      "epoch": 0.8124414145717938,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.6225019639585473e-06,
+      "loss": 0.7819,
+      "step": 9534
+    },
+    {
+      "epoch": 0.8125266297400937,
+      "grad_norm": 10.625,
+      "learning_rate": 1.6223415596741957e-06,
+      "loss": 0.3752,
+      "step": 9535
+    },
+    {
+      "epoch": 0.8126118449083937,
+      "grad_norm": 18.375,
+      "learning_rate": 1.6221811486612574e-06,
+      "loss": 1.1257,
+      "step": 9536
+    },
+    {
+      "epoch": 0.8126970600766936,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.6220207309226318e-06,
+      "loss": 0.7075,
+      "step": 9537
+    },
+    {
+      "epoch": 0.8127822752449936,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.6218603064612176e-06,
+      "loss": 0.2056,
+      "step": 9538
+    },
+    {
+      "epoch": 0.8128674904132935,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6216998752799132e-06,
+      "loss": 0.6924,
+      "step": 9539
+    },
+    {
+      "epoch": 0.8129527055815935,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.6215394373816187e-06,
+      "loss": 0.4378,
+      "step": 9540
+    },
+    {
+      "epoch": 0.8130379207498935,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6213789927692328e-06,
+      "loss": 0.7156,
+      "step": 9541
+    },
+    {
+      "epoch": 0.8131231359181934,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6212185414456554e-06,
+      "loss": 1.0168,
+      "step": 9542
+    },
+    {
+      "epoch": 0.8132083510864934,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.621058083413785e-06,
+      "loss": 0.3976,
+      "step": 9543
+    },
+    {
+      "epoch": 0.8132935662547933,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.6208976186765223e-06,
+      "loss": 0.7649,
+      "step": 9544
+    },
+    {
+      "epoch": 0.8133787814230933,
+      "grad_norm": 12.0,
+      "learning_rate": 1.6207371472367666e-06,
+      "loss": 0.4154,
+      "step": 9545
+    },
+    {
+      "epoch": 0.8134639965913932,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.6205766690974185e-06,
+      "loss": 0.6174,
+      "step": 9546
+    },
+    {
+      "epoch": 0.8135492117596932,
+      "grad_norm": 22.0,
+      "learning_rate": 1.6204161842613764e-06,
+      "loss": 0.9451,
+      "step": 9547
+    },
+    {
+      "epoch": 0.8136344269279931,
+      "grad_norm": 13.0,
+      "learning_rate": 1.6202556927315419e-06,
+      "loss": 0.4744,
+      "step": 9548
+    },
+    {
+      "epoch": 0.8137196420962931,
+      "grad_norm": 21.125,
+      "learning_rate": 1.6200951945108143e-06,
+      "loss": 0.8855,
+      "step": 9549
+    },
+    {
+      "epoch": 0.8138048572645931,
+      "grad_norm": 19.5,
+      "learning_rate": 1.6199346896020948e-06,
+      "loss": 1.2526,
+      "step": 9550
+    },
+    {
+      "epoch": 0.813890072432893,
+      "grad_norm": 33.25,
+      "learning_rate": 1.6197741780082832e-06,
+      "loss": 1.0598,
+      "step": 9551
+    },
+    {
+      "epoch": 0.813975287601193,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.6196136597322803e-06,
+      "loss": 0.7703,
+      "step": 9552
+    },
+    {
+      "epoch": 0.8140605027694929,
+      "grad_norm": 16.625,
+      "learning_rate": 1.619453134776987e-06,
+      "loss": 0.8769,
+      "step": 9553
+    },
+    {
+      "epoch": 0.8141457179377929,
+      "grad_norm": 11.125,
+      "learning_rate": 1.6192926031453041e-06,
+      "loss": 0.4337,
+      "step": 9554
+    },
+    {
+      "epoch": 0.8142309331060928,
+      "grad_norm": 34.5,
+      "learning_rate": 1.6191320648401325e-06,
+      "loss": 0.8267,
+      "step": 9555
+    },
+    {
+      "epoch": 0.8143161482743928,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6189715198643732e-06,
+      "loss": 0.5283,
+      "step": 9556
+    },
+    {
+      "epoch": 0.8144013634426928,
+      "grad_norm": 23.75,
+      "learning_rate": 1.6188109682209274e-06,
+      "loss": 0.7032,
+      "step": 9557
+    },
+    {
+      "epoch": 0.8144865786109927,
+      "grad_norm": 16.125,
+      "learning_rate": 1.6186504099126971e-06,
+      "loss": 0.755,
+      "step": 9558
+    },
+    {
+      "epoch": 0.8145717937792927,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.618489844942583e-06,
+      "loss": 0.9224,
+      "step": 9559
+    },
+    {
+      "epoch": 0.8146570089475926,
+      "grad_norm": 14.875,
+      "learning_rate": 1.618329273313486e-06,
+      "loss": 0.7357,
+      "step": 9560
+    },
+    {
+      "epoch": 0.8147422241158926,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.6181686950283096e-06,
+      "loss": 0.7173,
+      "step": 9561
+    },
+    {
+      "epoch": 0.8148274392841925,
+      "grad_norm": 13.125,
+      "learning_rate": 1.6180081100899543e-06,
+      "loss": 0.6681,
+      "step": 9562
+    },
+    {
+      "epoch": 0.8149126544524925,
+      "grad_norm": 11.875,
+      "learning_rate": 1.6178475185013226e-06,
+      "loss": 0.5094,
+      "step": 9563
+    },
+    {
+      "epoch": 0.8149978696207925,
+      "grad_norm": 15.625,
+      "learning_rate": 1.617686920265316e-06,
+      "loss": 0.9409,
+      "step": 9564
+    },
+    {
+      "epoch": 0.8150830847890924,
+      "grad_norm": 9.5,
+      "learning_rate": 1.6175263153848372e-06,
+      "loss": 0.3378,
+      "step": 9565
+    },
+    {
+      "epoch": 0.8151682999573924,
+      "grad_norm": 22.25,
+      "learning_rate": 1.6173657038627881e-06,
+      "loss": 0.8048,
+      "step": 9566
+    },
+    {
+      "epoch": 0.8152535151256923,
+      "grad_norm": 12.875,
+      "learning_rate": 1.6172050857020722e-06,
+      "loss": 0.6447,
+      "step": 9567
+    },
+    {
+      "epoch": 0.8153387302939923,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.61704446090559e-06,
+      "loss": 0.7769,
+      "step": 9568
+    },
+    {
+      "epoch": 0.8154239454622922,
+      "grad_norm": 16.125,
+      "learning_rate": 1.616883829476246e-06,
+      "loss": 0.6015,
+      "step": 9569
+    },
+    {
+      "epoch": 0.8155091606305922,
+      "grad_norm": 11.375,
+      "learning_rate": 1.6167231914169419e-06,
+      "loss": 0.3633,
+      "step": 9570
+    },
+    {
+      "epoch": 0.8155943757988922,
+      "grad_norm": 12.125,
+      "learning_rate": 1.6165625467305812e-06,
+      "loss": 0.4571,
+      "step": 9571
+    },
+    {
+      "epoch": 0.8156795909671921,
+      "grad_norm": 11.375,
+      "learning_rate": 1.6164018954200666e-06,
+      "loss": 0.4366,
+      "step": 9572
+    },
+    {
+      "epoch": 0.8157648061354921,
+      "grad_norm": 13.125,
+      "learning_rate": 1.616241237488301e-06,
+      "loss": 0.798,
+      "step": 9573
+    },
+    {
+      "epoch": 0.815850021303792,
+      "grad_norm": 18.75,
+      "learning_rate": 1.6160805729381884e-06,
+      "loss": 0.897,
+      "step": 9574
+    },
+    {
+      "epoch": 0.815935236472092,
+      "grad_norm": 19.625,
+      "learning_rate": 1.6159199017726316e-06,
+      "loss": 0.5949,
+      "step": 9575
+    },
+    {
+      "epoch": 0.8160204516403919,
+      "grad_norm": 18.0,
+      "learning_rate": 1.615759223994534e-06,
+      "loss": 0.6339,
+      "step": 9576
+    },
+    {
+      "epoch": 0.8161056668086919,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6155985396067997e-06,
+      "loss": 0.9377,
+      "step": 9577
+    },
+    {
+      "epoch": 0.8161908819769919,
+      "grad_norm": 19.0,
+      "learning_rate": 1.615437848612332e-06,
+      "loss": 0.7673,
+      "step": 9578
+    },
+    {
+      "epoch": 0.8162760971452918,
+      "grad_norm": 19.875,
+      "learning_rate": 1.6152771510140352e-06,
+      "loss": 0.9419,
+      "step": 9579
+    },
+    {
+      "epoch": 0.8163613123135918,
+      "grad_norm": 18.25,
+      "learning_rate": 1.6151164468148128e-06,
+      "loss": 1.1502,
+      "step": 9580
+    },
+    {
+      "epoch": 0.8164465274818917,
+      "grad_norm": 19.25,
+      "learning_rate": 1.6149557360175689e-06,
+      "loss": 0.889,
+      "step": 9581
+    },
+    {
+      "epoch": 0.8165317426501917,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.6147950186252078e-06,
+      "loss": 0.6464,
+      "step": 9582
+    },
+    {
+      "epoch": 0.8166169578184916,
+      "grad_norm": 18.875,
+      "learning_rate": 1.614634294640634e-06,
+      "loss": 0.6655,
+      "step": 9583
+    },
+    {
+      "epoch": 0.8167021729867916,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.614473564066752e-06,
+      "loss": 0.5105,
+      "step": 9584
+    },
+    {
+      "epoch": 0.8167873881550916,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.614312826906466e-06,
+      "loss": 0.6503,
+      "step": 9585
+    },
+    {
+      "epoch": 0.8168726033233915,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.6141520831626808e-06,
+      "loss": 0.4852,
+      "step": 9586
+    },
+    {
+      "epoch": 0.8169578184916915,
+      "grad_norm": 16.375,
+      "learning_rate": 1.6139913328383013e-06,
+      "loss": 0.559,
+      "step": 9587
+    },
+    {
+      "epoch": 0.8170430336599914,
+      "grad_norm": 20.5,
+      "learning_rate": 1.6138305759362325e-06,
+      "loss": 1.1046,
+      "step": 9588
+    },
+    {
+      "epoch": 0.8171282488282914,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6136698124593795e-06,
+      "loss": 1.0394,
+      "step": 9589
+    },
+    {
+      "epoch": 0.8172134639965913,
+      "grad_norm": 18.0,
+      "learning_rate": 1.613509042410647e-06,
+      "loss": 0.971,
+      "step": 9590
+    },
+    {
+      "epoch": 0.8172986791648913,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.613348265792941e-06,
+      "loss": 0.6667,
+      "step": 9591
+    },
+    {
+      "epoch": 0.8173838943331914,
+      "grad_norm": 19.0,
+      "learning_rate": 1.6131874826091655e-06,
+      "loss": 0.9719,
+      "step": 9592
+    },
+    {
+      "epoch": 0.8174691095014913,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.613026692862228e-06,
+      "loss": 0.813,
+      "step": 9593
+    },
+    {
+      "epoch": 0.8175543246697913,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.6128658965550323e-06,
+      "loss": 0.5615,
+      "step": 9594
+    },
+    {
+      "epoch": 0.8176395398380912,
+      "grad_norm": 25.375,
+      "learning_rate": 1.612705093690486e-06,
+      "loss": 1.2454,
+      "step": 9595
+    },
+    {
+      "epoch": 0.8177247550063912,
+      "grad_norm": 18.0,
+      "learning_rate": 1.6125442842714928e-06,
+      "loss": 0.8158,
+      "step": 9596
+    },
+    {
+      "epoch": 0.8178099701746911,
+      "grad_norm": 19.5,
+      "learning_rate": 1.6123834683009603e-06,
+      "loss": 0.8649,
+      "step": 9597
+    },
+    {
+      "epoch": 0.8178951853429911,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.6122226457817941e-06,
+      "loss": 1.1615,
+      "step": 9598
+    },
+    {
+      "epoch": 0.8179804005112911,
+      "grad_norm": 19.0,
+      "learning_rate": 1.612061816716901e-06,
+      "loss": 0.2495,
+      "step": 9599
+    },
+    {
+      "epoch": 0.818065615679591,
+      "grad_norm": 15.0,
+      "learning_rate": 1.611900981109186e-06,
+      "loss": 1.3293,
+      "step": 9600
+    },
+    {
+      "epoch": 0.818150830847891,
+      "grad_norm": 13.0,
+      "learning_rate": 1.6117401389615568e-06,
+      "loss": 0.5224,
+      "step": 9601
+    },
+    {
+      "epoch": 0.8182360460161909,
+      "grad_norm": 16.25,
+      "learning_rate": 1.611579290276919e-06,
+      "loss": 0.9867,
+      "step": 9602
+    },
+    {
+      "epoch": 0.8183212611844909,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6114184350581802e-06,
+      "loss": 0.7392,
+      "step": 9603
+    },
+    {
+      "epoch": 0.8184064763527908,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.611257573308247e-06,
+      "loss": 0.9066,
+      "step": 9604
+    },
+    {
+      "epoch": 0.8184916915210908,
+      "grad_norm": 13.5,
+      "learning_rate": 1.611096705030026e-06,
+      "loss": 0.6662,
+      "step": 9605
+    },
+    {
+      "epoch": 0.8185769066893908,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.6109358302264245e-06,
+      "loss": 0.409,
+      "step": 9606
+    },
+    {
+      "epoch": 0.8186621218576907,
+      "grad_norm": 10.375,
+      "learning_rate": 1.6107749489003493e-06,
+      "loss": 0.5051,
+      "step": 9607
+    },
+    {
+      "epoch": 0.8187473370259907,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6106140610547083e-06,
+      "loss": 0.8585,
+      "step": 9608
+    },
+    {
+      "epoch": 0.8188325521942906,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.6104531666924084e-06,
+      "loss": 0.6428,
+      "step": 9609
+    },
+    {
+      "epoch": 0.8189177673625906,
+      "grad_norm": 21.875,
+      "learning_rate": 1.6102922658163575e-06,
+      "loss": 0.8343,
+      "step": 9610
+    },
+    {
+      "epoch": 0.8190029825308905,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6101313584294632e-06,
+      "loss": 0.9247,
+      "step": 9611
+    },
+    {
+      "epoch": 0.8190881976991905,
+      "grad_norm": 13.875,
+      "learning_rate": 1.6099704445346328e-06,
+      "loss": 0.626,
+      "step": 9612
+    },
+    {
+      "epoch": 0.8191734128674905,
+      "grad_norm": 18.875,
+      "learning_rate": 1.6098095241347744e-06,
+      "loss": 1.1083,
+      "step": 9613
+    },
+    {
+      "epoch": 0.8192586280357904,
+      "grad_norm": 18.25,
+      "learning_rate": 1.6096485972327962e-06,
+      "loss": 0.761,
+      "step": 9614
+    },
+    {
+      "epoch": 0.8193438432040904,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.6094876638316064e-06,
+      "loss": 0.4208,
+      "step": 9615
+    },
+    {
+      "epoch": 0.8194290583723903,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.609326723934113e-06,
+      "loss": 0.603,
+      "step": 9616
+    },
+    {
+      "epoch": 0.8195142735406903,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.6091657775432243e-06,
+      "loss": 0.7158,
+      "step": 9617
+    },
+    {
+      "epoch": 0.8195994887089902,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.609004824661849e-06,
+      "loss": 0.6577,
+      "step": 9618
+    },
+    {
+      "epoch": 0.8196847038772902,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.608843865292895e-06,
+      "loss": 0.8638,
+      "step": 9619
+    },
+    {
+      "epoch": 0.8197699190455902,
+      "grad_norm": 19.75,
+      "learning_rate": 1.608682899439272e-06,
+      "loss": 1.058,
+      "step": 9620
+    },
+    {
+      "epoch": 0.8198551342138901,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.608521927103888e-06,
+      "loss": 0.3731,
+      "step": 9621
+    },
+    {
+      "epoch": 0.8199403493821901,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.6083609482896524e-06,
+      "loss": 0.6416,
+      "step": 9622
+    },
+    {
+      "epoch": 0.82002556455049,
+      "grad_norm": 13.875,
+      "learning_rate": 1.6081999629994743e-06,
+      "loss": 0.564,
+      "step": 9623
+    },
+    {
+      "epoch": 0.82011077971879,
+      "grad_norm": 17.75,
+      "learning_rate": 1.608038971236262e-06,
+      "loss": 0.8689,
+      "step": 9624
+    },
+    {
+      "epoch": 0.8201959948870899,
+      "grad_norm": 16.375,
+      "learning_rate": 1.6078779730029262e-06,
+      "loss": 0.5513,
+      "step": 9625
+    },
+    {
+      "epoch": 0.8202812100553899,
+      "grad_norm": 12.5,
+      "learning_rate": 1.6077169683023749e-06,
+      "loss": 0.6399,
+      "step": 9626
+    },
+    {
+      "epoch": 0.8203664252236899,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6075559571375188e-06,
+      "loss": 0.8342,
+      "step": 9627
+    },
+    {
+      "epoch": 0.8204516403919898,
+      "grad_norm": 13.75,
+      "learning_rate": 1.607394939511267e-06,
+      "loss": 0.7329,
+      "step": 9628
+    },
+    {
+      "epoch": 0.8205368555602898,
+      "grad_norm": 13.75,
+      "learning_rate": 1.607233915426529e-06,
+      "loss": 0.6268,
+      "step": 9629
+    },
+    {
+      "epoch": 0.8206220707285897,
+      "grad_norm": 11.625,
+      "learning_rate": 1.607072884886215e-06,
+      "loss": 0.3986,
+      "step": 9630
+    },
+    {
+      "epoch": 0.8207072858968897,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.606911847893235e-06,
+      "loss": 0.6255,
+      "step": 9631
+    },
+    {
+      "epoch": 0.8207925010651896,
+      "grad_norm": 14.125,
+      "learning_rate": 1.606750804450499e-06,
+      "loss": 0.7942,
+      "step": 9632
+    },
+    {
+      "epoch": 0.8208777162334896,
+      "grad_norm": 25.875,
+      "learning_rate": 1.6065897545609172e-06,
+      "loss": 0.9177,
+      "step": 9633
+    },
+    {
+      "epoch": 0.8209629314017896,
+      "grad_norm": 25.125,
+      "learning_rate": 1.6064286982273997e-06,
+      "loss": 1.2602,
+      "step": 9634
+    },
+    {
+      "epoch": 0.8210481465700895,
+      "grad_norm": 20.125,
+      "learning_rate": 1.6062676354528575e-06,
+      "loss": 1.3172,
+      "step": 9635
+    },
+    {
+      "epoch": 0.8211333617383895,
+      "grad_norm": 14.5,
+      "learning_rate": 1.6061065662402008e-06,
+      "loss": 0.5658,
+      "step": 9636
+    },
+    {
+      "epoch": 0.8212185769066894,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.6059454905923405e-06,
+      "loss": 0.7155,
+      "step": 9637
+    },
+    {
+      "epoch": 0.8213037920749894,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.6057844085121868e-06,
+      "loss": 0.7629,
+      "step": 9638
+    },
+    {
+      "epoch": 0.8213890072432893,
+      "grad_norm": 17.0,
+      "learning_rate": 1.6056233200026517e-06,
+      "loss": 0.7474,
+      "step": 9639
+    },
+    {
+      "epoch": 0.8214742224115893,
+      "grad_norm": 11.75,
+      "learning_rate": 1.6054622250666452e-06,
+      "loss": 0.5844,
+      "step": 9640
+    },
+    {
+      "epoch": 0.8215594375798893,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.6053011237070786e-06,
+      "loss": 0.9581,
+      "step": 9641
+    },
+    {
+      "epoch": 0.8216446527481892,
+      "grad_norm": 12.125,
+      "learning_rate": 1.6051400159268638e-06,
+      "loss": 0.5039,
+      "step": 9642
+    },
+    {
+      "epoch": 0.8217298679164892,
+      "grad_norm": 12.25,
+      "learning_rate": 1.604978901728912e-06,
+      "loss": 0.4849,
+      "step": 9643
+    },
+    {
+      "epoch": 0.8218150830847891,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.6048177811161342e-06,
+      "loss": 0.6487,
+      "step": 9644
+    },
+    {
+      "epoch": 0.8219002982530891,
+      "grad_norm": 12.75,
+      "learning_rate": 1.604656654091442e-06,
+      "loss": 0.7016,
+      "step": 9645
+    },
+    {
+      "epoch": 0.821985513421389,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.6044955206577479e-06,
+      "loss": 0.8086,
+      "step": 9646
+    },
+    {
+      "epoch": 0.822070728589689,
+      "grad_norm": 12.125,
+      "learning_rate": 1.6043343808179628e-06,
+      "loss": 0.5771,
+      "step": 9647
+    },
+    {
+      "epoch": 0.822155943757989,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.604173234575e-06,
+      "loss": 0.9748,
+      "step": 9648
+    },
+    {
+      "epoch": 0.8222411589262889,
+      "grad_norm": 16.125,
+      "learning_rate": 1.6040120819317698e-06,
+      "loss": 0.7285,
+      "step": 9649
+    },
+    {
+      "epoch": 0.8223263740945889,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6038509228911855e-06,
+      "loss": 0.7807,
+      "step": 9650
+    },
+    {
+      "epoch": 0.8224115892628888,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.6036897574561594e-06,
+      "loss": 1.0198,
+      "step": 9651
+    },
+    {
+      "epoch": 0.8224968044311888,
+      "grad_norm": 20.625,
+      "learning_rate": 1.603528585629604e-06,
+      "loss": 0.9035,
+      "step": 9652
+    },
+    {
+      "epoch": 0.8225820195994887,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.603367407414431e-06,
+      "loss": 0.7171,
+      "step": 9653
+    },
+    {
+      "epoch": 0.8226672347677887,
+      "grad_norm": 16.5,
+      "learning_rate": 1.6032062228135536e-06,
+      "loss": 0.6194,
+      "step": 9654
+    },
+    {
+      "epoch": 0.8227524499360886,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6030450318298847e-06,
+      "loss": 0.886,
+      "step": 9655
+    },
+    {
+      "epoch": 0.8228376651043886,
+      "grad_norm": 27.75,
+      "learning_rate": 1.6028838344663371e-06,
+      "loss": 1.0366,
+      "step": 9656
+    },
+    {
+      "epoch": 0.8229228802726886,
+      "grad_norm": 12.5,
+      "learning_rate": 1.6027226307258239e-06,
+      "loss": 0.4763,
+      "step": 9657
+    },
+    {
+      "epoch": 0.8230080954409885,
+      "grad_norm": 15.125,
+      "learning_rate": 1.6025614206112583e-06,
+      "loss": 0.5358,
+      "step": 9658
+    },
+    {
+      "epoch": 0.8230933106092885,
+      "grad_norm": 17.875,
+      "learning_rate": 1.6024002041255527e-06,
+      "loss": 1.0267,
+      "step": 9659
+    },
+    {
+      "epoch": 0.8231785257775884,
+      "grad_norm": 11.5,
+      "learning_rate": 1.6022389812716214e-06,
+      "loss": 0.4523,
+      "step": 9660
+    },
+    {
+      "epoch": 0.8232637409458884,
+      "grad_norm": 24.625,
+      "learning_rate": 1.6020777520523776e-06,
+      "loss": 0.8703,
+      "step": 9661
+    },
+    {
+      "epoch": 0.8233489561141883,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.6019165164707342e-06,
+      "loss": 0.1767,
+      "step": 9662
+    },
+    {
+      "epoch": 0.8234341712824883,
+      "grad_norm": 13.25,
+      "learning_rate": 1.6017552745296061e-06,
+      "loss": 0.6279,
+      "step": 9663
+    },
+    {
+      "epoch": 0.8235193864507883,
+      "grad_norm": 15.875,
+      "learning_rate": 1.6015940262319064e-06,
+      "loss": 0.7645,
+      "step": 9664
+    },
+    {
+      "epoch": 0.8236046016190882,
+      "grad_norm": 11.5,
+      "learning_rate": 1.601432771580549e-06,
+      "loss": 0.4265,
+      "step": 9665
+    },
+    {
+      "epoch": 0.8236898167873882,
+      "grad_norm": 18.0,
+      "learning_rate": 1.6012715105784477e-06,
+      "loss": 0.6059,
+      "step": 9666
+    },
+    {
+      "epoch": 0.8237750319556881,
+      "grad_norm": 19.625,
+      "learning_rate": 1.6011102432285172e-06,
+      "loss": 0.8687,
+      "step": 9667
+    },
+    {
+      "epoch": 0.8238602471239881,
+      "grad_norm": 17.625,
+      "learning_rate": 1.6009489695336717e-06,
+      "loss": 0.9237,
+      "step": 9668
+    },
+    {
+      "epoch": 0.823945462292288,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.6007876894968252e-06,
+      "loss": 0.7254,
+      "step": 9669
+    },
+    {
+      "epoch": 0.824030677460588,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6006264031208923e-06,
+      "loss": 0.8134,
+      "step": 9670
+    },
+    {
+      "epoch": 0.824115892628888,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.600465110408788e-06,
+      "loss": 0.571,
+      "step": 9671
+    },
+    {
+      "epoch": 0.8242011077971879,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6003038113634268e-06,
+      "loss": 0.892,
+      "step": 9672
+    },
+    {
+      "epoch": 0.8242863229654879,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6001425059877235e-06,
+      "loss": 0.7139,
+      "step": 9673
+    },
+    {
+      "epoch": 0.8243715381337878,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5999811942845924e-06,
+      "loss": 0.6816,
+      "step": 9674
+    },
+    {
+      "epoch": 0.8244567533020878,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.59981987625695e-06,
+      "loss": 0.4999,
+      "step": 9675
+    },
+    {
+      "epoch": 0.8245419684703877,
+      "grad_norm": 37.75,
+      "learning_rate": 1.5996585519077102e-06,
+      "loss": 1.0457,
+      "step": 9676
+    },
+    {
+      "epoch": 0.8246271836386877,
+      "grad_norm": 21.375,
+      "learning_rate": 1.5994972212397885e-06,
+      "loss": 0.7863,
+      "step": 9677
+    },
+    {
+      "epoch": 0.8247123988069877,
+      "grad_norm": 14.5,
+      "learning_rate": 1.599335884256101e-06,
+      "loss": 0.5265,
+      "step": 9678
+    },
+    {
+      "epoch": 0.8247976139752876,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5991745409595628e-06,
+      "loss": 0.689,
+      "step": 9679
+    },
+    {
+      "epoch": 0.8248828291435876,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.5990131913530897e-06,
+      "loss": 0.6493,
+      "step": 9680
+    },
+    {
+      "epoch": 0.8249680443118875,
+      "grad_norm": 14.75,
+      "learning_rate": 1.5988518354395965e-06,
+      "loss": 0.5115,
+      "step": 9681
+    },
+    {
+      "epoch": 0.8250532594801875,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5986904732220005e-06,
+      "loss": 0.8284,
+      "step": 9682
+    },
+    {
+      "epoch": 0.8251384746484874,
+      "grad_norm": 17.25,
+      "learning_rate": 1.5985291047032166e-06,
+      "loss": 0.928,
+      "step": 9683
+    },
+    {
+      "epoch": 0.8252236898167874,
+      "grad_norm": 21.875,
+      "learning_rate": 1.5983677298861622e-06,
+      "loss": 0.8874,
+      "step": 9684
+    },
+    {
+      "epoch": 0.8253089049850874,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5982063487737514e-06,
+      "loss": 0.565,
+      "step": 9685
+    },
+    {
+      "epoch": 0.8253941201533873,
+      "grad_norm": 11.875,
+      "learning_rate": 1.5980449613689025e-06,
+      "loss": 0.4842,
+      "step": 9686
+    },
+    {
+      "epoch": 0.8254793353216873,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5978835676745309e-06,
+      "loss": 0.8157,
+      "step": 9687
+    },
+    {
+      "epoch": 0.8255645504899872,
+      "grad_norm": 20.375,
+      "learning_rate": 1.5977221676935536e-06,
+      "loss": 0.4956,
+      "step": 9688
+    },
+    {
+      "epoch": 0.8256497656582872,
+      "grad_norm": 18.125,
+      "learning_rate": 1.597560761428887e-06,
+      "loss": 0.79,
+      "step": 9689
+    },
+    {
+      "epoch": 0.8257349808265871,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5973993488834476e-06,
+      "loss": 0.619,
+      "step": 9690
+    },
+    {
+      "epoch": 0.8258201959948871,
+      "grad_norm": 12.125,
+      "learning_rate": 1.597237930060153e-06,
+      "loss": 0.6217,
+      "step": 9691
+    },
+    {
+      "epoch": 0.825905411163187,
+      "grad_norm": 12.875,
+      "learning_rate": 1.5970765049619197e-06,
+      "loss": 0.6151,
+      "step": 9692
+    },
+    {
+      "epoch": 0.825990626331487,
+      "grad_norm": 12.875,
+      "learning_rate": 1.596915073591665e-06,
+      "loss": 0.6642,
+      "step": 9693
+    },
+    {
+      "epoch": 0.826075841499787,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.596753635952306e-06,
+      "loss": 0.3249,
+      "step": 9694
+    },
+    {
+      "epoch": 0.8261610566680869,
+      "grad_norm": 13.375,
+      "learning_rate": 1.59659219204676e-06,
+      "loss": 0.968,
+      "step": 9695
+    },
+    {
+      "epoch": 0.8262462718363869,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5964307418779445e-06,
+      "loss": 1.0474,
+      "step": 9696
+    },
+    {
+      "epoch": 0.8263314870046868,
+      "grad_norm": 18.25,
+      "learning_rate": 1.5962692854487777e-06,
+      "loss": 0.9244,
+      "step": 9697
+    },
+    {
+      "epoch": 0.8264167021729868,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.5961078227621762e-06,
+      "loss": 0.9517,
+      "step": 9698
+    },
+    {
+      "epoch": 0.8265019173412868,
+      "grad_norm": 20.125,
+      "learning_rate": 1.5959463538210584e-06,
+      "loss": 0.89,
+      "step": 9699
+    },
+    {
+      "epoch": 0.8265871325095867,
+      "grad_norm": 10.25,
+      "learning_rate": 1.595784878628342e-06,
+      "loss": 0.2895,
+      "step": 9700
+    },
+    {
+      "epoch": 0.8266723476778867,
+      "grad_norm": 17.0,
+      "learning_rate": 1.5956233971869455e-06,
+      "loss": 0.8022,
+      "step": 9701
+    },
+    {
+      "epoch": 0.8267575628461866,
+      "grad_norm": 12.75,
+      "learning_rate": 1.5954619094997865e-06,
+      "loss": 0.7376,
+      "step": 9702
+    },
+    {
+      "epoch": 0.8268427780144866,
+      "grad_norm": 20.5,
+      "learning_rate": 1.5953004155697835e-06,
+      "loss": 1.0007,
+      "step": 9703
+    },
+    {
+      "epoch": 0.8269279931827865,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.5951389153998542e-06,
+      "loss": 0.7068,
+      "step": 9704
+    },
+    {
+      "epoch": 0.8270132083510865,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5949774089929185e-06,
+      "loss": 0.643,
+      "step": 9705
+    },
+    {
+      "epoch": 0.8270984235193864,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5948158963518937e-06,
+      "loss": 1.1723,
+      "step": 9706
+    },
+    {
+      "epoch": 0.8271836386876864,
+      "grad_norm": 22.875,
+      "learning_rate": 1.5946543774796988e-06,
+      "loss": 0.9569,
+      "step": 9707
+    },
+    {
+      "epoch": 0.8272688538559864,
+      "grad_norm": 16.0,
+      "learning_rate": 1.5944928523792527e-06,
+      "loss": 0.6086,
+      "step": 9708
+    },
+    {
+      "epoch": 0.8273540690242863,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5943313210534744e-06,
+      "loss": 0.9288,
+      "step": 9709
+    },
+    {
+      "epoch": 0.8274392841925863,
+      "grad_norm": 12.75,
+      "learning_rate": 1.594169783505283e-06,
+      "loss": 0.4042,
+      "step": 9710
+    },
+    {
+      "epoch": 0.8275244993608862,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.5940082397375972e-06,
+      "loss": 0.7382,
+      "step": 9711
+    },
+    {
+      "epoch": 0.8276097145291862,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.5938466897533373e-06,
+      "loss": 0.6501,
+      "step": 9712
+    },
+    {
+      "epoch": 0.8276949296974861,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5936851335554212e-06,
+      "loss": 0.442,
+      "step": 9713
+    },
+    {
+      "epoch": 0.8277801448657861,
+      "grad_norm": 16.875,
+      "learning_rate": 1.5935235711467696e-06,
+      "loss": 0.79,
+      "step": 9714
+    },
+    {
+      "epoch": 0.8278653600340861,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.5933620025303012e-06,
+      "loss": 0.7064,
+      "step": 9715
+    },
+    {
+      "epoch": 0.827950575202386,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5932004277089365e-06,
+      "loss": 0.8353,
+      "step": 9716
+    },
+    {
+      "epoch": 0.828035790370686,
+      "grad_norm": 29.875,
+      "learning_rate": 1.593038846685595e-06,
+      "loss": 1.2232,
+      "step": 9717
+    },
+    {
+      "epoch": 0.8281210055389859,
+      "grad_norm": 11.875,
+      "learning_rate": 1.5928772594631964e-06,
+      "loss": 0.5594,
+      "step": 9718
+    },
+    {
+      "epoch": 0.8282062207072859,
+      "grad_norm": 19.875,
+      "learning_rate": 1.5927156660446608e-06,
+      "loss": 0.6795,
+      "step": 9719
+    },
+    {
+      "epoch": 0.8282914358755858,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.5925540664329086e-06,
+      "loss": 0.5582,
+      "step": 9720
+    },
+    {
+      "epoch": 0.8283766510438858,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.5923924606308601e-06,
+      "loss": 0.3808,
+      "step": 9721
+    },
+    {
+      "epoch": 0.8284618662121858,
+      "grad_norm": 14.5,
+      "learning_rate": 1.5922308486414357e-06,
+      "loss": 0.6694,
+      "step": 9722
+    },
+    {
+      "epoch": 0.8285470813804857,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5920692304675554e-06,
+      "loss": 0.8276,
+      "step": 9723
+    },
+    {
+      "epoch": 0.8286322965487857,
+      "grad_norm": 32.25,
+      "learning_rate": 1.5919076061121402e-06,
+      "loss": 0.861,
+      "step": 9724
+    },
+    {
+      "epoch": 0.8287175117170856,
+      "grad_norm": 20.875,
+      "learning_rate": 1.591745975578111e-06,
+      "loss": 0.6563,
+      "step": 9725
+    },
+    {
+      "epoch": 0.8288027268853856,
+      "grad_norm": 25.875,
+      "learning_rate": 1.5915843388683878e-06,
+      "loss": 1.0286,
+      "step": 9726
+    },
+    {
+      "epoch": 0.8288879420536855,
+      "grad_norm": 33.75,
+      "learning_rate": 1.5914226959858925e-06,
+      "loss": 1.3663,
+      "step": 9727
+    },
+    {
+      "epoch": 0.8289731572219855,
+      "grad_norm": 13.625,
+      "learning_rate": 1.5912610469335458e-06,
+      "loss": 0.7798,
+      "step": 9728
+    },
+    {
+      "epoch": 0.8290583723902855,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5910993917142687e-06,
+      "loss": 0.9191,
+      "step": 9729
+    },
+    {
+      "epoch": 0.8291435875585854,
+      "grad_norm": 19.125,
+      "learning_rate": 1.5909377303309823e-06,
+      "loss": 1.1412,
+      "step": 9730
+    },
+    {
+      "epoch": 0.8292288027268854,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.5907760627866087e-06,
+      "loss": 0.4271,
+      "step": 9731
+    },
+    {
+      "epoch": 0.8293140178951853,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5906143890840687e-06,
+      "loss": 0.4894,
+      "step": 9732
+    },
+    {
+      "epoch": 0.8293992330634853,
+      "grad_norm": 10.625,
+      "learning_rate": 1.5904527092262846e-06,
+      "loss": 1.027,
+      "step": 9733
+    },
+    {
+      "epoch": 0.8294844482317852,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5902910232161772e-06,
+      "loss": 0.9958,
+      "step": 9734
+    },
+    {
+      "epoch": 0.8295696634000852,
+      "grad_norm": 21.125,
+      "learning_rate": 1.590129331056669e-06,
+      "loss": 0.8219,
+      "step": 9735
+    },
+    {
+      "epoch": 0.8296548785683852,
+      "grad_norm": 20.25,
+      "learning_rate": 1.5899676327506816e-06,
+      "loss": 0.572,
+      "step": 9736
+    },
+    {
+      "epoch": 0.8297400937366851,
+      "grad_norm": 18.5,
+      "learning_rate": 1.5898059283011377e-06,
+      "loss": 0.9582,
+      "step": 9737
+    },
+    {
+      "epoch": 0.8298253089049851,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5896442177109584e-06,
+      "loss": 0.5103,
+      "step": 9738
+    },
+    {
+      "epoch": 0.829910524073285,
+      "grad_norm": 17.875,
+      "learning_rate": 1.589482500983067e-06,
+      "loss": 0.7345,
+      "step": 9739
+    },
+    {
+      "epoch": 0.829995739241585,
+      "grad_norm": 11.75,
+      "learning_rate": 1.5893207781203851e-06,
+      "loss": 0.58,
+      "step": 9740
+    },
+    {
+      "epoch": 0.8300809544098849,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.5891590491258358e-06,
+      "loss": 0.8765,
+      "step": 9741
+    },
+    {
+      "epoch": 0.8301661695781849,
+      "grad_norm": 14.25,
+      "learning_rate": 1.5889973140023414e-06,
+      "loss": 0.6009,
+      "step": 9742
+    },
+    {
+      "epoch": 0.8302513847464849,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5888355727528243e-06,
+      "loss": 0.7213,
+      "step": 9743
+    },
+    {
+      "epoch": 0.8303365999147848,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.5886738253802078e-06,
+      "loss": 0.5879,
+      "step": 9744
+    },
+    {
+      "epoch": 0.8304218150830848,
+      "grad_norm": 15.75,
+      "learning_rate": 1.588512071887415e-06,
+      "loss": 0.9383,
+      "step": 9745
+    },
+    {
+      "epoch": 0.8305070302513847,
+      "grad_norm": 17.0,
+      "learning_rate": 1.5883503122773686e-06,
+      "loss": 0.7263,
+      "step": 9746
+    },
+    {
+      "epoch": 0.8305922454196847,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.5881885465529914e-06,
+      "loss": 0.9516,
+      "step": 9747
+    },
+    {
+      "epoch": 0.8306774605879846,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5880267747172076e-06,
+      "loss": 0.5601,
+      "step": 9748
+    },
+    {
+      "epoch": 0.8307626757562846,
+      "grad_norm": 13.625,
+      "learning_rate": 1.5878649967729397e-06,
+      "loss": 0.7941,
+      "step": 9749
+    },
+    {
+      "epoch": 0.8308478909245846,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.587703212723112e-06,
+      "loss": 0.7843,
+      "step": 9750
+    },
+    {
+      "epoch": 0.8309331060928845,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.5875414225706471e-06,
+      "loss": 0.4843,
+      "step": 9751
+    },
+    {
+      "epoch": 0.8310183212611845,
+      "grad_norm": 20.5,
+      "learning_rate": 1.5873796263184698e-06,
+      "loss": 0.7809,
+      "step": 9752
+    },
+    {
+      "epoch": 0.8311035364294844,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5872178239695029e-06,
+      "loss": 0.5725,
+      "step": 9753
+    },
+    {
+      "epoch": 0.8311887515977844,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5870560155266711e-06,
+      "loss": 0.9388,
+      "step": 9754
+    },
+    {
+      "epoch": 0.8312739667660843,
+      "grad_norm": 22.75,
+      "learning_rate": 1.5868942009928983e-06,
+      "loss": 1.2478,
+      "step": 9755
+    },
+    {
+      "epoch": 0.8313591819343843,
+      "grad_norm": 14.25,
+      "learning_rate": 1.5867323803711081e-06,
+      "loss": 0.7398,
+      "step": 9756
+    },
+    {
+      "epoch": 0.8314443971026843,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5865705536642257e-06,
+      "loss": 0.8815,
+      "step": 9757
+    },
+    {
+      "epoch": 0.8315296122709842,
+      "grad_norm": 24.375,
+      "learning_rate": 1.586408720875175e-06,
+      "loss": 1.1097,
+      "step": 9758
+    },
+    {
+      "epoch": 0.8316148274392842,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5862468820068796e-06,
+      "loss": 0.5781,
+      "step": 9759
+    },
+    {
+      "epoch": 0.8317000426075841,
+      "grad_norm": 14.625,
+      "learning_rate": 1.5860850370622654e-06,
+      "loss": 0.3199,
+      "step": 9760
+    },
+    {
+      "epoch": 0.8317852577758841,
+      "grad_norm": 10.875,
+      "learning_rate": 1.5859231860442568e-06,
+      "loss": 0.5379,
+      "step": 9761
+    },
+    {
+      "epoch": 0.831870472944184,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5857613289557782e-06,
+      "loss": 0.5074,
+      "step": 9762
+    },
+    {
+      "epoch": 0.831955688112484,
+      "grad_norm": 18.875,
+      "learning_rate": 1.5855994657997547e-06,
+      "loss": 0.9243,
+      "step": 9763
+    },
+    {
+      "epoch": 0.832040903280784,
+      "grad_norm": 15.75,
+      "learning_rate": 1.5854375965791112e-06,
+      "loss": 0.9017,
+      "step": 9764
+    },
+    {
+      "epoch": 0.8321261184490839,
+      "grad_norm": 21.25,
+      "learning_rate": 1.5852757212967737e-06,
+      "loss": 0.9408,
+      "step": 9765
+    },
+    {
+      "epoch": 0.8322113336173839,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.5851138399556664e-06,
+      "loss": 0.4767,
+      "step": 9766
+    },
+    {
+      "epoch": 0.8322965487856838,
+      "grad_norm": 10.5,
+      "learning_rate": 1.584951952558715e-06,
+      "loss": 0.475,
+      "step": 9767
+    },
+    {
+      "epoch": 0.8323817639539838,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5847900591088449e-06,
+      "loss": 1.1705,
+      "step": 9768
+    },
+    {
+      "epoch": 0.8324669791222837,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5846281596089822e-06,
+      "loss": 0.7003,
+      "step": 9769
+    },
+    {
+      "epoch": 0.8325521942905837,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5844662540620517e-06,
+      "loss": 0.7125,
+      "step": 9770
+    },
+    {
+      "epoch": 0.8326374094588836,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.58430434247098e-06,
+      "loss": 0.3365,
+      "step": 9771
+    },
+    {
+      "epoch": 0.8327226246271836,
+      "grad_norm": 12.5,
+      "learning_rate": 1.5841424248386922e-06,
+      "loss": 0.6143,
+      "step": 9772
+    },
+    {
+      "epoch": 0.8328078397954836,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.5839805011681154e-06,
+      "loss": 0.6069,
+      "step": 9773
+    },
+    {
+      "epoch": 0.8328930549637835,
+      "grad_norm": 21.25,
+      "learning_rate": 1.5838185714621745e-06,
+      "loss": 0.653,
+      "step": 9774
+    },
+    {
+      "epoch": 0.8329782701320835,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5836566357237968e-06,
+      "loss": 0.8936,
+      "step": 9775
+    },
+    {
+      "epoch": 0.8330634853003834,
+      "grad_norm": 14.75,
+      "learning_rate": 1.5834946939559079e-06,
+      "loss": 0.5792,
+      "step": 9776
+    },
+    {
+      "epoch": 0.8331487004686834,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5833327461614345e-06,
+      "loss": 1.1301,
+      "step": 9777
+    },
+    {
+      "epoch": 0.8332339156369833,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.5831707923433034e-06,
+      "loss": 0.8825,
+      "step": 9778
+    },
+    {
+      "epoch": 0.8333191308052833,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.5830088325044407e-06,
+      "loss": 0.7664,
+      "step": 9779
+    },
+    {
+      "epoch": 0.8334043459735833,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.5828468666477738e-06,
+      "loss": 0.2657,
+      "step": 9780
+    },
+    {
+      "epoch": 0.8334895611418832,
+      "grad_norm": 13.25,
+      "learning_rate": 1.5826848947762292e-06,
+      "loss": 0.5132,
+      "step": 9781
+    },
+    {
+      "epoch": 0.8335747763101832,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.5825229168927344e-06,
+      "loss": 0.4001,
+      "step": 9782
+    },
+    {
+      "epoch": 0.8336599914784831,
+      "grad_norm": 14.875,
+      "learning_rate": 1.5823609330002155e-06,
+      "loss": 0.7679,
+      "step": 9783
+    },
+    {
+      "epoch": 0.8337452066467831,
+      "grad_norm": 11.0,
+      "learning_rate": 1.5821989431016007e-06,
+      "loss": 0.5385,
+      "step": 9784
+    },
+    {
+      "epoch": 0.833830421815083,
+      "grad_norm": 12.5,
+      "learning_rate": 1.5820369471998167e-06,
+      "loss": 0.5615,
+      "step": 9785
+    },
+    {
+      "epoch": 0.833915636983383,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5818749452977917e-06,
+      "loss": 0.6283,
+      "step": 9786
+    },
+    {
+      "epoch": 0.834000852151683,
+      "grad_norm": 19.375,
+      "learning_rate": 1.5817129373984522e-06,
+      "loss": 0.2593,
+      "step": 9787
+    },
+    {
+      "epoch": 0.8340860673199829,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5815509235047266e-06,
+      "loss": 0.6131,
+      "step": 9788
+    },
+    {
+      "epoch": 0.8341712824882829,
+      "grad_norm": 16.5,
+      "learning_rate": 1.581388903619542e-06,
+      "loss": 0.664,
+      "step": 9789
+    },
+    {
+      "epoch": 0.8342564976565828,
+      "grad_norm": 24.25,
+      "learning_rate": 1.5812268777458275e-06,
+      "loss": 0.8757,
+      "step": 9790
+    },
+    {
+      "epoch": 0.8343417128248828,
+      "grad_norm": 20.0,
+      "learning_rate": 1.5810648458865096e-06,
+      "loss": 1.0316,
+      "step": 9791
+    },
+    {
+      "epoch": 0.8344269279931827,
+      "grad_norm": 12.5,
+      "learning_rate": 1.580902808044517e-06,
+      "loss": 0.5827,
+      "step": 9792
+    },
+    {
+      "epoch": 0.8345121431614827,
+      "grad_norm": 13.875,
+      "learning_rate": 1.5807407642227782e-06,
+      "loss": 0.6134,
+      "step": 9793
+    },
+    {
+      "epoch": 0.8345973583297827,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.580578714424221e-06,
+      "loss": 0.9598,
+      "step": 9794
+    },
+    {
+      "epoch": 0.8346825734980826,
+      "grad_norm": 15.875,
+      "learning_rate": 1.580416658651774e-06,
+      "loss": 0.9506,
+      "step": 9795
+    },
+    {
+      "epoch": 0.8347677886663826,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.5802545969083657e-06,
+      "loss": 0.363,
+      "step": 9796
+    },
+    {
+      "epoch": 0.8348530038346825,
+      "grad_norm": 18.625,
+      "learning_rate": 1.5800925291969248e-06,
+      "loss": 0.8018,
+      "step": 9797
+    },
+    {
+      "epoch": 0.8349382190029825,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5799304555203804e-06,
+      "loss": 1.0559,
+      "step": 9798
+    },
+    {
+      "epoch": 0.8350234341712824,
+      "grad_norm": 19.125,
+      "learning_rate": 1.5797683758816606e-06,
+      "loss": 0.8637,
+      "step": 9799
+    },
+    {
+      "epoch": 0.8351086493395824,
+      "grad_norm": 14.75,
+      "learning_rate": 1.5796062902836942e-06,
+      "loss": 0.7914,
+      "step": 9800
+    },
+    {
+      "epoch": 0.8351938645078824,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.5794441987294113e-06,
+      "loss": 0.5058,
+      "step": 9801
+    },
+    {
+      "epoch": 0.8352790796761823,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.5792821012217401e-06,
+      "loss": 0.6316,
+      "step": 9802
+    },
+    {
+      "epoch": 0.8353642948444823,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5791199977636104e-06,
+      "loss": 0.7757,
+      "step": 9803
+    },
+    {
+      "epoch": 0.8354495100127822,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5789578883579512e-06,
+      "loss": 0.4614,
+      "step": 9804
+    },
+    {
+      "epoch": 0.8355347251810822,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.5787957730076922e-06,
+      "loss": 0.3779,
+      "step": 9805
+    },
+    {
+      "epoch": 0.8356199403493821,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.578633651715763e-06,
+      "loss": 0.5543,
+      "step": 9806
+    },
+    {
+      "epoch": 0.8357051555176821,
+      "grad_norm": 13.875,
+      "learning_rate": 1.5784715244850935e-06,
+      "loss": 0.5584,
+      "step": 9807
+    },
+    {
+      "epoch": 0.835790370685982,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5783093913186126e-06,
+      "loss": 1.0975,
+      "step": 9808
+    },
+    {
+      "epoch": 0.835875585854282,
+      "grad_norm": 39.5,
+      "learning_rate": 1.5781472522192514e-06,
+      "loss": 0.9335,
+      "step": 9809
+    },
+    {
+      "epoch": 0.835960801022582,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5779851071899394e-06,
+      "loss": 0.8782,
+      "step": 9810
+    },
+    {
+      "epoch": 0.8360460161908819,
+      "grad_norm": 12.625,
+      "learning_rate": 1.5778229562336064e-06,
+      "loss": 0.6672,
+      "step": 9811
+    },
+    {
+      "epoch": 0.8361312313591819,
+      "grad_norm": 12.625,
+      "learning_rate": 1.577660799353183e-06,
+      "loss": 0.4586,
+      "step": 9812
+    },
+    {
+      "epoch": 0.8362164465274818,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5774986365515998e-06,
+      "loss": 0.6113,
+      "step": 9813
+    },
+    {
+      "epoch": 0.8363016616957818,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.5773364678317865e-06,
+      "loss": 0.5669,
+      "step": 9814
+    },
+    {
+      "epoch": 0.8363868768640818,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.577174293196674e-06,
+      "loss": 0.7405,
+      "step": 9815
+    },
+    {
+      "epoch": 0.8364720920323817,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.5770121126491934e-06,
+      "loss": 0.9952,
+      "step": 9816
+    },
+    {
+      "epoch": 0.8365573072006818,
+      "grad_norm": 18.375,
+      "learning_rate": 1.576849926192275e-06,
+      "loss": 0.8584,
+      "step": 9817
+    },
+    {
+      "epoch": 0.8366425223689817,
+      "grad_norm": 19.125,
+      "learning_rate": 1.5766877338288496e-06,
+      "loss": 0.9105,
+      "step": 9818
+    },
+    {
+      "epoch": 0.8367277375372817,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.5765255355618484e-06,
+      "loss": 0.5223,
+      "step": 9819
+    },
+    {
+      "epoch": 0.8368129527055816,
+      "grad_norm": 22.25,
+      "learning_rate": 1.5763633313942025e-06,
+      "loss": 1.3057,
+      "step": 9820
+    },
+    {
+      "epoch": 0.8368981678738816,
+      "grad_norm": 17.75,
+      "learning_rate": 1.5762011213288427e-06,
+      "loss": 1.035,
+      "step": 9821
+    },
+    {
+      "epoch": 0.8369833830421816,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5760389053687014e-06,
+      "loss": 0.3214,
+      "step": 9822
+    },
+    {
+      "epoch": 0.8370685982104815,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5758766835167083e-06,
+      "loss": 0.4658,
+      "step": 9823
+    },
+    {
+      "epoch": 0.8371538133787815,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.5757144557757965e-06,
+      "loss": 0.5965,
+      "step": 9824
+    },
+    {
+      "epoch": 0.8372390285470814,
+      "grad_norm": 20.25,
+      "learning_rate": 1.5755522221488964e-06,
+      "loss": 0.8638,
+      "step": 9825
+    },
+    {
+      "epoch": 0.8373242437153814,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.5753899826389407e-06,
+      "loss": 0.6263,
+      "step": 9826
+    },
+    {
+      "epoch": 0.8374094588836813,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.5752277372488605e-06,
+      "loss": 0.6863,
+      "step": 9827
+    },
+    {
+      "epoch": 0.8374946740519813,
+      "grad_norm": 21.875,
+      "learning_rate": 1.5750654859815884e-06,
+      "loss": 0.6349,
+      "step": 9828
+    },
+    {
+      "epoch": 0.8375798892202813,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5749032288400557e-06,
+      "loss": 0.6991,
+      "step": 9829
+    },
+    {
+      "epoch": 0.8376651043885812,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5747409658271954e-06,
+      "loss": 0.9973,
+      "step": 9830
+    },
+    {
+      "epoch": 0.8377503195568812,
+      "grad_norm": 18.875,
+      "learning_rate": 1.574578696945939e-06,
+      "loss": 0.7255,
+      "step": 9831
+    },
+    {
+      "epoch": 0.8378355347251811,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.574416422199219e-06,
+      "loss": 0.8038,
+      "step": 9832
+    },
+    {
+      "epoch": 0.8379207498934811,
+      "grad_norm": 19.875,
+      "learning_rate": 1.574254141589968e-06,
+      "loss": 1.0336,
+      "step": 9833
+    },
+    {
+      "epoch": 0.838005965061781,
+      "grad_norm": 15.625,
+      "learning_rate": 1.574091855121119e-06,
+      "loss": 0.7533,
+      "step": 9834
+    },
+    {
+      "epoch": 0.838091180230081,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5739295627956045e-06,
+      "loss": 0.5521,
+      "step": 9835
+    },
+    {
+      "epoch": 0.838176395398381,
+      "grad_norm": 23.5,
+      "learning_rate": 1.5737672646163564e-06,
+      "loss": 0.8693,
+      "step": 9836
+    },
+    {
+      "epoch": 0.8382616105666809,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.5736049605863084e-06,
+      "loss": 0.8098,
+      "step": 9837
+    },
+    {
+      "epoch": 0.8383468257349809,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5734426507083933e-06,
+      "loss": 0.6288,
+      "step": 9838
+    },
+    {
+      "epoch": 0.8384320409032808,
+      "grad_norm": 17.5,
+      "learning_rate": 1.573280334985545e-06,
+      "loss": 0.9859,
+      "step": 9839
+    },
+    {
+      "epoch": 0.8385172560715808,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5731180134206953e-06,
+      "loss": 0.8724,
+      "step": 9840
+    },
+    {
+      "epoch": 0.8386024712398807,
+      "grad_norm": 17.625,
+      "learning_rate": 1.5729556860167782e-06,
+      "loss": 0.9944,
+      "step": 9841
+    },
+    {
+      "epoch": 0.8386876864081807,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.572793352776727e-06,
+      "loss": 0.6638,
+      "step": 9842
+    },
+    {
+      "epoch": 0.8387729015764807,
+      "grad_norm": 18.75,
+      "learning_rate": 1.5726310137034756e-06,
+      "loss": 1.1067,
+      "step": 9843
+    },
+    {
+      "epoch": 0.8388581167447806,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5724686687999572e-06,
+      "loss": 0.5252,
+      "step": 9844
+    },
+    {
+      "epoch": 0.8389433319130806,
+      "grad_norm": 30.75,
+      "learning_rate": 1.5723063180691058e-06,
+      "loss": 0.9885,
+      "step": 9845
+    },
+    {
+      "epoch": 0.8390285470813805,
+      "grad_norm": 25.125,
+      "learning_rate": 1.5721439615138547e-06,
+      "loss": 0.8916,
+      "step": 9846
+    },
+    {
+      "epoch": 0.8391137622496805,
+      "grad_norm": 16.875,
+      "learning_rate": 1.571981599137139e-06,
+      "loss": 1.0991,
+      "step": 9847
+    },
+    {
+      "epoch": 0.8391989774179804,
+      "grad_norm": 15.75,
+      "learning_rate": 1.5718192309418918e-06,
+      "loss": 1.0177,
+      "step": 9848
+    },
+    {
+      "epoch": 0.8392841925862804,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.5716568569310469e-06,
+      "loss": 0.704,
+      "step": 9849
+    },
+    {
+      "epoch": 0.8393694077545804,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.5714944771075396e-06,
+      "loss": 0.897,
+      "step": 9850
+    },
+    {
+      "epoch": 0.8394546229228803,
+      "grad_norm": 19.5,
+      "learning_rate": 1.571332091474304e-06,
+      "loss": 1.0276,
+      "step": 9851
+    },
+    {
+      "epoch": 0.8395398380911803,
+      "grad_norm": 17.25,
+      "learning_rate": 1.571169700034274e-06,
+      "loss": 0.7457,
+      "step": 9852
+    },
+    {
+      "epoch": 0.8396250532594802,
+      "grad_norm": 36.75,
+      "learning_rate": 1.5710073027903848e-06,
+      "loss": 0.8826,
+      "step": 9853
+    },
+    {
+      "epoch": 0.8397102684277802,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5708448997455709e-06,
+      "loss": 0.9323,
+      "step": 9854
+    },
+    {
+      "epoch": 0.8397954835960801,
+      "grad_norm": 20.75,
+      "learning_rate": 1.5706824909027667e-06,
+      "loss": 0.9471,
+      "step": 9855
+    },
+    {
+      "epoch": 0.8398806987643801,
+      "grad_norm": 9.75,
+      "learning_rate": 1.5705200762649079e-06,
+      "loss": 0.3089,
+      "step": 9856
+    },
+    {
+      "epoch": 0.83996591393268,
+      "grad_norm": 19.375,
+      "learning_rate": 1.5703576558349285e-06,
+      "loss": 0.7622,
+      "step": 9857
+    },
+    {
+      "epoch": 0.84005112910098,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5701952296157643e-06,
+      "loss": 0.7318,
+      "step": 9858
+    },
+    {
+      "epoch": 0.84013634426928,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5700327976103504e-06,
+      "loss": 0.9117,
+      "step": 9859
+    },
+    {
+      "epoch": 0.8402215594375799,
+      "grad_norm": 12.875,
+      "learning_rate": 1.5698703598216223e-06,
+      "loss": 0.6294,
+      "step": 9860
+    },
+    {
+      "epoch": 0.8403067746058799,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5697079162525148e-06,
+      "loss": 0.6508,
+      "step": 9861
+    },
+    {
+      "epoch": 0.8403919897741798,
+      "grad_norm": 10.125,
+      "learning_rate": 1.5695454669059637e-06,
+      "loss": 0.2965,
+      "step": 9862
+    },
+    {
+      "epoch": 0.8404772049424798,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5693830117849048e-06,
+      "loss": 0.7872,
+      "step": 9863
+    },
+    {
+      "epoch": 0.8405624201107798,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.5692205508922736e-06,
+      "loss": 1.0676,
+      "step": 9864
+    },
+    {
+      "epoch": 0.8406476352790797,
+      "grad_norm": 19.5,
+      "learning_rate": 1.569058084231006e-06,
+      "loss": 1.0584,
+      "step": 9865
+    },
+    {
+      "epoch": 0.8407328504473797,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5688956118040382e-06,
+      "loss": 0.7414,
+      "step": 9866
+    },
+    {
+      "epoch": 0.8408180656156796,
+      "grad_norm": 10.25,
+      "learning_rate": 1.5687331336143063e-06,
+      "loss": 0.3077,
+      "step": 9867
+    },
+    {
+      "epoch": 0.8409032807839796,
+      "grad_norm": 19.625,
+      "learning_rate": 1.5685706496647454e-06,
+      "loss": 0.7172,
+      "step": 9868
+    },
+    {
+      "epoch": 0.8409884959522795,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5684081599582932e-06,
+      "loss": 0.802,
+      "step": 9869
+    },
+    {
+      "epoch": 0.8410737111205795,
+      "grad_norm": 21.75,
+      "learning_rate": 1.568245664497885e-06,
+      "loss": 1.1287,
+      "step": 9870
+    },
+    {
+      "epoch": 0.8411589262888794,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.568083163286458e-06,
+      "loss": 0.8503,
+      "step": 9871
+    },
+    {
+      "epoch": 0.8412441414571794,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.5679206563269478e-06,
+      "loss": 0.4853,
+      "step": 9872
+    },
+    {
+      "epoch": 0.8413293566254794,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.567758143622292e-06,
+      "loss": 0.6183,
+      "step": 9873
+    },
+    {
+      "epoch": 0.8414145717937793,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.5675956251754268e-06,
+      "loss": 0.6364,
+      "step": 9874
+    },
+    {
+      "epoch": 0.8414997869620793,
+      "grad_norm": 17.625,
+      "learning_rate": 1.5674331009892896e-06,
+      "loss": 0.4606,
+      "step": 9875
+    },
+    {
+      "epoch": 0.8415850021303792,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.5672705710668168e-06,
+      "loss": 0.7307,
+      "step": 9876
+    },
+    {
+      "epoch": 0.8416702172986792,
+      "grad_norm": 19.375,
+      "learning_rate": 1.567108035410946e-06,
+      "loss": 0.7943,
+      "step": 9877
+    },
+    {
+      "epoch": 0.8417554324669791,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5669454940246138e-06,
+      "loss": 0.9012,
+      "step": 9878
+    },
+    {
+      "epoch": 0.8418406476352791,
+      "grad_norm": 19.5,
+      "learning_rate": 1.566782946910758e-06,
+      "loss": 1.0703,
+      "step": 9879
+    },
+    {
+      "epoch": 0.8419258628035791,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.5666203940723157e-06,
+      "loss": 0.4871,
+      "step": 9880
+    },
+    {
+      "epoch": 0.842011077971879,
+      "grad_norm": 14.5,
+      "learning_rate": 1.5664578355122245e-06,
+      "loss": 0.7322,
+      "step": 9881
+    },
+    {
+      "epoch": 0.842096293140179,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.566295271233422e-06,
+      "loss": 0.7332,
+      "step": 9882
+    },
+    {
+      "epoch": 0.8421815083084789,
+      "grad_norm": 12.5,
+      "learning_rate": 1.566132701238846e-06,
+      "loss": 0.7518,
+      "step": 9883
+    },
+    {
+      "epoch": 0.8422667234767789,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.5659701255314345e-06,
+      "loss": 0.5796,
+      "step": 9884
+    },
+    {
+      "epoch": 0.8423519386450788,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5658075441141243e-06,
+      "loss": 0.4041,
+      "step": 9885
+    },
+    {
+      "epoch": 0.8424371538133788,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5656449569898547e-06,
+      "loss": 0.768,
+      "step": 9886
+    },
+    {
+      "epoch": 0.8425223689816788,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.5654823641615636e-06,
+      "loss": 0.672,
+      "step": 9887
+    },
+    {
+      "epoch": 0.8426075841499787,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5653197656321886e-06,
+      "loss": 0.6673,
+      "step": 9888
+    },
+    {
+      "epoch": 0.8426927993182787,
+      "grad_norm": 12.0,
+      "learning_rate": 1.565157161404668e-06,
+      "loss": 0.6916,
+      "step": 9889
+    },
+    {
+      "epoch": 0.8427780144865786,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.564994551481941e-06,
+      "loss": 0.6301,
+      "step": 9890
+    },
+    {
+      "epoch": 0.8428632296548786,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.5648319358669458e-06,
+      "loss": 0.3635,
+      "step": 9891
+    },
+    {
+      "epoch": 0.8429484448231785,
+      "grad_norm": 14.875,
+      "learning_rate": 1.5646693145626208e-06,
+      "loss": 0.8302,
+      "step": 9892
+    },
+    {
+      "epoch": 0.8430336599914785,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.5645066875719048e-06,
+      "loss": 0.2821,
+      "step": 9893
+    },
+    {
+      "epoch": 0.8431188751597785,
+      "grad_norm": 20.5,
+      "learning_rate": 1.5643440548977367e-06,
+      "loss": 1.1424,
+      "step": 9894
+    },
+    {
+      "epoch": 0.8432040903280784,
+      "grad_norm": 13.125,
+      "learning_rate": 1.5641814165430552e-06,
+      "loss": 0.8648,
+      "step": 9895
+    },
+    {
+      "epoch": 0.8432893054963784,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.5640187725108e-06,
+      "loss": 0.8644,
+      "step": 9896
+    },
+    {
+      "epoch": 0.8433745206646783,
+      "grad_norm": 19.125,
+      "learning_rate": 1.56385612280391e-06,
+      "loss": 0.9258,
+      "step": 9897
+    },
+    {
+      "epoch": 0.8434597358329783,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5636934674253237e-06,
+      "loss": 0.5752,
+      "step": 9898
+    },
+    {
+      "epoch": 0.8435449510012782,
+      "grad_norm": 20.125,
+      "learning_rate": 1.563530806377981e-06,
+      "loss": 0.4719,
+      "step": 9899
+    },
+    {
+      "epoch": 0.8436301661695782,
+      "grad_norm": 13.125,
+      "learning_rate": 1.5633681396648214e-06,
+      "loss": 0.343,
+      "step": 9900
+    },
+    {
+      "epoch": 0.8437153813378782,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5632054672887848e-06,
+      "loss": 0.815,
+      "step": 9901
+    },
+    {
+      "epoch": 0.8438005965061781,
+      "grad_norm": 17.25,
+      "learning_rate": 1.56304278925281e-06,
+      "loss": 0.9937,
+      "step": 9902
+    },
+    {
+      "epoch": 0.8438858116744781,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5628801055598376e-06,
+      "loss": 0.6641,
+      "step": 9903
+    },
+    {
+      "epoch": 0.843971026842778,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5627174162128066e-06,
+      "loss": 0.9075,
+      "step": 9904
+    },
+    {
+      "epoch": 0.844056242011078,
+      "grad_norm": 10.25,
+      "learning_rate": 1.562554721214658e-06,
+      "loss": 0.5418,
+      "step": 9905
+    },
+    {
+      "epoch": 0.8441414571793779,
+      "grad_norm": 20.625,
+      "learning_rate": 1.5623920205683307e-06,
+      "loss": 0.8552,
+      "step": 9906
+    },
+    {
+      "epoch": 0.8442266723476779,
+      "grad_norm": 34.5,
+      "learning_rate": 1.5622293142767658e-06,
+      "loss": 0.9319,
+      "step": 9907
+    },
+    {
+      "epoch": 0.8443118875159779,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.5620666023429035e-06,
+      "loss": 0.6535,
+      "step": 9908
+    },
+    {
+      "epoch": 0.8443971026842778,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5619038847696835e-06,
+      "loss": 0.6715,
+      "step": 9909
+    },
+    {
+      "epoch": 0.8444823178525778,
+      "grad_norm": 18.625,
+      "learning_rate": 1.5617411615600465e-06,
+      "loss": 0.7355,
+      "step": 9910
+    },
+    {
+      "epoch": 0.8445675330208777,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5615784327169337e-06,
+      "loss": 0.9041,
+      "step": 9911
+    },
+    {
+      "epoch": 0.8446527481891777,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.5614156982432848e-06,
+      "loss": 0.6214,
+      "step": 9912
+    },
+    {
+      "epoch": 0.8447379633574776,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5612529581420415e-06,
+      "loss": 0.7267,
+      "step": 9913
+    },
+    {
+      "epoch": 0.8448231785257776,
+      "grad_norm": 11.375,
+      "learning_rate": 1.561090212416144e-06,
+      "loss": 0.565,
+      "step": 9914
+    },
+    {
+      "epoch": 0.8449083936940776,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5609274610685338e-06,
+      "loss": 0.5954,
+      "step": 9915
+    },
+    {
+      "epoch": 0.8449936088623775,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5607647041021516e-06,
+      "loss": 0.7366,
+      "step": 9916
+    },
+    {
+      "epoch": 0.8450788240306775,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.5606019415199383e-06,
+      "loss": 0.7262,
+      "step": 9917
+    },
+    {
+      "epoch": 0.8451640391989774,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5604391733248361e-06,
+      "loss": 0.96,
+      "step": 9918
+    },
+    {
+      "epoch": 0.8452492543672774,
+      "grad_norm": 11.625,
+      "learning_rate": 1.5602763995197853e-06,
+      "loss": 0.6817,
+      "step": 9919
+    },
+    {
+      "epoch": 0.8453344695355773,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5601136201077288e-06,
+      "loss": 0.9011,
+      "step": 9920
+    },
+    {
+      "epoch": 0.8454196847038773,
+      "grad_norm": 19.125,
+      "learning_rate": 1.5599508350916068e-06,
+      "loss": 0.7238,
+      "step": 9921
+    },
+    {
+      "epoch": 0.8455048998721773,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5597880444743617e-06,
+      "loss": 0.6033,
+      "step": 9922
+    },
+    {
+      "epoch": 0.8455901150404772,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5596252482589348e-06,
+      "loss": 0.4132,
+      "step": 9923
+    },
+    {
+      "epoch": 0.8456753302087772,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.5594624464482689e-06,
+      "loss": 0.4342,
+      "step": 9924
+    },
+    {
+      "epoch": 0.8457605453770771,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.5592996390453047e-06,
+      "loss": 0.6257,
+      "step": 9925
+    },
+    {
+      "epoch": 0.8458457605453771,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5591368260529855e-06,
+      "loss": 0.7901,
+      "step": 9926
+    },
+    {
+      "epoch": 0.845930975713677,
+      "grad_norm": 12.625,
+      "learning_rate": 1.5589740074742525e-06,
+      "loss": 0.5534,
+      "step": 9927
+    },
+    {
+      "epoch": 0.846016190881977,
+      "grad_norm": 24.875,
+      "learning_rate": 1.5588111833120493e-06,
+      "loss": 0.8871,
+      "step": 9928
+    },
+    {
+      "epoch": 0.846101406050277,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5586483535693166e-06,
+      "loss": 0.8796,
+      "step": 9929
+    },
+    {
+      "epoch": 0.8461866212185769,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5584855182489986e-06,
+      "loss": 1.4636,
+      "step": 9930
+    },
+    {
+      "epoch": 0.8462718363868769,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5583226773540364e-06,
+      "loss": 0.6934,
+      "step": 9931
+    },
+    {
+      "epoch": 0.8463570515551768,
+      "grad_norm": 18.25,
+      "learning_rate": 1.5581598308873735e-06,
+      "loss": 1.0093,
+      "step": 9932
+    },
+    {
+      "epoch": 0.8464422667234768,
+      "grad_norm": 22.375,
+      "learning_rate": 1.5579969788519531e-06,
+      "loss": 0.9159,
+      "step": 9933
+    },
+    {
+      "epoch": 0.8465274818917767,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.557834121250717e-06,
+      "loss": 0.6586,
+      "step": 9934
+    },
+    {
+      "epoch": 0.8466126970600767,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.5576712580866088e-06,
+      "loss": 0.6704,
+      "step": 9935
+    },
+    {
+      "epoch": 0.8466979122283766,
+      "grad_norm": 15.0,
+      "learning_rate": 1.557508389362572e-06,
+      "loss": 0.9263,
+      "step": 9936
+    },
+    {
+      "epoch": 0.8467831273966766,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.557345515081549e-06,
+      "loss": 0.893,
+      "step": 9937
+    },
+    {
+      "epoch": 0.8468683425649766,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.5571826352464835e-06,
+      "loss": 0.722,
+      "step": 9938
+    },
+    {
+      "epoch": 0.8469535577332765,
+      "grad_norm": 25.125,
+      "learning_rate": 1.5570197498603193e-06,
+      "loss": 0.9335,
+      "step": 9939
+    },
+    {
+      "epoch": 0.8470387729015765,
+      "grad_norm": 20.625,
+      "learning_rate": 1.5568568589259991e-06,
+      "loss": 0.9344,
+      "step": 9940
+    },
+    {
+      "epoch": 0.8471239880698764,
+      "grad_norm": 19.0,
+      "learning_rate": 1.5566939624464672e-06,
+      "loss": 0.4741,
+      "step": 9941
+    },
+    {
+      "epoch": 0.8472092032381764,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5565310604246664e-06,
+      "loss": 0.5916,
+      "step": 9942
+    },
+    {
+      "epoch": 0.8472944184064763,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.5563681528635415e-06,
+      "loss": 0.8187,
+      "step": 9943
+    },
+    {
+      "epoch": 0.8473796335747763,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.5562052397660363e-06,
+      "loss": 0.652,
+      "step": 9944
+    },
+    {
+      "epoch": 0.8474648487430763,
+      "grad_norm": 14.25,
+      "learning_rate": 1.5560423211350942e-06,
+      "loss": 0.5854,
+      "step": 9945
+    },
+    {
+      "epoch": 0.8475500639113762,
+      "grad_norm": 18.875,
+      "learning_rate": 1.5558793969736593e-06,
+      "loss": 0.8579,
+      "step": 9946
+    },
+    {
+      "epoch": 0.8476352790796762,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5557164672846765e-06,
+      "loss": 0.4096,
+      "step": 9947
+    },
+    {
+      "epoch": 0.8477204942479761,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5555535320710895e-06,
+      "loss": 0.6412,
+      "step": 9948
+    },
+    {
+      "epoch": 0.8478057094162761,
+      "grad_norm": 16.375,
+      "learning_rate": 1.555390591335843e-06,
+      "loss": 0.624,
+      "step": 9949
+    },
+    {
+      "epoch": 0.847890924584576,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5552276450818816e-06,
+      "loss": 0.8301,
+      "step": 9950
+    },
+    {
+      "epoch": 0.847976139752876,
+      "grad_norm": 19.625,
+      "learning_rate": 1.5550646933121497e-06,
+      "loss": 1.0591,
+      "step": 9951
+    },
+    {
+      "epoch": 0.848061354921176,
+      "grad_norm": 11.375,
+      "learning_rate": 1.554901736029592e-06,
+      "loss": 0.3938,
+      "step": 9952
+    },
+    {
+      "epoch": 0.8481465700894759,
+      "grad_norm": 21.0,
+      "learning_rate": 1.554738773237153e-06,
+      "loss": 0.8778,
+      "step": 9953
+    },
+    {
+      "epoch": 0.8482317852577759,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5545758049377781e-06,
+      "loss": 0.5937,
+      "step": 9954
+    },
+    {
+      "epoch": 0.8483170004260758,
+      "grad_norm": 22.375,
+      "learning_rate": 1.554412831134412e-06,
+      "loss": 0.9817,
+      "step": 9955
+    },
+    {
+      "epoch": 0.8484022155943758,
+      "grad_norm": 23.375,
+      "learning_rate": 1.5542498518300006e-06,
+      "loss": 1.1974,
+      "step": 9956
+    },
+    {
+      "epoch": 0.8484874307626757,
+      "grad_norm": 11.375,
+      "learning_rate": 1.5540868670274879e-06,
+      "loss": 0.502,
+      "step": 9957
+    },
+    {
+      "epoch": 0.8485726459309757,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5539238767298197e-06,
+      "loss": 0.3593,
+      "step": 9958
+    },
+    {
+      "epoch": 0.8486578610992757,
+      "grad_norm": 23.5,
+      "learning_rate": 1.5537608809399413e-06,
+      "loss": 1.1276,
+      "step": 9959
+    },
+    {
+      "epoch": 0.8487430762675756,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5535978796607989e-06,
+      "loss": 0.7329,
+      "step": 9960
+    },
+    {
+      "epoch": 0.8488282914358756,
+      "grad_norm": 9.25,
+      "learning_rate": 1.5534348728953368e-06,
+      "loss": 0.3222,
+      "step": 9961
+    },
+    {
+      "epoch": 0.8489135066041755,
+      "grad_norm": 15.0,
+      "learning_rate": 1.553271860646502e-06,
+      "loss": 0.8051,
+      "step": 9962
+    },
+    {
+      "epoch": 0.8489987217724755,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5531088429172392e-06,
+      "loss": 0.8001,
+      "step": 9963
+    },
+    {
+      "epoch": 0.8490839369407754,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5529458197104951e-06,
+      "loss": 0.7866,
+      "step": 9964
+    },
+    {
+      "epoch": 0.8491691521090754,
+      "grad_norm": 35.75,
+      "learning_rate": 1.5527827910292156e-06,
+      "loss": 0.697,
+      "step": 9965
+    },
+    {
+      "epoch": 0.8492543672773754,
+      "grad_norm": 32.0,
+      "learning_rate": 1.5526197568763465e-06,
+      "loss": 1.2927,
+      "step": 9966
+    },
+    {
+      "epoch": 0.8493395824456753,
+      "grad_norm": 12.0,
+      "learning_rate": 1.5524567172548339e-06,
+      "loss": 0.6935,
+      "step": 9967
+    },
+    {
+      "epoch": 0.8494247976139753,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5522936721676243e-06,
+      "loss": 0.7174,
+      "step": 9968
+    },
+    {
+      "epoch": 0.8495100127822752,
+      "grad_norm": 15.875,
+      "learning_rate": 1.5521306216176644e-06,
+      "loss": 0.6199,
+      "step": 9969
+    },
+    {
+      "epoch": 0.8495952279505752,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5519675656078998e-06,
+      "loss": 0.7787,
+      "step": 9970
+    },
+    {
+      "epoch": 0.8496804431188751,
+      "grad_norm": 17.375,
+      "learning_rate": 1.551804504141278e-06,
+      "loss": 0.8508,
+      "step": 9971
+    },
+    {
+      "epoch": 0.8497656582871751,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.551641437220745e-06,
+      "loss": 0.884,
+      "step": 9972
+    },
+    {
+      "epoch": 0.849850873455475,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5514783648492485e-06,
+      "loss": 0.8434,
+      "step": 9973
+    },
+    {
+      "epoch": 0.849936088623775,
+      "grad_norm": 13.5,
+      "learning_rate": 1.551315287029734e-06,
+      "loss": 0.6282,
+      "step": 9974
+    },
+    {
+      "epoch": 0.850021303792075,
+      "grad_norm": 12.25,
+      "learning_rate": 1.5511522037651499e-06,
+      "loss": 0.6725,
+      "step": 9975
+    },
+    {
+      "epoch": 0.8501065189603749,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.5509891150584422e-06,
+      "loss": 0.341,
+      "step": 9976
+    },
+    {
+      "epoch": 0.8501917341286749,
+      "grad_norm": 10.25,
+      "learning_rate": 1.5508260209125592e-06,
+      "loss": 0.3871,
+      "step": 9977
+    },
+    {
+      "epoch": 0.8502769492969748,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.550662921330447e-06,
+      "loss": 0.6555,
+      "step": 9978
+    },
+    {
+      "epoch": 0.8503621644652748,
+      "grad_norm": 20.25,
+      "learning_rate": 1.5504998163150533e-06,
+      "loss": 1.0452,
+      "step": 9979
+    },
+    {
+      "epoch": 0.8504473796335748,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.5503367058693259e-06,
+      "loss": 0.5478,
+      "step": 9980
+    },
+    {
+      "epoch": 0.8505325948018747,
+      "grad_norm": 18.25,
+      "learning_rate": 1.5501735899962122e-06,
+      "loss": 0.6974,
+      "step": 9981
+    },
+    {
+      "epoch": 0.8506178099701747,
+      "grad_norm": 13.625,
+      "learning_rate": 1.5500104686986604e-06,
+      "loss": 0.7311,
+      "step": 9982
+    },
+    {
+      "epoch": 0.8507030251384746,
+      "grad_norm": 35.25,
+      "learning_rate": 1.5498473419796172e-06,
+      "loss": 0.8109,
+      "step": 9983
+    },
+    {
+      "epoch": 0.8507882403067746,
+      "grad_norm": 19.25,
+      "learning_rate": 1.549684209842031e-06,
+      "loss": 0.922,
+      "step": 9984
+    },
+    {
+      "epoch": 0.8508734554750745,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5495210722888498e-06,
+      "loss": 0.7558,
+      "step": 9985
+    },
+    {
+      "epoch": 0.8509586706433745,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.549357929323022e-06,
+      "loss": 0.8149,
+      "step": 9986
+    },
+    {
+      "epoch": 0.8510438858116744,
+      "grad_norm": 26.25,
+      "learning_rate": 1.549194780947495e-06,
+      "loss": 0.4548,
+      "step": 9987
+    },
+    {
+      "epoch": 0.8511291009799744,
+      "grad_norm": 23.125,
+      "learning_rate": 1.5490316271652176e-06,
+      "loss": 1.4007,
+      "step": 9988
+    },
+    {
+      "epoch": 0.8512143161482744,
+      "grad_norm": 16.875,
+      "learning_rate": 1.548868467979138e-06,
+      "loss": 0.6879,
+      "step": 9989
+    },
+    {
+      "epoch": 0.8512995313165743,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5487053033922048e-06,
+      "loss": 0.7199,
+      "step": 9990
+    },
+    {
+      "epoch": 0.8513847464848743,
+      "grad_norm": 12.5,
+      "learning_rate": 1.5485421334073658e-06,
+      "loss": 0.7627,
+      "step": 9991
+    },
+    {
+      "epoch": 0.8514699616531742,
+      "grad_norm": 12.625,
+      "learning_rate": 1.5483789580275707e-06,
+      "loss": 0.903,
+      "step": 9992
+    },
+    {
+      "epoch": 0.8515551768214742,
+      "grad_norm": 15.375,
+      "learning_rate": 1.548215777255768e-06,
+      "loss": 1.1291,
+      "step": 9993
+    },
+    {
+      "epoch": 0.8516403919897741,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.548052591094906e-06,
+      "loss": 0.6761,
+      "step": 9994
+    },
+    {
+      "epoch": 0.8517256071580741,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.5478893995479338e-06,
+      "loss": 0.4795,
+      "step": 9995
+    },
+    {
+      "epoch": 0.8518108223263741,
+      "grad_norm": 18.5,
+      "learning_rate": 1.5477262026178008e-06,
+      "loss": 0.7059,
+      "step": 9996
+    },
+    {
+      "epoch": 0.851896037494674,
+      "grad_norm": 19.375,
+      "learning_rate": 1.5475630003074563e-06,
+      "loss": 0.8677,
+      "step": 9997
+    },
+    {
+      "epoch": 0.851981252662974,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.5473997926198488e-06,
+      "loss": 0.3925,
+      "step": 9998
+    },
+    {
+      "epoch": 0.8520664678312739,
+      "grad_norm": 10.375,
+      "learning_rate": 1.5472365795579277e-06,
+      "loss": 0.421,
+      "step": 9999
+    },
+    {
+      "epoch": 0.8521516829995739,
+      "grad_norm": 12.5,
+      "learning_rate": 1.5470733611246432e-06,
+      "loss": 0.4204,
+      "step": 10000
+    },
+    {
+      "epoch": 0.8522368981678738,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.5469101373229442e-06,
+      "loss": 0.6935,
+      "step": 10001
+    },
+    {
+      "epoch": 0.8523221133361738,
+      "grad_norm": 16.125,
+      "learning_rate": 1.54674690815578e-06,
+      "loss": 0.9133,
+      "step": 10002
+    },
+    {
+      "epoch": 0.8524073285044738,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5465836736261015e-06,
+      "loss": 0.982,
+      "step": 10003
+    },
+    {
+      "epoch": 0.8524925436727737,
+      "grad_norm": 16.125,
+      "learning_rate": 1.546420433736857e-06,
+      "loss": 0.79,
+      "step": 10004
+    },
+    {
+      "epoch": 0.8525777588410737,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5462571884909984e-06,
+      "loss": 0.2605,
+      "step": 10005
+    },
+    {
+      "epoch": 0.8526629740093736,
+      "grad_norm": 47.0,
+      "learning_rate": 1.5460939378914732e-06,
+      "loss": 1.1058,
+      "step": 10006
+    },
+    {
+      "epoch": 0.8527481891776736,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5459306819412337e-06,
+      "loss": 0.7308,
+      "step": 10007
+    },
+    {
+      "epoch": 0.8528334043459735,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5457674206432286e-06,
+      "loss": 0.2821,
+      "step": 10008
+    },
+    {
+      "epoch": 0.8529186195142735,
+      "grad_norm": 12.0,
+      "learning_rate": 1.5456041540004094e-06,
+      "loss": 0.7543,
+      "step": 10009
+    },
+    {
+      "epoch": 0.8530038346825735,
+      "grad_norm": 15.625,
+      "learning_rate": 1.545440882015725e-06,
+      "loss": 0.8549,
+      "step": 10010
+    },
+    {
+      "epoch": 0.8530890498508734,
+      "grad_norm": 15.125,
+      "learning_rate": 1.5452776046921275e-06,
+      "loss": 0.8032,
+      "step": 10011
+    },
+    {
+      "epoch": 0.8531742650191734,
+      "grad_norm": 11.625,
+      "learning_rate": 1.5451143220325664e-06,
+      "loss": 0.5344,
+      "step": 10012
+    },
+    {
+      "epoch": 0.8532594801874733,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5449510340399932e-06,
+      "loss": 0.6638,
+      "step": 10013
+    },
+    {
+      "epoch": 0.8533446953557733,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5447877407173577e-06,
+      "loss": 0.882,
+      "step": 10014
+    },
+    {
+      "epoch": 0.8534299105240732,
+      "grad_norm": 15.75,
+      "learning_rate": 1.5446244420676116e-06,
+      "loss": 0.8087,
+      "step": 10015
+    },
+    {
+      "epoch": 0.8535151256923732,
+      "grad_norm": 11.375,
+      "learning_rate": 1.5444611380937052e-06,
+      "loss": 0.4129,
+      "step": 10016
+    },
+    {
+      "epoch": 0.8536003408606732,
+      "grad_norm": 20.0,
+      "learning_rate": 1.5442978287985902e-06,
+      "loss": 0.9496,
+      "step": 10017
+    },
+    {
+      "epoch": 0.8536855560289731,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5441345141852177e-06,
+      "loss": 1.0274,
+      "step": 10018
+    },
+    {
+      "epoch": 0.8537707711972731,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.5439711942565384e-06,
+      "loss": 0.3175,
+      "step": 10019
+    },
+    {
+      "epoch": 0.853855986365573,
+      "grad_norm": 19.5,
+      "learning_rate": 1.5438078690155043e-06,
+      "loss": 0.7375,
+      "step": 10020
+    },
+    {
+      "epoch": 0.853941201533873,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5436445384650663e-06,
+      "loss": 0.9234,
+      "step": 10021
+    },
+    {
+      "epoch": 0.8540264167021729,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.5434812026081763e-06,
+      "loss": 0.6828,
+      "step": 10022
+    },
+    {
+      "epoch": 0.8541116318704729,
+      "grad_norm": 11.625,
+      "learning_rate": 1.5433178614477857e-06,
+      "loss": 0.4363,
+      "step": 10023
+    },
+    {
+      "epoch": 0.8541968470387729,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5431545149868466e-06,
+      "loss": 0.408,
+      "step": 10024
+    },
+    {
+      "epoch": 0.8542820622070728,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5429911632283106e-06,
+      "loss": 0.7951,
+      "step": 10025
+    },
+    {
+      "epoch": 0.8543672773753728,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5428278061751297e-06,
+      "loss": 0.826,
+      "step": 10026
+    },
+    {
+      "epoch": 0.8544524925436727,
+      "grad_norm": 19.625,
+      "learning_rate": 1.5426644438302557e-06,
+      "loss": 0.8248,
+      "step": 10027
+    },
+    {
+      "epoch": 0.8545377077119727,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5425010761966412e-06,
+      "loss": 0.9161,
+      "step": 10028
+    },
+    {
+      "epoch": 0.8546229228802726,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.5423377032772376e-06,
+      "loss": 0.6557,
+      "step": 10029
+    },
+    {
+      "epoch": 0.8547081380485726,
+      "grad_norm": 20.625,
+      "learning_rate": 1.5421743250749984e-06,
+      "loss": 1.3825,
+      "step": 10030
+    },
+    {
+      "epoch": 0.8547933532168726,
+      "grad_norm": 11.5,
+      "learning_rate": 1.542010941592875e-06,
+      "loss": 0.2564,
+      "step": 10031
+    },
+    {
+      "epoch": 0.8548785683851725,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5418475528338204e-06,
+      "loss": 0.552,
+      "step": 10032
+    },
+    {
+      "epoch": 0.8549637835534725,
+      "grad_norm": 13.25,
+      "learning_rate": 1.5416841588007863e-06,
+      "loss": 0.5562,
+      "step": 10033
+    },
+    {
+      "epoch": 0.8550489987217724,
+      "grad_norm": 14.5,
+      "learning_rate": 1.5415207594967272e-06,
+      "loss": 0.5113,
+      "step": 10034
+    },
+    {
+      "epoch": 0.8551342138900724,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.5413573549245941e-06,
+      "loss": 0.6859,
+      "step": 10035
+    },
+    {
+      "epoch": 0.8552194290583723,
+      "grad_norm": 12.375,
+      "learning_rate": 1.541193945087341e-06,
+      "loss": 0.685,
+      "step": 10036
+    },
+    {
+      "epoch": 0.8553046442266723,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5410305299879202e-06,
+      "loss": 0.8904,
+      "step": 10037
+    },
+    {
+      "epoch": 0.8553898593949723,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.5408671096292855e-06,
+      "loss": 0.4668,
+      "step": 10038
+    },
+    {
+      "epoch": 0.8554750745632722,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5407036840143893e-06,
+      "loss": 0.843,
+      "step": 10039
+    },
+    {
+      "epoch": 0.8555602897315722,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5405402531461853e-06,
+      "loss": 0.4669,
+      "step": 10040
+    },
+    {
+      "epoch": 0.8556455048998721,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5403768170276268e-06,
+      "loss": 0.7578,
+      "step": 10041
+    },
+    {
+      "epoch": 0.8557307200681721,
+      "grad_norm": 19.5,
+      "learning_rate": 1.5402133756616672e-06,
+      "loss": 0.8379,
+      "step": 10042
+    },
+    {
+      "epoch": 0.8558159352364721,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.54004992905126e-06,
+      "loss": 0.7742,
+      "step": 10043
+    },
+    {
+      "epoch": 0.8559011504047721,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.5398864771993588e-06,
+      "loss": 1.0037,
+      "step": 10044
+    },
+    {
+      "epoch": 0.8559863655730721,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.5397230201089178e-06,
+      "loss": 0.6259,
+      "step": 10045
+    },
+    {
+      "epoch": 0.856071580741372,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.5395595577828903e-06,
+      "loss": 0.5707,
+      "step": 10046
+    },
+    {
+      "epoch": 0.856156795909672,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5393960902242306e-06,
+      "loss": 0.629,
+      "step": 10047
+    },
+    {
+      "epoch": 0.8562420110779719,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.5392326174358921e-06,
+      "loss": 0.5655,
+      "step": 10048
+    },
+    {
+      "epoch": 0.8563272262462719,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5390691394208297e-06,
+      "loss": 0.1799,
+      "step": 10049
+    },
+    {
+      "epoch": 0.8564124414145718,
+      "grad_norm": 10.875,
+      "learning_rate": 1.5389056561819973e-06,
+      "loss": 0.4715,
+      "step": 10050
+    },
+    {
+      "epoch": 0.8564976565828718,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.538742167722349e-06,
+      "loss": 0.4801,
+      "step": 10051
+    },
+    {
+      "epoch": 0.8565828717511718,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.5385786740448395e-06,
+      "loss": 0.3631,
+      "step": 10052
+    },
+    {
+      "epoch": 0.8566680869194717,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5384151751524228e-06,
+      "loss": 0.9585,
+      "step": 10053
+    },
+    {
+      "epoch": 0.8567533020877717,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5382516710480544e-06,
+      "loss": 0.6098,
+      "step": 10054
+    },
+    {
+      "epoch": 0.8568385172560716,
+      "grad_norm": 19.875,
+      "learning_rate": 1.5380881617346879e-06,
+      "loss": 1.1762,
+      "step": 10055
+    },
+    {
+      "epoch": 0.8569237324243716,
+      "grad_norm": 21.25,
+      "learning_rate": 1.5379246472152792e-06,
+      "loss": 1.0837,
+      "step": 10056
+    },
+    {
+      "epoch": 0.8570089475926715,
+      "grad_norm": 12.625,
+      "learning_rate": 1.537761127492782e-06,
+      "loss": 0.6825,
+      "step": 10057
+    },
+    {
+      "epoch": 0.8570941627609715,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5375976025701526e-06,
+      "loss": 0.5408,
+      "step": 10058
+    },
+    {
+      "epoch": 0.8571793779292715,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5374340724503448e-06,
+      "loss": 0.9366,
+      "step": 10059
+    },
+    {
+      "epoch": 0.8572645930975714,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5372705371363145e-06,
+      "loss": 0.6548,
+      "step": 10060
+    },
+    {
+      "epoch": 0.8573498082658714,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5371069966310163e-06,
+      "loss": 0.6377,
+      "step": 10061
+    },
+    {
+      "epoch": 0.8574350234341713,
+      "grad_norm": 13.125,
+      "learning_rate": 1.5369434509374064e-06,
+      "loss": 0.8494,
+      "step": 10062
+    },
+    {
+      "epoch": 0.8575202386024713,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.5367799000584395e-06,
+      "loss": 0.6824,
+      "step": 10063
+    },
+    {
+      "epoch": 0.8576054537707712,
+      "grad_norm": 12.25,
+      "learning_rate": 1.5366163439970718e-06,
+      "loss": 0.4818,
+      "step": 10064
+    },
+    {
+      "epoch": 0.8576906689390712,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5364527827562582e-06,
+      "loss": 0.497,
+      "step": 10065
+    },
+    {
+      "epoch": 0.8577758841073712,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.5362892163389553e-06,
+      "loss": 0.3624,
+      "step": 10066
+    },
+    {
+      "epoch": 0.8578610992756711,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5361256447481176e-06,
+      "loss": 0.6826,
+      "step": 10067
+    },
+    {
+      "epoch": 0.8579463144439711,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.5359620679867025e-06,
+      "loss": 0.7534,
+      "step": 10068
+    },
+    {
+      "epoch": 0.858031529612271,
+      "grad_norm": 20.25,
+      "learning_rate": 1.5357984860576647e-06,
+      "loss": 0.8449,
+      "step": 10069
+    },
+    {
+      "epoch": 0.858116744780571,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5356348989639611e-06,
+      "loss": 0.5237,
+      "step": 10070
+    },
+    {
+      "epoch": 0.8582019599488709,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5354713067085476e-06,
+      "loss": 0.6756,
+      "step": 10071
+    },
+    {
+      "epoch": 0.8582871751171709,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5353077092943804e-06,
+      "loss": 0.6356,
+      "step": 10072
+    },
+    {
+      "epoch": 0.8583723902854709,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5351441067244163e-06,
+      "loss": 0.6461,
+      "step": 10073
+    },
+    {
+      "epoch": 0.8584576054537708,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5349804990016115e-06,
+      "loss": 0.8606,
+      "step": 10074
+    },
+    {
+      "epoch": 0.8585428206220708,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5348168861289222e-06,
+      "loss": 0.1709,
+      "step": 10075
+    },
+    {
+      "epoch": 0.8586280357903707,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.5346532681093054e-06,
+      "loss": 0.3805,
+      "step": 10076
+    },
+    {
+      "epoch": 0.8587132509586707,
+      "grad_norm": 13.25,
+      "learning_rate": 1.534489644945718e-06,
+      "loss": 0.543,
+      "step": 10077
+    },
+    {
+      "epoch": 0.8587984661269706,
+      "grad_norm": 10.375,
+      "learning_rate": 1.5343260166411165e-06,
+      "loss": 0.4412,
+      "step": 10078
+    },
+    {
+      "epoch": 0.8588836812952706,
+      "grad_norm": 24.125,
+      "learning_rate": 1.5341623831984582e-06,
+      "loss": 0.7599,
+      "step": 10079
+    },
+    {
+      "epoch": 0.8589688964635706,
+      "grad_norm": 10.75,
+      "learning_rate": 1.5339987446206993e-06,
+      "loss": 0.9472,
+      "step": 10080
+    },
+    {
+      "epoch": 0.8590541116318705,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.5338351009107979e-06,
+      "loss": 0.9006,
+      "step": 10081
+    },
+    {
+      "epoch": 0.8591393268001705,
+      "grad_norm": 15.0,
+      "learning_rate": 1.533671452071711e-06,
+      "loss": 0.8534,
+      "step": 10082
+    },
+    {
+      "epoch": 0.8592245419684704,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.5335077981063955e-06,
+      "loss": 0.5402,
+      "step": 10083
+    },
+    {
+      "epoch": 0.8593097571367704,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.5333441390178088e-06,
+      "loss": 0.7779,
+      "step": 10084
+    },
+    {
+      "epoch": 0.8593949723050703,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5331804748089092e-06,
+      "loss": 0.8625,
+      "step": 10085
+    },
+    {
+      "epoch": 0.8594801874733703,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.533016805482653e-06,
+      "loss": 0.4704,
+      "step": 10086
+    },
+    {
+      "epoch": 0.8595654026416703,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5328531310419985e-06,
+      "loss": 0.5451,
+      "step": 10087
+    },
+    {
+      "epoch": 0.8596506178099702,
+      "grad_norm": 14.375,
+      "learning_rate": 1.532689451489904e-06,
+      "loss": 0.5704,
+      "step": 10088
+    },
+    {
+      "epoch": 0.8597358329782702,
+      "grad_norm": 14.625,
+      "learning_rate": 1.5325257668293264e-06,
+      "loss": 0.7277,
+      "step": 10089
+    },
+    {
+      "epoch": 0.8598210481465701,
+      "grad_norm": 21.5,
+      "learning_rate": 1.5323620770632248e-06,
+      "loss": 0.7453,
+      "step": 10090
+    },
+    {
+      "epoch": 0.8599062633148701,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5321983821945558e-06,
+      "loss": 0.7724,
+      "step": 10091
+    },
+    {
+      "epoch": 0.85999147848317,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5320346822262786e-06,
+      "loss": 0.9587,
+      "step": 10092
+    },
+    {
+      "epoch": 0.86007669365147,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.5318709771613508e-06,
+      "loss": 0.297,
+      "step": 10093
+    },
+    {
+      "epoch": 0.86016190881977,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5317072670027316e-06,
+      "loss": 0.7796,
+      "step": 10094
+    },
+    {
+      "epoch": 0.8602471239880699,
+      "grad_norm": 16.75,
+      "learning_rate": 1.531543551753378e-06,
+      "loss": 0.967,
+      "step": 10095
+    },
+    {
+      "epoch": 0.8603323391563699,
+      "grad_norm": 12.875,
+      "learning_rate": 1.5313798314162498e-06,
+      "loss": 0.5251,
+      "step": 10096
+    },
+    {
+      "epoch": 0.8604175543246698,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.5312161059943048e-06,
+      "loss": 0.4213,
+      "step": 10097
+    },
+    {
+      "epoch": 0.8605027694929698,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5310523754905024e-06,
+      "loss": 0.9128,
+      "step": 10098
+    },
+    {
+      "epoch": 0.8605879846612697,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5308886399078008e-06,
+      "loss": 0.4332,
+      "step": 10099
+    },
+    {
+      "epoch": 0.8606731998295697,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5307248992491588e-06,
+      "loss": 0.6996,
+      "step": 10100
+    },
+    {
+      "epoch": 0.8607584149978696,
+      "grad_norm": 10.625,
+      "learning_rate": 1.5305611535175357e-06,
+      "loss": 0.5253,
+      "step": 10101
+    },
+    {
+      "epoch": 0.8608436301661696,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5303974027158905e-06,
+      "loss": 0.7075,
+      "step": 10102
+    },
+    {
+      "epoch": 0.8609288453344696,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5302336468471823e-06,
+      "loss": 0.5526,
+      "step": 10103
+    },
+    {
+      "epoch": 0.8610140605027695,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.53006988591437e-06,
+      "loss": 0.8887,
+      "step": 10104
+    },
+    {
+      "epoch": 0.8610992756710695,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.5299061199204136e-06,
+      "loss": 0.6814,
+      "step": 10105
+    },
+    {
+      "epoch": 0.8611844908393694,
+      "grad_norm": 15.125,
+      "learning_rate": 1.5297423488682721e-06,
+      "loss": 0.9171,
+      "step": 10106
+    },
+    {
+      "epoch": 0.8612697060076694,
+      "grad_norm": 18.125,
+      "learning_rate": 1.529578572760905e-06,
+      "loss": 1.0592,
+      "step": 10107
+    },
+    {
+      "epoch": 0.8613549211759693,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.5294147916012717e-06,
+      "loss": 0.2861,
+      "step": 10108
+    },
+    {
+      "epoch": 0.8614401363442693,
+      "grad_norm": 21.375,
+      "learning_rate": 1.5292510053923327e-06,
+      "loss": 1.1863,
+      "step": 10109
+    },
+    {
+      "epoch": 0.8615253515125693,
+      "grad_norm": 15.625,
+      "learning_rate": 1.5290872141370466e-06,
+      "loss": 0.9118,
+      "step": 10110
+    },
+    {
+      "epoch": 0.8616105666808692,
+      "grad_norm": 13.875,
+      "learning_rate": 1.5289234178383744e-06,
+      "loss": 0.5765,
+      "step": 10111
+    },
+    {
+      "epoch": 0.8616957818491692,
+      "grad_norm": 24.5,
+      "learning_rate": 1.5287596164992752e-06,
+      "loss": 1.1467,
+      "step": 10112
+    },
+    {
+      "epoch": 0.8617809970174691,
+      "grad_norm": 19.5,
+      "learning_rate": 1.5285958101227098e-06,
+      "loss": 1.1988,
+      "step": 10113
+    },
+    {
+      "epoch": 0.8618662121857691,
+      "grad_norm": 43.75,
+      "learning_rate": 1.5284319987116371e-06,
+      "loss": 0.4794,
+      "step": 10114
+    },
+    {
+      "epoch": 0.861951427354069,
+      "grad_norm": 23.375,
+      "learning_rate": 1.5282681822690194e-06,
+      "loss": 0.6724,
+      "step": 10115
+    },
+    {
+      "epoch": 0.862036642522369,
+      "grad_norm": 8.875,
+      "learning_rate": 1.528104360797815e-06,
+      "loss": 0.3069,
+      "step": 10116
+    },
+    {
+      "epoch": 0.862121857690669,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5279405343009854e-06,
+      "loss": 0.7261,
+      "step": 10117
+    },
+    {
+      "epoch": 0.8622070728589689,
+      "grad_norm": 14.5,
+      "learning_rate": 1.5277767027814908e-06,
+      "loss": 0.8507,
+      "step": 10118
+    },
+    {
+      "epoch": 0.8622922880272689,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.5276128662422917e-06,
+      "loss": 0.4278,
+      "step": 10119
+    },
+    {
+      "epoch": 0.8623775031955688,
+      "grad_norm": 16.75,
+      "learning_rate": 1.5274490246863495e-06,
+      "loss": 1.0737,
+      "step": 10120
+    },
+    {
+      "epoch": 0.8624627183638688,
+      "grad_norm": 10.0,
+      "learning_rate": 1.527285178116624e-06,
+      "loss": 0.453,
+      "step": 10121
+    },
+    {
+      "epoch": 0.8625479335321687,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5271213265360766e-06,
+      "loss": 0.6308,
+      "step": 10122
+    },
+    {
+      "epoch": 0.8626331487004687,
+      "grad_norm": 17.25,
+      "learning_rate": 1.526957469947668e-06,
+      "loss": 0.6245,
+      "step": 10123
+    },
+    {
+      "epoch": 0.8627183638687687,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5267936083543597e-06,
+      "loss": 0.5867,
+      "step": 10124
+    },
+    {
+      "epoch": 0.8628035790370686,
+      "grad_norm": 12.75,
+      "learning_rate": 1.5266297417591121e-06,
+      "loss": 0.5257,
+      "step": 10125
+    },
+    {
+      "epoch": 0.8628887942053686,
+      "grad_norm": 12.0,
+      "learning_rate": 1.5264658701648874e-06,
+      "loss": 0.4022,
+      "step": 10126
+    },
+    {
+      "epoch": 0.8629740093736685,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.5263019935746462e-06,
+      "loss": 0.7267,
+      "step": 10127
+    },
+    {
+      "epoch": 0.8630592245419685,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.5261381119913504e-06,
+      "loss": 0.8053,
+      "step": 10128
+    },
+    {
+      "epoch": 0.8631444397102684,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5259742254179607e-06,
+      "loss": 0.7908,
+      "step": 10129
+    },
+    {
+      "epoch": 0.8632296548785684,
+      "grad_norm": 22.625,
+      "learning_rate": 1.5258103338574396e-06,
+      "loss": 0.8839,
+      "step": 10130
+    },
+    {
+      "epoch": 0.8633148700468684,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.5256464373127483e-06,
+      "loss": 0.6898,
+      "step": 10131
+    },
+    {
+      "epoch": 0.8634000852151683,
+      "grad_norm": 12.875,
+      "learning_rate": 1.5254825357868488e-06,
+      "loss": 0.5738,
+      "step": 10132
+    },
+    {
+      "epoch": 0.8634853003834683,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5253186292827027e-06,
+      "loss": 0.7751,
+      "step": 10133
+    },
+    {
+      "epoch": 0.8635705155517682,
+      "grad_norm": 14.25,
+      "learning_rate": 1.5251547178032722e-06,
+      "loss": 0.634,
+      "step": 10134
+    },
+    {
+      "epoch": 0.8636557307200682,
+      "grad_norm": 15.875,
+      "learning_rate": 1.5249908013515193e-06,
+      "loss": 0.4709,
+      "step": 10135
+    },
+    {
+      "epoch": 0.8637409458883681,
+      "grad_norm": 19.5,
+      "learning_rate": 1.5248268799304062e-06,
+      "loss": 0.9359,
+      "step": 10136
+    },
+    {
+      "epoch": 0.8638261610566681,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5246629535428947e-06,
+      "loss": 0.6998,
+      "step": 10137
+    },
+    {
+      "epoch": 0.863911376224968,
+      "grad_norm": 11.25,
+      "learning_rate": 1.5244990221919479e-06,
+      "loss": 0.4152,
+      "step": 10138
+    },
+    {
+      "epoch": 0.863996591393268,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5243350858805272e-06,
+      "loss": 0.5428,
+      "step": 10139
+    },
+    {
+      "epoch": 0.864081806561568,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5241711446115958e-06,
+      "loss": 0.8772,
+      "step": 10140
+    },
+    {
+      "epoch": 0.8641670217298679,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5240071983881164e-06,
+      "loss": 0.9802,
+      "step": 10141
+    },
+    {
+      "epoch": 0.8642522368981679,
+      "grad_norm": 17.625,
+      "learning_rate": 1.523843247213051e-06,
+      "loss": 0.8427,
+      "step": 10142
+    },
+    {
+      "epoch": 0.8643374520664678,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.5236792910893633e-06,
+      "loss": 0.8581,
+      "step": 10143
+    },
+    {
+      "epoch": 0.8644226672347678,
+      "grad_norm": 18.625,
+      "learning_rate": 1.523515330020015e-06,
+      "loss": 0.812,
+      "step": 10144
+    },
+    {
+      "epoch": 0.8645078824030678,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.52335136400797e-06,
+      "loss": 0.858,
+      "step": 10145
+    },
+    {
+      "epoch": 0.8645930975713677,
+      "grad_norm": 13.0,
+      "learning_rate": 1.5231873930561909e-06,
+      "loss": 0.5582,
+      "step": 10146
+    },
+    {
+      "epoch": 0.8646783127396677,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5230234171676412e-06,
+      "loss": 0.7779,
+      "step": 10147
+    },
+    {
+      "epoch": 0.8647635279079676,
+      "grad_norm": 21.375,
+      "learning_rate": 1.5228594363452831e-06,
+      "loss": 0.6991,
+      "step": 10148
+    },
+    {
+      "epoch": 0.8648487430762676,
+      "grad_norm": 12.875,
+      "learning_rate": 1.522695450592081e-06,
+      "loss": 0.5775,
+      "step": 10149
+    },
+    {
+      "epoch": 0.8649339582445675,
+      "grad_norm": 18.625,
+      "learning_rate": 1.5225314599109978e-06,
+      "loss": 0.8759,
+      "step": 10150
+    },
+    {
+      "epoch": 0.8650191734128675,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5223674643049975e-06,
+      "loss": 0.8011,
+      "step": 10151
+    },
+    {
+      "epoch": 0.8651043885811674,
+      "grad_norm": 9.875,
+      "learning_rate": 1.5222034637770424e-06,
+      "loss": 0.2941,
+      "step": 10152
+    },
+    {
+      "epoch": 0.8651896037494674,
+      "grad_norm": 17.75,
+      "learning_rate": 1.5220394583300976e-06,
+      "loss": 0.5878,
+      "step": 10153
+    },
+    {
+      "epoch": 0.8652748189177674,
+      "grad_norm": 13.625,
+      "learning_rate": 1.521875447967126e-06,
+      "loss": 0.8522,
+      "step": 10154
+    },
+    {
+      "epoch": 0.8653600340860673,
+      "grad_norm": 11.25,
+      "learning_rate": 1.5217114326910918e-06,
+      "loss": 0.5354,
+      "step": 10155
+    },
+    {
+      "epoch": 0.8654452492543673,
+      "grad_norm": 16.875,
+      "learning_rate": 1.521547412504959e-06,
+      "loss": 0.8246,
+      "step": 10156
+    },
+    {
+      "epoch": 0.8655304644226672,
+      "grad_norm": 12.75,
+      "learning_rate": 1.5213833874116908e-06,
+      "loss": 0.6896,
+      "step": 10157
+    },
+    {
+      "epoch": 0.8656156795909672,
+      "grad_norm": 14.625,
+      "learning_rate": 1.5212193574142522e-06,
+      "loss": 0.7967,
+      "step": 10158
+    },
+    {
+      "epoch": 0.8657008947592671,
+      "grad_norm": 15.75,
+      "learning_rate": 1.5210553225156072e-06,
+      "loss": 0.6133,
+      "step": 10159
+    },
+    {
+      "epoch": 0.8657861099275671,
+      "grad_norm": 17.375,
+      "learning_rate": 1.52089128271872e-06,
+      "loss": 0.8595,
+      "step": 10160
+    },
+    {
+      "epoch": 0.8658713250958671,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5207272380265545e-06,
+      "loss": 1.028,
+      "step": 10161
+    },
+    {
+      "epoch": 0.865956540264167,
+      "grad_norm": 21.0,
+      "learning_rate": 1.520563188442076e-06,
+      "loss": 0.7477,
+      "step": 10162
+    },
+    {
+      "epoch": 0.866041755432467,
+      "grad_norm": 18.5,
+      "learning_rate": 1.5203991339682487e-06,
+      "loss": 1.1012,
+      "step": 10163
+    },
+    {
+      "epoch": 0.8661269706007669,
+      "grad_norm": 19.5,
+      "learning_rate": 1.520235074608037e-06,
+      "loss": 0.7039,
+      "step": 10164
+    },
+    {
+      "epoch": 0.8662121857690669,
+      "grad_norm": 15.875,
+      "learning_rate": 1.5200710103644054e-06,
+      "loss": 1.0405,
+      "step": 10165
+    },
+    {
+      "epoch": 0.8662974009373668,
+      "grad_norm": 19.0,
+      "learning_rate": 1.5199069412403193e-06,
+      "loss": 1.412,
+      "step": 10166
+    },
+    {
+      "epoch": 0.8663826161056668,
+      "grad_norm": 11.75,
+      "learning_rate": 1.5197428672387432e-06,
+      "loss": 0.4241,
+      "step": 10167
+    },
+    {
+      "epoch": 0.8664678312739668,
+      "grad_norm": 15.625,
+      "learning_rate": 1.519578788362643e-06,
+      "loss": 0.8056,
+      "step": 10168
+    },
+    {
+      "epoch": 0.8665530464422667,
+      "grad_norm": 13.875,
+      "learning_rate": 1.5194147046149821e-06,
+      "loss": 1.0708,
+      "step": 10169
+    },
+    {
+      "epoch": 0.8666382616105667,
+      "grad_norm": 17.875,
+      "learning_rate": 1.519250615998727e-06,
+      "loss": 0.7453,
+      "step": 10170
+    },
+    {
+      "epoch": 0.8667234767788666,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.5190865225168422e-06,
+      "loss": 0.3066,
+      "step": 10171
+    },
+    {
+      "epoch": 0.8668086919471666,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.5189224241722935e-06,
+      "loss": 0.4125,
+      "step": 10172
+    },
+    {
+      "epoch": 0.8668939071154665,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.5187583209680464e-06,
+      "loss": 0.6304,
+      "step": 10173
+    },
+    {
+      "epoch": 0.8669791222837665,
+      "grad_norm": 12.125,
+      "learning_rate": 1.5185942129070657e-06,
+      "loss": 0.9896,
+      "step": 10174
+    },
+    {
+      "epoch": 0.8670643374520665,
+      "grad_norm": 17.0,
+      "learning_rate": 1.518430099992318e-06,
+      "loss": 1.027,
+      "step": 10175
+    },
+    {
+      "epoch": 0.8671495526203664,
+      "grad_norm": 32.0,
+      "learning_rate": 1.5182659822267685e-06,
+      "loss": 0.6872,
+      "step": 10176
+    },
+    {
+      "epoch": 0.8672347677886664,
+      "grad_norm": 12.25,
+      "learning_rate": 1.5181018596133825e-06,
+      "loss": 0.5029,
+      "step": 10177
+    },
+    {
+      "epoch": 0.8673199829569663,
+      "grad_norm": 12.0,
+      "learning_rate": 1.5179377321551264e-06,
+      "loss": 0.5507,
+      "step": 10178
+    },
+    {
+      "epoch": 0.8674051981252663,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5177735998549664e-06,
+      "loss": 0.9249,
+      "step": 10179
+    },
+    {
+      "epoch": 0.8674904132935662,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.517609462715868e-06,
+      "loss": 0.6407,
+      "step": 10180
+    },
+    {
+      "epoch": 0.8675756284618662,
+      "grad_norm": 16.875,
+      "learning_rate": 1.5174453207407974e-06,
+      "loss": 0.9847,
+      "step": 10181
+    },
+    {
+      "epoch": 0.8676608436301662,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.517281173932721e-06,
+      "loss": 0.8747,
+      "step": 10182
+    },
+    {
+      "epoch": 0.8677460587984661,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.517117022294605e-06,
+      "loss": 0.8621,
+      "step": 10183
+    },
+    {
+      "epoch": 0.8678312739667661,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5169528658294158e-06,
+      "loss": 0.5167,
+      "step": 10184
+    },
+    {
+      "epoch": 0.867916489135066,
+      "grad_norm": 13.0,
+      "learning_rate": 1.51678870454012e-06,
+      "loss": 0.4209,
+      "step": 10185
+    },
+    {
+      "epoch": 0.868001704303366,
+      "grad_norm": 13.125,
+      "learning_rate": 1.5166245384296838e-06,
+      "loss": 0.7921,
+      "step": 10186
+    },
+    {
+      "epoch": 0.8680869194716659,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5164603675010741e-06,
+      "loss": 0.6952,
+      "step": 10187
+    },
+    {
+      "epoch": 0.8681721346399659,
+      "grad_norm": 16.0,
+      "learning_rate": 1.5162961917572577e-06,
+      "loss": 0.8782,
+      "step": 10188
+    },
+    {
+      "epoch": 0.8682573498082659,
+      "grad_norm": 18.625,
+      "learning_rate": 1.5161320112012012e-06,
+      "loss": 0.8164,
+      "step": 10189
+    },
+    {
+      "epoch": 0.8683425649765658,
+      "grad_norm": 10.625,
+      "learning_rate": 1.5159678258358718e-06,
+      "loss": 0.6189,
+      "step": 10190
+    },
+    {
+      "epoch": 0.8684277801448658,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5158036356642364e-06,
+      "loss": 0.7026,
+      "step": 10191
+    },
+    {
+      "epoch": 0.8685129953131657,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.515639440689262e-06,
+      "loss": 0.6142,
+      "step": 10192
+    },
+    {
+      "epoch": 0.8685982104814657,
+      "grad_norm": 12.5,
+      "learning_rate": 1.5154752409139156e-06,
+      "loss": 0.6408,
+      "step": 10193
+    },
+    {
+      "epoch": 0.8686834256497656,
+      "grad_norm": 11.875,
+      "learning_rate": 1.5153110363411647e-06,
+      "loss": 0.3683,
+      "step": 10194
+    },
+    {
+      "epoch": 0.8687686408180656,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5151468269739764e-06,
+      "loss": 0.8836,
+      "step": 10195
+    },
+    {
+      "epoch": 0.8688538559863656,
+      "grad_norm": 14.625,
+      "learning_rate": 1.514982612815319e-06,
+      "loss": 0.5157,
+      "step": 10196
+    },
+    {
+      "epoch": 0.8689390711546655,
+      "grad_norm": 12.125,
+      "learning_rate": 1.5148183938681586e-06,
+      "loss": 0.6051,
+      "step": 10197
+    },
+    {
+      "epoch": 0.8690242863229655,
+      "grad_norm": 12.875,
+      "learning_rate": 1.514654170135464e-06,
+      "loss": 0.6138,
+      "step": 10198
+    },
+    {
+      "epoch": 0.8691095014912654,
+      "grad_norm": 11.25,
+      "learning_rate": 1.5144899416202019e-06,
+      "loss": 0.4368,
+      "step": 10199
+    },
+    {
+      "epoch": 0.8691947166595654,
+      "grad_norm": 12.625,
+      "learning_rate": 1.5143257083253413e-06,
+      "loss": 0.6164,
+      "step": 10200
+    },
+    {
+      "epoch": 0.8692799318278653,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.5141614702538488e-06,
+      "loss": 0.4718,
+      "step": 10201
+    },
+    {
+      "epoch": 0.8693651469961653,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.5139972274086928e-06,
+      "loss": 0.4012,
+      "step": 10202
+    },
+    {
+      "epoch": 0.8694503621644653,
+      "grad_norm": 14.625,
+      "learning_rate": 1.5138329797928416e-06,
+      "loss": 0.8543,
+      "step": 10203
+    },
+    {
+      "epoch": 0.8695355773327652,
+      "grad_norm": 18.875,
+      "learning_rate": 1.5136687274092632e-06,
+      "loss": 1.0437,
+      "step": 10204
+    },
+    {
+      "epoch": 0.8696207925010652,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5135044702609259e-06,
+      "loss": 0.4893,
+      "step": 10205
+    },
+    {
+      "epoch": 0.8697060076693651,
+      "grad_norm": 13.625,
+      "learning_rate": 1.5133402083507978e-06,
+      "loss": 0.8573,
+      "step": 10206
+    },
+    {
+      "epoch": 0.8697912228376651,
+      "grad_norm": 12.625,
+      "learning_rate": 1.513175941681847e-06,
+      "loss": 0.6931,
+      "step": 10207
+    },
+    {
+      "epoch": 0.869876438005965,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5130116702570427e-06,
+      "loss": 0.6333,
+      "step": 10208
+    },
+    {
+      "epoch": 0.869961653174265,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5128473940793531e-06,
+      "loss": 0.4928,
+      "step": 10209
+    },
+    {
+      "epoch": 0.870046868342565,
+      "grad_norm": 25.5,
+      "learning_rate": 1.5126831131517464e-06,
+      "loss": 0.9812,
+      "step": 10210
+    },
+    {
+      "epoch": 0.8701320835108649,
+      "grad_norm": 16.0,
+      "learning_rate": 1.5125188274771918e-06,
+      "loss": 0.8645,
+      "step": 10211
+    },
+    {
+      "epoch": 0.8702172986791649,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.5123545370586581e-06,
+      "loss": 0.7024,
+      "step": 10212
+    },
+    {
+      "epoch": 0.8703025138474648,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.5121902418991142e-06,
+      "loss": 0.6007,
+      "step": 10213
+    },
+    {
+      "epoch": 0.8703877290157648,
+      "grad_norm": 13.375,
+      "learning_rate": 1.5120259420015286e-06,
+      "loss": 0.6857,
+      "step": 10214
+    },
+    {
+      "epoch": 0.8704729441840647,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.511861637368871e-06,
+      "loss": 0.6279,
+      "step": 10215
+    },
+    {
+      "epoch": 0.8705581593523647,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5116973280041103e-06,
+      "loss": 1.0007,
+      "step": 10216
+    },
+    {
+      "epoch": 0.8706433745206646,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5115330139102157e-06,
+      "loss": 0.9044,
+      "step": 10217
+    },
+    {
+      "epoch": 0.8707285896889646,
+      "grad_norm": 14.5,
+      "learning_rate": 1.5113686950901566e-06,
+      "loss": 0.5351,
+      "step": 10218
+    },
+    {
+      "epoch": 0.8708138048572646,
+      "grad_norm": 13.375,
+      "learning_rate": 1.5112043715469022e-06,
+      "loss": 0.7688,
+      "step": 10219
+    },
+    {
+      "epoch": 0.8708990200255645,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5110400432834224e-06,
+      "loss": 1.0671,
+      "step": 10220
+    },
+    {
+      "epoch": 0.8709842351938645,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.5108757103026864e-06,
+      "loss": 0.7939,
+      "step": 10221
+    },
+    {
+      "epoch": 0.8710694503621644,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.510711372607664e-06,
+      "loss": 0.5829,
+      "step": 10222
+    },
+    {
+      "epoch": 0.8711546655304644,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.5105470302013248e-06,
+      "loss": 0.7145,
+      "step": 10223
+    },
+    {
+      "epoch": 0.8712398806987643,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.510382683086639e-06,
+      "loss": 0.3441,
+      "step": 10224
+    },
+    {
+      "epoch": 0.8713250958670643,
+      "grad_norm": 17.75,
+      "learning_rate": 1.5102183312665757e-06,
+      "loss": 0.8196,
+      "step": 10225
+    },
+    {
+      "epoch": 0.8714103110353643,
+      "grad_norm": 17.75,
+      "learning_rate": 1.510053974744106e-06,
+      "loss": 0.9774,
+      "step": 10226
+    },
+    {
+      "epoch": 0.8714955262036642,
+      "grad_norm": 10.375,
+      "learning_rate": 1.509889613522199e-06,
+      "loss": 0.3194,
+      "step": 10227
+    },
+    {
+      "epoch": 0.8715807413719642,
+      "grad_norm": 10.25,
+      "learning_rate": 1.5097252476038262e-06,
+      "loss": 0.2893,
+      "step": 10228
+    },
+    {
+      "epoch": 0.8716659565402641,
+      "grad_norm": 13.625,
+      "learning_rate": 1.5095608769919566e-06,
+      "loss": 0.8017,
+      "step": 10229
+    },
+    {
+      "epoch": 0.8717511717085641,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5093965016895606e-06,
+      "loss": 0.8652,
+      "step": 10230
+    },
+    {
+      "epoch": 0.871836386876864,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.5092321216996092e-06,
+      "loss": 0.4625,
+      "step": 10231
+    },
+    {
+      "epoch": 0.871921602045164,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5090677370250729e-06,
+      "loss": 0.4071,
+      "step": 10232
+    },
+    {
+      "epoch": 0.872006817213464,
+      "grad_norm": 17.0,
+      "learning_rate": 1.5089033476689219e-06,
+      "loss": 0.9149,
+      "step": 10233
+    },
+    {
+      "epoch": 0.8720920323817639,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.508738953634127e-06,
+      "loss": 1.024,
+      "step": 10234
+    },
+    {
+      "epoch": 0.8721772475500639,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5085745549236588e-06,
+      "loss": 0.8391,
+      "step": 10235
+    },
+    {
+      "epoch": 0.8722624627183638,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.5084101515404889e-06,
+      "loss": 0.5198,
+      "step": 10236
+    },
+    {
+      "epoch": 0.8723476778866638,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5082457434875874e-06,
+      "loss": 0.8274,
+      "step": 10237
+    },
+    {
+      "epoch": 0.8724328930549637,
+      "grad_norm": 16.125,
+      "learning_rate": 1.508081330767926e-06,
+      "loss": 0.9801,
+      "step": 10238
+    },
+    {
+      "epoch": 0.8725181082232637,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.5079169133844749e-06,
+      "loss": 0.591,
+      "step": 10239
+    },
+    {
+      "epoch": 0.8726033233915637,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.507752491340206e-06,
+      "loss": 0.6914,
+      "step": 10240
+    },
+    {
+      "epoch": 0.8726885385598636,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.5075880646380905e-06,
+      "loss": 1.0586,
+      "step": 10241
+    },
+    {
+      "epoch": 0.8727737537281636,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.5074236332810993e-06,
+      "loss": 0.3778,
+      "step": 10242
+    },
+    {
+      "epoch": 0.8728589688964635,
+      "grad_norm": 18.25,
+      "learning_rate": 1.5072591972722047e-06,
+      "loss": 0.8658,
+      "step": 10243
+    },
+    {
+      "epoch": 0.8729441840647635,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5070947566143775e-06,
+      "loss": 0.8338,
+      "step": 10244
+    },
+    {
+      "epoch": 0.8730293992330634,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5069303113105895e-06,
+      "loss": 0.8188,
+      "step": 10245
+    },
+    {
+      "epoch": 0.8731146144013634,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.5067658613638124e-06,
+      "loss": 0.5875,
+      "step": 10246
+    },
+    {
+      "epoch": 0.8731998295696634,
+      "grad_norm": 15.125,
+      "learning_rate": 1.506601406777018e-06,
+      "loss": 0.9176,
+      "step": 10247
+    },
+    {
+      "epoch": 0.8732850447379633,
+      "grad_norm": 21.0,
+      "learning_rate": 1.506436947553178e-06,
+      "loss": 0.6382,
+      "step": 10248
+    },
+    {
+      "epoch": 0.8733702599062633,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.5062724836952649e-06,
+      "loss": 0.6083,
+      "step": 10249
+    },
+    {
+      "epoch": 0.8734554750745632,
+      "grad_norm": 14.875,
+      "learning_rate": 1.5061080152062498e-06,
+      "loss": 0.6199,
+      "step": 10250
+    },
+    {
+      "epoch": 0.8735406902428632,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5059435420891058e-06,
+      "loss": 0.6179,
+      "step": 10251
+    },
+    {
+      "epoch": 0.8736259054111631,
+      "grad_norm": 15.625,
+      "learning_rate": 1.5057790643468042e-06,
+      "loss": 0.7413,
+      "step": 10252
+    },
+    {
+      "epoch": 0.8737111205794631,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5056145819823182e-06,
+      "loss": 0.6481,
+      "step": 10253
+    },
+    {
+      "epoch": 0.873796335747763,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.5054500949986193e-06,
+      "loss": 0.5957,
+      "step": 10254
+    },
+    {
+      "epoch": 0.873881550916063,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5052856033986804e-06,
+      "loss": 1.1521,
+      "step": 10255
+    },
+    {
+      "epoch": 0.873966766084363,
+      "grad_norm": 13.375,
+      "learning_rate": 1.5051211071854738e-06,
+      "loss": 0.636,
+      "step": 10256
+    },
+    {
+      "epoch": 0.8740519812526629,
+      "grad_norm": 15.125,
+      "learning_rate": 1.5049566063619725e-06,
+      "loss": 0.7734,
+      "step": 10257
+    },
+    {
+      "epoch": 0.8741371964209629,
+      "grad_norm": 16.375,
+      "learning_rate": 1.504792100931149e-06,
+      "loss": 0.6933,
+      "step": 10258
+    },
+    {
+      "epoch": 0.8742224115892628,
+      "grad_norm": 14.5,
+      "learning_rate": 1.5046275908959755e-06,
+      "loss": 0.5732,
+      "step": 10259
+    },
+    {
+      "epoch": 0.8743076267575628,
+      "grad_norm": 13.875,
+      "learning_rate": 1.5044630762594257e-06,
+      "loss": 0.6248,
+      "step": 10260
+    },
+    {
+      "epoch": 0.8743928419258628,
+      "grad_norm": 15.875,
+      "learning_rate": 1.5042985570244722e-06,
+      "loss": 0.9475,
+      "step": 10261
+    },
+    {
+      "epoch": 0.8744780570941627,
+      "grad_norm": 14.25,
+      "learning_rate": 1.5041340331940885e-06,
+      "loss": 0.9059,
+      "step": 10262
+    },
+    {
+      "epoch": 0.8745632722624627,
+      "grad_norm": 21.5,
+      "learning_rate": 1.5039695047712466e-06,
+      "loss": 0.7358,
+      "step": 10263
+    },
+    {
+      "epoch": 0.8746484874307626,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5038049717589206e-06,
+      "loss": 1.0387,
+      "step": 10264
+    },
+    {
+      "epoch": 0.8747337025990626,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.5036404341600835e-06,
+      "loss": 0.5354,
+      "step": 10265
+    },
+    {
+      "epoch": 0.8748189177673625,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.503475891977709e-06,
+      "loss": 0.5545,
+      "step": 10266
+    },
+    {
+      "epoch": 0.8749041329356625,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.50331134521477e-06,
+      "loss": 0.688,
+      "step": 10267
+    },
+    {
+      "epoch": 0.8749893481039624,
+      "grad_norm": 11.125,
+      "learning_rate": 1.5031467938742404e-06,
+      "loss": 0.4511,
+      "step": 10268
+    },
+    {
+      "epoch": 0.8750745632722625,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5029822379590936e-06,
+      "loss": 0.4503,
+      "step": 10269
+    },
+    {
+      "epoch": 0.8751597784405625,
+      "grad_norm": 16.75,
+      "learning_rate": 1.5028176774723033e-06,
+      "loss": 0.9331,
+      "step": 10270
+    },
+    {
+      "epoch": 0.8752449936088624,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5026531124168431e-06,
+      "loss": 0.8639,
+      "step": 10271
+    },
+    {
+      "epoch": 0.8753302087771624,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.5024885427956877e-06,
+      "loss": 0.7644,
+      "step": 10272
+    },
+    {
+      "epoch": 0.8754154239454623,
+      "grad_norm": 12.75,
+      "learning_rate": 1.50232396861181e-06,
+      "loss": 0.4953,
+      "step": 10273
+    },
+    {
+      "epoch": 0.8755006391137623,
+      "grad_norm": 24.375,
+      "learning_rate": 1.5021593898681844e-06,
+      "loss": 0.9405,
+      "step": 10274
+    },
+    {
+      "epoch": 0.8755858542820623,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.5019948065677852e-06,
+      "loss": 0.5444,
+      "step": 10275
+    },
+    {
+      "epoch": 0.8756710694503622,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5018302187135866e-06,
+      "loss": 0.5449,
+      "step": 10276
+    },
+    {
+      "epoch": 0.8757562846186622,
+      "grad_norm": 13.375,
+      "learning_rate": 1.5016656263085627e-06,
+      "loss": 0.608,
+      "step": 10277
+    },
+    {
+      "epoch": 0.8758414997869621,
+      "grad_norm": 25.375,
+      "learning_rate": 1.5015010293556875e-06,
+      "loss": 1.1416,
+      "step": 10278
+    },
+    {
+      "epoch": 0.8759267149552621,
+      "grad_norm": 12.75,
+      "learning_rate": 1.5013364278579362e-06,
+      "loss": 0.5875,
+      "step": 10279
+    },
+    {
+      "epoch": 0.876011930123562,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5011718218182824e-06,
+      "loss": 0.338,
+      "step": 10280
+    },
+    {
+      "epoch": 0.876097145291862,
+      "grad_norm": 17.0,
+      "learning_rate": 1.501007211239702e-06,
+      "loss": 0.7929,
+      "step": 10281
+    },
+    {
+      "epoch": 0.876182360460162,
+      "grad_norm": 20.75,
+      "learning_rate": 1.5008425961251682e-06,
+      "loss": 0.8831,
+      "step": 10282
+    },
+    {
+      "epoch": 0.8762675756284619,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5006779764776565e-06,
+      "loss": 1.1731,
+      "step": 10283
+    },
+    {
+      "epoch": 0.8763527907967619,
+      "grad_norm": 18.5,
+      "learning_rate": 1.500513352300142e-06,
+      "loss": 0.9694,
+      "step": 10284
+    },
+    {
+      "epoch": 0.8764380059650618,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5003487235955994e-06,
+      "loss": 0.9148,
+      "step": 10285
+    },
+    {
+      "epoch": 0.8765232211333618,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5001840903670034e-06,
+      "loss": 0.9419,
+      "step": 10286
+    },
+    {
+      "epoch": 0.8766084363016617,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5000194526173291e-06,
+      "loss": 0.7039,
+      "step": 10287
+    },
+    {
+      "epoch": 0.8766936514699617,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4998548103495519e-06,
+      "loss": 0.6037,
+      "step": 10288
+    },
+    {
+      "epoch": 0.8767788666382617,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4996901635666475e-06,
+      "loss": 0.8228,
+      "step": 10289
+    },
+    {
+      "epoch": 0.8768640818065616,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.4995255122715905e-06,
+      "loss": 0.6379,
+      "step": 10290
+    },
+    {
+      "epoch": 0.8769492969748616,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.4993608564673567e-06,
+      "loss": 0.5076,
+      "step": 10291
+    },
+    {
+      "epoch": 0.8770345121431615,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.499196196156921e-06,
+      "loss": 0.6565,
+      "step": 10292
+    },
+    {
+      "epoch": 0.8771197273114615,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.49903153134326e-06,
+      "loss": 0.9863,
+      "step": 10293
+    },
+    {
+      "epoch": 0.8772049424797614,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.4988668620293487e-06,
+      "loss": 0.9442,
+      "step": 10294
+    },
+    {
+      "epoch": 0.8772901576480614,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.4987021882181625e-06,
+      "loss": 0.5681,
+      "step": 10295
+    },
+    {
+      "epoch": 0.8773753728163614,
+      "grad_norm": 12.5,
+      "learning_rate": 1.4985375099126781e-06,
+      "loss": 0.5828,
+      "step": 10296
+    },
+    {
+      "epoch": 0.8774605879846613,
+      "grad_norm": 13.875,
+      "learning_rate": 1.4983728271158709e-06,
+      "loss": 0.6703,
+      "step": 10297
+    },
+    {
+      "epoch": 0.8775458031529613,
+      "grad_norm": 18.0,
+      "learning_rate": 1.498208139830717e-06,
+      "loss": 0.68,
+      "step": 10298
+    },
+    {
+      "epoch": 0.8776310183212612,
+      "grad_norm": 12.125,
+      "learning_rate": 1.4980434480601922e-06,
+      "loss": 0.7133,
+      "step": 10299
+    },
+    {
+      "epoch": 0.8777162334895612,
+      "grad_norm": 17.625,
+      "learning_rate": 1.4978787518072732e-06,
+      "loss": 0.7827,
+      "step": 10300
+    },
+    {
+      "epoch": 0.8778014486578611,
+      "grad_norm": 15.75,
+      "learning_rate": 1.4977140510749354e-06,
+      "loss": 0.6571,
+      "step": 10301
+    },
+    {
+      "epoch": 0.8778866638261611,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4975493458661564e-06,
+      "loss": 0.7491,
+      "step": 10302
+    },
+    {
+      "epoch": 0.877971878994461,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.497384636183911e-06,
+      "loss": 0.5453,
+      "step": 10303
+    },
+    {
+      "epoch": 0.878057094162761,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4972199220311766e-06,
+      "loss": 0.9451,
+      "step": 10304
+    },
+    {
+      "epoch": 0.878142309331061,
+      "grad_norm": 12.75,
+      "learning_rate": 1.4970552034109298e-06,
+      "loss": 0.6084,
+      "step": 10305
+    },
+    {
+      "epoch": 0.8782275244993609,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.4968904803261475e-06,
+      "loss": 0.8222,
+      "step": 10306
+    },
+    {
+      "epoch": 0.8783127396676609,
+      "grad_norm": 19.625,
+      "learning_rate": 1.4967257527798052e-06,
+      "loss": 0.845,
+      "step": 10307
+    },
+    {
+      "epoch": 0.8783979548359608,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.4965610207748806e-06,
+      "loss": 0.2869,
+      "step": 10308
+    },
+    {
+      "epoch": 0.8784831700042608,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.4963962843143503e-06,
+      "loss": 1.0366,
+      "step": 10309
+    },
+    {
+      "epoch": 0.8785683851725608,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.4962315434011918e-06,
+      "loss": 0.705,
+      "step": 10310
+    },
+    {
+      "epoch": 0.8786536003408607,
+      "grad_norm": 15.25,
+      "learning_rate": 1.4960667980383816e-06,
+      "loss": 0.4886,
+      "step": 10311
+    },
+    {
+      "epoch": 0.8787388155091607,
+      "grad_norm": 18.375,
+      "learning_rate": 1.4959020482288966e-06,
+      "loss": 0.8341,
+      "step": 10312
+    },
+    {
+      "epoch": 0.8788240306774606,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.4957372939757148e-06,
+      "loss": 0.5601,
+      "step": 10313
+    },
+    {
+      "epoch": 0.8789092458457606,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4955725352818127e-06,
+      "loss": 0.8358,
+      "step": 10314
+    },
+    {
+      "epoch": 0.8789944610140605,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.495407772150168e-06,
+      "loss": 0.76,
+      "step": 10315
+    },
+    {
+      "epoch": 0.8790796761823605,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.495243004583758e-06,
+      "loss": 0.7222,
+      "step": 10316
+    },
+    {
+      "epoch": 0.8791648913506604,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4950782325855603e-06,
+      "loss": 0.8994,
+      "step": 10317
+    },
+    {
+      "epoch": 0.8792501065189604,
+      "grad_norm": 15.5,
+      "learning_rate": 1.4949134561585527e-06,
+      "loss": 0.936,
+      "step": 10318
+    },
+    {
+      "epoch": 0.8793353216872604,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.4947486753057125e-06,
+      "loss": 0.5492,
+      "step": 10319
+    },
+    {
+      "epoch": 0.8794205368555603,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.4945838900300174e-06,
+      "loss": 0.5649,
+      "step": 10320
+    },
+    {
+      "epoch": 0.8795057520238603,
+      "grad_norm": 12.75,
+      "learning_rate": 1.4944191003344457e-06,
+      "loss": 0.7471,
+      "step": 10321
+    },
+    {
+      "epoch": 0.8795909671921602,
+      "grad_norm": 11.75,
+      "learning_rate": 1.4942543062219752e-06,
+      "loss": 0.6514,
+      "step": 10322
+    },
+    {
+      "epoch": 0.8796761823604602,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4940895076955838e-06,
+      "loss": 0.8471,
+      "step": 10323
+    },
+    {
+      "epoch": 0.8797613975287601,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4939247047582491e-06,
+      "loss": 0.8896,
+      "step": 10324
+    },
+    {
+      "epoch": 0.8798466126970601,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4937598974129503e-06,
+      "loss": 1.0708,
+      "step": 10325
+    },
+    {
+      "epoch": 0.8799318278653601,
+      "grad_norm": 13.75,
+      "learning_rate": 1.4935950856626646e-06,
+      "loss": 0.781,
+      "step": 10326
+    },
+    {
+      "epoch": 0.88001704303366,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4934302695103708e-06,
+      "loss": 0.5999,
+      "step": 10327
+    },
+    {
+      "epoch": 0.88010225820196,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4932654489590473e-06,
+      "loss": 0.496,
+      "step": 10328
+    },
+    {
+      "epoch": 0.8801874733702599,
+      "grad_norm": 12.375,
+      "learning_rate": 1.4931006240116726e-06,
+      "loss": 0.5902,
+      "step": 10329
+    },
+    {
+      "epoch": 0.8802726885385599,
+      "grad_norm": 12.125,
+      "learning_rate": 1.4929357946712253e-06,
+      "loss": 0.3672,
+      "step": 10330
+    },
+    {
+      "epoch": 0.8803579037068598,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.4927709609406835e-06,
+      "loss": 0.6805,
+      "step": 10331
+    },
+    {
+      "epoch": 0.8804431188751598,
+      "grad_norm": 19.0,
+      "learning_rate": 1.4926061228230264e-06,
+      "loss": 1.146,
+      "step": 10332
+    },
+    {
+      "epoch": 0.8805283340434598,
+      "grad_norm": 12.375,
+      "learning_rate": 1.4924412803212327e-06,
+      "loss": 0.5561,
+      "step": 10333
+    },
+    {
+      "epoch": 0.8806135492117597,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4922764334382816e-06,
+      "loss": 0.8491,
+      "step": 10334
+    },
+    {
+      "epoch": 0.8806987643800597,
+      "grad_norm": 10.375,
+      "learning_rate": 1.4921115821771514e-06,
+      "loss": 0.3205,
+      "step": 10335
+    },
+    {
+      "epoch": 0.8807839795483596,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4919467265408216e-06,
+      "loss": 0.6154,
+      "step": 10336
+    },
+    {
+      "epoch": 0.8808691947166596,
+      "grad_norm": 16.125,
+      "learning_rate": 1.4917818665322708e-06,
+      "loss": 0.8127,
+      "step": 10337
+    },
+    {
+      "epoch": 0.8809544098849595,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4916170021544792e-06,
+      "loss": 0.6141,
+      "step": 10338
+    },
+    {
+      "epoch": 0.8810396250532595,
+      "grad_norm": 19.75,
+      "learning_rate": 1.491452133410425e-06,
+      "loss": 1.012,
+      "step": 10339
+    },
+    {
+      "epoch": 0.8811248402215595,
+      "grad_norm": 16.875,
+      "learning_rate": 1.491287260303088e-06,
+      "loss": 0.8337,
+      "step": 10340
+    },
+    {
+      "epoch": 0.8812100553898594,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4911223828354473e-06,
+      "loss": 1.0503,
+      "step": 10341
+    },
+    {
+      "epoch": 0.8812952705581594,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4909575010104832e-06,
+      "loss": 0.8615,
+      "step": 10342
+    },
+    {
+      "epoch": 0.8813804857264593,
+      "grad_norm": 23.25,
+      "learning_rate": 1.4907926148311747e-06,
+      "loss": 0.7213,
+      "step": 10343
+    },
+    {
+      "epoch": 0.8814657008947593,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4906277243005012e-06,
+      "loss": 0.7568,
+      "step": 10344
+    },
+    {
+      "epoch": 0.8815509160630592,
+      "grad_norm": 23.25,
+      "learning_rate": 1.4904628294214426e-06,
+      "loss": 1.1367,
+      "step": 10345
+    },
+    {
+      "epoch": 0.8816361312313592,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4902979301969795e-06,
+      "loss": 0.9534,
+      "step": 10346
+    },
+    {
+      "epoch": 0.8817213463996592,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4901330266300905e-06,
+      "loss": 1.0428,
+      "step": 10347
+    },
+    {
+      "epoch": 0.8818065615679591,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.4899681187237567e-06,
+      "loss": 0.5236,
+      "step": 10348
+    },
+    {
+      "epoch": 0.8818917767362591,
+      "grad_norm": 25.125,
+      "learning_rate": 1.4898032064809575e-06,
+      "loss": 0.7212,
+      "step": 10349
+    },
+    {
+      "epoch": 0.881976991904559,
+      "grad_norm": 22.5,
+      "learning_rate": 1.4896382899046733e-06,
+      "loss": 1.0081,
+      "step": 10350
+    },
+    {
+      "epoch": 0.882062207072859,
+      "grad_norm": 39.0,
+      "learning_rate": 1.489473368997884e-06,
+      "loss": 0.8309,
+      "step": 10351
+    },
+    {
+      "epoch": 0.8821474222411589,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4893084437635702e-06,
+      "loss": 0.6007,
+      "step": 10352
+    },
+    {
+      "epoch": 0.8822326374094589,
+      "grad_norm": 12.25,
+      "learning_rate": 1.4891435142047123e-06,
+      "loss": 0.553,
+      "step": 10353
+    },
+    {
+      "epoch": 0.8823178525777589,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4889785803242907e-06,
+      "loss": 0.8502,
+      "step": 10354
+    },
+    {
+      "epoch": 0.8824030677460588,
+      "grad_norm": 15.5,
+      "learning_rate": 1.4888136421252857e-06,
+      "loss": 0.7555,
+      "step": 10355
+    },
+    {
+      "epoch": 0.8824882829143588,
+      "grad_norm": 19.625,
+      "learning_rate": 1.4886486996106781e-06,
+      "loss": 0.8821,
+      "step": 10356
+    },
+    {
+      "epoch": 0.8825734980826587,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.4884837527834485e-06,
+      "loss": 0.4516,
+      "step": 10357
+    },
+    {
+      "epoch": 0.8826587132509587,
+      "grad_norm": 17.75,
+      "learning_rate": 1.4883188016465776e-06,
+      "loss": 0.7758,
+      "step": 10358
+    },
+    {
+      "epoch": 0.8827439284192586,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4881538462030466e-06,
+      "loss": 0.9124,
+      "step": 10359
+    },
+    {
+      "epoch": 0.8828291435875586,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.4879888864558356e-06,
+      "loss": 0.56,
+      "step": 10360
+    },
+    {
+      "epoch": 0.8829143587558586,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.4878239224079265e-06,
+      "loss": 0.7051,
+      "step": 10361
+    },
+    {
+      "epoch": 0.8829995739241585,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.4876589540622998e-06,
+      "loss": 0.6376,
+      "step": 10362
+    },
+    {
+      "epoch": 0.8830847890924585,
+      "grad_norm": 12.125,
+      "learning_rate": 1.487493981421937e-06,
+      "loss": 0.6026,
+      "step": 10363
+    },
+    {
+      "epoch": 0.8831700042607584,
+      "grad_norm": 18.125,
+      "learning_rate": 1.487329004489819e-06,
+      "loss": 0.9058,
+      "step": 10364
+    },
+    {
+      "epoch": 0.8832552194290584,
+      "grad_norm": 20.375,
+      "learning_rate": 1.487164023268927e-06,
+      "loss": 0.748,
+      "step": 10365
+    },
+    {
+      "epoch": 0.8833404345973583,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4869990377622432e-06,
+      "loss": 0.6584,
+      "step": 10366
+    },
+    {
+      "epoch": 0.8834256497656583,
+      "grad_norm": 29.375,
+      "learning_rate": 1.4868340479727477e-06,
+      "loss": 1.1751,
+      "step": 10367
+    },
+    {
+      "epoch": 0.8835108649339583,
+      "grad_norm": 20.375,
+      "learning_rate": 1.4866690539034235e-06,
+      "loss": 0.9223,
+      "step": 10368
+    },
+    {
+      "epoch": 0.8835960801022582,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.486504055557251e-06,
+      "loss": 0.6066,
+      "step": 10369
+    },
+    {
+      "epoch": 0.8836812952705582,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4863390529372126e-06,
+      "loss": 0.8881,
+      "step": 10370
+    },
+    {
+      "epoch": 0.8837665104388581,
+      "grad_norm": 18.0,
+      "learning_rate": 1.48617404604629e-06,
+      "loss": 1.0259,
+      "step": 10371
+    },
+    {
+      "epoch": 0.8838517256071581,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4860090348874648e-06,
+      "loss": 0.8101,
+      "step": 10372
+    },
+    {
+      "epoch": 0.883936940775458,
+      "grad_norm": 19.375,
+      "learning_rate": 1.485844019463719e-06,
+      "loss": 0.7419,
+      "step": 10373
+    },
+    {
+      "epoch": 0.884022155943758,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.4856789997780347e-06,
+      "loss": 0.482,
+      "step": 10374
+    },
+    {
+      "epoch": 0.884107371112058,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4855139758333936e-06,
+      "loss": 0.4415,
+      "step": 10375
+    },
+    {
+      "epoch": 0.8841925862803579,
+      "grad_norm": 12.875,
+      "learning_rate": 1.4853489476327785e-06,
+      "loss": 0.7432,
+      "step": 10376
+    },
+    {
+      "epoch": 0.8842778014486579,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4851839151791707e-06,
+      "loss": 0.6941,
+      "step": 10377
+    },
+    {
+      "epoch": 0.8843630166169578,
+      "grad_norm": 13.125,
+      "learning_rate": 1.4850188784755538e-06,
+      "loss": 0.6133,
+      "step": 10378
+    },
+    {
+      "epoch": 0.8844482317852578,
+      "grad_norm": 22.25,
+      "learning_rate": 1.4848538375249088e-06,
+      "loss": 0.6276,
+      "step": 10379
+    },
+    {
+      "epoch": 0.8845334469535577,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.484688792330219e-06,
+      "loss": 0.3604,
+      "step": 10380
+    },
+    {
+      "epoch": 0.8846186621218577,
+      "grad_norm": 19.375,
+      "learning_rate": 1.484523742894467e-06,
+      "loss": 0.4631,
+      "step": 10381
+    },
+    {
+      "epoch": 0.8847038772901576,
+      "grad_norm": 12.625,
+      "learning_rate": 1.4843586892206346e-06,
+      "loss": 0.5561,
+      "step": 10382
+    },
+    {
+      "epoch": 0.8847890924584576,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4841936313117056e-06,
+      "loss": 0.5532,
+      "step": 10383
+    },
+    {
+      "epoch": 0.8848743076267576,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4840285691706618e-06,
+      "loss": 0.8556,
+      "step": 10384
+    },
+    {
+      "epoch": 0.8849595227950575,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4838635028004866e-06,
+      "loss": 0.7676,
+      "step": 10385
+    },
+    {
+      "epoch": 0.8850447379633575,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.4836984322041626e-06,
+      "loss": 0.5567,
+      "step": 10386
+    },
+    {
+      "epoch": 0.8851299531316574,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4835333573846733e-06,
+      "loss": 1.0572,
+      "step": 10387
+    },
+    {
+      "epoch": 0.8852151682999574,
+      "grad_norm": 11.0,
+      "learning_rate": 1.4833682783450007e-06,
+      "loss": 0.4456,
+      "step": 10388
+    },
+    {
+      "epoch": 0.8853003834682573,
+      "grad_norm": 11.125,
+      "learning_rate": 1.4832031950881292e-06,
+      "loss": 0.4612,
+      "step": 10389
+    },
+    {
+      "epoch": 0.8853855986365573,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4830381076170411e-06,
+      "loss": 0.7626,
+      "step": 10390
+    },
+    {
+      "epoch": 0.8854708138048573,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4828730159347204e-06,
+      "loss": 0.5575,
+      "step": 10391
+    },
+    {
+      "epoch": 0.8855560289731572,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4827079200441498e-06,
+      "loss": 0.8416,
+      "step": 10392
+    },
+    {
+      "epoch": 0.8856412441414572,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.4825428199483135e-06,
+      "loss": 0.6601,
+      "step": 10393
+    },
+    {
+      "epoch": 0.8857264593097571,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.4823777156501943e-06,
+      "loss": 0.6939,
+      "step": 10394
+    },
+    {
+      "epoch": 0.8858116744780571,
+      "grad_norm": 17.125,
+      "learning_rate": 1.482212607152776e-06,
+      "loss": 0.943,
+      "step": 10395
+    },
+    {
+      "epoch": 0.885896889646357,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.4820474944590426e-06,
+      "loss": 0.6323,
+      "step": 10396
+    },
+    {
+      "epoch": 0.885982104814657,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4818823775719777e-06,
+      "loss": 0.815,
+      "step": 10397
+    },
+    {
+      "epoch": 0.886067319982957,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4817172564945648e-06,
+      "loss": 0.7762,
+      "step": 10398
+    },
+    {
+      "epoch": 0.8861525351512569,
+      "grad_norm": 17.875,
+      "learning_rate": 1.4815521312297882e-06,
+      "loss": 0.71,
+      "step": 10399
+    },
+    {
+      "epoch": 0.8862377503195569,
+      "grad_norm": 16.625,
+      "learning_rate": 1.481387001780632e-06,
+      "loss": 0.7437,
+      "step": 10400
+    },
+    {
+      "epoch": 0.8863229654878568,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4812218681500794e-06,
+      "loss": 0.9264,
+      "step": 10401
+    },
+    {
+      "epoch": 0.8864081806561568,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4810567303411155e-06,
+      "loss": 0.8544,
+      "step": 10402
+    },
+    {
+      "epoch": 0.8864933958244567,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4808915883567244e-06,
+      "loss": 1.0451,
+      "step": 10403
+    },
+    {
+      "epoch": 0.8865786109927567,
+      "grad_norm": 15.125,
+      "learning_rate": 1.4807264421998899e-06,
+      "loss": 0.851,
+      "step": 10404
+    },
+    {
+      "epoch": 0.8866638261610567,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4805612918735963e-06,
+      "loss": 0.9114,
+      "step": 10405
+    },
+    {
+      "epoch": 0.8867490413293566,
+      "grad_norm": 18.75,
+      "learning_rate": 1.4803961373808285e-06,
+      "loss": 0.7644,
+      "step": 10406
+    },
+    {
+      "epoch": 0.8868342564976566,
+      "grad_norm": 14.375,
+      "learning_rate": 1.4802309787245708e-06,
+      "loss": 0.5669,
+      "step": 10407
+    },
+    {
+      "epoch": 0.8869194716659565,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.480065815907808e-06,
+      "loss": 0.4577,
+      "step": 10408
+    },
+    {
+      "epoch": 0.8870046868342565,
+      "grad_norm": 17.875,
+      "learning_rate": 1.4799006489335244e-06,
+      "loss": 1.0599,
+      "step": 10409
+    },
+    {
+      "epoch": 0.8870899020025564,
+      "grad_norm": 10.0,
+      "learning_rate": 1.4797354778047048e-06,
+      "loss": 0.3546,
+      "step": 10410
+    },
+    {
+      "epoch": 0.8871751171708564,
+      "grad_norm": 22.5,
+      "learning_rate": 1.4795703025243344e-06,
+      "loss": 1.0385,
+      "step": 10411
+    },
+    {
+      "epoch": 0.8872603323391564,
+      "grad_norm": 11.0,
+      "learning_rate": 1.4794051230953973e-06,
+      "loss": 0.5235,
+      "step": 10412
+    },
+    {
+      "epoch": 0.8873455475074563,
+      "grad_norm": 13.75,
+      "learning_rate": 1.4792399395208792e-06,
+      "loss": 0.7556,
+      "step": 10413
+    },
+    {
+      "epoch": 0.8874307626757563,
+      "grad_norm": 13.375,
+      "learning_rate": 1.4790747518037652e-06,
+      "loss": 0.7138,
+      "step": 10414
+    },
+    {
+      "epoch": 0.8875159778440562,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4789095599470396e-06,
+      "loss": 0.9096,
+      "step": 10415
+    },
+    {
+      "epoch": 0.8876011930123562,
+      "grad_norm": 17.125,
+      "learning_rate": 1.478744363953688e-06,
+      "loss": 0.8506,
+      "step": 10416
+    },
+    {
+      "epoch": 0.8876864081806561,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4785791638266961e-06,
+      "loss": 0.5188,
+      "step": 10417
+    },
+    {
+      "epoch": 0.8877716233489561,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4784139595690486e-06,
+      "loss": 0.7408,
+      "step": 10418
+    },
+    {
+      "epoch": 0.887856838517256,
+      "grad_norm": 20.875,
+      "learning_rate": 1.4782487511837314e-06,
+      "loss": 0.9588,
+      "step": 10419
+    },
+    {
+      "epoch": 0.887942053685556,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.4780835386737297e-06,
+      "loss": 0.7271,
+      "step": 10420
+    },
+    {
+      "epoch": 0.888027268853856,
+      "grad_norm": 12.75,
+      "learning_rate": 1.477918322042029e-06,
+      "loss": 0.6001,
+      "step": 10421
+    },
+    {
+      "epoch": 0.8881124840221559,
+      "grad_norm": 13.0,
+      "learning_rate": 1.4777531012916147e-06,
+      "loss": 0.5802,
+      "step": 10422
+    },
+    {
+      "epoch": 0.8881976991904559,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.4775878764254737e-06,
+      "loss": 0.4183,
+      "step": 10423
+    },
+    {
+      "epoch": 0.8882829143587558,
+      "grad_norm": 28.375,
+      "learning_rate": 1.4774226474465902e-06,
+      "loss": 1.2556,
+      "step": 10424
+    },
+    {
+      "epoch": 0.8883681295270558,
+      "grad_norm": 14.5,
+      "learning_rate": 1.477257414357951e-06,
+      "loss": 0.9505,
+      "step": 10425
+    },
+    {
+      "epoch": 0.8884533446953558,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4770921771625415e-06,
+      "loss": 0.6335,
+      "step": 10426
+    },
+    {
+      "epoch": 0.8885385598636557,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4769269358633485e-06,
+      "loss": 0.6803,
+      "step": 10427
+    },
+    {
+      "epoch": 0.8886237750319557,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4767616904633572e-06,
+      "loss": 0.6328,
+      "step": 10428
+    },
+    {
+      "epoch": 0.8887089902002556,
+      "grad_norm": 19.75,
+      "learning_rate": 1.4765964409655544e-06,
+      "loss": 0.8183,
+      "step": 10429
+    },
+    {
+      "epoch": 0.8887942053685556,
+      "grad_norm": 18.75,
+      "learning_rate": 1.4764311873729259e-06,
+      "loss": 0.8082,
+      "step": 10430
+    },
+    {
+      "epoch": 0.8888794205368555,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.4762659296884581e-06,
+      "loss": 0.6917,
+      "step": 10431
+    },
+    {
+      "epoch": 0.8889646357051555,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4761006679151377e-06,
+      "loss": 0.4813,
+      "step": 10432
+    },
+    {
+      "epoch": 0.8890498508734554,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4759354020559501e-06,
+      "loss": 0.9395,
+      "step": 10433
+    },
+    {
+      "epoch": 0.8891350660417554,
+      "grad_norm": 13.375,
+      "learning_rate": 1.4757701321138834e-06,
+      "loss": 0.4614,
+      "step": 10434
+    },
+    {
+      "epoch": 0.8892202812100554,
+      "grad_norm": 19.875,
+      "learning_rate": 1.475604858091923e-06,
+      "loss": 0.5981,
+      "step": 10435
+    },
+    {
+      "epoch": 0.8893054963783553,
+      "grad_norm": 11.75,
+      "learning_rate": 1.4754395799930565e-06,
+      "loss": 0.4949,
+      "step": 10436
+    },
+    {
+      "epoch": 0.8893907115466553,
+      "grad_norm": 15.75,
+      "learning_rate": 1.4752742978202694e-06,
+      "loss": 0.7965,
+      "step": 10437
+    },
+    {
+      "epoch": 0.8894759267149552,
+      "grad_norm": 16.125,
+      "learning_rate": 1.4751090115765495e-06,
+      "loss": 0.8804,
+      "step": 10438
+    },
+    {
+      "epoch": 0.8895611418832552,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4749437212648832e-06,
+      "loss": 0.8737,
+      "step": 10439
+    },
+    {
+      "epoch": 0.8896463570515551,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.4747784268882584e-06,
+      "loss": 0.5247,
+      "step": 10440
+    },
+    {
+      "epoch": 0.8897315722198551,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.4746131284496606e-06,
+      "loss": 0.2444,
+      "step": 10441
+    },
+    {
+      "epoch": 0.8898167873881551,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4744478259520777e-06,
+      "loss": 0.4327,
+      "step": 10442
+    },
+    {
+      "epoch": 0.889902002556455,
+      "grad_norm": 19.375,
+      "learning_rate": 1.4742825193984973e-06,
+      "loss": 0.9235,
+      "step": 10443
+    },
+    {
+      "epoch": 0.889987217724755,
+      "grad_norm": 16.25,
+      "learning_rate": 1.474117208791906e-06,
+      "loss": 0.8009,
+      "step": 10444
+    },
+    {
+      "epoch": 0.8900724328930549,
+      "grad_norm": 12.25,
+      "learning_rate": 1.4739518941352915e-06,
+      "loss": 0.5349,
+      "step": 10445
+    },
+    {
+      "epoch": 0.8901576480613549,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.473786575431641e-06,
+      "loss": 0.7068,
+      "step": 10446
+    },
+    {
+      "epoch": 0.8902428632296548,
+      "grad_norm": 21.625,
+      "learning_rate": 1.4736212526839418e-06,
+      "loss": 0.6457,
+      "step": 10447
+    },
+    {
+      "epoch": 0.8903280783979548,
+      "grad_norm": 35.5,
+      "learning_rate": 1.4734559258951822e-06,
+      "loss": 1.0247,
+      "step": 10448
+    },
+    {
+      "epoch": 0.8904132935662548,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.473290595068349e-06,
+      "loss": 0.7623,
+      "step": 10449
+    },
+    {
+      "epoch": 0.8904985087345547,
+      "grad_norm": 10.75,
+      "learning_rate": 1.4731252602064303e-06,
+      "loss": 0.3704,
+      "step": 10450
+    },
+    {
+      "epoch": 0.8905837239028547,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4729599213124138e-06,
+      "loss": 0.8652,
+      "step": 10451
+    },
+    {
+      "epoch": 0.8906689390711546,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.4727945783892872e-06,
+      "loss": 1.0799,
+      "step": 10452
+    },
+    {
+      "epoch": 0.8907541542394546,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.4726292314400386e-06,
+      "loss": 0.5508,
+      "step": 10453
+    },
+    {
+      "epoch": 0.8908393694077545,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4724638804676557e-06,
+      "loss": 0.9287,
+      "step": 10454
+    },
+    {
+      "epoch": 0.8909245845760545,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4722985254751274e-06,
+      "loss": 0.7868,
+      "step": 10455
+    },
+    {
+      "epoch": 0.8910097997443545,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4721331664654407e-06,
+      "loss": 0.8529,
+      "step": 10456
+    },
+    {
+      "epoch": 0.8910950149126544,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.4719678034415846e-06,
+      "loss": 0.6699,
+      "step": 10457
+    },
+    {
+      "epoch": 0.8911802300809544,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.4718024364065468e-06,
+      "loss": 0.7054,
+      "step": 10458
+    },
+    {
+      "epoch": 0.8912654452492543,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.471637065363316e-06,
+      "loss": 0.6142,
+      "step": 10459
+    },
+    {
+      "epoch": 0.8913506604175543,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4714716903148807e-06,
+      "loss": 0.7312,
+      "step": 10460
+    },
+    {
+      "epoch": 0.8914358755858542,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4713063112642289e-06,
+      "loss": 0.4958,
+      "step": 10461
+    },
+    {
+      "epoch": 0.8915210907541542,
+      "grad_norm": 24.5,
+      "learning_rate": 1.4711409282143495e-06,
+      "loss": 0.79,
+      "step": 10462
+    },
+    {
+      "epoch": 0.8916063059224542,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4709755411682316e-06,
+      "loss": 0.6834,
+      "step": 10463
+    },
+    {
+      "epoch": 0.8916915210907541,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4708101501288635e-06,
+      "loss": 0.8516,
+      "step": 10464
+    },
+    {
+      "epoch": 0.8917767362590541,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4706447550992333e-06,
+      "loss": 0.4842,
+      "step": 10465
+    },
+    {
+      "epoch": 0.891861951427354,
+      "grad_norm": 16.75,
+      "learning_rate": 1.470479356082331e-06,
+      "loss": 0.7353,
+      "step": 10466
+    },
+    {
+      "epoch": 0.891947166595654,
+      "grad_norm": 13.25,
+      "learning_rate": 1.470313953081145e-06,
+      "loss": 0.6872,
+      "step": 10467
+    },
+    {
+      "epoch": 0.8920323817639539,
+      "grad_norm": 21.5,
+      "learning_rate": 1.4701485460986646e-06,
+      "loss": 1.0796,
+      "step": 10468
+    },
+    {
+      "epoch": 0.8921175969322539,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.469983135137878e-06,
+      "loss": 0.8546,
+      "step": 10469
+    },
+    {
+      "epoch": 0.8922028121005539,
+      "grad_norm": 15.125,
+      "learning_rate": 1.4698177202017751e-06,
+      "loss": 1.0181,
+      "step": 10470
+    },
+    {
+      "epoch": 0.8922880272688538,
+      "grad_norm": 15.0,
+      "learning_rate": 1.469652301293345e-06,
+      "loss": 0.5778,
+      "step": 10471
+    },
+    {
+      "epoch": 0.8923732424371538,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.4694868784155776e-06,
+      "loss": 0.793,
+      "step": 10472
+    },
+    {
+      "epoch": 0.8924584576054537,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.4693214515714604e-06,
+      "loss": 0.9842,
+      "step": 10473
+    },
+    {
+      "epoch": 0.8925436727737537,
+      "grad_norm": 16.5,
+      "learning_rate": 1.469156020763985e-06,
+      "loss": 0.9131,
+      "step": 10474
+    },
+    {
+      "epoch": 0.8926288879420536,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4689905859961392e-06,
+      "loss": 1.0074,
+      "step": 10475
+    },
+    {
+      "epoch": 0.8927141031103536,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.4688251472709138e-06,
+      "loss": 0.2624,
+      "step": 10476
+    },
+    {
+      "epoch": 0.8927993182786536,
+      "grad_norm": 15.125,
+      "learning_rate": 1.4686597045912978e-06,
+      "loss": 0.7598,
+      "step": 10477
+    },
+    {
+      "epoch": 0.8928845334469535,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4684942579602809e-06,
+      "loss": 0.5933,
+      "step": 10478
+    },
+    {
+      "epoch": 0.8929697486152535,
+      "grad_norm": 13.375,
+      "learning_rate": 1.4683288073808533e-06,
+      "loss": 0.6608,
+      "step": 10479
+    },
+    {
+      "epoch": 0.8930549637835534,
+      "grad_norm": 17.75,
+      "learning_rate": 1.4681633528560043e-06,
+      "loss": 0.7335,
+      "step": 10480
+    },
+    {
+      "epoch": 0.8931401789518534,
+      "grad_norm": 19.125,
+      "learning_rate": 1.4679978943887245e-06,
+      "loss": 1.1301,
+      "step": 10481
+    },
+    {
+      "epoch": 0.8932253941201533,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4678324319820032e-06,
+      "loss": 0.8879,
+      "step": 10482
+    },
+    {
+      "epoch": 0.8933106092884533,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4676669656388308e-06,
+      "loss": 0.793,
+      "step": 10483
+    },
+    {
+      "epoch": 0.8933958244567533,
+      "grad_norm": 17.75,
+      "learning_rate": 1.467501495362198e-06,
+      "loss": 0.6204,
+      "step": 10484
+    },
+    {
+      "epoch": 0.8934810396250532,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.467336021155094e-06,
+      "loss": 0.6806,
+      "step": 10485
+    },
+    {
+      "epoch": 0.8935662547933532,
+      "grad_norm": 19.125,
+      "learning_rate": 1.4671705430205093e-06,
+      "loss": 1.0269,
+      "step": 10486
+    },
+    {
+      "epoch": 0.8936514699616531,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.467005060961435e-06,
+      "loss": 0.7571,
+      "step": 10487
+    },
+    {
+      "epoch": 0.8937366851299531,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.4668395749808613e-06,
+      "loss": 0.6106,
+      "step": 10488
+    },
+    {
+      "epoch": 0.893821900298253,
+      "grad_norm": 15.25,
+      "learning_rate": 1.4666740850817778e-06,
+      "loss": 0.9196,
+      "step": 10489
+    },
+    {
+      "epoch": 0.893907115466553,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4665085912671756e-06,
+      "loss": 0.5685,
+      "step": 10490
+    },
+    {
+      "epoch": 0.893992330634853,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4663430935400457e-06,
+      "loss": 0.5427,
+      "step": 10491
+    },
+    {
+      "epoch": 0.8940775458031529,
+      "grad_norm": 28.5,
+      "learning_rate": 1.4661775919033788e-06,
+      "loss": 0.7841,
+      "step": 10492
+    },
+    {
+      "epoch": 0.8941627609714529,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.4660120863601654e-06,
+      "loss": 0.5337,
+      "step": 10493
+    },
+    {
+      "epoch": 0.8942479761397529,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4658465769133958e-06,
+      "loss": 0.7457,
+      "step": 10494
+    },
+    {
+      "epoch": 0.8943331913080529,
+      "grad_norm": 26.625,
+      "learning_rate": 1.465681063566062e-06,
+      "loss": 1.1096,
+      "step": 10495
+    },
+    {
+      "epoch": 0.8944184064763528,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.4655155463211545e-06,
+      "loss": 0.5296,
+      "step": 10496
+    },
+    {
+      "epoch": 0.8945036216446528,
+      "grad_norm": 19.25,
+      "learning_rate": 1.465350025181664e-06,
+      "loss": 0.5372,
+      "step": 10497
+    },
+    {
+      "epoch": 0.8945888368129528,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4651845001505824e-06,
+      "loss": 0.5976,
+      "step": 10498
+    },
+    {
+      "epoch": 0.8946740519812527,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4650189712309004e-06,
+      "loss": 0.5368,
+      "step": 10499
+    },
+    {
+      "epoch": 0.8947592671495527,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4648534384256093e-06,
+      "loss": 0.8193,
+      "step": 10500
+    },
+    {
+      "epoch": 0.8948444823178526,
+      "grad_norm": 11.25,
+      "learning_rate": 1.4646879017377005e-06,
+      "loss": 0.4957,
+      "step": 10501
+    },
+    {
+      "epoch": 0.8949296974861526,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.4645223611701656e-06,
+      "loss": 0.6787,
+      "step": 10502
+    },
+    {
+      "epoch": 0.8950149126544525,
+      "grad_norm": 15.0,
+      "learning_rate": 1.4643568167259955e-06,
+      "loss": 0.5931,
+      "step": 10503
+    },
+    {
+      "epoch": 0.8951001278227525,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4641912684081826e-06,
+      "loss": 0.747,
+      "step": 10504
+    },
+    {
+      "epoch": 0.8951853429910525,
+      "grad_norm": 11.375,
+      "learning_rate": 1.464025716219718e-06,
+      "loss": 0.6261,
+      "step": 10505
+    },
+    {
+      "epoch": 0.8952705581593524,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4638601601635936e-06,
+      "loss": 0.3864,
+      "step": 10506
+    },
+    {
+      "epoch": 0.8953557733276524,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4636946002428007e-06,
+      "loss": 1.0076,
+      "step": 10507
+    },
+    {
+      "epoch": 0.8954409884959523,
+      "grad_norm": 18.0,
+      "learning_rate": 1.4635290364603316e-06,
+      "loss": 0.7457,
+      "step": 10508
+    },
+    {
+      "epoch": 0.8955262036642523,
+      "grad_norm": 15.0,
+      "learning_rate": 1.4633634688191784e-06,
+      "loss": 0.763,
+      "step": 10509
+    },
+    {
+      "epoch": 0.8956114188325522,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4631978973223326e-06,
+      "loss": 0.7147,
+      "step": 10510
+    },
+    {
+      "epoch": 0.8956966340008522,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.463032321972786e-06,
+      "loss": 0.9861,
+      "step": 10511
+    },
+    {
+      "epoch": 0.8957818491691522,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4628667427735315e-06,
+      "loss": 0.8539,
+      "step": 10512
+    },
+    {
+      "epoch": 0.8958670643374521,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4627011597275611e-06,
+      "loss": 0.7863,
+      "step": 10513
+    },
+    {
+      "epoch": 0.8959522795057521,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4625355728378666e-06,
+      "loss": 0.6729,
+      "step": 10514
+    },
+    {
+      "epoch": 0.896037494674052,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4623699821074405e-06,
+      "loss": 0.8672,
+      "step": 10515
+    },
+    {
+      "epoch": 0.896122709842352,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.4622043875392755e-06,
+      "loss": 0.4958,
+      "step": 10516
+    },
+    {
+      "epoch": 0.8962079250106519,
+      "grad_norm": 15.25,
+      "learning_rate": 1.4620387891363636e-06,
+      "loss": 0.8627,
+      "step": 10517
+    },
+    {
+      "epoch": 0.8962931401789519,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4618731869016976e-06,
+      "loss": 0.667,
+      "step": 10518
+    },
+    {
+      "epoch": 0.8963783553472519,
+      "grad_norm": 19.875,
+      "learning_rate": 1.4617075808382703e-06,
+      "loss": 1.2579,
+      "step": 10519
+    },
+    {
+      "epoch": 0.8964635705155518,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.4615419709490736e-06,
+      "loss": 0.5814,
+      "step": 10520
+    },
+    {
+      "epoch": 0.8965487856838518,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4613763572371015e-06,
+      "loss": 0.8294,
+      "step": 10521
+    },
+    {
+      "epoch": 0.8966340008521517,
+      "grad_norm": 12.25,
+      "learning_rate": 1.4612107397053455e-06,
+      "loss": 0.6992,
+      "step": 10522
+    },
+    {
+      "epoch": 0.8967192160204517,
+      "grad_norm": 13.375,
+      "learning_rate": 1.4610451183567991e-06,
+      "loss": 0.6981,
+      "step": 10523
+    },
+    {
+      "epoch": 0.8968044311887516,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.4608794931944554e-06,
+      "loss": 0.2513,
+      "step": 10524
+    },
+    {
+      "epoch": 0.8968896463570516,
+      "grad_norm": 15.5,
+      "learning_rate": 1.4607138642213072e-06,
+      "loss": 0.7937,
+      "step": 10525
+    },
+    {
+      "epoch": 0.8969748615253516,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.4605482314403474e-06,
+      "loss": 0.7128,
+      "step": 10526
+    },
+    {
+      "epoch": 0.8970600766936515,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4603825948545694e-06,
+      "loss": 0.9293,
+      "step": 10527
+    },
+    {
+      "epoch": 0.8971452918619515,
+      "grad_norm": 19.125,
+      "learning_rate": 1.4602169544669662e-06,
+      "loss": 1.0681,
+      "step": 10528
+    },
+    {
+      "epoch": 0.8972305070302514,
+      "grad_norm": 17.75,
+      "learning_rate": 1.460051310280532e-06,
+      "loss": 0.8572,
+      "step": 10529
+    },
+    {
+      "epoch": 0.8973157221985514,
+      "grad_norm": 21.0,
+      "learning_rate": 1.4598856622982586e-06,
+      "loss": 0.9751,
+      "step": 10530
+    },
+    {
+      "epoch": 0.8974009373668513,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.4597200105231404e-06,
+      "loss": 0.7405,
+      "step": 10531
+    },
+    {
+      "epoch": 0.8974861525351513,
+      "grad_norm": 17.25,
+      "learning_rate": 1.4595543549581706e-06,
+      "loss": 0.7768,
+      "step": 10532
+    },
+    {
+      "epoch": 0.8975713677034513,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.459388695606343e-06,
+      "loss": 0.5918,
+      "step": 10533
+    },
+    {
+      "epoch": 0.8976565828717512,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4592230324706514e-06,
+      "loss": 0.4957,
+      "step": 10534
+    },
+    {
+      "epoch": 0.8977417980400512,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4590573655540887e-06,
+      "loss": 0.9917,
+      "step": 10535
+    },
+    {
+      "epoch": 0.8978270132083511,
+      "grad_norm": 18.5,
+      "learning_rate": 1.4588916948596496e-06,
+      "loss": 0.942,
+      "step": 10536
+    },
+    {
+      "epoch": 0.8979122283766511,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4587260203903274e-06,
+      "loss": 0.9032,
+      "step": 10537
+    },
+    {
+      "epoch": 0.897997443544951,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4585603421491162e-06,
+      "loss": 0.7249,
+      "step": 10538
+    },
+    {
+      "epoch": 0.898082658713251,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4583946601390095e-06,
+      "loss": 0.7009,
+      "step": 10539
+    },
+    {
+      "epoch": 0.898167873881551,
+      "grad_norm": 17.0,
+      "learning_rate": 1.458228974363002e-06,
+      "loss": 0.9629,
+      "step": 10540
+    },
+    {
+      "epoch": 0.8982530890498509,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.4580632848240877e-06,
+      "loss": 0.4967,
+      "step": 10541
+    },
+    {
+      "epoch": 0.8983383042181509,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.4578975915252607e-06,
+      "loss": 0.6625,
+      "step": 10542
+    },
+    {
+      "epoch": 0.8984235193864508,
+      "grad_norm": 15.375,
+      "learning_rate": 1.4577318944695145e-06,
+      "loss": 0.949,
+      "step": 10543
+    },
+    {
+      "epoch": 0.8985087345547508,
+      "grad_norm": 12.625,
+      "learning_rate": 1.4575661936598447e-06,
+      "loss": 0.5023,
+      "step": 10544
+    },
+    {
+      "epoch": 0.8985939497230507,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4574004890992451e-06,
+      "loss": 0.7043,
+      "step": 10545
+    },
+    {
+      "epoch": 0.8986791648913507,
+      "grad_norm": 20.125,
+      "learning_rate": 1.45723478079071e-06,
+      "loss": 0.6524,
+      "step": 10546
+    },
+    {
+      "epoch": 0.8987643800596506,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4570690687372338e-06,
+      "loss": 0.7655,
+      "step": 10547
+    },
+    {
+      "epoch": 0.8988495952279506,
+      "grad_norm": 25.25,
+      "learning_rate": 1.4569033529418113e-06,
+      "loss": 1.3418,
+      "step": 10548
+    },
+    {
+      "epoch": 0.8989348103962506,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4567376334074372e-06,
+      "loss": 0.6286,
+      "step": 10549
+    },
+    {
+      "epoch": 0.8990200255645505,
+      "grad_norm": 11.75,
+      "learning_rate": 1.4565719101371062e-06,
+      "loss": 0.4998,
+      "step": 10550
+    },
+    {
+      "epoch": 0.8991052407328505,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4564061831338132e-06,
+      "loss": 0.8654,
+      "step": 10551
+    },
+    {
+      "epoch": 0.8991904559011504,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4562404524005527e-06,
+      "loss": 0.9501,
+      "step": 10552
+    },
+    {
+      "epoch": 0.8992756710694504,
+      "grad_norm": 12.875,
+      "learning_rate": 1.4560747179403203e-06,
+      "loss": 0.5746,
+      "step": 10553
+    },
+    {
+      "epoch": 0.8993608862377503,
+      "grad_norm": 18.0,
+      "learning_rate": 1.4559089797561099e-06,
+      "loss": 1.26,
+      "step": 10554
+    },
+    {
+      "epoch": 0.8994461014060503,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4557432378509176e-06,
+      "loss": 0.4765,
+      "step": 10555
+    },
+    {
+      "epoch": 0.8995313165743503,
+      "grad_norm": 15.75,
+      "learning_rate": 1.4555774922277377e-06,
+      "loss": 0.3654,
+      "step": 10556
+    },
+    {
+      "epoch": 0.8996165317426502,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4554117428895664e-06,
+      "loss": 0.679,
+      "step": 10557
+    },
+    {
+      "epoch": 0.8997017469109502,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.4552459898393977e-06,
+      "loss": 0.4775,
+      "step": 10558
+    },
+    {
+      "epoch": 0.8997869620792501,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.455080233080228e-06,
+      "loss": 0.3762,
+      "step": 10559
+    },
+    {
+      "epoch": 0.8998721772475501,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4549144726150522e-06,
+      "loss": 0.7119,
+      "step": 10560
+    },
+    {
+      "epoch": 0.89995739241585,
+      "grad_norm": 19.625,
+      "learning_rate": 1.4547487084468659e-06,
+      "loss": 0.9674,
+      "step": 10561
+    },
+    {
+      "epoch": 0.90004260758415,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4545829405786643e-06,
+      "loss": 0.7943,
+      "step": 10562
+    },
+    {
+      "epoch": 0.90012782275245,
+      "grad_norm": 12.125,
+      "learning_rate": 1.4544171690134434e-06,
+      "loss": 0.5507,
+      "step": 10563
+    },
+    {
+      "epoch": 0.9002130379207499,
+      "grad_norm": 11.375,
+      "learning_rate": 1.4542513937541983e-06,
+      "loss": 0.3777,
+      "step": 10564
+    },
+    {
+      "epoch": 0.9002982530890499,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4540856148039256e-06,
+      "loss": 0.6855,
+      "step": 10565
+    },
+    {
+      "epoch": 0.9003834682573498,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4539198321656207e-06,
+      "loss": 0.6312,
+      "step": 10566
+    },
+    {
+      "epoch": 0.9004686834256498,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4537540458422788e-06,
+      "loss": 0.8699,
+      "step": 10567
+    },
+    {
+      "epoch": 0.9005538985939497,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.453588255836897e-06,
+      "loss": 0.7147,
+      "step": 10568
+    },
+    {
+      "epoch": 0.9006391137622497,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.4534224621524703e-06,
+      "loss": 0.6076,
+      "step": 10569
+    },
+    {
+      "epoch": 0.9007243289305497,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4532566647919956e-06,
+      "loss": 1.1227,
+      "step": 10570
+    },
+    {
+      "epoch": 0.9008095440988496,
+      "grad_norm": 20.125,
+      "learning_rate": 1.453090863758468e-06,
+      "loss": 0.8747,
+      "step": 10571
+    },
+    {
+      "epoch": 0.9008947592671496,
+      "grad_norm": 16.125,
+      "learning_rate": 1.4529250590548846e-06,
+      "loss": 0.6937,
+      "step": 10572
+    },
+    {
+      "epoch": 0.9009799744354495,
+      "grad_norm": 15.75,
+      "learning_rate": 1.4527592506842411e-06,
+      "loss": 1.0556,
+      "step": 10573
+    },
+    {
+      "epoch": 0.9010651896037495,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.4525934386495343e-06,
+      "loss": 0.666,
+      "step": 10574
+    },
+    {
+      "epoch": 0.9011504047720494,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4524276229537605e-06,
+      "loss": 0.8782,
+      "step": 10575
+    },
+    {
+      "epoch": 0.9012356199403494,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4522618035999156e-06,
+      "loss": 0.5147,
+      "step": 10576
+    },
+    {
+      "epoch": 0.9013208351086494,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.4520959805909966e-06,
+      "loss": 0.8113,
+      "step": 10577
+    },
+    {
+      "epoch": 0.9014060502769493,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4519301539300005e-06,
+      "loss": 0.3953,
+      "step": 10578
+    },
+    {
+      "epoch": 0.9014912654452493,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4517643236199228e-06,
+      "loss": 0.5396,
+      "step": 10579
+    },
+    {
+      "epoch": 0.9015764806135492,
+      "grad_norm": 13.75,
+      "learning_rate": 1.4515984896637612e-06,
+      "loss": 0.7874,
+      "step": 10580
+    },
+    {
+      "epoch": 0.9016616957818492,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4514326520645116e-06,
+      "loss": 0.6227,
+      "step": 10581
+    },
+    {
+      "epoch": 0.9017469109501491,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4512668108251722e-06,
+      "loss": 0.928,
+      "step": 10582
+    },
+    {
+      "epoch": 0.9018321261184491,
+      "grad_norm": 13.125,
+      "learning_rate": 1.4511009659487391e-06,
+      "loss": 0.4735,
+      "step": 10583
+    },
+    {
+      "epoch": 0.901917341286749,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.450935117438209e-06,
+      "loss": 0.3723,
+      "step": 10584
+    },
+    {
+      "epoch": 0.902002556455049,
+      "grad_norm": 13.25,
+      "learning_rate": 1.450769265296579e-06,
+      "loss": 0.6483,
+      "step": 10585
+    },
+    {
+      "epoch": 0.902087771623349,
+      "grad_norm": 15.875,
+      "learning_rate": 1.450603409526847e-06,
+      "loss": 0.6936,
+      "step": 10586
+    },
+    {
+      "epoch": 0.9021729867916489,
+      "grad_norm": 24.0,
+      "learning_rate": 1.4504375501320098e-06,
+      "loss": 0.9486,
+      "step": 10587
+    },
+    {
+      "epoch": 0.9022582019599489,
+      "grad_norm": 10.625,
+      "learning_rate": 1.4502716871150637e-06,
+      "loss": 0.3607,
+      "step": 10588
+    },
+    {
+      "epoch": 0.9023434171282488,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.4501058204790075e-06,
+      "loss": 0.8044,
+      "step": 10589
+    },
+    {
+      "epoch": 0.9024286322965488,
+      "grad_norm": 16.0,
+      "learning_rate": 1.449939950226838e-06,
+      "loss": 0.7381,
+      "step": 10590
+    },
+    {
+      "epoch": 0.9025138474648488,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.4497740763615523e-06,
+      "loss": 0.5229,
+      "step": 10591
+    },
+    {
+      "epoch": 0.9025990626331487,
+      "grad_norm": 11.0,
+      "learning_rate": 1.449608198886148e-06,
+      "loss": 0.4299,
+      "step": 10592
+    },
+    {
+      "epoch": 0.9026842778014487,
+      "grad_norm": 20.0,
+      "learning_rate": 1.4494423178036229e-06,
+      "loss": 1.2262,
+      "step": 10593
+    },
+    {
+      "epoch": 0.9027694929697486,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4492764331169748e-06,
+      "loss": 0.942,
+      "step": 10594
+    },
+    {
+      "epoch": 0.9028547081380486,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.4491105448292012e-06,
+      "loss": 0.611,
+      "step": 10595
+    },
+    {
+      "epoch": 0.9029399233063485,
+      "grad_norm": 12.75,
+      "learning_rate": 1.4489446529432993e-06,
+      "loss": 0.5748,
+      "step": 10596
+    },
+    {
+      "epoch": 0.9030251384746485,
+      "grad_norm": 12.875,
+      "learning_rate": 1.4487787574622677e-06,
+      "loss": 0.4048,
+      "step": 10597
+    },
+    {
+      "epoch": 0.9031103536429484,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4486128583891043e-06,
+      "loss": 0.9708,
+      "step": 10598
+    },
+    {
+      "epoch": 0.9031955688112484,
+      "grad_norm": 14.5,
+      "learning_rate": 1.448446955726807e-06,
+      "loss": 0.7211,
+      "step": 10599
+    },
+    {
+      "epoch": 0.9032807839795484,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.448281049478373e-06,
+      "loss": 0.6689,
+      "step": 10600
+    },
+    {
+      "epoch": 0.9033659991478483,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4481151396468015e-06,
+      "loss": 0.5666,
+      "step": 10601
+    },
+    {
+      "epoch": 0.9034512143161483,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.4479492262350902e-06,
+      "loss": 0.3463,
+      "step": 10602
+    },
+    {
+      "epoch": 0.9035364294844482,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.447783309246237e-06,
+      "loss": 0.2918,
+      "step": 10603
+    },
+    {
+      "epoch": 0.9036216446527482,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.447617388683241e-06,
+      "loss": 0.5017,
+      "step": 10604
+    },
+    {
+      "epoch": 0.9037068598210481,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4474514645490995e-06,
+      "loss": 0.6685,
+      "step": 10605
+    },
+    {
+      "epoch": 0.9037920749893481,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4472855368468122e-06,
+      "loss": 0.793,
+      "step": 10606
+    },
+    {
+      "epoch": 0.9038772901576481,
+      "grad_norm": 14.125,
+      "learning_rate": 1.4471196055793762e-06,
+      "loss": 0.5274,
+      "step": 10607
+    },
+    {
+      "epoch": 0.903962505325948,
+      "grad_norm": 18.625,
+      "learning_rate": 1.4469536707497912e-06,
+      "loss": 0.7363,
+      "step": 10608
+    },
+    {
+      "epoch": 0.904047720494248,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4467877323610546e-06,
+      "loss": 0.8751,
+      "step": 10609
+    },
+    {
+      "epoch": 0.9041329356625479,
+      "grad_norm": 12.625,
+      "learning_rate": 1.4466217904161667e-06,
+      "loss": 0.7177,
+      "step": 10610
+    },
+    {
+      "epoch": 0.9042181508308479,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4464558449181248e-06,
+      "loss": 0.6765,
+      "step": 10611
+    },
+    {
+      "epoch": 0.9043033659991478,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.4462898958699284e-06,
+      "loss": 0.7753,
+      "step": 10612
+    },
+    {
+      "epoch": 0.9043885811674478,
+      "grad_norm": 15.25,
+      "learning_rate": 1.4461239432745758e-06,
+      "loss": 0.6275,
+      "step": 10613
+    },
+    {
+      "epoch": 0.9044737963357478,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.445957987135067e-06,
+      "loss": 0.5641,
+      "step": 10614
+    },
+    {
+      "epoch": 0.9045590115040477,
+      "grad_norm": 15.375,
+      "learning_rate": 1.4457920274544e-06,
+      "loss": 0.9015,
+      "step": 10615
+    },
+    {
+      "epoch": 0.9046442266723477,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4456260642355741e-06,
+      "loss": 0.7694,
+      "step": 10616
+    },
+    {
+      "epoch": 0.9047294418406476,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.4454600974815884e-06,
+      "loss": 0.8692,
+      "step": 10617
+    },
+    {
+      "epoch": 0.9048146570089476,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4452941271954424e-06,
+      "loss": 0.7682,
+      "step": 10618
+    },
+    {
+      "epoch": 0.9048998721772475,
+      "grad_norm": 13.125,
+      "learning_rate": 1.4451281533801351e-06,
+      "loss": 0.8098,
+      "step": 10619
+    },
+    {
+      "epoch": 0.9049850873455475,
+      "grad_norm": 12.5,
+      "learning_rate": 1.4449621760386657e-06,
+      "loss": 0.6211,
+      "step": 10620
+    },
+    {
+      "epoch": 0.9050703025138475,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.444796195174034e-06,
+      "loss": 0.518,
+      "step": 10621
+    },
+    {
+      "epoch": 0.9051555176821474,
+      "grad_norm": 14.375,
+      "learning_rate": 1.444630210789239e-06,
+      "loss": 0.6669,
+      "step": 10622
+    },
+    {
+      "epoch": 0.9052407328504474,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.4444642228872807e-06,
+      "loss": 0.5608,
+      "step": 10623
+    },
+    {
+      "epoch": 0.9053259480187473,
+      "grad_norm": 18.875,
+      "learning_rate": 1.444298231471158e-06,
+      "loss": 0.6591,
+      "step": 10624
+    },
+    {
+      "epoch": 0.9054111631870473,
+      "grad_norm": 33.5,
+      "learning_rate": 1.4441322365438709e-06,
+      "loss": 0.9929,
+      "step": 10625
+    },
+    {
+      "epoch": 0.9054963783553472,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4439662381084191e-06,
+      "loss": 0.598,
+      "step": 10626
+    },
+    {
+      "epoch": 0.9055815935236472,
+      "grad_norm": 16.125,
+      "learning_rate": 1.4438002361678024e-06,
+      "loss": 0.9534,
+      "step": 10627
+    },
+    {
+      "epoch": 0.9056668086919472,
+      "grad_norm": 19.375,
+      "learning_rate": 1.4436342307250203e-06,
+      "loss": 1.1651,
+      "step": 10628
+    },
+    {
+      "epoch": 0.9057520238602471,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4434682217830731e-06,
+      "loss": 0.8231,
+      "step": 10629
+    },
+    {
+      "epoch": 0.9058372390285471,
+      "grad_norm": 19.0,
+      "learning_rate": 1.4433022093449608e-06,
+      "loss": 1.0404,
+      "step": 10630
+    },
+    {
+      "epoch": 0.905922454196847,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4431361934136832e-06,
+      "loss": 0.7809,
+      "step": 10631
+    },
+    {
+      "epoch": 0.906007669365147,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4429701739922397e-06,
+      "loss": 0.7415,
+      "step": 10632
+    },
+    {
+      "epoch": 0.9060928845334469,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.442804151083632e-06,
+      "loss": 0.4831,
+      "step": 10633
+    },
+    {
+      "epoch": 0.9061780997017469,
+      "grad_norm": 17.875,
+      "learning_rate": 1.442638124690859e-06,
+      "loss": 0.7217,
+      "step": 10634
+    },
+    {
+      "epoch": 0.9062633148700469,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.442472094816921e-06,
+      "loss": 1.0018,
+      "step": 10635
+    },
+    {
+      "epoch": 0.9063485300383468,
+      "grad_norm": 16.0,
+      "learning_rate": 1.442306061464819e-06,
+      "loss": 0.9099,
+      "step": 10636
+    },
+    {
+      "epoch": 0.9064337452066468,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.4421400246375533e-06,
+      "loss": 0.4625,
+      "step": 10637
+    },
+    {
+      "epoch": 0.9065189603749467,
+      "grad_norm": 15.25,
+      "learning_rate": 1.4419739843381239e-06,
+      "loss": 0.905,
+      "step": 10638
+    },
+    {
+      "epoch": 0.9066041755432467,
+      "grad_norm": 15.75,
+      "learning_rate": 1.4418079405695317e-06,
+      "loss": 0.9072,
+      "step": 10639
+    },
+    {
+      "epoch": 0.9066893907115466,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4416418933347772e-06,
+      "loss": 0.5375,
+      "step": 10640
+    },
+    {
+      "epoch": 0.9067746058798466,
+      "grad_norm": 13.5,
+      "learning_rate": 1.441475842636861e-06,
+      "loss": 0.5738,
+      "step": 10641
+    },
+    {
+      "epoch": 0.9068598210481466,
+      "grad_norm": 20.0,
+      "learning_rate": 1.4413097884787835e-06,
+      "loss": 0.976,
+      "step": 10642
+    },
+    {
+      "epoch": 0.9069450362164465,
+      "grad_norm": 24.75,
+      "learning_rate": 1.441143730863546e-06,
+      "loss": 0.9979,
+      "step": 10643
+    },
+    {
+      "epoch": 0.9070302513847465,
+      "grad_norm": 37.75,
+      "learning_rate": 1.4409776697941491e-06,
+      "loss": 0.8972,
+      "step": 10644
+    },
+    {
+      "epoch": 0.9071154665530464,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4408116052735932e-06,
+      "loss": 0.8906,
+      "step": 10645
+    },
+    {
+      "epoch": 0.9072006817213464,
+      "grad_norm": 11.125,
+      "learning_rate": 1.44064553730488e-06,
+      "loss": 0.4891,
+      "step": 10646
+    },
+    {
+      "epoch": 0.9072858968896463,
+      "grad_norm": 11.375,
+      "learning_rate": 1.4404794658910104e-06,
+      "loss": 0.4495,
+      "step": 10647
+    },
+    {
+      "epoch": 0.9073711120579463,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.4403133910349854e-06,
+      "loss": 0.6476,
+      "step": 10648
+    },
+    {
+      "epoch": 0.9074563272262463,
+      "grad_norm": 10.875,
+      "learning_rate": 1.4401473127398054e-06,
+      "loss": 0.3401,
+      "step": 10649
+    },
+    {
+      "epoch": 0.9075415423945462,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.4399812310084728e-06,
+      "loss": 1.1216,
+      "step": 10650
+    },
+    {
+      "epoch": 0.9076267575628462,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4398151458439884e-06,
+      "loss": 0.6491,
+      "step": 10651
+    },
+    {
+      "epoch": 0.9077119727311461,
+      "grad_norm": 20.125,
+      "learning_rate": 1.4396490572493535e-06,
+      "loss": 0.4412,
+      "step": 10652
+    },
+    {
+      "epoch": 0.9077971878994461,
+      "grad_norm": 11.375,
+      "learning_rate": 1.4394829652275688e-06,
+      "loss": 0.4415,
+      "step": 10653
+    },
+    {
+      "epoch": 0.907882403067746,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.4393168697816367e-06,
+      "loss": 0.6512,
+      "step": 10654
+    },
+    {
+      "epoch": 0.907967618236046,
+      "grad_norm": 14.25,
+      "learning_rate": 1.439150770914559e-06,
+      "loss": 0.9739,
+      "step": 10655
+    },
+    {
+      "epoch": 0.908052833404346,
+      "grad_norm": 13.125,
+      "learning_rate": 1.4389846686293357e-06,
+      "loss": 0.51,
+      "step": 10656
+    },
+    {
+      "epoch": 0.9081380485726459,
+      "grad_norm": 12.5,
+      "learning_rate": 1.4388185629289697e-06,
+      "loss": 0.7369,
+      "step": 10657
+    },
+    {
+      "epoch": 0.9082232637409459,
+      "grad_norm": 20.375,
+      "learning_rate": 1.4386524538164626e-06,
+      "loss": 0.856,
+      "step": 10658
+    },
+    {
+      "epoch": 0.9083084789092458,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4384863412948161e-06,
+      "loss": 0.669,
+      "step": 10659
+    },
+    {
+      "epoch": 0.9083936940775458,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.4383202253670315e-06,
+      "loss": 0.3547,
+      "step": 10660
+    },
+    {
+      "epoch": 0.9084789092458457,
+      "grad_norm": 12.625,
+      "learning_rate": 1.4381541060361111e-06,
+      "loss": 0.5269,
+      "step": 10661
+    },
+    {
+      "epoch": 0.9085641244141457,
+      "grad_norm": 13.0,
+      "learning_rate": 1.4379879833050568e-06,
+      "loss": 0.4384,
+      "step": 10662
+    },
+    {
+      "epoch": 0.9086493395824456,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4378218571768709e-06,
+      "loss": 0.6667,
+      "step": 10663
+    },
+    {
+      "epoch": 0.9087345547507456,
+      "grad_norm": 13.125,
+      "learning_rate": 1.437655727654555e-06,
+      "loss": 0.564,
+      "step": 10664
+    },
+    {
+      "epoch": 0.9088197699190456,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4374895947411114e-06,
+      "loss": 1.0638,
+      "step": 10665
+    },
+    {
+      "epoch": 0.9089049850873455,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.437323458439542e-06,
+      "loss": 0.6386,
+      "step": 10666
+    },
+    {
+      "epoch": 0.9089902002556455,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.43715731875285e-06,
+      "loss": 0.5608,
+      "step": 10667
+    },
+    {
+      "epoch": 0.9090754154239454,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.4369911756840365e-06,
+      "loss": 0.8517,
+      "step": 10668
+    },
+    {
+      "epoch": 0.9091606305922454,
+      "grad_norm": 16.125,
+      "learning_rate": 1.4368250292361047e-06,
+      "loss": 0.8857,
+      "step": 10669
+    },
+    {
+      "epoch": 0.9092458457605453,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.436658879412056e-06,
+      "loss": 0.6905,
+      "step": 10670
+    },
+    {
+      "epoch": 0.9093310609288453,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.4364927262148948e-06,
+      "loss": 0.6195,
+      "step": 10671
+    },
+    {
+      "epoch": 0.9094162760971453,
+      "grad_norm": 22.0,
+      "learning_rate": 1.436326569647622e-06,
+      "loss": 0.9549,
+      "step": 10672
+    },
+    {
+      "epoch": 0.9095014912654452,
+      "grad_norm": 18.5,
+      "learning_rate": 1.4361604097132402e-06,
+      "loss": 0.5507,
+      "step": 10673
+    },
+    {
+      "epoch": 0.9095867064337452,
+      "grad_norm": 45.0,
+      "learning_rate": 1.4359942464147533e-06,
+      "loss": 0.9017,
+      "step": 10674
+    },
+    {
+      "epoch": 0.9096719216020451,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4358280797551633e-06,
+      "loss": 0.7672,
+      "step": 10675
+    },
+    {
+      "epoch": 0.9097571367703451,
+      "grad_norm": 18.0,
+      "learning_rate": 1.435661909737473e-06,
+      "loss": 0.7525,
+      "step": 10676
+    },
+    {
+      "epoch": 0.909842351938645,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4354957363646848e-06,
+      "loss": 0.757,
+      "step": 10677
+    },
+    {
+      "epoch": 0.909927567106945,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4353295596398025e-06,
+      "loss": 0.9639,
+      "step": 10678
+    },
+    {
+      "epoch": 0.910012782275245,
+      "grad_norm": 12.5,
+      "learning_rate": 1.4351633795658287e-06,
+      "loss": 0.2647,
+      "step": 10679
+    },
+    {
+      "epoch": 0.9100979974435449,
+      "grad_norm": 31.875,
+      "learning_rate": 1.4349971961457666e-06,
+      "loss": 0.8639,
+      "step": 10680
+    },
+    {
+      "epoch": 0.9101832126118449,
+      "grad_norm": 12.875,
+      "learning_rate": 1.4348310093826187e-06,
+      "loss": 0.3345,
+      "step": 10681
+    },
+    {
+      "epoch": 0.9102684277801448,
+      "grad_norm": 18.75,
+      "learning_rate": 1.434664819279389e-06,
+      "loss": 1.0251,
+      "step": 10682
+    },
+    {
+      "epoch": 0.9103536429484448,
+      "grad_norm": 20.5,
+      "learning_rate": 1.4344986258390803e-06,
+      "loss": 0.8824,
+      "step": 10683
+    },
+    {
+      "epoch": 0.9104388581167447,
+      "grad_norm": 17.625,
+      "learning_rate": 1.4343324290646957e-06,
+      "loss": 1.1389,
+      "step": 10684
+    },
+    {
+      "epoch": 0.9105240732850447,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4341662289592386e-06,
+      "loss": 0.8872,
+      "step": 10685
+    },
+    {
+      "epoch": 0.9106092884533447,
+      "grad_norm": 12.0,
+      "learning_rate": 1.4340000255257128e-06,
+      "loss": 0.3582,
+      "step": 10686
+    },
+    {
+      "epoch": 0.9106945036216446,
+      "grad_norm": 20.25,
+      "learning_rate": 1.4338338187671215e-06,
+      "loss": 0.8789,
+      "step": 10687
+    },
+    {
+      "epoch": 0.9107797187899446,
+      "grad_norm": 29.5,
+      "learning_rate": 1.433667608686468e-06,
+      "loss": 1.428,
+      "step": 10688
+    },
+    {
+      "epoch": 0.9108649339582445,
+      "grad_norm": 24.75,
+      "learning_rate": 1.4335013952867566e-06,
+      "loss": 0.9267,
+      "step": 10689
+    },
+    {
+      "epoch": 0.9109501491265445,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.4333351785709898e-06,
+      "loss": 0.4721,
+      "step": 10690
+    },
+    {
+      "epoch": 0.9110353642948444,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.4331689585421726e-06,
+      "loss": 0.3674,
+      "step": 10691
+    },
+    {
+      "epoch": 0.9111205794631444,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.433002735203308e-06,
+      "loss": 0.2565,
+      "step": 10692
+    },
+    {
+      "epoch": 0.9112057946314444,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.4328365085573997e-06,
+      "loss": 0.5713,
+      "step": 10693
+    },
+    {
+      "epoch": 0.9112910097997443,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4326702786074518e-06,
+      "loss": 0.9657,
+      "step": 10694
+    },
+    {
+      "epoch": 0.9113762249680443,
+      "grad_norm": 25.5,
+      "learning_rate": 1.4325040453564687e-06,
+      "loss": 1.1145,
+      "step": 10695
+    },
+    {
+      "epoch": 0.9114614401363442,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4323378088074536e-06,
+      "loss": 0.7592,
+      "step": 10696
+    },
+    {
+      "epoch": 0.9115466553046442,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.432171568963411e-06,
+      "loss": 0.5014,
+      "step": 10697
+    },
+    {
+      "epoch": 0.9116318704729441,
+      "grad_norm": 20.875,
+      "learning_rate": 1.432005325827345e-06,
+      "loss": 1.0508,
+      "step": 10698
+    },
+    {
+      "epoch": 0.9117170856412441,
+      "grad_norm": 11.5,
+      "learning_rate": 1.4318390794022602e-06,
+      "loss": 0.4104,
+      "step": 10699
+    },
+    {
+      "epoch": 0.911802300809544,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.4316728296911598e-06,
+      "loss": 0.5074,
+      "step": 10700
+    },
+    {
+      "epoch": 0.911887515977844,
+      "grad_norm": 19.25,
+      "learning_rate": 1.4315065766970487e-06,
+      "loss": 1.1956,
+      "step": 10701
+    },
+    {
+      "epoch": 0.911972731146144,
+      "grad_norm": 12.875,
+      "learning_rate": 1.4313403204229313e-06,
+      "loss": 0.6013,
+      "step": 10702
+    },
+    {
+      "epoch": 0.9120579463144439,
+      "grad_norm": 24.5,
+      "learning_rate": 1.431174060871812e-06,
+      "loss": 0.9155,
+      "step": 10703
+    },
+    {
+      "epoch": 0.9121431614827439,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4310077980466953e-06,
+      "loss": 1.2339,
+      "step": 10704
+    },
+    {
+      "epoch": 0.9122283766510438,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4308415319505856e-06,
+      "loss": 0.8029,
+      "step": 10705
+    },
+    {
+      "epoch": 0.9123135918193438,
+      "grad_norm": 11.625,
+      "learning_rate": 1.4306752625864875e-06,
+      "loss": 0.4112,
+      "step": 10706
+    },
+    {
+      "epoch": 0.9123988069876438,
+      "grad_norm": 13.25,
+      "learning_rate": 1.430508989957406e-06,
+      "loss": 0.7702,
+      "step": 10707
+    },
+    {
+      "epoch": 0.9124840221559437,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4303427140663454e-06,
+      "loss": 1.094,
+      "step": 10708
+    },
+    {
+      "epoch": 0.9125692373242437,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4301764349163102e-06,
+      "loss": 0.706,
+      "step": 10709
+    },
+    {
+      "epoch": 0.9126544524925436,
+      "grad_norm": 15.75,
+      "learning_rate": 1.4300101525103063e-06,
+      "loss": 0.6238,
+      "step": 10710
+    },
+    {
+      "epoch": 0.9127396676608436,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4298438668513373e-06,
+      "loss": 0.6509,
+      "step": 10711
+    },
+    {
+      "epoch": 0.9128248828291435,
+      "grad_norm": 18.5,
+      "learning_rate": 1.4296775779424096e-06,
+      "loss": 0.8594,
+      "step": 10712
+    },
+    {
+      "epoch": 0.9129100979974435,
+      "grad_norm": 19.75,
+      "learning_rate": 1.4295112857865268e-06,
+      "loss": 1.0079,
+      "step": 10713
+    },
+    {
+      "epoch": 0.9129953131657434,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4293449903866946e-06,
+      "loss": 0.8087,
+      "step": 10714
+    },
+    {
+      "epoch": 0.9130805283340434,
+      "grad_norm": 23.25,
+      "learning_rate": 1.429178691745918e-06,
+      "loss": 1.0736,
+      "step": 10715
+    },
+    {
+      "epoch": 0.9131657435023434,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.4290123898672026e-06,
+      "loss": 0.7095,
+      "step": 10716
+    },
+    {
+      "epoch": 0.9132509586706433,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.428846084753553e-06,
+      "loss": 0.7069,
+      "step": 10717
+    },
+    {
+      "epoch": 0.9133361738389433,
+      "grad_norm": 17.0,
+      "learning_rate": 1.428679776407975e-06,
+      "loss": 0.8269,
+      "step": 10718
+    },
+    {
+      "epoch": 0.9134213890072432,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4285134648334733e-06,
+      "loss": 0.7666,
+      "step": 10719
+    },
+    {
+      "epoch": 0.9135066041755433,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4283471500330543e-06,
+      "loss": 0.6026,
+      "step": 10720
+    },
+    {
+      "epoch": 0.9135918193438433,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.428180832009723e-06,
+      "loss": 0.971,
+      "step": 10721
+    },
+    {
+      "epoch": 0.9136770345121432,
+      "grad_norm": 16.125,
+      "learning_rate": 1.4280145107664848e-06,
+      "loss": 0.7324,
+      "step": 10722
+    },
+    {
+      "epoch": 0.9137622496804432,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.4278481863063447e-06,
+      "loss": 0.8911,
+      "step": 10723
+    },
+    {
+      "epoch": 0.9138474648487431,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4276818586323098e-06,
+      "loss": 0.8858,
+      "step": 10724
+    },
+    {
+      "epoch": 0.9139326800170431,
+      "grad_norm": 11.875,
+      "learning_rate": 1.427515527747385e-06,
+      "loss": 0.4187,
+      "step": 10725
+    },
+    {
+      "epoch": 0.914017895185343,
+      "grad_norm": 77.5,
+      "learning_rate": 1.4273491936545757e-06,
+      "loss": 0.636,
+      "step": 10726
+    },
+    {
+      "epoch": 0.914103110353643,
+      "grad_norm": 11.625,
+      "learning_rate": 1.4271828563568884e-06,
+      "loss": 0.3706,
+      "step": 10727
+    },
+    {
+      "epoch": 0.914188325521943,
+      "grad_norm": 15.375,
+      "learning_rate": 1.427016515857329e-06,
+      "loss": 0.9849,
+      "step": 10728
+    },
+    {
+      "epoch": 0.9142735406902429,
+      "grad_norm": 11.125,
+      "learning_rate": 1.4268501721589029e-06,
+      "loss": 0.5107,
+      "step": 10729
+    },
+    {
+      "epoch": 0.9143587558585429,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.4266838252646162e-06,
+      "loss": 0.566,
+      "step": 10730
+    },
+    {
+      "epoch": 0.9144439710268428,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4265174751774752e-06,
+      "loss": 1.0135,
+      "step": 10731
+    },
+    {
+      "epoch": 0.9145291861951428,
+      "grad_norm": 13.125,
+      "learning_rate": 1.426351121900486e-06,
+      "loss": 0.5943,
+      "step": 10732
+    },
+    {
+      "epoch": 0.9146144013634427,
+      "grad_norm": 20.0,
+      "learning_rate": 1.4261847654366547e-06,
+      "loss": 1.2011,
+      "step": 10733
+    },
+    {
+      "epoch": 0.9146996165317427,
+      "grad_norm": 13.0,
+      "learning_rate": 1.4260184057889872e-06,
+      "loss": 0.452,
+      "step": 10734
+    },
+    {
+      "epoch": 0.9147848317000427,
+      "grad_norm": 11.375,
+      "learning_rate": 1.4258520429604905e-06,
+      "loss": 0.9703,
+      "step": 10735
+    },
+    {
+      "epoch": 0.9148700468683426,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.4256856769541704e-06,
+      "loss": 0.4704,
+      "step": 10736
+    },
+    {
+      "epoch": 0.9149552620366426,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.4255193077730336e-06,
+      "loss": 0.6168,
+      "step": 10737
+    },
+    {
+      "epoch": 0.9150404772049425,
+      "grad_norm": 14.375,
+      "learning_rate": 1.4253529354200862e-06,
+      "loss": 0.4867,
+      "step": 10738
+    },
+    {
+      "epoch": 0.9151256923732425,
+      "grad_norm": 21.5,
+      "learning_rate": 1.425186559898335e-06,
+      "loss": 0.7618,
+      "step": 10739
+    },
+    {
+      "epoch": 0.9152109075415424,
+      "grad_norm": 25.75,
+      "learning_rate": 1.4250201812107866e-06,
+      "loss": 0.843,
+      "step": 10740
+    },
+    {
+      "epoch": 0.9152961227098424,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4248537993604472e-06,
+      "loss": 0.5797,
+      "step": 10741
+    },
+    {
+      "epoch": 0.9153813378781424,
+      "grad_norm": 31.5,
+      "learning_rate": 1.4246874143503241e-06,
+      "loss": 0.8654,
+      "step": 10742
+    },
+    {
+      "epoch": 0.9154665530464423,
+      "grad_norm": 39.75,
+      "learning_rate": 1.4245210261834236e-06,
+      "loss": 1.0468,
+      "step": 10743
+    },
+    {
+      "epoch": 0.9155517682147423,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4243546348627529e-06,
+      "loss": 0.8127,
+      "step": 10744
+    },
+    {
+      "epoch": 0.9156369833830422,
+      "grad_norm": 13.875,
+      "learning_rate": 1.4241882403913184e-06,
+      "loss": 0.5697,
+      "step": 10745
+    },
+    {
+      "epoch": 0.9157221985513422,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.4240218427721276e-06,
+      "loss": 0.4631,
+      "step": 10746
+    },
+    {
+      "epoch": 0.9158074137196421,
+      "grad_norm": 10.375,
+      "learning_rate": 1.4238554420081865e-06,
+      "loss": 0.6388,
+      "step": 10747
+    },
+    {
+      "epoch": 0.9158926288879421,
+      "grad_norm": 15.875,
+      "learning_rate": 1.4236890381025033e-06,
+      "loss": 0.7782,
+      "step": 10748
+    },
+    {
+      "epoch": 0.915977844056242,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.4235226310580843e-06,
+      "loss": 0.2891,
+      "step": 10749
+    },
+    {
+      "epoch": 0.916063059224542,
+      "grad_norm": 13.875,
+      "learning_rate": 1.423356220877937e-06,
+      "loss": 0.6451,
+      "step": 10750
+    },
+    {
+      "epoch": 0.916148274392842,
+      "grad_norm": 12.875,
+      "learning_rate": 1.423189807565068e-06,
+      "loss": 0.6233,
+      "step": 10751
+    },
+    {
+      "epoch": 0.9162334895611419,
+      "grad_norm": 11.625,
+      "learning_rate": 1.4230233911224858e-06,
+      "loss": 0.5776,
+      "step": 10752
+    },
+    {
+      "epoch": 0.9163187047294419,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4228569715531962e-06,
+      "loss": 0.8446,
+      "step": 10753
+    },
+    {
+      "epoch": 0.9164039198977418,
+      "grad_norm": 19.125,
+      "learning_rate": 1.4226905488602077e-06,
+      "loss": 0.8129,
+      "step": 10754
+    },
+    {
+      "epoch": 0.9164891350660418,
+      "grad_norm": 12.25,
+      "learning_rate": 1.422524123046527e-06,
+      "loss": 0.4956,
+      "step": 10755
+    },
+    {
+      "epoch": 0.9165743502343418,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4223576941151624e-06,
+      "loss": 0.7465,
+      "step": 10756
+    },
+    {
+      "epoch": 0.9166595654026417,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4221912620691206e-06,
+      "loss": 0.6157,
+      "step": 10757
+    },
+    {
+      "epoch": 0.9167447805709417,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4220248269114093e-06,
+      "loss": 0.7655,
+      "step": 10758
+    },
+    {
+      "epoch": 0.9168299957392416,
+      "grad_norm": 18.375,
+      "learning_rate": 1.421858388645037e-06,
+      "loss": 0.9627,
+      "step": 10759
+    },
+    {
+      "epoch": 0.9169152109075416,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4216919472730104e-06,
+      "loss": 0.6259,
+      "step": 10760
+    },
+    {
+      "epoch": 0.9170004260758415,
+      "grad_norm": 17.125,
+      "learning_rate": 1.421525502798338e-06,
+      "loss": 0.8324,
+      "step": 10761
+    },
+    {
+      "epoch": 0.9170856412441415,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4213590552240267e-06,
+      "loss": 0.8552,
+      "step": 10762
+    },
+    {
+      "epoch": 0.9171708564124414,
+      "grad_norm": 12.0,
+      "learning_rate": 1.4211926045530855e-06,
+      "loss": 0.4953,
+      "step": 10763
+    },
+    {
+      "epoch": 0.9172560715807414,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4210261507885214e-06,
+      "loss": 0.6579,
+      "step": 10764
+    },
+    {
+      "epoch": 0.9173412867490414,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4208596939333433e-06,
+      "loss": 0.6735,
+      "step": 10765
+    },
+    {
+      "epoch": 0.9174265019173413,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.4206932339905582e-06,
+      "loss": 0.6644,
+      "step": 10766
+    },
+    {
+      "epoch": 0.9175117170856413,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4205267709631748e-06,
+      "loss": 0.7312,
+      "step": 10767
+    },
+    {
+      "epoch": 0.9175969322539412,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4203603048542014e-06,
+      "loss": 0.6361,
+      "step": 10768
+    },
+    {
+      "epoch": 0.9176821474222412,
+      "grad_norm": 14.375,
+      "learning_rate": 1.4201938356666462e-06,
+      "loss": 0.6674,
+      "step": 10769
+    },
+    {
+      "epoch": 0.9177673625905411,
+      "grad_norm": 15.5,
+      "learning_rate": 1.4200273634035166e-06,
+      "loss": 0.9394,
+      "step": 10770
+    },
+    {
+      "epoch": 0.9178525777588411,
+      "grad_norm": 10.625,
+      "learning_rate": 1.4198608880678219e-06,
+      "loss": 0.934,
+      "step": 10771
+    },
+    {
+      "epoch": 0.9179377929271411,
+      "grad_norm": 11.75,
+      "learning_rate": 1.4196944096625698e-06,
+      "loss": 0.4415,
+      "step": 10772
+    },
+    {
+      "epoch": 0.918023008095441,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.4195279281907696e-06,
+      "loss": 0.7838,
+      "step": 10773
+    },
+    {
+      "epoch": 0.918108223263741,
+      "grad_norm": 27.625,
+      "learning_rate": 1.4193614436554293e-06,
+      "loss": 0.7849,
+      "step": 10774
+    },
+    {
+      "epoch": 0.9181934384320409,
+      "grad_norm": 10.625,
+      "learning_rate": 1.419194956059557e-06,
+      "loss": 0.4729,
+      "step": 10775
+    },
+    {
+      "epoch": 0.9182786536003409,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4190284654061617e-06,
+      "loss": 0.7227,
+      "step": 10776
+    },
+    {
+      "epoch": 0.9183638687686408,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4188619716982523e-06,
+      "loss": 0.8909,
+      "step": 10777
+    },
+    {
+      "epoch": 0.9184490839369408,
+      "grad_norm": 14.375,
+      "learning_rate": 1.4186954749388373e-06,
+      "loss": 0.6256,
+      "step": 10778
+    },
+    {
+      "epoch": 0.9185342991052408,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4185289751309253e-06,
+      "loss": 0.7952,
+      "step": 10779
+    },
+    {
+      "epoch": 0.9186195142735407,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.4183624722775251e-06,
+      "loss": 0.2722,
+      "step": 10780
+    },
+    {
+      "epoch": 0.9187047294418407,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4181959663816462e-06,
+      "loss": 0.4637,
+      "step": 10781
+    },
+    {
+      "epoch": 0.9187899446101406,
+      "grad_norm": 14.75,
+      "learning_rate": 1.4180294574462966e-06,
+      "loss": 0.9549,
+      "step": 10782
+    },
+    {
+      "epoch": 0.9188751597784406,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4178629454744858e-06,
+      "loss": 1.1805,
+      "step": 10783
+    },
+    {
+      "epoch": 0.9189603749467405,
+      "grad_norm": 22.375,
+      "learning_rate": 1.4176964304692232e-06,
+      "loss": 0.6905,
+      "step": 10784
+    },
+    {
+      "epoch": 0.9190455901150405,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.417529912433517e-06,
+      "loss": 0.6286,
+      "step": 10785
+    },
+    {
+      "epoch": 0.9191308052833405,
+      "grad_norm": 16.0,
+      "learning_rate": 1.417363391370377e-06,
+      "loss": 1.0318,
+      "step": 10786
+    },
+    {
+      "epoch": 0.9192160204516404,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4171968672828122e-06,
+      "loss": 0.7355,
+      "step": 10787
+    },
+    {
+      "epoch": 0.9193012356199404,
+      "grad_norm": 20.375,
+      "learning_rate": 1.4170303401738316e-06,
+      "loss": 0.7971,
+      "step": 10788
+    },
+    {
+      "epoch": 0.9193864507882403,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.4168638100464455e-06,
+      "loss": 0.2897,
+      "step": 10789
+    },
+    {
+      "epoch": 0.9194716659565403,
+      "grad_norm": 25.0,
+      "learning_rate": 1.416697276903662e-06,
+      "loss": 1.0276,
+      "step": 10790
+    },
+    {
+      "epoch": 0.9195568811248402,
+      "grad_norm": 13.125,
+      "learning_rate": 1.416530740748491e-06,
+      "loss": 0.7483,
+      "step": 10791
+    },
+    {
+      "epoch": 0.9196420962931402,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4163642015839423e-06,
+      "loss": 0.4395,
+      "step": 10792
+    },
+    {
+      "epoch": 0.9197273114614402,
+      "grad_norm": 19.5,
+      "learning_rate": 1.4161976594130253e-06,
+      "loss": 0.7301,
+      "step": 10793
+    },
+    {
+      "epoch": 0.9198125266297401,
+      "grad_norm": 20.5,
+      "learning_rate": 1.416031114238749e-06,
+      "loss": 1.1139,
+      "step": 10794
+    },
+    {
+      "epoch": 0.9198977417980401,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.415864566064124e-06,
+      "loss": 0.4253,
+      "step": 10795
+    },
+    {
+      "epoch": 0.91998295696634,
+      "grad_norm": 12.0,
+      "learning_rate": 1.4156980148921589e-06,
+      "loss": 0.511,
+      "step": 10796
+    },
+    {
+      "epoch": 0.92006817213464,
+      "grad_norm": 13.625,
+      "learning_rate": 1.415531460725865e-06,
+      "loss": 0.7488,
+      "step": 10797
+    },
+    {
+      "epoch": 0.9201533873029399,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4153649035682504e-06,
+      "loss": 0.8851,
+      "step": 10798
+    },
+    {
+      "epoch": 0.9202386024712399,
+      "grad_norm": 18.125,
+      "learning_rate": 1.415198343422326e-06,
+      "loss": 0.9737,
+      "step": 10799
+    },
+    {
+      "epoch": 0.9203238176395399,
+      "grad_norm": 16.375,
+      "learning_rate": 1.415031780291101e-06,
+      "loss": 0.764,
+      "step": 10800
+    },
+    {
+      "epoch": 0.9204090328078398,
+      "grad_norm": 10.125,
+      "learning_rate": 1.4148652141775864e-06,
+      "loss": 0.4451,
+      "step": 10801
+    },
+    {
+      "epoch": 0.9204942479761398,
+      "grad_norm": 19.375,
+      "learning_rate": 1.4146986450847912e-06,
+      "loss": 0.9216,
+      "step": 10802
+    },
+    {
+      "epoch": 0.9205794631444397,
+      "grad_norm": 12.875,
+      "learning_rate": 1.4145320730157262e-06,
+      "loss": 0.6393,
+      "step": 10803
+    },
+    {
+      "epoch": 0.9206646783127397,
+      "grad_norm": 13.5,
+      "learning_rate": 1.4143654979734006e-06,
+      "loss": 0.8285,
+      "step": 10804
+    },
+    {
+      "epoch": 0.9207498934810396,
+      "grad_norm": 15.875,
+      "learning_rate": 1.4141989199608256e-06,
+      "loss": 0.7608,
+      "step": 10805
+    },
+    {
+      "epoch": 0.9208351086493396,
+      "grad_norm": 20.75,
+      "learning_rate": 1.414032338981011e-06,
+      "loss": 0.9688,
+      "step": 10806
+    },
+    {
+      "epoch": 0.9209203238176396,
+      "grad_norm": 13.75,
+      "learning_rate": 1.4138657550369672e-06,
+      "loss": 0.357,
+      "step": 10807
+    },
+    {
+      "epoch": 0.9210055389859395,
+      "grad_norm": 25.375,
+      "learning_rate": 1.413699168131704e-06,
+      "loss": 1.0686,
+      "step": 10808
+    },
+    {
+      "epoch": 0.9210907541542395,
+      "grad_norm": 20.75,
+      "learning_rate": 1.4135325782682324e-06,
+      "loss": 0.8688,
+      "step": 10809
+    },
+    {
+      "epoch": 0.9211759693225394,
+      "grad_norm": 14.125,
+      "learning_rate": 1.413365985449563e-06,
+      "loss": 0.6263,
+      "step": 10810
+    },
+    {
+      "epoch": 0.9212611844908394,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4131993896787056e-06,
+      "loss": 0.5667,
+      "step": 10811
+    },
+    {
+      "epoch": 0.9213463996591393,
+      "grad_norm": 11.375,
+      "learning_rate": 1.413032790958671e-06,
+      "loss": 0.8596,
+      "step": 10812
+    },
+    {
+      "epoch": 0.9214316148274393,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4128661892924702e-06,
+      "loss": 1.1177,
+      "step": 10813
+    },
+    {
+      "epoch": 0.9215168299957393,
+      "grad_norm": 12.5,
+      "learning_rate": 1.4126995846831137e-06,
+      "loss": 0.4265,
+      "step": 10814
+    },
+    {
+      "epoch": 0.9216020451640392,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.4125329771336115e-06,
+      "loss": 0.8378,
+      "step": 10815
+    },
+    {
+      "epoch": 0.9216872603323392,
+      "grad_norm": 20.375,
+      "learning_rate": 1.4123663666469756e-06,
+      "loss": 0.9733,
+      "step": 10816
+    },
+    {
+      "epoch": 0.9217724755006391,
+      "grad_norm": 12.0,
+      "learning_rate": 1.412199753226216e-06,
+      "loss": 0.6782,
+      "step": 10817
+    },
+    {
+      "epoch": 0.9218576906689391,
+      "grad_norm": 13.5,
+      "learning_rate": 1.412033136874344e-06,
+      "loss": 0.9881,
+      "step": 10818
+    },
+    {
+      "epoch": 0.921942905837239,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.4118665175943696e-06,
+      "loss": 0.7595,
+      "step": 10819
+    },
+    {
+      "epoch": 0.922028121005539,
+      "grad_norm": 11.75,
+      "learning_rate": 1.4116998953893051e-06,
+      "loss": 0.705,
+      "step": 10820
+    },
+    {
+      "epoch": 0.922113336173839,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4115332702621607e-06,
+      "loss": 0.9031,
+      "step": 10821
+    },
+    {
+      "epoch": 0.9221985513421389,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4113666422159477e-06,
+      "loss": 0.5239,
+      "step": 10822
+    },
+    {
+      "epoch": 0.9222837665104389,
+      "grad_norm": 21.0,
+      "learning_rate": 1.411200011253677e-06,
+      "loss": 0.977,
+      "step": 10823
+    },
+    {
+      "epoch": 0.9223689816787388,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.4110333773783603e-06,
+      "loss": 0.9753,
+      "step": 10824
+    },
+    {
+      "epoch": 0.9224541968470388,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.4108667405930082e-06,
+      "loss": 0.3109,
+      "step": 10825
+    },
+    {
+      "epoch": 0.9225394120153387,
+      "grad_norm": 13.125,
+      "learning_rate": 1.4107001009006324e-06,
+      "loss": 0.544,
+      "step": 10826
+    },
+    {
+      "epoch": 0.9226246271836387,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4105334583042444e-06,
+      "loss": 0.8261,
+      "step": 10827
+    },
+    {
+      "epoch": 0.9227098423519386,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.4103668128068548e-06,
+      "loss": 0.352,
+      "step": 10828
+    },
+    {
+      "epoch": 0.9227950575202386,
+      "grad_norm": 13.0,
+      "learning_rate": 1.4102001644114766e-06,
+      "loss": 0.3697,
+      "step": 10829
+    },
+    {
+      "epoch": 0.9228802726885386,
+      "grad_norm": 18.375,
+      "learning_rate": 1.4100335131211195e-06,
+      "loss": 0.9283,
+      "step": 10830
+    },
+    {
+      "epoch": 0.9229654878568385,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4098668589387966e-06,
+      "loss": 0.8075,
+      "step": 10831
+    },
+    {
+      "epoch": 0.9230507030251385,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.409700201867518e-06,
+      "loss": 0.5521,
+      "step": 10832
+    },
+    {
+      "epoch": 0.9231359181934384,
+      "grad_norm": 13.625,
+      "learning_rate": 1.409533541910297e-06,
+      "loss": 0.2999,
+      "step": 10833
+    },
+    {
+      "epoch": 0.9232211333617384,
+      "grad_norm": 11.25,
+      "learning_rate": 1.4093668790701437e-06,
+      "loss": 0.5054,
+      "step": 10834
+    },
+    {
+      "epoch": 0.9233063485300383,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4092002133500713e-06,
+      "loss": 0.7719,
+      "step": 10835
+    },
+    {
+      "epoch": 0.9233915636983383,
+      "grad_norm": 12.5,
+      "learning_rate": 1.4090335447530903e-06,
+      "loss": 0.4164,
+      "step": 10836
+    },
+    {
+      "epoch": 0.9234767788666383,
+      "grad_norm": 20.25,
+      "learning_rate": 1.408866873282214e-06,
+      "loss": 0.7232,
+      "step": 10837
+    },
+    {
+      "epoch": 0.9235619940349382,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.4087001989404529e-06,
+      "loss": 0.6379,
+      "step": 10838
+    },
+    {
+      "epoch": 0.9236472092032382,
+      "grad_norm": 13.625,
+      "learning_rate": 1.40853352173082e-06,
+      "loss": 0.664,
+      "step": 10839
+    },
+    {
+      "epoch": 0.9237324243715381,
+      "grad_norm": 21.5,
+      "learning_rate": 1.4083668416563265e-06,
+      "loss": 1.0119,
+      "step": 10840
+    },
+    {
+      "epoch": 0.9238176395398381,
+      "grad_norm": 28.75,
+      "learning_rate": 1.4082001587199853e-06,
+      "loss": 0.9788,
+      "step": 10841
+    },
+    {
+      "epoch": 0.923902854708138,
+      "grad_norm": 16.125,
+      "learning_rate": 1.408033472924808e-06,
+      "loss": 0.7658,
+      "step": 10842
+    },
+    {
+      "epoch": 0.923988069876438,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.407866784273807e-06,
+      "loss": 0.6037,
+      "step": 10843
+    },
+    {
+      "epoch": 0.924073285044738,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4077000927699945e-06,
+      "loss": 0.4422,
+      "step": 10844
+    },
+    {
+      "epoch": 0.9241585002130379,
+      "grad_norm": 13.625,
+      "learning_rate": 1.4075333984163824e-06,
+      "loss": 0.6748,
+      "step": 10845
+    },
+    {
+      "epoch": 0.9242437153813379,
+      "grad_norm": 15.5,
+      "learning_rate": 1.407366701215984e-06,
+      "loss": 0.3819,
+      "step": 10846
+    },
+    {
+      "epoch": 0.9243289305496378,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4072000011718106e-06,
+      "loss": 0.6826,
+      "step": 10847
+    },
+    {
+      "epoch": 0.9244141457179378,
+      "grad_norm": 14.625,
+      "learning_rate": 1.4070332982868754e-06,
+      "loss": 0.7509,
+      "step": 10848
+    },
+    {
+      "epoch": 0.9244993608862377,
+      "grad_norm": 19.5,
+      "learning_rate": 1.40686659256419e-06,
+      "loss": 0.2675,
+      "step": 10849
+    },
+    {
+      "epoch": 0.9245845760545377,
+      "grad_norm": 10.25,
+      "learning_rate": 1.4066998840067682e-06,
+      "loss": 0.3213,
+      "step": 10850
+    },
+    {
+      "epoch": 0.9246697912228377,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4065331726176218e-06,
+      "loss": 0.6964,
+      "step": 10851
+    },
+    {
+      "epoch": 0.9247550063911376,
+      "grad_norm": 12.25,
+      "learning_rate": 1.4063664583997632e-06,
+      "loss": 0.4816,
+      "step": 10852
+    },
+    {
+      "epoch": 0.9248402215594376,
+      "grad_norm": 18.75,
+      "learning_rate": 1.4061997413562057e-06,
+      "loss": 0.7428,
+      "step": 10853
+    },
+    {
+      "epoch": 0.9249254367277375,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.406033021489962e-06,
+      "loss": 0.9001,
+      "step": 10854
+    },
+    {
+      "epoch": 0.9250106518960375,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4058662988040444e-06,
+      "loss": 0.7291,
+      "step": 10855
+    },
+    {
+      "epoch": 0.9250958670643374,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4056995733014664e-06,
+      "loss": 0.8586,
+      "step": 10856
+    },
+    {
+      "epoch": 0.9251810822326374,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.4055328449852398e-06,
+      "loss": 0.7642,
+      "step": 10857
+    },
+    {
+      "epoch": 0.9252662974009374,
+      "grad_norm": 13.0,
+      "learning_rate": 1.405366113858379e-06,
+      "loss": 0.592,
+      "step": 10858
+    },
+    {
+      "epoch": 0.9253515125692373,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.4051993799238962e-06,
+      "loss": 0.6228,
+      "step": 10859
+    },
+    {
+      "epoch": 0.9254367277375373,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.405032643184804e-06,
+      "loss": 0.4445,
+      "step": 10860
+    },
+    {
+      "epoch": 0.9255219429058372,
+      "grad_norm": 20.25,
+      "learning_rate": 1.4048659036441166e-06,
+      "loss": 0.9376,
+      "step": 10861
+    },
+    {
+      "epoch": 0.9256071580741372,
+      "grad_norm": 17.5,
+      "learning_rate": 1.4046991613048466e-06,
+      "loss": 0.8778,
+      "step": 10862
+    },
+    {
+      "epoch": 0.9256923732424371,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.4045324161700073e-06,
+      "loss": 0.3592,
+      "step": 10863
+    },
+    {
+      "epoch": 0.9257775884107371,
+      "grad_norm": 13.875,
+      "learning_rate": 1.4043656682426112e-06,
+      "loss": 0.4612,
+      "step": 10864
+    },
+    {
+      "epoch": 0.925862803579037,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4041989175256727e-06,
+      "loss": 0.5827,
+      "step": 10865
+    },
+    {
+      "epoch": 0.925948018747337,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.4040321640222046e-06,
+      "loss": 0.2933,
+      "step": 10866
+    },
+    {
+      "epoch": 0.926033233915637,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4038654077352206e-06,
+      "loss": 0.8013,
+      "step": 10867
+    },
+    {
+      "epoch": 0.9261184490839369,
+      "grad_norm": 17.625,
+      "learning_rate": 1.4036986486677336e-06,
+      "loss": 0.8841,
+      "step": 10868
+    },
+    {
+      "epoch": 0.9262036642522369,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4035318868227577e-06,
+      "loss": 0.6938,
+      "step": 10869
+    },
+    {
+      "epoch": 0.9262888794205368,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4033651222033062e-06,
+      "loss": 1.1186,
+      "step": 10870
+    },
+    {
+      "epoch": 0.9263740945888368,
+      "grad_norm": 12.25,
+      "learning_rate": 1.4031983548123928e-06,
+      "loss": 0.717,
+      "step": 10871
+    },
+    {
+      "epoch": 0.9264593097571368,
+      "grad_norm": 26.5,
+      "learning_rate": 1.403031584653031e-06,
+      "loss": 0.4043,
+      "step": 10872
+    },
+    {
+      "epoch": 0.9265445249254367,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.4028648117282345e-06,
+      "loss": 0.3551,
+      "step": 10873
+    },
+    {
+      "epoch": 0.9266297400937367,
+      "grad_norm": 21.0,
+      "learning_rate": 1.4026980360410172e-06,
+      "loss": 0.9203,
+      "step": 10874
+    },
+    {
+      "epoch": 0.9267149552620366,
+      "grad_norm": 14.0,
+      "learning_rate": 1.4025312575943928e-06,
+      "loss": 0.8876,
+      "step": 10875
+    },
+    {
+      "epoch": 0.9268001704303366,
+      "grad_norm": 19.0,
+      "learning_rate": 1.4023644763913752e-06,
+      "loss": 0.9873,
+      "step": 10876
+    },
+    {
+      "epoch": 0.9268853855986365,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.4021976924349783e-06,
+      "loss": 0.553,
+      "step": 10877
+    },
+    {
+      "epoch": 0.9269706007669365,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.402030905728216e-06,
+      "loss": 0.5433,
+      "step": 10878
+    },
+    {
+      "epoch": 0.9270558159352364,
+      "grad_norm": 14.375,
+      "learning_rate": 1.4018641162741023e-06,
+      "loss": 0.9289,
+      "step": 10879
+    },
+    {
+      "epoch": 0.9271410311035364,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.4016973240756516e-06,
+      "loss": 0.6297,
+      "step": 10880
+    },
+    {
+      "epoch": 0.9272262462718364,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4015305291358774e-06,
+      "loss": 0.6875,
+      "step": 10881
+    },
+    {
+      "epoch": 0.9273114614401363,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4013637314577947e-06,
+      "loss": 0.9845,
+      "step": 10882
+    },
+    {
+      "epoch": 0.9273966766084363,
+      "grad_norm": 28.375,
+      "learning_rate": 1.4011969310444165e-06,
+      "loss": 1.8797,
+      "step": 10883
+    },
+    {
+      "epoch": 0.9274818917767362,
+      "grad_norm": 13.75,
+      "learning_rate": 1.4010301278987582e-06,
+      "loss": 0.6481,
+      "step": 10884
+    },
+    {
+      "epoch": 0.9275671069450362,
+      "grad_norm": 22.875,
+      "learning_rate": 1.400863322023833e-06,
+      "loss": 1.3978,
+      "step": 10885
+    },
+    {
+      "epoch": 0.9276523221133361,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4006965134226567e-06,
+      "loss": 1.0267,
+      "step": 10886
+    },
+    {
+      "epoch": 0.9277375372816361,
+      "grad_norm": 11.125,
+      "learning_rate": 1.4005297020982422e-06,
+      "loss": 0.6056,
+      "step": 10887
+    },
+    {
+      "epoch": 0.9278227524499361,
+      "grad_norm": 13.5,
+      "learning_rate": 1.400362888053605e-06,
+      "loss": 0.8652,
+      "step": 10888
+    },
+    {
+      "epoch": 0.927907967618236,
+      "grad_norm": 19.625,
+      "learning_rate": 1.4001960712917586e-06,
+      "loss": 0.9383,
+      "step": 10889
+    },
+    {
+      "epoch": 0.927993182786536,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.4000292518157188e-06,
+      "loss": 0.5327,
+      "step": 10890
+    },
+    {
+      "epoch": 0.9280783979548359,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3998624296284996e-06,
+      "loss": 0.7557,
+      "step": 10891
+    },
+    {
+      "epoch": 0.9281636131231359,
+      "grad_norm": 15.25,
+      "learning_rate": 1.3996956047331153e-06,
+      "loss": 0.7795,
+      "step": 10892
+    },
+    {
+      "epoch": 0.9282488282914358,
+      "grad_norm": 12.875,
+      "learning_rate": 1.3995287771325808e-06,
+      "loss": 0.5039,
+      "step": 10893
+    },
+    {
+      "epoch": 0.9283340434597358,
+      "grad_norm": 16.5,
+      "learning_rate": 1.3993619468299112e-06,
+      "loss": 0.6502,
+      "step": 10894
+    },
+    {
+      "epoch": 0.9284192586280358,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3991951138281209e-06,
+      "loss": 0.8852,
+      "step": 10895
+    },
+    {
+      "epoch": 0.9285044737963357,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3990282781302247e-06,
+      "loss": 0.5781,
+      "step": 10896
+    },
+    {
+      "epoch": 0.9285896889646357,
+      "grad_norm": 14.625,
+      "learning_rate": 1.398861439739238e-06,
+      "loss": 0.4403,
+      "step": 10897
+    },
+    {
+      "epoch": 0.9286749041329356,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.3986945986581752e-06,
+      "loss": 0.6279,
+      "step": 10898
+    },
+    {
+      "epoch": 0.9287601193012356,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.3985277548900518e-06,
+      "loss": 0.5392,
+      "step": 10899
+    },
+    {
+      "epoch": 0.9288453344695355,
+      "grad_norm": 11.5,
+      "learning_rate": 1.3983609084378818e-06,
+      "loss": 0.4736,
+      "step": 10900
+    },
+    {
+      "epoch": 0.9289305496378355,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3981940593046817e-06,
+      "loss": 0.7413,
+      "step": 10901
+    },
+    {
+      "epoch": 0.9290157648061355,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.3980272074934656e-06,
+      "loss": 0.7464,
+      "step": 10902
+    },
+    {
+      "epoch": 0.9291009799744354,
+      "grad_norm": 13.625,
+      "learning_rate": 1.397860353007249e-06,
+      "loss": 0.7442,
+      "step": 10903
+    },
+    {
+      "epoch": 0.9291861951427354,
+      "grad_norm": 18.75,
+      "learning_rate": 1.397693495849047e-06,
+      "loss": 0.9192,
+      "step": 10904
+    },
+    {
+      "epoch": 0.9292714103110353,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3975266360218753e-06,
+      "loss": 0.8816,
+      "step": 10905
+    },
+    {
+      "epoch": 0.9293566254793353,
+      "grad_norm": 10.875,
+      "learning_rate": 1.397359773528749e-06,
+      "loss": 0.5126,
+      "step": 10906
+    },
+    {
+      "epoch": 0.9294418406476352,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3971929083726831e-06,
+      "loss": 0.5135,
+      "step": 10907
+    },
+    {
+      "epoch": 0.9295270558159352,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3970260405566935e-06,
+      "loss": 0.6412,
+      "step": 10908
+    },
+    {
+      "epoch": 0.9296122709842352,
+      "grad_norm": 13.5,
+      "learning_rate": 1.3968591700837955e-06,
+      "loss": 0.432,
+      "step": 10909
+    },
+    {
+      "epoch": 0.9296974861525351,
+      "grad_norm": 21.75,
+      "learning_rate": 1.3966922969570044e-06,
+      "loss": 0.8647,
+      "step": 10910
+    },
+    {
+      "epoch": 0.9297827013208351,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3965254211793364e-06,
+      "loss": 0.5473,
+      "step": 10911
+    },
+    {
+      "epoch": 0.929867916489135,
+      "grad_norm": 13.875,
+      "learning_rate": 1.3963585427538064e-06,
+      "loss": 0.5646,
+      "step": 10912
+    },
+    {
+      "epoch": 0.929953131657435,
+      "grad_norm": 20.125,
+      "learning_rate": 1.3961916616834304e-06,
+      "loss": 1.0701,
+      "step": 10913
+    },
+    {
+      "epoch": 0.9300383468257349,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.3960247779712241e-06,
+      "loss": 0.6506,
+      "step": 10914
+    },
+    {
+      "epoch": 0.9301235619940349,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3958578916202031e-06,
+      "loss": 0.794,
+      "step": 10915
+    },
+    {
+      "epoch": 0.9302087771623349,
+      "grad_norm": 28.75,
+      "learning_rate": 1.3956910026333837e-06,
+      "loss": 1.3003,
+      "step": 10916
+    },
+    {
+      "epoch": 0.9302939923306348,
+      "grad_norm": 16.625,
+      "learning_rate": 1.395524111013781e-06,
+      "loss": 0.8453,
+      "step": 10917
+    },
+    {
+      "epoch": 0.9303792074989348,
+      "grad_norm": 15.875,
+      "learning_rate": 1.3953572167644115e-06,
+      "loss": 0.7645,
+      "step": 10918
+    },
+    {
+      "epoch": 0.9304644226672347,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3951903198882908e-06,
+      "loss": 0.9429,
+      "step": 10919
+    },
+    {
+      "epoch": 0.9305496378355347,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.3950234203884351e-06,
+      "loss": 0.7103,
+      "step": 10920
+    },
+    {
+      "epoch": 0.9306348530038346,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3948565182678602e-06,
+      "loss": 0.4285,
+      "step": 10921
+    },
+    {
+      "epoch": 0.9307200681721346,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3946896135295827e-06,
+      "loss": 0.4695,
+      "step": 10922
+    },
+    {
+      "epoch": 0.9308052833404346,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3945227061766182e-06,
+      "loss": 0.9451,
+      "step": 10923
+    },
+    {
+      "epoch": 0.9308904985087345,
+      "grad_norm": 13.625,
+      "learning_rate": 1.394355796211983e-06,
+      "loss": 0.6517,
+      "step": 10924
+    },
+    {
+      "epoch": 0.9309757136770345,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.3941888836386933e-06,
+      "loss": 0.4722,
+      "step": 10925
+    },
+    {
+      "epoch": 0.9310609288453344,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.3940219684597656e-06,
+      "loss": 0.632,
+      "step": 10926
+    },
+    {
+      "epoch": 0.9311461440136344,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3938550506782162e-06,
+      "loss": 0.9764,
+      "step": 10927
+    },
+    {
+      "epoch": 0.9312313591819343,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3936881302970608e-06,
+      "loss": 0.9851,
+      "step": 10928
+    },
+    {
+      "epoch": 0.9313165743502343,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3935212073193173e-06,
+      "loss": 1.0548,
+      "step": 10929
+    },
+    {
+      "epoch": 0.9314017895185343,
+      "grad_norm": 19.0,
+      "learning_rate": 1.393354281748001e-06,
+      "loss": 0.9416,
+      "step": 10930
+    },
+    {
+      "epoch": 0.9314870046868342,
+      "grad_norm": 19.75,
+      "learning_rate": 1.3931873535861284e-06,
+      "loss": 1.1657,
+      "step": 10931
+    },
+    {
+      "epoch": 0.9315722198551342,
+      "grad_norm": 15.25,
+      "learning_rate": 1.393020422836716e-06,
+      "loss": 0.824,
+      "step": 10932
+    },
+    {
+      "epoch": 0.9316574350234341,
+      "grad_norm": 16.25,
+      "learning_rate": 1.3928534895027816e-06,
+      "loss": 0.95,
+      "step": 10933
+    },
+    {
+      "epoch": 0.9317426501917341,
+      "grad_norm": 13.5,
+      "learning_rate": 1.3926865535873401e-06,
+      "loss": 0.7342,
+      "step": 10934
+    },
+    {
+      "epoch": 0.931827865360034,
+      "grad_norm": 14.625,
+      "learning_rate": 1.39251961509341e-06,
+      "loss": 0.7263,
+      "step": 10935
+    },
+    {
+      "epoch": 0.931913080528334,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3923526740240067e-06,
+      "loss": 0.6572,
+      "step": 10936
+    },
+    {
+      "epoch": 0.931998295696634,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3921857303821477e-06,
+      "loss": 0.4874,
+      "step": 10937
+    },
+    {
+      "epoch": 0.9320835108649339,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3920187841708488e-06,
+      "loss": 0.5382,
+      "step": 10938
+    },
+    {
+      "epoch": 0.9321687260332339,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3918518353931288e-06,
+      "loss": 0.8582,
+      "step": 10939
+    },
+    {
+      "epoch": 0.9322539412015338,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.3916848840520025e-06,
+      "loss": 0.4581,
+      "step": 10940
+    },
+    {
+      "epoch": 0.9323391563698338,
+      "grad_norm": 13.75,
+      "learning_rate": 1.3915179301504883e-06,
+      "loss": 0.4752,
+      "step": 10941
+    },
+    {
+      "epoch": 0.9324243715381337,
+      "grad_norm": 21.25,
+      "learning_rate": 1.3913509736916025e-06,
+      "loss": 1.1669,
+      "step": 10942
+    },
+    {
+      "epoch": 0.9325095867064337,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.3911840146783628e-06,
+      "loss": 0.593,
+      "step": 10943
+    },
+    {
+      "epoch": 0.9325948018747336,
+      "grad_norm": 19.625,
+      "learning_rate": 1.3910170531137862e-06,
+      "loss": 1.05,
+      "step": 10944
+    },
+    {
+      "epoch": 0.9326800170430336,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3908500890008894e-06,
+      "loss": 0.5251,
+      "step": 10945
+    },
+    {
+      "epoch": 0.9327652322113337,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3906831223426895e-06,
+      "loss": 0.8289,
+      "step": 10946
+    },
+    {
+      "epoch": 0.9328504473796336,
+      "grad_norm": 14.875,
+      "learning_rate": 1.3905161531422047e-06,
+      "loss": 0.9262,
+      "step": 10947
+    },
+    {
+      "epoch": 0.9329356625479336,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.3903491814024515e-06,
+      "loss": 0.4337,
+      "step": 10948
+    },
+    {
+      "epoch": 0.9330208777162335,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.390182207126447e-06,
+      "loss": 0.5801,
+      "step": 10949
+    },
+    {
+      "epoch": 0.9331060928845335,
+      "grad_norm": 24.375,
+      "learning_rate": 1.3900152303172096e-06,
+      "loss": 1.1195,
+      "step": 10950
+    },
+    {
+      "epoch": 0.9331913080528335,
+      "grad_norm": 26.75,
+      "learning_rate": 1.389848250977756e-06,
+      "loss": 0.784,
+      "step": 10951
+    },
+    {
+      "epoch": 0.9332765232211334,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.389681269111104e-06,
+      "loss": 0.9963,
+      "step": 10952
+    },
+    {
+      "epoch": 0.9333617383894334,
+      "grad_norm": 13.125,
+      "learning_rate": 1.3895142847202706e-06,
+      "loss": 0.5923,
+      "step": 10953
+    },
+    {
+      "epoch": 0.9334469535577333,
+      "grad_norm": 10.875,
+      "learning_rate": 1.3893472978082741e-06,
+      "loss": 0.3826,
+      "step": 10954
+    },
+    {
+      "epoch": 0.9335321687260333,
+      "grad_norm": 12.125,
+      "learning_rate": 1.389180308378132e-06,
+      "loss": 0.6673,
+      "step": 10955
+    },
+    {
+      "epoch": 0.9336173838943332,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3890133164328618e-06,
+      "loss": 1.3411,
+      "step": 10956
+    },
+    {
+      "epoch": 0.9337025990626332,
+      "grad_norm": 13.125,
+      "learning_rate": 1.3888463219754806e-06,
+      "loss": 0.6956,
+      "step": 10957
+    },
+    {
+      "epoch": 0.9337878142309332,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3886793250090071e-06,
+      "loss": 0.6626,
+      "step": 10958
+    },
+    {
+      "epoch": 0.9338730293992331,
+      "grad_norm": 15.875,
+      "learning_rate": 1.3885123255364588e-06,
+      "loss": 0.9476,
+      "step": 10959
+    },
+    {
+      "epoch": 0.9339582445675331,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3883453235608532e-06,
+      "loss": 0.5006,
+      "step": 10960
+    },
+    {
+      "epoch": 0.934043459735833,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.3881783190852089e-06,
+      "loss": 0.452,
+      "step": 10961
+    },
+    {
+      "epoch": 0.934128674904133,
+      "grad_norm": 12.875,
+      "learning_rate": 1.3880113121125432e-06,
+      "loss": 0.5628,
+      "step": 10962
+    },
+    {
+      "epoch": 0.9342138900724329,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.3878443026458745e-06,
+      "loss": 0.699,
+      "step": 10963
+    },
+    {
+      "epoch": 0.9342991052407329,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.3876772906882203e-06,
+      "loss": 0.5204,
+      "step": 10964
+    },
+    {
+      "epoch": 0.9343843204090329,
+      "grad_norm": 20.125,
+      "learning_rate": 1.3875102762425994e-06,
+      "loss": 1.1444,
+      "step": 10965
+    },
+    {
+      "epoch": 0.9344695355773328,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3873432593120289e-06,
+      "loss": 0.7788,
+      "step": 10966
+    },
+    {
+      "epoch": 0.9345547507456328,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3871762398995286e-06,
+      "loss": 0.8329,
+      "step": 10967
+    },
+    {
+      "epoch": 0.9346399659139327,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3870092180081148e-06,
+      "loss": 0.5616,
+      "step": 10968
+    },
+    {
+      "epoch": 0.9347251810822327,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3868421936408072e-06,
+      "loss": 0.8149,
+      "step": 10969
+    },
+    {
+      "epoch": 0.9348103962505326,
+      "grad_norm": 14.5,
+      "learning_rate": 1.386675166800623e-06,
+      "loss": 0.8236,
+      "step": 10970
+    },
+    {
+      "epoch": 0.9348956114188326,
+      "grad_norm": 15.875,
+      "learning_rate": 1.386508137490582e-06,
+      "loss": 0.7371,
+      "step": 10971
+    },
+    {
+      "epoch": 0.9349808265871326,
+      "grad_norm": 18.25,
+      "learning_rate": 1.3863411057137007e-06,
+      "loss": 1.2087,
+      "step": 10972
+    },
+    {
+      "epoch": 0.9350660417554325,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3861740714729991e-06,
+      "loss": 0.7364,
+      "step": 10973
+    },
+    {
+      "epoch": 0.9351512569237325,
+      "grad_norm": 28.625,
+      "learning_rate": 1.3860070347714948e-06,
+      "loss": 0.7654,
+      "step": 10974
+    },
+    {
+      "epoch": 0.9352364720920324,
+      "grad_norm": 23.375,
+      "learning_rate": 1.3858399956122067e-06,
+      "loss": 0.9621,
+      "step": 10975
+    },
+    {
+      "epoch": 0.9353216872603324,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3856729539981533e-06,
+      "loss": 0.8979,
+      "step": 10976
+    },
+    {
+      "epoch": 0.9354069024286323,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3855059099323534e-06,
+      "loss": 0.9275,
+      "step": 10977
+    },
+    {
+      "epoch": 0.9354921175969323,
+      "grad_norm": 21.125,
+      "learning_rate": 1.3853388634178252e-06,
+      "loss": 0.8401,
+      "step": 10978
+    },
+    {
+      "epoch": 0.9355773327652323,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3851718144575877e-06,
+      "loss": 0.5417,
+      "step": 10979
+    },
+    {
+      "epoch": 0.9356625479335322,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3850047630546598e-06,
+      "loss": 0.9337,
+      "step": 10980
+    },
+    {
+      "epoch": 0.9357477631018322,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3848377092120595e-06,
+      "loss": 0.7117,
+      "step": 10981
+    },
+    {
+      "epoch": 0.9358329782701321,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.3846706529328068e-06,
+      "loss": 0.4003,
+      "step": 10982
+    },
+    {
+      "epoch": 0.9359181934384321,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3845035942199196e-06,
+      "loss": 0.9912,
+      "step": 10983
+    },
+    {
+      "epoch": 0.936003408606732,
+      "grad_norm": 69.0,
+      "learning_rate": 1.3843365330764177e-06,
+      "loss": 1.1518,
+      "step": 10984
+    },
+    {
+      "epoch": 0.936088623775032,
+      "grad_norm": 17.25,
+      "learning_rate": 1.384169469505319e-06,
+      "loss": 0.9458,
+      "step": 10985
+    },
+    {
+      "epoch": 0.936173838943332,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3840024035096434e-06,
+      "loss": 0.8911,
+      "step": 10986
+    },
+    {
+      "epoch": 0.9362590541116319,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3838353350924093e-06,
+      "loss": 0.9467,
+      "step": 10987
+    },
+    {
+      "epoch": 0.9363442692799319,
+      "grad_norm": 13.75,
+      "learning_rate": 1.3836682642566369e-06,
+      "loss": 0.6473,
+      "step": 10988
+    },
+    {
+      "epoch": 0.9364294844482318,
+      "grad_norm": 15.0,
+      "learning_rate": 1.3835011910053437e-06,
+      "loss": 1.0352,
+      "step": 10989
+    },
+    {
+      "epoch": 0.9365146996165318,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.38333411534155e-06,
+      "loss": 0.5928,
+      "step": 10990
+    },
+    {
+      "epoch": 0.9365999147848317,
+      "grad_norm": 12.25,
+      "learning_rate": 1.3831670372682745e-06,
+      "loss": 0.6214,
+      "step": 10991
+    },
+    {
+      "epoch": 0.9366851299531317,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.3829999567885375e-06,
+      "loss": 0.4847,
+      "step": 10992
+    },
+    {
+      "epoch": 0.9367703451214316,
+      "grad_norm": 20.875,
+      "learning_rate": 1.3828328739053567e-06,
+      "loss": 1.021,
+      "step": 10993
+    },
+    {
+      "epoch": 0.9368555602897316,
+      "grad_norm": 22.125,
+      "learning_rate": 1.3826657886217526e-06,
+      "loss": 1.0854,
+      "step": 10994
+    },
+    {
+      "epoch": 0.9369407754580316,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3824987009407443e-06,
+      "loss": 0.4298,
+      "step": 10995
+    },
+    {
+      "epoch": 0.9370259906263315,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.382331610865351e-06,
+      "loss": 0.519,
+      "step": 10996
+    },
+    {
+      "epoch": 0.9371112057946315,
+      "grad_norm": 14.625,
+      "learning_rate": 1.3821645183985929e-06,
+      "loss": 0.493,
+      "step": 10997
+    },
+    {
+      "epoch": 0.9371964209629314,
+      "grad_norm": 12.625,
+      "learning_rate": 1.3819974235434889e-06,
+      "loss": 0.454,
+      "step": 10998
+    },
+    {
+      "epoch": 0.9372816361312314,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3818303263030586e-06,
+      "loss": 0.9092,
+      "step": 10999
+    },
+    {
+      "epoch": 0.9373668512995313,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3816632266803217e-06,
+      "loss": 1.0118,
+      "step": 11000
+    },
+    {
+      "epoch": 0.9374520664678313,
+      "grad_norm": 12.75,
+      "learning_rate": 1.3814961246782982e-06,
+      "loss": 0.5042,
+      "step": 11001
+    },
+    {
+      "epoch": 0.9375372816361313,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.381329020300007e-06,
+      "loss": 0.6875,
+      "step": 11002
+    },
+    {
+      "epoch": 0.9376224968044312,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3811619135484689e-06,
+      "loss": 0.7561,
+      "step": 11003
+    },
+    {
+      "epoch": 0.9377077119727312,
+      "grad_norm": 18.0,
+      "learning_rate": 1.3809948044267028e-06,
+      "loss": 0.8927,
+      "step": 11004
+    },
+    {
+      "epoch": 0.9377929271410311,
+      "grad_norm": 32.75,
+      "learning_rate": 1.380827692937729e-06,
+      "loss": 0.9235,
+      "step": 11005
+    },
+    {
+      "epoch": 0.9378781423093311,
+      "grad_norm": 14.75,
+      "learning_rate": 1.380660579084567e-06,
+      "loss": 0.7896,
+      "step": 11006
+    },
+    {
+      "epoch": 0.937963357477631,
+      "grad_norm": 25.0,
+      "learning_rate": 1.3804934628702372e-06,
+      "loss": 1.4595,
+      "step": 11007
+    },
+    {
+      "epoch": 0.938048572645931,
+      "grad_norm": 12.875,
+      "learning_rate": 1.3803263442977592e-06,
+      "loss": 0.5708,
+      "step": 11008
+    },
+    {
+      "epoch": 0.938133787814231,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3801592233701533e-06,
+      "loss": 0.9006,
+      "step": 11009
+    },
+    {
+      "epoch": 0.9382190029825309,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.3799921000904386e-06,
+      "loss": 0.4296,
+      "step": 11010
+    },
+    {
+      "epoch": 0.9383042181508309,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3798249744616368e-06,
+      "loss": 1.0577,
+      "step": 11011
+    },
+    {
+      "epoch": 0.9383894333191308,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3796578464867668e-06,
+      "loss": 0.8821,
+      "step": 11012
+    },
+    {
+      "epoch": 0.9384746484874308,
+      "grad_norm": 15.75,
+      "learning_rate": 1.379490716168849e-06,
+      "loss": 0.6517,
+      "step": 11013
+    },
+    {
+      "epoch": 0.9385598636557307,
+      "grad_norm": 15.75,
+      "learning_rate": 1.379323583510904e-06,
+      "loss": 0.9636,
+      "step": 11014
+    },
+    {
+      "epoch": 0.9386450788240307,
+      "grad_norm": 17.125,
+      "learning_rate": 1.379156448515952e-06,
+      "loss": 0.7771,
+      "step": 11015
+    },
+    {
+      "epoch": 0.9387302939923307,
+      "grad_norm": 19.0,
+      "learning_rate": 1.378989311187013e-06,
+      "loss": 1.2333,
+      "step": 11016
+    },
+    {
+      "epoch": 0.9388155091606306,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3788221715271068e-06,
+      "loss": 0.6809,
+      "step": 11017
+    },
+    {
+      "epoch": 0.9389007243289306,
+      "grad_norm": 18.625,
+      "learning_rate": 1.3786550295392551e-06,
+      "loss": 0.6074,
+      "step": 11018
+    },
+    {
+      "epoch": 0.9389859394972305,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.3784878852264771e-06,
+      "loss": 0.53,
+      "step": 11019
+    },
+    {
+      "epoch": 0.9390711546655305,
+      "grad_norm": 16.25,
+      "learning_rate": 1.3783207385917947e-06,
+      "loss": 0.7652,
+      "step": 11020
+    },
+    {
+      "epoch": 0.9391563698338304,
+      "grad_norm": 21.875,
+      "learning_rate": 1.3781535896382265e-06,
+      "loss": 0.9202,
+      "step": 11021
+    },
+    {
+      "epoch": 0.9392415850021304,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.3779864383687947e-06,
+      "loss": 0.4956,
+      "step": 11022
+    },
+    {
+      "epoch": 0.9393268001704304,
+      "grad_norm": 11.5,
+      "learning_rate": 1.3778192847865188e-06,
+      "loss": 0.4734,
+      "step": 11023
+    },
+    {
+      "epoch": 0.9394120153387303,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.3776521288944206e-06,
+      "loss": 0.7378,
+      "step": 11024
+    },
+    {
+      "epoch": 0.9394972305070303,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.3774849706955193e-06,
+      "loss": 0.3357,
+      "step": 11025
+    },
+    {
+      "epoch": 0.9395824456753302,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.377317810192837e-06,
+      "loss": 0.6553,
+      "step": 11026
+    },
+    {
+      "epoch": 0.9396676608436302,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.3771506473893933e-06,
+      "loss": 0.8324,
+      "step": 11027
+    },
+    {
+      "epoch": 0.9397528760119301,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3769834822882096e-06,
+      "loss": 1.2539,
+      "step": 11028
+    },
+    {
+      "epoch": 0.9398380911802301,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3768163148923073e-06,
+      "loss": 0.9988,
+      "step": 11029
+    },
+    {
+      "epoch": 0.93992330634853,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3766491452047062e-06,
+      "loss": 1.0352,
+      "step": 11030
+    },
+    {
+      "epoch": 0.94000852151683,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3764819732284276e-06,
+      "loss": 0.6993,
+      "step": 11031
+    },
+    {
+      "epoch": 0.94009373668513,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3763147989664927e-06,
+      "loss": 0.5443,
+      "step": 11032
+    },
+    {
+      "epoch": 0.9401789518534299,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3761476224219226e-06,
+      "loss": 0.6295,
+      "step": 11033
+    },
+    {
+      "epoch": 0.9402641670217299,
+      "grad_norm": 19.375,
+      "learning_rate": 1.3759804435977375e-06,
+      "loss": 1.2088,
+      "step": 11034
+    },
+    {
+      "epoch": 0.9403493821900298,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3758132624969594e-06,
+      "loss": 1.0608,
+      "step": 11035
+    },
+    {
+      "epoch": 0.9404345973583298,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.3756460791226095e-06,
+      "loss": 0.444,
+      "step": 11036
+    },
+    {
+      "epoch": 0.9405198125266298,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.3754788934777084e-06,
+      "loss": 0.8103,
+      "step": 11037
+    },
+    {
+      "epoch": 0.9406050276949297,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.3753117055652774e-06,
+      "loss": 0.4821,
+      "step": 11038
+    },
+    {
+      "epoch": 0.9406902428632297,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3751445153883377e-06,
+      "loss": 0.9522,
+      "step": 11039
+    },
+    {
+      "epoch": 0.9407754580315296,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.374977322949911e-06,
+      "loss": 0.8709,
+      "step": 11040
+    },
+    {
+      "epoch": 0.9408606731998296,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3748101282530185e-06,
+      "loss": 0.6893,
+      "step": 11041
+    },
+    {
+      "epoch": 0.9409458883681295,
+      "grad_norm": 15.125,
+      "learning_rate": 1.3746429313006812e-06,
+      "loss": 0.8334,
+      "step": 11042
+    },
+    {
+      "epoch": 0.9410311035364295,
+      "grad_norm": 14.75,
+      "learning_rate": 1.374475732095921e-06,
+      "loss": 0.5598,
+      "step": 11043
+    },
+    {
+      "epoch": 0.9411163187047294,
+      "grad_norm": 14.25,
+      "learning_rate": 1.374308530641759e-06,
+      "loss": 0.4919,
+      "step": 11044
+    },
+    {
+      "epoch": 0.9412015338730294,
+      "grad_norm": 24.75,
+      "learning_rate": 1.3741413269412172e-06,
+      "loss": 1.2341,
+      "step": 11045
+    },
+    {
+      "epoch": 0.9412867490413294,
+      "grad_norm": 16.375,
+      "learning_rate": 1.373974120997316e-06,
+      "loss": 0.765,
+      "step": 11046
+    },
+    {
+      "epoch": 0.9413719642096293,
+      "grad_norm": 18.0,
+      "learning_rate": 1.3738069128130787e-06,
+      "loss": 0.7991,
+      "step": 11047
+    },
+    {
+      "epoch": 0.9414571793779293,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.3736397023915254e-06,
+      "loss": 0.468,
+      "step": 11048
+    },
+    {
+      "epoch": 0.9415423945462292,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3734724897356785e-06,
+      "loss": 0.8174,
+      "step": 11049
+    },
+    {
+      "epoch": 0.9416276097145292,
+      "grad_norm": 13.625,
+      "learning_rate": 1.3733052748485598e-06,
+      "loss": 0.5951,
+      "step": 11050
+    },
+    {
+      "epoch": 0.9417128248828291,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3731380577331901e-06,
+      "loss": 0.9719,
+      "step": 11051
+    },
+    {
+      "epoch": 0.9417980400511291,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3729708383925925e-06,
+      "loss": 0.5042,
+      "step": 11052
+    },
+    {
+      "epoch": 0.9418832552194291,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3728036168297882e-06,
+      "loss": 0.8386,
+      "step": 11053
+    },
+    {
+      "epoch": 0.941968470387729,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.372636393047799e-06,
+      "loss": 0.6466,
+      "step": 11054
+    },
+    {
+      "epoch": 0.942053685556029,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.3724691670496466e-06,
+      "loss": 0.757,
+      "step": 11055
+    },
+    {
+      "epoch": 0.9421389007243289,
+      "grad_norm": 16.25,
+      "learning_rate": 1.3723019388383532e-06,
+      "loss": 0.8277,
+      "step": 11056
+    },
+    {
+      "epoch": 0.9422241158926289,
+      "grad_norm": 17.875,
+      "learning_rate": 1.3721347084169412e-06,
+      "loss": 0.9954,
+      "step": 11057
+    },
+    {
+      "epoch": 0.9423093310609288,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3719674757884318e-06,
+      "loss": 0.7858,
+      "step": 11058
+    },
+    {
+      "epoch": 0.9423945462292288,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3718002409558476e-06,
+      "loss": 0.9745,
+      "step": 11059
+    },
+    {
+      "epoch": 0.9424797613975288,
+      "grad_norm": 19.125,
+      "learning_rate": 1.3716330039222106e-06,
+      "loss": 1.0373,
+      "step": 11060
+    },
+    {
+      "epoch": 0.9425649765658287,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.3714657646905428e-06,
+      "loss": 0.7585,
+      "step": 11061
+    },
+    {
+      "epoch": 0.9426501917341287,
+      "grad_norm": 15.375,
+      "learning_rate": 1.3712985232638668e-06,
+      "loss": 0.8033,
+      "step": 11062
+    },
+    {
+      "epoch": 0.9427354069024286,
+      "grad_norm": 23.125,
+      "learning_rate": 1.371131279645204e-06,
+      "loss": 0.8723,
+      "step": 11063
+    },
+    {
+      "epoch": 0.9428206220707286,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3709640338375775e-06,
+      "loss": 0.6407,
+      "step": 11064
+    },
+    {
+      "epoch": 0.9429058372390285,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3707967858440092e-06,
+      "loss": 0.8283,
+      "step": 11065
+    },
+    {
+      "epoch": 0.9429910524073285,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3706295356675214e-06,
+      "loss": 0.8207,
+      "step": 11066
+    },
+    {
+      "epoch": 0.9430762675756285,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.3704622833111367e-06,
+      "loss": 0.8268,
+      "step": 11067
+    },
+    {
+      "epoch": 0.9431614827439284,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3702950287778772e-06,
+      "loss": 0.8447,
+      "step": 11068
+    },
+    {
+      "epoch": 0.9432466979122284,
+      "grad_norm": 12.875,
+      "learning_rate": 1.3701277720707662e-06,
+      "loss": 0.6439,
+      "step": 11069
+    },
+    {
+      "epoch": 0.9433319130805283,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3699605131928247e-06,
+      "loss": 0.5469,
+      "step": 11070
+    },
+    {
+      "epoch": 0.9434171282488283,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3697932521470766e-06,
+      "loss": 0.7244,
+      "step": 11071
+    },
+    {
+      "epoch": 0.9435023434171282,
+      "grad_norm": 22.0,
+      "learning_rate": 1.3696259889365438e-06,
+      "loss": 0.9343,
+      "step": 11072
+    },
+    {
+      "epoch": 0.9435875585854282,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.3694587235642496e-06,
+      "loss": 0.6349,
+      "step": 11073
+    },
+    {
+      "epoch": 0.9436727737537282,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.3692914560332154e-06,
+      "loss": 0.5881,
+      "step": 11074
+    },
+    {
+      "epoch": 0.9437579889220281,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3691241863464653e-06,
+      "loss": 0.8899,
+      "step": 11075
+    },
+    {
+      "epoch": 0.9438432040903281,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3689569145070205e-06,
+      "loss": 0.7084,
+      "step": 11076
+    },
+    {
+      "epoch": 0.943928419258628,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3687896405179058e-06,
+      "loss": 0.6732,
+      "step": 11077
+    },
+    {
+      "epoch": 0.944013634426928,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.368622364382142e-06,
+      "loss": 0.5679,
+      "step": 11078
+    },
+    {
+      "epoch": 0.9440988495952279,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.368455086102753e-06,
+      "loss": 0.6421,
+      "step": 11079
+    },
+    {
+      "epoch": 0.9441840647635279,
+      "grad_norm": 17.25,
+      "learning_rate": 1.368287805682761e-06,
+      "loss": 0.83,
+      "step": 11080
+    },
+    {
+      "epoch": 0.9442692799318279,
+      "grad_norm": 13.5,
+      "learning_rate": 1.36812052312519e-06,
+      "loss": 0.4536,
+      "step": 11081
+    },
+    {
+      "epoch": 0.9443544951001278,
+      "grad_norm": 14.25,
+      "learning_rate": 1.3679532384330624e-06,
+      "loss": 0.7111,
+      "step": 11082
+    },
+    {
+      "epoch": 0.9444397102684278,
+      "grad_norm": 14.375,
+      "learning_rate": 1.3677859516094008e-06,
+      "loss": 0.5433,
+      "step": 11083
+    },
+    {
+      "epoch": 0.9445249254367277,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3676186626572291e-06,
+      "loss": 0.7989,
+      "step": 11084
+    },
+    {
+      "epoch": 0.9446101406050277,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3674513715795695e-06,
+      "loss": 0.5892,
+      "step": 11085
+    },
+    {
+      "epoch": 0.9446953557733276,
+      "grad_norm": 21.0,
+      "learning_rate": 1.3672840783794458e-06,
+      "loss": 0.8751,
+      "step": 11086
+    },
+    {
+      "epoch": 0.9447805709416276,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3671167830598806e-06,
+      "loss": 0.9412,
+      "step": 11087
+    },
+    {
+      "epoch": 0.9448657861099276,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.3669494856238974e-06,
+      "loss": 0.5189,
+      "step": 11088
+    },
+    {
+      "epoch": 0.9449510012782275,
+      "grad_norm": 12.625,
+      "learning_rate": 1.3667821860745196e-06,
+      "loss": 0.4717,
+      "step": 11089
+    },
+    {
+      "epoch": 0.9450362164465275,
+      "grad_norm": 26.125,
+      "learning_rate": 1.36661488441477e-06,
+      "loss": 0.9389,
+      "step": 11090
+    },
+    {
+      "epoch": 0.9451214316148274,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.366447580647672e-06,
+      "loss": 0.5008,
+      "step": 11091
+    },
+    {
+      "epoch": 0.9452066467831274,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3662802747762495e-06,
+      "loss": 0.9154,
+      "step": 11092
+    },
+    {
+      "epoch": 0.9452918619514273,
+      "grad_norm": 21.625,
+      "learning_rate": 1.3661129668035255e-06,
+      "loss": 1.2268,
+      "step": 11093
+    },
+    {
+      "epoch": 0.9453770771197273,
+      "grad_norm": 28.375,
+      "learning_rate": 1.3659456567325236e-06,
+      "loss": 1.0453,
+      "step": 11094
+    },
+    {
+      "epoch": 0.9454622922880273,
+      "grad_norm": 9.75,
+      "learning_rate": 1.3657783445662664e-06,
+      "loss": 1.2264,
+      "step": 11095
+    },
+    {
+      "epoch": 0.9455475074563272,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.3656110303077788e-06,
+      "loss": 0.6336,
+      "step": 11096
+    },
+    {
+      "epoch": 0.9456327226246272,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3654437139600834e-06,
+      "loss": 0.6112,
+      "step": 11097
+    },
+    {
+      "epoch": 0.9457179377929271,
+      "grad_norm": 13.875,
+      "learning_rate": 1.3652763955262039e-06,
+      "loss": 0.5786,
+      "step": 11098
+    },
+    {
+      "epoch": 0.9458031529612271,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.365109075009164e-06,
+      "loss": 0.6727,
+      "step": 11099
+    },
+    {
+      "epoch": 0.945888368129527,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3649417524119874e-06,
+      "loss": 0.8454,
+      "step": 11100
+    },
+    {
+      "epoch": 0.945973583297827,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3647744277376979e-06,
+      "loss": 0.7791,
+      "step": 11101
+    },
+    {
+      "epoch": 0.946058798466127,
+      "grad_norm": 15.125,
+      "learning_rate": 1.3646071009893188e-06,
+      "loss": 0.6159,
+      "step": 11102
+    },
+    {
+      "epoch": 0.9461440136344269,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3644397721698742e-06,
+      "loss": 0.9871,
+      "step": 11103
+    },
+    {
+      "epoch": 0.9462292288027269,
+      "grad_norm": 10.625,
+      "learning_rate": 1.364272441282388e-06,
+      "loss": 0.4288,
+      "step": 11104
+    },
+    {
+      "epoch": 0.9463144439710268,
+      "grad_norm": 13.5,
+      "learning_rate": 1.364105108329884e-06,
+      "loss": 0.5338,
+      "step": 11105
+    },
+    {
+      "epoch": 0.9463996591393268,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3639377733153856e-06,
+      "loss": 0.6392,
+      "step": 11106
+    },
+    {
+      "epoch": 0.9464848743076267,
+      "grad_norm": 14.375,
+      "learning_rate": 1.3637704362419171e-06,
+      "loss": 0.6271,
+      "step": 11107
+    },
+    {
+      "epoch": 0.9465700894759267,
+      "grad_norm": 16.25,
+      "learning_rate": 1.3636030971125026e-06,
+      "loss": 0.896,
+      "step": 11108
+    },
+    {
+      "epoch": 0.9466553046442266,
+      "grad_norm": 25.625,
+      "learning_rate": 1.3634357559301662e-06,
+      "loss": 1.2843,
+      "step": 11109
+    },
+    {
+      "epoch": 0.9467405198125266,
+      "grad_norm": 14.25,
+      "learning_rate": 1.363268412697931e-06,
+      "loss": 0.7364,
+      "step": 11110
+    },
+    {
+      "epoch": 0.9468257349808266,
+      "grad_norm": 12.125,
+      "learning_rate": 1.3631010674188219e-06,
+      "loss": 0.4524,
+      "step": 11111
+    },
+    {
+      "epoch": 0.9469109501491265,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.3629337200958626e-06,
+      "loss": 0.505,
+      "step": 11112
+    },
+    {
+      "epoch": 0.9469961653174265,
+      "grad_norm": 13.0,
+      "learning_rate": 1.3627663707320775e-06,
+      "loss": 0.6783,
+      "step": 11113
+    },
+    {
+      "epoch": 0.9470813804857264,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3625990193304911e-06,
+      "loss": 0.5561,
+      "step": 11114
+    },
+    {
+      "epoch": 0.9471665956540264,
+      "grad_norm": 21.75,
+      "learning_rate": 1.362431665894127e-06,
+      "loss": 0.8594,
+      "step": 11115
+    },
+    {
+      "epoch": 0.9472518108223263,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3622643104260095e-06,
+      "loss": 1.1498,
+      "step": 11116
+    },
+    {
+      "epoch": 0.9473370259906263,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.362096952929163e-06,
+      "loss": 0.2352,
+      "step": 11117
+    },
+    {
+      "epoch": 0.9474222411589263,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.3619295934066117e-06,
+      "loss": 0.8051,
+      "step": 11118
+    },
+    {
+      "epoch": 0.9475074563272262,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3617622318613806e-06,
+      "loss": 0.8366,
+      "step": 11119
+    },
+    {
+      "epoch": 0.9475926714955262,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.361594868296493e-06,
+      "loss": 0.6228,
+      "step": 11120
+    },
+    {
+      "epoch": 0.9476778866638261,
+      "grad_norm": 11.125,
+      "learning_rate": 1.361427502714974e-06,
+      "loss": 0.423,
+      "step": 11121
+    },
+    {
+      "epoch": 0.9477631018321261,
+      "grad_norm": 23.875,
+      "learning_rate": 1.3612601351198485e-06,
+      "loss": 1.1474,
+      "step": 11122
+    },
+    {
+      "epoch": 0.947848317000426,
+      "grad_norm": 14.625,
+      "learning_rate": 1.3610927655141402e-06,
+      "loss": 0.6743,
+      "step": 11123
+    },
+    {
+      "epoch": 0.947933532168726,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.3609253939008738e-06,
+      "loss": 0.4045,
+      "step": 11124
+    },
+    {
+      "epoch": 0.948018747337026,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.3607580202830739e-06,
+      "loss": 0.4785,
+      "step": 11125
+    },
+    {
+      "epoch": 0.9481039625053259,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3605906446637656e-06,
+      "loss": 0.9845,
+      "step": 11126
+    },
+    {
+      "epoch": 0.9481891776736259,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3604232670459727e-06,
+      "loss": 0.6442,
+      "step": 11127
+    },
+    {
+      "epoch": 0.9482743928419258,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.3602558874327206e-06,
+      "loss": 0.8672,
+      "step": 11128
+    },
+    {
+      "epoch": 0.9483596080102258,
+      "grad_norm": 19.125,
+      "learning_rate": 1.3600885058270335e-06,
+      "loss": 0.7797,
+      "step": 11129
+    },
+    {
+      "epoch": 0.9484448231785257,
+      "grad_norm": 11.25,
+      "learning_rate": 1.359921122231937e-06,
+      "loss": 0.543,
+      "step": 11130
+    },
+    {
+      "epoch": 0.9485300383468257,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.3597537366504546e-06,
+      "loss": 0.5551,
+      "step": 11131
+    },
+    {
+      "epoch": 0.9486152535151257,
+      "grad_norm": 11.5,
+      "learning_rate": 1.359586349085612e-06,
+      "loss": 0.6685,
+      "step": 11132
+    },
+    {
+      "epoch": 0.9487004686834256,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3594189595404338e-06,
+      "loss": 0.7485,
+      "step": 11133
+    },
+    {
+      "epoch": 0.9487856838517256,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.3592515680179452e-06,
+      "loss": 1.0117,
+      "step": 11134
+    },
+    {
+      "epoch": 0.9488708990200255,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3590841745211707e-06,
+      "loss": 0.7559,
+      "step": 11135
+    },
+    {
+      "epoch": 0.9489561141883255,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3589167790531354e-06,
+      "loss": 0.996,
+      "step": 11136
+    },
+    {
+      "epoch": 0.9490413293566254,
+      "grad_norm": 14.25,
+      "learning_rate": 1.3587493816168648e-06,
+      "loss": 0.4351,
+      "step": 11137
+    },
+    {
+      "epoch": 0.9491265445249254,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3585819822153834e-06,
+      "loss": 0.5481,
+      "step": 11138
+    },
+    {
+      "epoch": 0.9492117596932254,
+      "grad_norm": 11.625,
+      "learning_rate": 1.3584145808517163e-06,
+      "loss": 0.5167,
+      "step": 11139
+    },
+    {
+      "epoch": 0.9492969748615253,
+      "grad_norm": 23.5,
+      "learning_rate": 1.3582471775288883e-06,
+      "loss": 0.8805,
+      "step": 11140
+    },
+    {
+      "epoch": 0.9493821900298253,
+      "grad_norm": 12.625,
+      "learning_rate": 1.3580797722499256e-06,
+      "loss": 0.8834,
+      "step": 11141
+    },
+    {
+      "epoch": 0.9494674051981252,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3579123650178525e-06,
+      "loss": 1.0889,
+      "step": 11142
+    },
+    {
+      "epoch": 0.9495526203664252,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.3577449558356944e-06,
+      "loss": 0.5831,
+      "step": 11143
+    },
+    {
+      "epoch": 0.9496378355347251,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.3575775447064765e-06,
+      "loss": 0.5942,
+      "step": 11144
+    },
+    {
+      "epoch": 0.9497230507030251,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3574101316332244e-06,
+      "loss": 0.4823,
+      "step": 11145
+    },
+    {
+      "epoch": 0.949808265871325,
+      "grad_norm": 11.5,
+      "learning_rate": 1.357242716618963e-06,
+      "loss": 0.6503,
+      "step": 11146
+    },
+    {
+      "epoch": 0.949893481039625,
+      "grad_norm": 13.5,
+      "learning_rate": 1.357075299666718e-06,
+      "loss": 0.8248,
+      "step": 11147
+    },
+    {
+      "epoch": 0.949978696207925,
+      "grad_norm": 23.75,
+      "learning_rate": 1.3569078807795148e-06,
+      "loss": 1.0501,
+      "step": 11148
+    },
+    {
+      "epoch": 0.9500639113762249,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3567404599603784e-06,
+      "loss": 0.724,
+      "step": 11149
+    },
+    {
+      "epoch": 0.9501491265445249,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3565730372123348e-06,
+      "loss": 0.8842,
+      "step": 11150
+    },
+    {
+      "epoch": 0.9502343417128248,
+      "grad_norm": 16.0,
+      "learning_rate": 1.356405612538409e-06,
+      "loss": 1.0217,
+      "step": 11151
+    },
+    {
+      "epoch": 0.9503195568811248,
+      "grad_norm": 18.125,
+      "learning_rate": 1.3562381859416268e-06,
+      "loss": 0.6596,
+      "step": 11152
+    },
+    {
+      "epoch": 0.9504047720494248,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3560707574250137e-06,
+      "loss": 1.1156,
+      "step": 11153
+    },
+    {
+      "epoch": 0.9504899872177247,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.3559033269915959e-06,
+      "loss": 0.5993,
+      "step": 11154
+    },
+    {
+      "epoch": 0.9505752023860247,
+      "grad_norm": 13.75,
+      "learning_rate": 1.355735894644398e-06,
+      "loss": 0.6029,
+      "step": 11155
+    },
+    {
+      "epoch": 0.9506604175543246,
+      "grad_norm": 15.5,
+      "learning_rate": 1.3555684603864464e-06,
+      "loss": 0.8165,
+      "step": 11156
+    },
+    {
+      "epoch": 0.9507456327226246,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.355401024220766e-06,
+      "loss": 0.7239,
+      "step": 11157
+    },
+    {
+      "epoch": 0.9508308478909245,
+      "grad_norm": 17.375,
+      "learning_rate": 1.355233586150384e-06,
+      "loss": 1.0024,
+      "step": 11158
+    },
+    {
+      "epoch": 0.9509160630592245,
+      "grad_norm": 21.875,
+      "learning_rate": 1.3550661461783245e-06,
+      "loss": 0.6146,
+      "step": 11159
+    },
+    {
+      "epoch": 0.9510012782275244,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.3548987043076145e-06,
+      "loss": 0.5588,
+      "step": 11160
+    },
+    {
+      "epoch": 0.9510864933958244,
+      "grad_norm": 12.125,
+      "learning_rate": 1.3547312605412793e-06,
+      "loss": 0.5698,
+      "step": 11161
+    },
+    {
+      "epoch": 0.9511717085641244,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.3545638148823453e-06,
+      "loss": 0.8529,
+      "step": 11162
+    },
+    {
+      "epoch": 0.9512569237324243,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3543963673338372e-06,
+      "loss": 0.442,
+      "step": 11163
+    },
+    {
+      "epoch": 0.9513421389007243,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3542289178987825e-06,
+      "loss": 0.8318,
+      "step": 11164
+    },
+    {
+      "epoch": 0.9514273540690242,
+      "grad_norm": 25.5,
+      "learning_rate": 1.354061466580206e-06,
+      "loss": 0.491,
+      "step": 11165
+    },
+    {
+      "epoch": 0.9515125692373242,
+      "grad_norm": 40.0,
+      "learning_rate": 1.3538940133811344e-06,
+      "loss": 0.9467,
+      "step": 11166
+    },
+    {
+      "epoch": 0.9515977844056241,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.3537265583045933e-06,
+      "loss": 0.5424,
+      "step": 11167
+    },
+    {
+      "epoch": 0.9516829995739241,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.353559101353609e-06,
+      "loss": 0.8793,
+      "step": 11168
+    },
+    {
+      "epoch": 0.9517682147422241,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3533916425312076e-06,
+      "loss": 0.8525,
+      "step": 11169
+    },
+    {
+      "epoch": 0.951853429910524,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3532241818404157e-06,
+      "loss": 1.0974,
+      "step": 11170
+    },
+    {
+      "epoch": 0.9519386450788241,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3530567192842586e-06,
+      "loss": 0.9203,
+      "step": 11171
+    },
+    {
+      "epoch": 0.952023860247124,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.352889254865763e-06,
+      "loss": 0.9527,
+      "step": 11172
+    },
+    {
+      "epoch": 0.952109075415424,
+      "grad_norm": 26.0,
+      "learning_rate": 1.352721788587955e-06,
+      "loss": 1.0209,
+      "step": 11173
+    },
+    {
+      "epoch": 0.952194290583724,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3525543204538613e-06,
+      "loss": 0.681,
+      "step": 11174
+    },
+    {
+      "epoch": 0.9522795057520239,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3523868504665077e-06,
+      "loss": 0.6908,
+      "step": 11175
+    },
+    {
+      "epoch": 0.9523647209203239,
+      "grad_norm": 25.375,
+      "learning_rate": 1.3522193786289206e-06,
+      "loss": 0.9578,
+      "step": 11176
+    },
+    {
+      "epoch": 0.9524499360886238,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3520519049441268e-06,
+      "loss": 0.9875,
+      "step": 11177
+    },
+    {
+      "epoch": 0.9525351512569238,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3518844294151523e-06,
+      "loss": 0.8847,
+      "step": 11178
+    },
+    {
+      "epoch": 0.9526203664252237,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.351716952045024e-06,
+      "loss": 0.7185,
+      "step": 11179
+    },
+    {
+      "epoch": 0.9527055815935237,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3515494728367677e-06,
+      "loss": 0.797,
+      "step": 11180
+    },
+    {
+      "epoch": 0.9527907967618237,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3513819917934108e-06,
+      "loss": 0.7569,
+      "step": 11181
+    },
+    {
+      "epoch": 0.9528760119301236,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.3512145089179787e-06,
+      "loss": 0.5086,
+      "step": 11182
+    },
+    {
+      "epoch": 0.9529612270984236,
+      "grad_norm": 13.125,
+      "learning_rate": 1.351047024213499e-06,
+      "loss": 0.7267,
+      "step": 11183
+    },
+    {
+      "epoch": 0.9530464422667235,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.350879537682998e-06,
+      "loss": 0.673,
+      "step": 11184
+    },
+    {
+      "epoch": 0.9531316574350235,
+      "grad_norm": 23.5,
+      "learning_rate": 1.3507120493295026e-06,
+      "loss": 0.9373,
+      "step": 11185
+    },
+    {
+      "epoch": 0.9532168726033234,
+      "grad_norm": 11.0,
+      "learning_rate": 1.3505445591560387e-06,
+      "loss": 0.4915,
+      "step": 11186
+    },
+    {
+      "epoch": 0.9533020877716234,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3503770671656336e-06,
+      "loss": 0.9672,
+      "step": 11187
+    },
+    {
+      "epoch": 0.9533873029399234,
+      "grad_norm": 17.625,
+      "learning_rate": 1.350209573361314e-06,
+      "loss": 1.1056,
+      "step": 11188
+    },
+    {
+      "epoch": 0.9534725181082233,
+      "grad_norm": 20.125,
+      "learning_rate": 1.3500420777461065e-06,
+      "loss": 0.5954,
+      "step": 11189
+    },
+    {
+      "epoch": 0.9535577332765233,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3498745803230383e-06,
+      "loss": 0.9035,
+      "step": 11190
+    },
+    {
+      "epoch": 0.9536429484448232,
+      "grad_norm": 20.25,
+      "learning_rate": 1.3497070810951363e-06,
+      "loss": 0.8066,
+      "step": 11191
+    },
+    {
+      "epoch": 0.9537281636131232,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3495395800654265e-06,
+      "loss": 0.9601,
+      "step": 11192
+    },
+    {
+      "epoch": 0.9538133787814231,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3493720772369365e-06,
+      "loss": 1.07,
+      "step": 11193
+    },
+    {
+      "epoch": 0.9538985939497231,
+      "grad_norm": 21.875,
+      "learning_rate": 1.3492045726126934e-06,
+      "loss": 0.6457,
+      "step": 11194
+    },
+    {
+      "epoch": 0.953983809118023,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.3490370661957238e-06,
+      "loss": 0.5155,
+      "step": 11195
+    },
+    {
+      "epoch": 0.954069024286323,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3488695579890549e-06,
+      "loss": 0.9245,
+      "step": 11196
+    },
+    {
+      "epoch": 0.954154239454623,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3487020479957133e-06,
+      "loss": 0.5774,
+      "step": 11197
+    },
+    {
+      "epoch": 0.9542394546229229,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3485345362187268e-06,
+      "loss": 0.9729,
+      "step": 11198
+    },
+    {
+      "epoch": 0.9543246697912229,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.348367022661122e-06,
+      "loss": 0.8422,
+      "step": 11199
+    },
+    {
+      "epoch": 0.9544098849595228,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.3481995073259263e-06,
+      "loss": 0.3893,
+      "step": 11200
+    },
+    {
+      "epoch": 0.9544951001278228,
+      "grad_norm": 12.625,
+      "learning_rate": 1.3480319902161665e-06,
+      "loss": 0.5194,
+      "step": 11201
+    },
+    {
+      "epoch": 0.9545803152961228,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3478644713348705e-06,
+      "loss": 1.1465,
+      "step": 11202
+    },
+    {
+      "epoch": 0.9546655304644227,
+      "grad_norm": 12.625,
+      "learning_rate": 1.3476969506850648e-06,
+      "loss": 0.6907,
+      "step": 11203
+    },
+    {
+      "epoch": 0.9547507456327227,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.3475294282697769e-06,
+      "loss": 0.7228,
+      "step": 11204
+    },
+    {
+      "epoch": 0.9548359608010226,
+      "grad_norm": 21.75,
+      "learning_rate": 1.347361904092034e-06,
+      "loss": 0.8344,
+      "step": 11205
+    },
+    {
+      "epoch": 0.9549211759693226,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.3471943781548638e-06,
+      "loss": 0.4806,
+      "step": 11206
+    },
+    {
+      "epoch": 0.9550063911376225,
+      "grad_norm": 14.375,
+      "learning_rate": 1.3470268504612937e-06,
+      "loss": 0.526,
+      "step": 11207
+    },
+    {
+      "epoch": 0.9550916063059225,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3468593210143508e-06,
+      "loss": 0.4478,
+      "step": 11208
+    },
+    {
+      "epoch": 0.9551768214742224,
+      "grad_norm": 37.5,
+      "learning_rate": 1.3466917898170623e-06,
+      "loss": 1.4774,
+      "step": 11209
+    },
+    {
+      "epoch": 0.9552620366425224,
+      "grad_norm": 20.75,
+      "learning_rate": 1.3465242568724556e-06,
+      "loss": 0.9637,
+      "step": 11210
+    },
+    {
+      "epoch": 0.9553472518108224,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3463567221835593e-06,
+      "loss": 0.5761,
+      "step": 11211
+    },
+    {
+      "epoch": 0.9554324669791223,
+      "grad_norm": 15.625,
+      "learning_rate": 1.3461891857533998e-06,
+      "loss": 0.9901,
+      "step": 11212
+    },
+    {
+      "epoch": 0.9555176821474223,
+      "grad_norm": 16.375,
+      "learning_rate": 1.346021647585005e-06,
+      "loss": 1.0439,
+      "step": 11213
+    },
+    {
+      "epoch": 0.9556028973157222,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3458541076814022e-06,
+      "loss": 0.927,
+      "step": 11214
+    },
+    {
+      "epoch": 0.9556881124840222,
+      "grad_norm": 13.0,
+      "learning_rate": 1.34568656604562e-06,
+      "loss": 0.6717,
+      "step": 11215
+    },
+    {
+      "epoch": 0.9557733276523221,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3455190226806846e-06,
+      "loss": 1.05,
+      "step": 11216
+    },
+    {
+      "epoch": 0.9558585428206221,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3453514775896248e-06,
+      "loss": 0.7356,
+      "step": 11217
+    },
+    {
+      "epoch": 0.9559437579889221,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3451839307754673e-06,
+      "loss": 1.0657,
+      "step": 11218
+    },
+    {
+      "epoch": 0.956028973157222,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.3450163822412413e-06,
+      "loss": 0.4628,
+      "step": 11219
+    },
+    {
+      "epoch": 0.956114188325522,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3448488319899734e-06,
+      "loss": 0.6879,
+      "step": 11220
+    },
+    {
+      "epoch": 0.9561994034938219,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3446812800246913e-06,
+      "loss": 0.4198,
+      "step": 11221
+    },
+    {
+      "epoch": 0.9562846186621219,
+      "grad_norm": 13.5,
+      "learning_rate": 1.344513726348424e-06,
+      "loss": 0.5079,
+      "step": 11222
+    },
+    {
+      "epoch": 0.9563698338304218,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3443461709641983e-06,
+      "loss": 0.6333,
+      "step": 11223
+    },
+    {
+      "epoch": 0.9564550489987218,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3441786138750424e-06,
+      "loss": 0.7613,
+      "step": 11224
+    },
+    {
+      "epoch": 0.9565402641670218,
+      "grad_norm": 44.25,
+      "learning_rate": 1.344011055083984e-06,
+      "loss": 1.1897,
+      "step": 11225
+    },
+    {
+      "epoch": 0.9566254793353217,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3438434945940515e-06,
+      "loss": 0.7008,
+      "step": 11226
+    },
+    {
+      "epoch": 0.9567106945036217,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.343675932408273e-06,
+      "loss": 0.9346,
+      "step": 11227
+    },
+    {
+      "epoch": 0.9567959096719216,
+      "grad_norm": 17.875,
+      "learning_rate": 1.343508368529676e-06,
+      "loss": 0.7028,
+      "step": 11228
+    },
+    {
+      "epoch": 0.9568811248402216,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3433408029612885e-06,
+      "loss": 0.8707,
+      "step": 11229
+    },
+    {
+      "epoch": 0.9569663400085215,
+      "grad_norm": 22.25,
+      "learning_rate": 1.3431732357061389e-06,
+      "loss": 1.4393,
+      "step": 11230
+    },
+    {
+      "epoch": 0.9570515551768215,
+      "grad_norm": 11.125,
+      "learning_rate": 1.3430056667672555e-06,
+      "loss": 0.3387,
+      "step": 11231
+    },
+    {
+      "epoch": 0.9571367703451215,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3428380961476662e-06,
+      "loss": 0.5938,
+      "step": 11232
+    },
+    {
+      "epoch": 0.9572219855134214,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3426705238503986e-06,
+      "loss": 0.9805,
+      "step": 11233
+    },
+    {
+      "epoch": 0.9573072006817214,
+      "grad_norm": 18.875,
+      "learning_rate": 1.3425029498784819e-06,
+      "loss": 1.0721,
+      "step": 11234
+    },
+    {
+      "epoch": 0.9573924158500213,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.3423353742349441e-06,
+      "loss": 0.355,
+      "step": 11235
+    },
+    {
+      "epoch": 0.9574776310183213,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3421677969228124e-06,
+      "loss": 0.7045,
+      "step": 11236
+    },
+    {
+      "epoch": 0.9575628461866212,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3420002179451167e-06,
+      "loss": 0.6337,
+      "step": 11237
+    },
+    {
+      "epoch": 0.9576480613549212,
+      "grad_norm": 22.0,
+      "learning_rate": 1.341832637304884e-06,
+      "loss": 0.9191,
+      "step": 11238
+    },
+    {
+      "epoch": 0.9577332765232212,
+      "grad_norm": 21.0,
+      "learning_rate": 1.3416650550051438e-06,
+      "loss": 1.2782,
+      "step": 11239
+    },
+    {
+      "epoch": 0.9578184916915211,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3414974710489234e-06,
+      "loss": 0.7834,
+      "step": 11240
+    },
+    {
+      "epoch": 0.9579037068598211,
+      "grad_norm": 13.0,
+      "learning_rate": 1.341329885439252e-06,
+      "loss": 0.7518,
+      "step": 11241
+    },
+    {
+      "epoch": 0.957988922028121,
+      "grad_norm": 10.75,
+      "learning_rate": 1.3411622981791574e-06,
+      "loss": 0.3337,
+      "step": 11242
+    },
+    {
+      "epoch": 0.958074137196421,
+      "grad_norm": 16.5,
+      "learning_rate": 1.340994709271669e-06,
+      "loss": 0.6011,
+      "step": 11243
+    },
+    {
+      "epoch": 0.9581593523647209,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3408271187198142e-06,
+      "loss": 0.5868,
+      "step": 11244
+    },
+    {
+      "epoch": 0.9582445675330209,
+      "grad_norm": 9.5,
+      "learning_rate": 1.340659526526622e-06,
+      "loss": 0.4471,
+      "step": 11245
+    },
+    {
+      "epoch": 0.9583297827013209,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.340491932695121e-06,
+      "loss": 0.4826,
+      "step": 11246
+    },
+    {
+      "epoch": 0.9584149978696208,
+      "grad_norm": 12.375,
+      "learning_rate": 1.34032433722834e-06,
+      "loss": 0.6259,
+      "step": 11247
+    },
+    {
+      "epoch": 0.9585002130379208,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3401567401293075e-06,
+      "loss": 0.9066,
+      "step": 11248
+    },
+    {
+      "epoch": 0.9585854282062207,
+      "grad_norm": 26.0,
+      "learning_rate": 1.3399891414010519e-06,
+      "loss": 1.1396,
+      "step": 11249
+    },
+    {
+      "epoch": 0.9586706433745207,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3398215410466018e-06,
+      "loss": 1.0114,
+      "step": 11250
+    },
+    {
+      "epoch": 0.9587558585428206,
+      "grad_norm": 15.375,
+      "learning_rate": 1.3396539390689863e-06,
+      "loss": 0.9949,
+      "step": 11251
+    },
+    {
+      "epoch": 0.9588410737111206,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.339486335471234e-06,
+      "loss": 0.6976,
+      "step": 11252
+    },
+    {
+      "epoch": 0.9589262888794206,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3393187302563734e-06,
+      "loss": 0.698,
+      "step": 11253
+    },
+    {
+      "epoch": 0.9590115040477205,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.3391511234274337e-06,
+      "loss": 0.8465,
+      "step": 11254
+    },
+    {
+      "epoch": 0.9590967192160205,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3389835149874438e-06,
+      "loss": 0.7536,
+      "step": 11255
+    },
+    {
+      "epoch": 0.9591819343843204,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3388159049394323e-06,
+      "loss": 1.0718,
+      "step": 11256
+    },
+    {
+      "epoch": 0.9592671495526204,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.3386482932864275e-06,
+      "loss": 0.4184,
+      "step": 11257
+    },
+    {
+      "epoch": 0.9593523647209203,
+      "grad_norm": 31.125,
+      "learning_rate": 1.3384806800314592e-06,
+      "loss": 0.7978,
+      "step": 11258
+    },
+    {
+      "epoch": 0.9594375798892203,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3383130651775561e-06,
+      "loss": 0.6583,
+      "step": 11259
+    },
+    {
+      "epoch": 0.9595227950575203,
+      "grad_norm": 15.625,
+      "learning_rate": 1.3381454487277473e-06,
+      "loss": 0.4511,
+      "step": 11260
+    },
+    {
+      "epoch": 0.9596080102258202,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3379778306850614e-06,
+      "loss": 0.8283,
+      "step": 11261
+    },
+    {
+      "epoch": 0.9596932253941202,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.3378102110525277e-06,
+      "loss": 0.3871,
+      "step": 11262
+    },
+    {
+      "epoch": 0.9597784405624201,
+      "grad_norm": 13.625,
+      "learning_rate": 1.337642589833175e-06,
+      "loss": 0.4974,
+      "step": 11263
+    },
+    {
+      "epoch": 0.9598636557307201,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.337474967030033e-06,
+      "loss": 0.8338,
+      "step": 11264
+    },
+    {
+      "epoch": 0.95994887089902,
+      "grad_norm": 15.875,
+      "learning_rate": 1.33730734264613e-06,
+      "loss": 0.714,
+      "step": 11265
+    },
+    {
+      "epoch": 0.96003408606732,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.3371397166844955e-06,
+      "loss": 1.0014,
+      "step": 11266
+    },
+    {
+      "epoch": 0.96011930123562,
+      "grad_norm": 20.875,
+      "learning_rate": 1.3369720891481585e-06,
+      "loss": 1.0857,
+      "step": 11267
+    },
+    {
+      "epoch": 0.9602045164039199,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.3368044600401489e-06,
+      "loss": 0.6234,
+      "step": 11268
+    },
+    {
+      "epoch": 0.9602897315722199,
+      "grad_norm": 13.125,
+      "learning_rate": 1.3366368293634954e-06,
+      "loss": 0.5609,
+      "step": 11269
+    },
+    {
+      "epoch": 0.9603749467405198,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.3364691971212271e-06,
+      "loss": 0.5431,
+      "step": 11270
+    },
+    {
+      "epoch": 0.9604601619088198,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.336301563316373e-06,
+      "loss": 0.5444,
+      "step": 11271
+    },
+    {
+      "epoch": 0.9605453770771197,
+      "grad_norm": 12.25,
+      "learning_rate": 1.3361339279519635e-06,
+      "loss": 0.6693,
+      "step": 11272
+    },
+    {
+      "epoch": 0.9606305922454197,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3359662910310273e-06,
+      "loss": 1.0053,
+      "step": 11273
+    },
+    {
+      "epoch": 0.9607158074137196,
+      "grad_norm": 15.0,
+      "learning_rate": 1.3357986525565934e-06,
+      "loss": 0.8832,
+      "step": 11274
+    },
+    {
+      "epoch": 0.9608010225820196,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3356310125316917e-06,
+      "loss": 0.6233,
+      "step": 11275
+    },
+    {
+      "epoch": 0.9608862377503196,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3354633709593517e-06,
+      "loss": 0.5333,
+      "step": 11276
+    },
+    {
+      "epoch": 0.9609714529186195,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.3352957278426026e-06,
+      "loss": 0.7211,
+      "step": 11277
+    },
+    {
+      "epoch": 0.9610566680869195,
+      "grad_norm": 14.375,
+      "learning_rate": 1.3351280831844737e-06,
+      "loss": 0.5842,
+      "step": 11278
+    },
+    {
+      "epoch": 0.9611418832552194,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.334960436987995e-06,
+      "loss": 0.3551,
+      "step": 11279
+    },
+    {
+      "epoch": 0.9612270984235194,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.3347927892561956e-06,
+      "loss": 0.4117,
+      "step": 11280
+    },
+    {
+      "epoch": 0.9613123135918193,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3346251399921053e-06,
+      "loss": 1.1344,
+      "step": 11281
+    },
+    {
+      "epoch": 0.9613975287601193,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3344574891987535e-06,
+      "loss": 0.9437,
+      "step": 11282
+    },
+    {
+      "epoch": 0.9614827439284193,
+      "grad_norm": 20.625,
+      "learning_rate": 1.33428983687917e-06,
+      "loss": 1.0353,
+      "step": 11283
+    },
+    {
+      "epoch": 0.9615679590967192,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3341221830363845e-06,
+      "loss": 0.8744,
+      "step": 11284
+    },
+    {
+      "epoch": 0.9616531742650192,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.3339545276734267e-06,
+      "loss": 0.3047,
+      "step": 11285
+    },
+    {
+      "epoch": 0.9617383894333191,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3337868707933258e-06,
+      "loss": 0.8173,
+      "step": 11286
+    },
+    {
+      "epoch": 0.9618236046016191,
+      "grad_norm": 10.375,
+      "learning_rate": 1.333619212399112e-06,
+      "loss": 0.4057,
+      "step": 11287
+    },
+    {
+      "epoch": 0.961908819769919,
+      "grad_norm": 12.25,
+      "learning_rate": 1.3334515524938152e-06,
+      "loss": 0.4672,
+      "step": 11288
+    },
+    {
+      "epoch": 0.961994034938219,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3332838910804644e-06,
+      "loss": 0.5211,
+      "step": 11289
+    },
+    {
+      "epoch": 0.962079250106519,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3331162281620905e-06,
+      "loss": 0.4687,
+      "step": 11290
+    },
+    {
+      "epoch": 0.9621644652748189,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.3329485637417223e-06,
+      "loss": 0.6322,
+      "step": 11291
+    },
+    {
+      "epoch": 0.9622496804431189,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.3327808978223905e-06,
+      "loss": 0.7581,
+      "step": 11292
+    },
+    {
+      "epoch": 0.9623348956114188,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3326132304071244e-06,
+      "loss": 0.846,
+      "step": 11293
+    },
+    {
+      "epoch": 0.9624201107797188,
+      "grad_norm": 12.125,
+      "learning_rate": 1.3324455614989542e-06,
+      "loss": 0.534,
+      "step": 11294
+    },
+    {
+      "epoch": 0.9625053259480187,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3322778911009099e-06,
+      "loss": 0.693,
+      "step": 11295
+    },
+    {
+      "epoch": 0.9625905411163187,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3321102192160216e-06,
+      "loss": 0.5275,
+      "step": 11296
+    },
+    {
+      "epoch": 0.9626757562846187,
+      "grad_norm": 18.25,
+      "learning_rate": 1.3319425458473188e-06,
+      "loss": 1.0927,
+      "step": 11297
+    },
+    {
+      "epoch": 0.9627609714529186,
+      "grad_norm": 34.25,
+      "learning_rate": 1.3317748709978317e-06,
+      "loss": 0.9722,
+      "step": 11298
+    },
+    {
+      "epoch": 0.9628461866212186,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3316071946705902e-06,
+      "loss": 1.2774,
+      "step": 11299
+    },
+    {
+      "epoch": 0.9629314017895185,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3314395168686256e-06,
+      "loss": 0.7155,
+      "step": 11300
+    },
+    {
+      "epoch": 0.9630166169578185,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.331271837594966e-06,
+      "loss": 0.7423,
+      "step": 11301
+    },
+    {
+      "epoch": 0.9631018321261184,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3311041568526432e-06,
+      "loss": 0.8511,
+      "step": 11302
+    },
+    {
+      "epoch": 0.9631870472944184,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3309364746446863e-06,
+      "loss": 1.085,
+      "step": 11303
+    },
+    {
+      "epoch": 0.9632722624627184,
+      "grad_norm": 12.0,
+      "learning_rate": 1.3307687909741263e-06,
+      "loss": 0.4792,
+      "step": 11304
+    },
+    {
+      "epoch": 0.9633574776310183,
+      "grad_norm": 13.0,
+      "learning_rate": 1.3306011058439926e-06,
+      "loss": 0.7825,
+      "step": 11305
+    },
+    {
+      "epoch": 0.9634426927993183,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3304334192573163e-06,
+      "loss": 0.6797,
+      "step": 11306
+    },
+    {
+      "epoch": 0.9635279079676182,
+      "grad_norm": 13.5,
+      "learning_rate": 1.3302657312171268e-06,
+      "loss": 0.649,
+      "step": 11307
+    },
+    {
+      "epoch": 0.9636131231359182,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3300980417264552e-06,
+      "loss": 0.5308,
+      "step": 11308
+    },
+    {
+      "epoch": 0.9636983383042181,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3299303507883316e-06,
+      "loss": 0.7404,
+      "step": 11309
+    },
+    {
+      "epoch": 0.9637835534725181,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.3297626584057853e-06,
+      "loss": 0.9554,
+      "step": 11310
+    },
+    {
+      "epoch": 0.963868768640818,
+      "grad_norm": 21.5,
+      "learning_rate": 1.3295949645818484e-06,
+      "loss": 0.9217,
+      "step": 11311
+    },
+    {
+      "epoch": 0.963953983809118,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.3294272693195504e-06,
+      "loss": 0.9075,
+      "step": 11312
+    },
+    {
+      "epoch": 0.964039198977418,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3292595726219215e-06,
+      "loss": 0.753,
+      "step": 11313
+    },
+    {
+      "epoch": 0.9641244141457179,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3290918744919926e-06,
+      "loss": 0.8299,
+      "step": 11314
+    },
+    {
+      "epoch": 0.9642096293140179,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.3289241749327939e-06,
+      "loss": 0.7705,
+      "step": 11315
+    },
+    {
+      "epoch": 0.9642948444823178,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3287564739473558e-06,
+      "loss": 0.4943,
+      "step": 11316
+    },
+    {
+      "epoch": 0.9643800596506178,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.3285887715387095e-06,
+      "loss": 0.2655,
+      "step": 11317
+    },
+    {
+      "epoch": 0.9644652748189178,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3284210677098847e-06,
+      "loss": 0.8507,
+      "step": 11318
+    },
+    {
+      "epoch": 0.9645504899872177,
+      "grad_norm": 12.75,
+      "learning_rate": 1.3282533624639124e-06,
+      "loss": 0.6636,
+      "step": 11319
+    },
+    {
+      "epoch": 0.9646357051555177,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.328085655803823e-06,
+      "loss": 0.8428,
+      "step": 11320
+    },
+    {
+      "epoch": 0.9647209203238176,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.3279179477326475e-06,
+      "loss": 0.9147,
+      "step": 11321
+    },
+    {
+      "epoch": 0.9648061354921176,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3277502382534163e-06,
+      "loss": 0.7329,
+      "step": 11322
+    },
+    {
+      "epoch": 0.9648913506604175,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3275825273691603e-06,
+      "loss": 0.5594,
+      "step": 11323
+    },
+    {
+      "epoch": 0.9649765658287175,
+      "grad_norm": 10.75,
+      "learning_rate": 1.3274148150829098e-06,
+      "loss": 0.5454,
+      "step": 11324
+    },
+    {
+      "epoch": 0.9650617809970174,
+      "grad_norm": 20.5,
+      "learning_rate": 1.3272471013976956e-06,
+      "loss": 1.0577,
+      "step": 11325
+    },
+    {
+      "epoch": 0.9651469961653174,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3270793863165489e-06,
+      "loss": 0.8825,
+      "step": 11326
+    },
+    {
+      "epoch": 0.9652322113336174,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.3269116698424997e-06,
+      "loss": 0.6828,
+      "step": 11327
+    },
+    {
+      "epoch": 0.9653174265019173,
+      "grad_norm": 18.375,
+      "learning_rate": 1.32674395197858e-06,
+      "loss": 0.714,
+      "step": 11328
+    },
+    {
+      "epoch": 0.9654026416702173,
+      "grad_norm": 15.125,
+      "learning_rate": 1.3265762327278195e-06,
+      "loss": 0.682,
+      "step": 11329
+    },
+    {
+      "epoch": 0.9654878568385172,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.3264085120932495e-06,
+      "loss": 0.6552,
+      "step": 11330
+    },
+    {
+      "epoch": 0.9655730720068172,
+      "grad_norm": 13.75,
+      "learning_rate": 1.3262407900779006e-06,
+      "loss": 0.4089,
+      "step": 11331
+    },
+    {
+      "epoch": 0.9656582871751171,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3260730666848043e-06,
+      "loss": 0.5968,
+      "step": 11332
+    },
+    {
+      "epoch": 0.9657435023434171,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.3259053419169912e-06,
+      "loss": 0.6452,
+      "step": 11333
+    },
+    {
+      "epoch": 0.9658287175117171,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.3257376157774924e-06,
+      "loss": 0.3799,
+      "step": 11334
+    },
+    {
+      "epoch": 0.965913932680017,
+      "grad_norm": 18.625,
+      "learning_rate": 1.3255698882693382e-06,
+      "loss": 1.0815,
+      "step": 11335
+    },
+    {
+      "epoch": 0.965999147848317,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3254021593955605e-06,
+      "loss": 1.021,
+      "step": 11336
+    },
+    {
+      "epoch": 0.9660843630166169,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3252344291591898e-06,
+      "loss": 0.9381,
+      "step": 11337
+    },
+    {
+      "epoch": 0.9661695781849169,
+      "grad_norm": 11.5,
+      "learning_rate": 1.3250666975632577e-06,
+      "loss": 0.4586,
+      "step": 11338
+    },
+    {
+      "epoch": 0.9662547933532168,
+      "grad_norm": 21.125,
+      "learning_rate": 1.3248989646107943e-06,
+      "loss": 1.1198,
+      "step": 11339
+    },
+    {
+      "epoch": 0.9663400085215168,
+      "grad_norm": 13.75,
+      "learning_rate": 1.3247312303048315e-06,
+      "loss": 0.5648,
+      "step": 11340
+    },
+    {
+      "epoch": 0.9664252236898168,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3245634946484005e-06,
+      "loss": 0.7224,
+      "step": 11341
+    },
+    {
+      "epoch": 0.9665104388581167,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3243957576445314e-06,
+      "loss": 0.7446,
+      "step": 11342
+    },
+    {
+      "epoch": 0.9665956540264167,
+      "grad_norm": 12.125,
+      "learning_rate": 1.324228019296257e-06,
+      "loss": 0.5621,
+      "step": 11343
+    },
+    {
+      "epoch": 0.9666808691947166,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.324060279606607e-06,
+      "loss": 0.7114,
+      "step": 11344
+    },
+    {
+      "epoch": 0.9667660843630166,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3238925385786141e-06,
+      "loss": 0.8529,
+      "step": 11345
+    },
+    {
+      "epoch": 0.9668512995313165,
+      "grad_norm": 13.0,
+      "learning_rate": 1.323724796215308e-06,
+      "loss": 0.6608,
+      "step": 11346
+    },
+    {
+      "epoch": 0.9669365146996165,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.3235570525197212e-06,
+      "loss": 0.3913,
+      "step": 11347
+    },
+    {
+      "epoch": 0.9670217298679165,
+      "grad_norm": 25.0,
+      "learning_rate": 1.3233893074948838e-06,
+      "loss": 0.9876,
+      "step": 11348
+    },
+    {
+      "epoch": 0.9671069450362164,
+      "grad_norm": 21.125,
+      "learning_rate": 1.3232215611438288e-06,
+      "loss": 0.9292,
+      "step": 11349
+    },
+    {
+      "epoch": 0.9671921602045164,
+      "grad_norm": 13.375,
+      "learning_rate": 1.323053813469586e-06,
+      "loss": 0.692,
+      "step": 11350
+    },
+    {
+      "epoch": 0.9672773753728163,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.322886064475187e-06,
+      "loss": 0.8455,
+      "step": 11351
+    },
+    {
+      "epoch": 0.9673625905411163,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3227183141636637e-06,
+      "loss": 0.7569,
+      "step": 11352
+    },
+    {
+      "epoch": 0.9674478057094162,
+      "grad_norm": 12.75,
+      "learning_rate": 1.3225505625380481e-06,
+      "loss": 0.6274,
+      "step": 11353
+    },
+    {
+      "epoch": 0.9675330208777162,
+      "grad_norm": 16.125,
+      "learning_rate": 1.32238280960137e-06,
+      "loss": 0.6565,
+      "step": 11354
+    },
+    {
+      "epoch": 0.9676182360460162,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.322215055356662e-06,
+      "loss": 0.7698,
+      "step": 11355
+    },
+    {
+      "epoch": 0.9677034512143161,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3220472998069553e-06,
+      "loss": 0.9246,
+      "step": 11356
+    },
+    {
+      "epoch": 0.9677886663826161,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.3218795429552813e-06,
+      "loss": 0.6619,
+      "step": 11357
+    },
+    {
+      "epoch": 0.967873881550916,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.3217117848046721e-06,
+      "loss": 0.9119,
+      "step": 11358
+    },
+    {
+      "epoch": 0.967959096719216,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3215440253581585e-06,
+      "loss": 0.796,
+      "step": 11359
+    },
+    {
+      "epoch": 0.9680443118875159,
+      "grad_norm": 15.75,
+      "learning_rate": 1.3213762646187724e-06,
+      "loss": 0.7491,
+      "step": 11360
+    },
+    {
+      "epoch": 0.9681295270558159,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3212085025895458e-06,
+      "loss": 0.674,
+      "step": 11361
+    },
+    {
+      "epoch": 0.9682147422241159,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3210407392735097e-06,
+      "loss": 0.5801,
+      "step": 11362
+    },
+    {
+      "epoch": 0.9682999573924158,
+      "grad_norm": 20.375,
+      "learning_rate": 1.3208729746736958e-06,
+      "loss": 0.9625,
+      "step": 11363
+    },
+    {
+      "epoch": 0.9683851725607158,
+      "grad_norm": 18.25,
+      "learning_rate": 1.3207052087931362e-06,
+      "loss": 0.8111,
+      "step": 11364
+    },
+    {
+      "epoch": 0.9684703877290157,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3205374416348627e-06,
+      "loss": 0.5216,
+      "step": 11365
+    },
+    {
+      "epoch": 0.9685556028973157,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3203696732019065e-06,
+      "loss": 0.7154,
+      "step": 11366
+    },
+    {
+      "epoch": 0.9686408180656156,
+      "grad_norm": 19.75,
+      "learning_rate": 1.320201903497299e-06,
+      "loss": 0.8924,
+      "step": 11367
+    },
+    {
+      "epoch": 0.9687260332339156,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.320034132524073e-06,
+      "loss": 0.4112,
+      "step": 11368
+    },
+    {
+      "epoch": 0.9688112484022156,
+      "grad_norm": 11.875,
+      "learning_rate": 1.31986636028526e-06,
+      "loss": 0.5414,
+      "step": 11369
+    },
+    {
+      "epoch": 0.9688964635705155,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3196985867838913e-06,
+      "loss": 0.7155,
+      "step": 11370
+    },
+    {
+      "epoch": 0.9689816787388155,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.3195308120229988e-06,
+      "loss": 0.3473,
+      "step": 11371
+    },
+    {
+      "epoch": 0.9690668939071154,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.319363036005615e-06,
+      "loss": 0.8222,
+      "step": 11372
+    },
+    {
+      "epoch": 0.9691521090754154,
+      "grad_norm": 20.375,
+      "learning_rate": 1.3191952587347714e-06,
+      "loss": 0.978,
+      "step": 11373
+    },
+    {
+      "epoch": 0.9692373242437153,
+      "grad_norm": 18.0,
+      "learning_rate": 1.3190274802134995e-06,
+      "loss": 0.6053,
+      "step": 11374
+    },
+    {
+      "epoch": 0.9693225394120153,
+      "grad_norm": 22.125,
+      "learning_rate": 1.318859700444832e-06,
+      "loss": 0.9879,
+      "step": 11375
+    },
+    {
+      "epoch": 0.9694077545803153,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3186919194318002e-06,
+      "loss": 0.8097,
+      "step": 11376
+    },
+    {
+      "epoch": 0.9694929697486152,
+      "grad_norm": 14.625,
+      "learning_rate": 1.3185241371774368e-06,
+      "loss": 1.3046,
+      "step": 11377
+    },
+    {
+      "epoch": 0.9695781849169152,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.3183563536847729e-06,
+      "loss": 0.7367,
+      "step": 11378
+    },
+    {
+      "epoch": 0.9696634000852151,
+      "grad_norm": 20.5,
+      "learning_rate": 1.318188568956841e-06,
+      "loss": 0.8635,
+      "step": 11379
+    },
+    {
+      "epoch": 0.9697486152535151,
+      "grad_norm": 16.0,
+      "learning_rate": 1.318020782996673e-06,
+      "loss": 0.8515,
+      "step": 11380
+    },
+    {
+      "epoch": 0.969833830421815,
+      "grad_norm": 14.625,
+      "learning_rate": 1.3178529958073014e-06,
+      "loss": 0.8373,
+      "step": 11381
+    },
+    {
+      "epoch": 0.969919045590115,
+      "grad_norm": 13.75,
+      "learning_rate": 1.317685207391758e-06,
+      "loss": 0.7394,
+      "step": 11382
+    },
+    {
+      "epoch": 0.970004260758415,
+      "grad_norm": 11.75,
+      "learning_rate": 1.3175174177530744e-06,
+      "loss": 0.5973,
+      "step": 11383
+    },
+    {
+      "epoch": 0.9700894759267149,
+      "grad_norm": 15.0,
+      "learning_rate": 1.3173496268942835e-06,
+      "loss": 0.6504,
+      "step": 11384
+    },
+    {
+      "epoch": 0.9701746910950149,
+      "grad_norm": 15.375,
+      "learning_rate": 1.3171818348184168e-06,
+      "loss": 0.7858,
+      "step": 11385
+    },
+    {
+      "epoch": 0.9702599062633148,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3170140415285071e-06,
+      "loss": 0.843,
+      "step": 11386
+    },
+    {
+      "epoch": 0.9703451214316148,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.3168462470275864e-06,
+      "loss": 0.494,
+      "step": 11387
+    },
+    {
+      "epoch": 0.9704303365999147,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.316678451318686e-06,
+      "loss": 0.4649,
+      "step": 11388
+    },
+    {
+      "epoch": 0.9705155517682147,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3165106544048397e-06,
+      "loss": 0.7248,
+      "step": 11389
+    },
+    {
+      "epoch": 0.9706007669365146,
+      "grad_norm": 19.25,
+      "learning_rate": 1.316342856289079e-06,
+      "loss": 0.9645,
+      "step": 11390
+    },
+    {
+      "epoch": 0.9706859821048146,
+      "grad_norm": 10.25,
+      "learning_rate": 1.3161750569744357e-06,
+      "loss": 0.6843,
+      "step": 11391
+    },
+    {
+      "epoch": 0.9707711972731146,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.3160072564639429e-06,
+      "loss": 0.4417,
+      "step": 11392
+    },
+    {
+      "epoch": 0.9708564124414145,
+      "grad_norm": 20.375,
+      "learning_rate": 1.3158394547606326e-06,
+      "loss": 0.8439,
+      "step": 11393
+    },
+    {
+      "epoch": 0.9709416276097145,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3156716518675372e-06,
+      "loss": 0.4711,
+      "step": 11394
+    },
+    {
+      "epoch": 0.9710268427780144,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.3155038477876886e-06,
+      "loss": 0.4872,
+      "step": 11395
+    },
+    {
+      "epoch": 0.9711120579463144,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.31533604252412e-06,
+      "loss": 0.7636,
+      "step": 11396
+    },
+    {
+      "epoch": 0.9711972731146145,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3151682360798632e-06,
+      "loss": 0.5719,
+      "step": 11397
+    },
+    {
+      "epoch": 0.9712824882829144,
+      "grad_norm": 13.125,
+      "learning_rate": 1.3150004284579513e-06,
+      "loss": 0.402,
+      "step": 11398
+    },
+    {
+      "epoch": 0.9713677034512144,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3148326196614158e-06,
+      "loss": 0.7881,
+      "step": 11399
+    },
+    {
+      "epoch": 0.9714529186195143,
+      "grad_norm": 17.5,
+      "learning_rate": 1.31466480969329e-06,
+      "loss": 0.8346,
+      "step": 11400
+    },
+    {
+      "epoch": 0.9715381337878143,
+      "grad_norm": 15.125,
+      "learning_rate": 1.3144969985566057e-06,
+      "loss": 0.9679,
+      "step": 11401
+    },
+    {
+      "epoch": 0.9716233489561142,
+      "grad_norm": 25.625,
+      "learning_rate": 1.314329186254396e-06,
+      "loss": 0.7881,
+      "step": 11402
+    },
+    {
+      "epoch": 0.9717085641244142,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.3141613727896932e-06,
+      "loss": 0.4283,
+      "step": 11403
+    },
+    {
+      "epoch": 0.9717937792927142,
+      "grad_norm": 21.0,
+      "learning_rate": 1.3139935581655297e-06,
+      "loss": 1.1608,
+      "step": 11404
+    },
+    {
+      "epoch": 0.9718789944610141,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.3138257423849382e-06,
+      "loss": 0.7783,
+      "step": 11405
+    },
+    {
+      "epoch": 0.9719642096293141,
+      "grad_norm": 20.875,
+      "learning_rate": 1.3136579254509518e-06,
+      "loss": 0.8851,
+      "step": 11406
+    },
+    {
+      "epoch": 0.972049424797614,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.3134901073666022e-06,
+      "loss": 0.5869,
+      "step": 11407
+    },
+    {
+      "epoch": 0.972134639965914,
+      "grad_norm": 12.875,
+      "learning_rate": 1.313322288134923e-06,
+      "loss": 0.7989,
+      "step": 11408
+    },
+    {
+      "epoch": 0.9722198551342139,
+      "grad_norm": 14.75,
+      "learning_rate": 1.313154467758946e-06,
+      "loss": 0.6269,
+      "step": 11409
+    },
+    {
+      "epoch": 0.9723050703025139,
+      "grad_norm": 25.375,
+      "learning_rate": 1.3129866462417047e-06,
+      "loss": 1.1742,
+      "step": 11410
+    },
+    {
+      "epoch": 0.9723902854708139,
+      "grad_norm": 13.125,
+      "learning_rate": 1.3128188235862308e-06,
+      "loss": 0.5059,
+      "step": 11411
+    },
+    {
+      "epoch": 0.9724755006391138,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3126509997955579e-06,
+      "loss": 1.0356,
+      "step": 11412
+    },
+    {
+      "epoch": 0.9725607158074138,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3124831748727183e-06,
+      "loss": 0.7453,
+      "step": 11413
+    },
+    {
+      "epoch": 0.9726459309757137,
+      "grad_norm": 14.875,
+      "learning_rate": 1.312315348820745e-06,
+      "loss": 0.6555,
+      "step": 11414
+    },
+    {
+      "epoch": 0.9727311461440137,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3121475216426708e-06,
+      "loss": 0.7213,
+      "step": 11415
+    },
+    {
+      "epoch": 0.9728163613123136,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.311979693341528e-06,
+      "loss": 0.7463,
+      "step": 11416
+    },
+    {
+      "epoch": 0.9729015764806136,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.31181186392035e-06,
+      "loss": 0.66,
+      "step": 11417
+    },
+    {
+      "epoch": 0.9729867916489136,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3116440333821699e-06,
+      "loss": 0.9052,
+      "step": 11418
+    },
+    {
+      "epoch": 0.9730720068172135,
+      "grad_norm": 16.5,
+      "learning_rate": 1.3114762017300194e-06,
+      "loss": 0.9488,
+      "step": 11419
+    },
+    {
+      "epoch": 0.9731572219855135,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3113083689669325e-06,
+      "loss": 0.7577,
+      "step": 11420
+    },
+    {
+      "epoch": 0.9732424371538134,
+      "grad_norm": 11.125,
+      "learning_rate": 1.3111405350959417e-06,
+      "loss": 0.3497,
+      "step": 11421
+    },
+    {
+      "epoch": 0.9733276523221134,
+      "grad_norm": 16.5,
+      "learning_rate": 1.3109727001200801e-06,
+      "loss": 0.622,
+      "step": 11422
+    },
+    {
+      "epoch": 0.9734128674904133,
+      "grad_norm": 14.5,
+      "learning_rate": 1.3108048640423803e-06,
+      "loss": 0.5931,
+      "step": 11423
+    },
+    {
+      "epoch": 0.9734980826587133,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3106370268658752e-06,
+      "loss": 0.8682,
+      "step": 11424
+    },
+    {
+      "epoch": 0.9735832978270132,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.3104691885935986e-06,
+      "loss": 0.5822,
+      "step": 11425
+    },
+    {
+      "epoch": 0.9736685129953132,
+      "grad_norm": 27.25,
+      "learning_rate": 1.3103013492285827e-06,
+      "loss": 1.0992,
+      "step": 11426
+    },
+    {
+      "epoch": 0.9737537281636132,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.3101335087738607e-06,
+      "loss": 0.7458,
+      "step": 11427
+    },
+    {
+      "epoch": 0.9738389433319131,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.3099656672324659e-06,
+      "loss": 0.6046,
+      "step": 11428
+    },
+    {
+      "epoch": 0.9739241585002131,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.309797824607431e-06,
+      "loss": 0.9732,
+      "step": 11429
+    },
+    {
+      "epoch": 0.974009373668513,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.3096299809017896e-06,
+      "loss": 0.8161,
+      "step": 11430
+    },
+    {
+      "epoch": 0.974094588836813,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3094621361185738e-06,
+      "loss": 0.5617,
+      "step": 11431
+    },
+    {
+      "epoch": 0.974179804005113,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.309294290260818e-06,
+      "loss": 0.6842,
+      "step": 11432
+    },
+    {
+      "epoch": 0.9742650191734129,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.3091264433315545e-06,
+      "loss": 0.8921,
+      "step": 11433
+    },
+    {
+      "epoch": 0.9743502343417129,
+      "grad_norm": 11.75,
+      "learning_rate": 1.308958595333817e-06,
+      "loss": 0.4538,
+      "step": 11434
+    },
+    {
+      "epoch": 0.9744354495100128,
+      "grad_norm": 13.25,
+      "learning_rate": 1.3087907462706384e-06,
+      "loss": 0.5017,
+      "step": 11435
+    },
+    {
+      "epoch": 0.9745206646783128,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.3086228961450517e-06,
+      "loss": 0.928,
+      "step": 11436
+    },
+    {
+      "epoch": 0.9746058798466127,
+      "grad_norm": 18.875,
+      "learning_rate": 1.3084550449600903e-06,
+      "loss": 0.677,
+      "step": 11437
+    },
+    {
+      "epoch": 0.9746910950149127,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3082871927187877e-06,
+      "loss": 0.6583,
+      "step": 11438
+    },
+    {
+      "epoch": 0.9747763101832126,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.3081193394241765e-06,
+      "loss": 0.7183,
+      "step": 11439
+    },
+    {
+      "epoch": 0.9748615253515126,
+      "grad_norm": 16.5,
+      "learning_rate": 1.3079514850792906e-06,
+      "loss": 0.9364,
+      "step": 11440
+    },
+    {
+      "epoch": 0.9749467405198126,
+      "grad_norm": 11.5,
+      "learning_rate": 1.307783629687163e-06,
+      "loss": 0.369,
+      "step": 11441
+    },
+    {
+      "epoch": 0.9750319556881125,
+      "grad_norm": 15.0,
+      "learning_rate": 1.3076157732508271e-06,
+      "loss": 0.5203,
+      "step": 11442
+    },
+    {
+      "epoch": 0.9751171708564125,
+      "grad_norm": 21.625,
+      "learning_rate": 1.3074479157733165e-06,
+      "loss": 0.7108,
+      "step": 11443
+    },
+    {
+      "epoch": 0.9752023860247124,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3072800572576638e-06,
+      "loss": 0.8176,
+      "step": 11444
+    },
+    {
+      "epoch": 0.9752876011930124,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3071121977069035e-06,
+      "loss": 0.8551,
+      "step": 11445
+    },
+    {
+      "epoch": 0.9753728163613123,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.306944337124068e-06,
+      "loss": 0.492,
+      "step": 11446
+    },
+    {
+      "epoch": 0.9754580315296123,
+      "grad_norm": 11.625,
+      "learning_rate": 1.3067764755121911e-06,
+      "loss": 0.4828,
+      "step": 11447
+    },
+    {
+      "epoch": 0.9755432466979123,
+      "grad_norm": 13.0,
+      "learning_rate": 1.3066086128743062e-06,
+      "loss": 0.5425,
+      "step": 11448
+    },
+    {
+      "epoch": 0.9756284618662122,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3064407492134468e-06,
+      "loss": 0.8659,
+      "step": 11449
+    },
+    {
+      "epoch": 0.9757136770345122,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.306272884532646e-06,
+      "loss": 0.4714,
+      "step": 11450
+    },
+    {
+      "epoch": 0.9757988922028121,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3061050188349382e-06,
+      "loss": 0.7596,
+      "step": 11451
+    },
+    {
+      "epoch": 0.9758841073711121,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3059371521233557e-06,
+      "loss": 0.8078,
+      "step": 11452
+    },
+    {
+      "epoch": 0.975969322539412,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.3057692844009331e-06,
+      "loss": 0.4329,
+      "step": 11453
+    },
+    {
+      "epoch": 0.976054537707712,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3056014156707031e-06,
+      "loss": 0.8311,
+      "step": 11454
+    },
+    {
+      "epoch": 0.976139752876012,
+      "grad_norm": 14.875,
+      "learning_rate": 1.3054335459356998e-06,
+      "loss": 0.486,
+      "step": 11455
+    },
+    {
+      "epoch": 0.9762249680443119,
+      "grad_norm": 13.625,
+      "learning_rate": 1.3052656751989565e-06,
+      "loss": 0.5052,
+      "step": 11456
+    },
+    {
+      "epoch": 0.9763101832126119,
+      "grad_norm": 22.125,
+      "learning_rate": 1.305097803463507e-06,
+      "loss": 0.9126,
+      "step": 11457
+    },
+    {
+      "epoch": 0.9763953983809118,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.3049299307323842e-06,
+      "loss": 0.7523,
+      "step": 11458
+    },
+    {
+      "epoch": 0.9764806135492118,
+      "grad_norm": 23.125,
+      "learning_rate": 1.304762057008623e-06,
+      "loss": 0.6702,
+      "step": 11459
+    },
+    {
+      "epoch": 0.9765658287175117,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3045941822952562e-06,
+      "loss": 0.7718,
+      "step": 11460
+    },
+    {
+      "epoch": 0.9766510438858117,
+      "grad_norm": 14.375,
+      "learning_rate": 1.3044263065953173e-06,
+      "loss": 0.8307,
+      "step": 11461
+    },
+    {
+      "epoch": 0.9767362590541117,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.3042584299118407e-06,
+      "loss": 0.5395,
+      "step": 11462
+    },
+    {
+      "epoch": 0.9768214742224116,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.3040905522478598e-06,
+      "loss": 0.4624,
+      "step": 11463
+    },
+    {
+      "epoch": 0.9769066893907116,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.303922673606408e-06,
+      "loss": 0.7776,
+      "step": 11464
+    },
+    {
+      "epoch": 0.9769919045590115,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.303754793990519e-06,
+      "loss": 0.6031,
+      "step": 11465
+    },
+    {
+      "epoch": 0.9770771197273115,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3035869134032272e-06,
+      "loss": 1.0215,
+      "step": 11466
+    },
+    {
+      "epoch": 0.9771623348956114,
+      "grad_norm": 20.375,
+      "learning_rate": 1.303419031847566e-06,
+      "loss": 0.8326,
+      "step": 11467
+    },
+    {
+      "epoch": 0.9772475500639114,
+      "grad_norm": 16.75,
+      "learning_rate": 1.303251149326569e-06,
+      "loss": 0.9276,
+      "step": 11468
+    },
+    {
+      "epoch": 0.9773327652322114,
+      "grad_norm": 14.375,
+      "learning_rate": 1.3030832658432702e-06,
+      "loss": 0.7195,
+      "step": 11469
+    },
+    {
+      "epoch": 0.9774179804005113,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3029153814007034e-06,
+      "loss": 0.6745,
+      "step": 11470
+    },
+    {
+      "epoch": 0.9775031955688113,
+      "grad_norm": 10.375,
+      "learning_rate": 1.3027474960019026e-06,
+      "loss": 0.3661,
+      "step": 11471
+    },
+    {
+      "epoch": 0.9775884107371112,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3025796096499016e-06,
+      "loss": 0.5098,
+      "step": 11472
+    },
+    {
+      "epoch": 0.9776736259054112,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3024117223477337e-06,
+      "loss": 0.7546,
+      "step": 11473
+    },
+    {
+      "epoch": 0.9777588410737111,
+      "grad_norm": 19.375,
+      "learning_rate": 1.3022438340984337e-06,
+      "loss": 0.9314,
+      "step": 11474
+    },
+    {
+      "epoch": 0.9778440562420111,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.302075944905035e-06,
+      "loss": 0.6504,
+      "step": 11475
+    },
+    {
+      "epoch": 0.977929271410311,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.3019080547705713e-06,
+      "loss": 0.478,
+      "step": 11476
+    },
+    {
+      "epoch": 0.978014486578611,
+      "grad_norm": 13.75,
+      "learning_rate": 1.3017401636980772e-06,
+      "loss": 0.4074,
+      "step": 11477
+    },
+    {
+      "epoch": 0.978099701746911,
+      "grad_norm": 15.75,
+      "learning_rate": 1.3015722716905862e-06,
+      "loss": 1.022,
+      "step": 11478
+    },
+    {
+      "epoch": 0.9781849169152109,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.3014043787511327e-06,
+      "loss": 0.4357,
+      "step": 11479
+    },
+    {
+      "epoch": 0.9782701320835109,
+      "grad_norm": 13.5,
+      "learning_rate": 1.3012364848827497e-06,
+      "loss": 0.9168,
+      "step": 11480
+    },
+    {
+      "epoch": 0.9783553472518108,
+      "grad_norm": 15.125,
+      "learning_rate": 1.3010685900884723e-06,
+      "loss": 0.853,
+      "step": 11481
+    },
+    {
+      "epoch": 0.9784405624201108,
+      "grad_norm": 26.375,
+      "learning_rate": 1.300900694371334e-06,
+      "loss": 0.9446,
+      "step": 11482
+    },
+    {
+      "epoch": 0.9785257775884108,
+      "grad_norm": 18.625,
+      "learning_rate": 1.3007327977343695e-06,
+      "loss": 0.9474,
+      "step": 11483
+    },
+    {
+      "epoch": 0.9786109927567107,
+      "grad_norm": 18.625,
+      "learning_rate": 1.3005649001806118e-06,
+      "loss": 0.8374,
+      "step": 11484
+    },
+    {
+      "epoch": 0.9786962079250107,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3003970017130958e-06,
+      "loss": 0.4643,
+      "step": 11485
+    },
+    {
+      "epoch": 0.9787814230933106,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3002291023348545e-06,
+      "loss": 0.6401,
+      "step": 11486
+    },
+    {
+      "epoch": 0.9788666382616106,
+      "grad_norm": 11.875,
+      "learning_rate": 1.300061202048924e-06,
+      "loss": 0.5251,
+      "step": 11487
+    },
+    {
+      "epoch": 0.9789518534299105,
+      "grad_norm": 20.0,
+      "learning_rate": 1.2998933008583365e-06,
+      "loss": 1.1831,
+      "step": 11488
+    },
+    {
+      "epoch": 0.9790370685982105,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2997253987661272e-06,
+      "loss": 0.9766,
+      "step": 11489
+    },
+    {
+      "epoch": 0.9791222837665104,
+      "grad_norm": 22.375,
+      "learning_rate": 1.2995574957753294e-06,
+      "loss": 1.1832,
+      "step": 11490
+    },
+    {
+      "epoch": 0.9792074989348104,
+      "grad_norm": 18.375,
+      "learning_rate": 1.2993895918889783e-06,
+      "loss": 0.8554,
+      "step": 11491
+    },
+    {
+      "epoch": 0.9792927141031104,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2992216871101077e-06,
+      "loss": 0.9373,
+      "step": 11492
+    },
+    {
+      "epoch": 0.9793779292714103,
+      "grad_norm": 13.125,
+      "learning_rate": 1.2990537814417513e-06,
+      "loss": 0.8313,
+      "step": 11493
+    },
+    {
+      "epoch": 0.9794631444397103,
+      "grad_norm": 14.0,
+      "learning_rate": 1.298885874886944e-06,
+      "loss": 0.8489,
+      "step": 11494
+    },
+    {
+      "epoch": 0.9795483596080102,
+      "grad_norm": 15.375,
+      "learning_rate": 1.29871796744872e-06,
+      "loss": 0.978,
+      "step": 11495
+    },
+    {
+      "epoch": 0.9796335747763102,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.298550059130113e-06,
+      "loss": 0.6944,
+      "step": 11496
+    },
+    {
+      "epoch": 0.9797187899446101,
+      "grad_norm": 15.75,
+      "learning_rate": 1.2983821499341576e-06,
+      "loss": 1.0332,
+      "step": 11497
+    },
+    {
+      "epoch": 0.9798040051129101,
+      "grad_norm": 23.875,
+      "learning_rate": 1.298214239863888e-06,
+      "loss": 1.2074,
+      "step": 11498
+    },
+    {
+      "epoch": 0.9798892202812101,
+      "grad_norm": 38.75,
+      "learning_rate": 1.298046328922339e-06,
+      "loss": 1.1794,
+      "step": 11499
+    },
+    {
+      "epoch": 0.97997443544951,
+      "grad_norm": 11.0,
+      "learning_rate": 1.2978784171125445e-06,
+      "loss": 0.5342,
+      "step": 11500
+    },
+    {
+      "epoch": 0.98005965061781,
+      "grad_norm": 17.875,
+      "learning_rate": 1.2977105044375385e-06,
+      "loss": 0.836,
+      "step": 11501
+    },
+    {
+      "epoch": 0.9801448657861099,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2975425909003558e-06,
+      "loss": 0.6023,
+      "step": 11502
+    },
+    {
+      "epoch": 0.9802300809544099,
+      "grad_norm": 13.375,
+      "learning_rate": 1.297374676504031e-06,
+      "loss": 0.4707,
+      "step": 11503
+    },
+    {
+      "epoch": 0.9803152961227098,
+      "grad_norm": 12.0,
+      "learning_rate": 1.297206761251598e-06,
+      "loss": 0.471,
+      "step": 11504
+    },
+    {
+      "epoch": 0.9804005112910098,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2970388451460912e-06,
+      "loss": 0.6333,
+      "step": 11505
+    },
+    {
+      "epoch": 0.9804857264593098,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.296870928190545e-06,
+      "loss": 0.7306,
+      "step": 11506
+    },
+    {
+      "epoch": 0.9805709416276097,
+      "grad_norm": 19.25,
+      "learning_rate": 1.2967030103879946e-06,
+      "loss": 1.1319,
+      "step": 11507
+    },
+    {
+      "epoch": 0.9806561567959097,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.2965350917414735e-06,
+      "loss": 0.795,
+      "step": 11508
+    },
+    {
+      "epoch": 0.9807413719642096,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.2963671722540164e-06,
+      "loss": 0.8793,
+      "step": 11509
+    },
+    {
+      "epoch": 0.9808265871325096,
+      "grad_norm": 11.875,
+      "learning_rate": 1.2961992519286578e-06,
+      "loss": 0.3831,
+      "step": 11510
+    },
+    {
+      "epoch": 0.9809118023008095,
+      "grad_norm": 14.0,
+      "learning_rate": 1.2960313307684324e-06,
+      "loss": 0.8376,
+      "step": 11511
+    },
+    {
+      "epoch": 0.9809970174691095,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2958634087763747e-06,
+      "loss": 0.8669,
+      "step": 11512
+    },
+    {
+      "epoch": 0.9810822326374095,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2956954859555188e-06,
+      "loss": 0.8526,
+      "step": 11513
+    },
+    {
+      "epoch": 0.9811674478057094,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2955275623088995e-06,
+      "loss": 0.6765,
+      "step": 11514
+    },
+    {
+      "epoch": 0.9812526629740094,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2953596378395517e-06,
+      "loss": 1.0282,
+      "step": 11515
+    },
+    {
+      "epoch": 0.9813378781423093,
+      "grad_norm": 19.0,
+      "learning_rate": 1.2951917125505095e-06,
+      "loss": 0.6992,
+      "step": 11516
+    },
+    {
+      "epoch": 0.9814230933106093,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2950237864448077e-06,
+      "loss": 0.7495,
+      "step": 11517
+    },
+    {
+      "epoch": 0.9815083084789092,
+      "grad_norm": 13.0,
+      "learning_rate": 1.2948558595254803e-06,
+      "loss": 0.7088,
+      "step": 11518
+    },
+    {
+      "epoch": 0.9815935236472092,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.294687931795563e-06,
+      "loss": 0.5023,
+      "step": 11519
+    },
+    {
+      "epoch": 0.9816787388155092,
+      "grad_norm": 15.0,
+      "learning_rate": 1.2945200032580898e-06,
+      "loss": 0.812,
+      "step": 11520
+    },
+    {
+      "epoch": 0.9817639539838091,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2943520739160953e-06,
+      "loss": 0.9937,
+      "step": 11521
+    },
+    {
+      "epoch": 0.9818491691521091,
+      "grad_norm": 18.5,
+      "learning_rate": 1.294184143772614e-06,
+      "loss": 1.0094,
+      "step": 11522
+    },
+    {
+      "epoch": 0.981934384320409,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.294016212830681e-06,
+      "loss": 0.5905,
+      "step": 11523
+    },
+    {
+      "epoch": 0.982019599488709,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2938482810933307e-06,
+      "loss": 0.8264,
+      "step": 11524
+    },
+    {
+      "epoch": 0.9821048146570089,
+      "grad_norm": 11.25,
+      "learning_rate": 1.293680348563598e-06,
+      "loss": 0.3662,
+      "step": 11525
+    },
+    {
+      "epoch": 0.9821900298253089,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2935124152445173e-06,
+      "loss": 0.5752,
+      "step": 11526
+    },
+    {
+      "epoch": 0.9822752449936089,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2933444811391235e-06,
+      "loss": 0.698,
+      "step": 11527
+    },
+    {
+      "epoch": 0.9823604601619088,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2931765462504513e-06,
+      "loss": 0.5559,
+      "step": 11528
+    },
+    {
+      "epoch": 0.9824456753302088,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2930086105815354e-06,
+      "loss": 0.9016,
+      "step": 11529
+    },
+    {
+      "epoch": 0.9825308904985087,
+      "grad_norm": 13.25,
+      "learning_rate": 1.292840674135411e-06,
+      "loss": 0.4944,
+      "step": 11530
+    },
+    {
+      "epoch": 0.9826161056668087,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.292672736915112e-06,
+      "loss": 0.6946,
+      "step": 11531
+    },
+    {
+      "epoch": 0.9827013208351086,
+      "grad_norm": 11.875,
+      "learning_rate": 1.2925047989236748e-06,
+      "loss": 0.4064,
+      "step": 11532
+    },
+    {
+      "epoch": 0.9827865360034086,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.292336860164132e-06,
+      "loss": 0.6902,
+      "step": 11533
+    },
+    {
+      "epoch": 0.9828717511717086,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2921689206395203e-06,
+      "loss": 0.7277,
+      "step": 11534
+    },
+    {
+      "epoch": 0.9829569663400085,
+      "grad_norm": 20.75,
+      "learning_rate": 1.2920009803528733e-06,
+      "loss": 1.0278,
+      "step": 11535
+    },
+    {
+      "epoch": 0.9830421815083085,
+      "grad_norm": 20.875,
+      "learning_rate": 1.2918330393072269e-06,
+      "loss": 1.1821,
+      "step": 11536
+    },
+    {
+      "epoch": 0.9831273966766084,
+      "grad_norm": 11.25,
+      "learning_rate": 1.2916650975056147e-06,
+      "loss": 0.4462,
+      "step": 11537
+    },
+    {
+      "epoch": 0.9832126118449084,
+      "grad_norm": 18.0,
+      "learning_rate": 1.291497154951073e-06,
+      "loss": 1.034,
+      "step": 11538
+    },
+    {
+      "epoch": 0.9832978270132083,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2913292116466353e-06,
+      "loss": 0.6812,
+      "step": 11539
+    },
+    {
+      "epoch": 0.9833830421815083,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.2911612675953378e-06,
+      "loss": 0.3512,
+      "step": 11540
+    },
+    {
+      "epoch": 0.9834682573498083,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.2909933228002144e-06,
+      "loss": 0.5021,
+      "step": 11541
+    },
+    {
+      "epoch": 0.9835534725181082,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2908253772643006e-06,
+      "loss": 0.4651,
+      "step": 11542
+    },
+    {
+      "epoch": 0.9836386876864082,
+      "grad_norm": 13.625,
+      "learning_rate": 1.2906574309906312e-06,
+      "loss": 0.6579,
+      "step": 11543
+    },
+    {
+      "epoch": 0.9837239028547081,
+      "grad_norm": 17.0,
+      "learning_rate": 1.290489483982241e-06,
+      "loss": 0.7014,
+      "step": 11544
+    },
+    {
+      "epoch": 0.9838091180230081,
+      "grad_norm": 19.625,
+      "learning_rate": 1.2903215362421654e-06,
+      "loss": 1.1158,
+      "step": 11545
+    },
+    {
+      "epoch": 0.983894333191308,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.2901535877734385e-06,
+      "loss": 0.8292,
+      "step": 11546
+    },
+    {
+      "epoch": 0.983979548359608,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2899856385790964e-06,
+      "loss": 0.6922,
+      "step": 11547
+    },
+    {
+      "epoch": 0.984064763527908,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2898176886621736e-06,
+      "loss": 1.3498,
+      "step": 11548
+    },
+    {
+      "epoch": 0.9841499786962079,
+      "grad_norm": 19.125,
+      "learning_rate": 1.289649738025705e-06,
+      "loss": 0.9013,
+      "step": 11549
+    },
+    {
+      "epoch": 0.9842351938645079,
+      "grad_norm": 17.75,
+      "learning_rate": 1.2894817866727257e-06,
+      "loss": 0.8277,
+      "step": 11550
+    },
+    {
+      "epoch": 0.9843204090328078,
+      "grad_norm": 13.375,
+      "learning_rate": 1.289313834606271e-06,
+      "loss": 0.5901,
+      "step": 11551
+    },
+    {
+      "epoch": 0.9844056242011078,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2891458818293757e-06,
+      "loss": 0.7029,
+      "step": 11552
+    },
+    {
+      "epoch": 0.9844908393694077,
+      "grad_norm": 18.5,
+      "learning_rate": 1.288977928345075e-06,
+      "loss": 1.0259,
+      "step": 11553
+    },
+    {
+      "epoch": 0.9845760545377077,
+      "grad_norm": 14.875,
+      "learning_rate": 1.2888099741564036e-06,
+      "loss": 0.6933,
+      "step": 11554
+    },
+    {
+      "epoch": 0.9846612697060076,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2886420192663973e-06,
+      "loss": 1.0109,
+      "step": 11555
+    },
+    {
+      "epoch": 0.9847464848743076,
+      "grad_norm": 19.0,
+      "learning_rate": 1.288474063678091e-06,
+      "loss": 0.9019,
+      "step": 11556
+    },
+    {
+      "epoch": 0.9848317000426076,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2883061073945194e-06,
+      "loss": 0.7381,
+      "step": 11557
+    },
+    {
+      "epoch": 0.9849169152109075,
+      "grad_norm": 29.125,
+      "learning_rate": 1.2881381504187177e-06,
+      "loss": 1.1664,
+      "step": 11558
+    },
+    {
+      "epoch": 0.9850021303792075,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2879701927537218e-06,
+      "loss": 0.617,
+      "step": 11559
+    },
+    {
+      "epoch": 0.9850873455475074,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.287802234402566e-06,
+      "loss": 0.6309,
+      "step": 11560
+    },
+    {
+      "epoch": 0.9851725607158074,
+      "grad_norm": 11.25,
+      "learning_rate": 1.2876342753682856e-06,
+      "loss": 0.299,
+      "step": 11561
+    },
+    {
+      "epoch": 0.9852577758841073,
+      "grad_norm": 14.75,
+      "learning_rate": 1.2874663156539163e-06,
+      "loss": 0.4828,
+      "step": 11562
+    },
+    {
+      "epoch": 0.9853429910524073,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2872983552624932e-06,
+      "loss": 0.2755,
+      "step": 11563
+    },
+    {
+      "epoch": 0.9854282062207073,
+      "grad_norm": 24.125,
+      "learning_rate": 1.287130394197051e-06,
+      "loss": 1.0892,
+      "step": 11564
+    },
+    {
+      "epoch": 0.9855134213890072,
+      "grad_norm": 15.75,
+      "learning_rate": 1.2869624324606252e-06,
+      "loss": 0.3957,
+      "step": 11565
+    },
+    {
+      "epoch": 0.9855986365573072,
+      "grad_norm": 12.0,
+      "learning_rate": 1.2867944700562512e-06,
+      "loss": 0.58,
+      "step": 11566
+    },
+    {
+      "epoch": 0.9856838517256071,
+      "grad_norm": 18.0,
+      "learning_rate": 1.286626506986964e-06,
+      "loss": 0.8879,
+      "step": 11567
+    },
+    {
+      "epoch": 0.9857690668939071,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.2864585432557995e-06,
+      "loss": 0.8339,
+      "step": 11568
+    },
+    {
+      "epoch": 0.985854282062207,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2862905788657918e-06,
+      "loss": 0.5279,
+      "step": 11569
+    },
+    {
+      "epoch": 0.985939497230507,
+      "grad_norm": 54.25,
+      "learning_rate": 1.2861226138199773e-06,
+      "loss": 0.5374,
+      "step": 11570
+    },
+    {
+      "epoch": 0.986024712398807,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.2859546481213905e-06,
+      "loss": 0.7737,
+      "step": 11571
+    },
+    {
+      "epoch": 0.9861099275671069,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2857866817730674e-06,
+      "loss": 0.7286,
+      "step": 11572
+    },
+    {
+      "epoch": 0.9861951427354069,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.2856187147780427e-06,
+      "loss": 0.8106,
+      "step": 11573
+    },
+    {
+      "epoch": 0.9862803579037068,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2854507471393521e-06,
+      "loss": 0.7582,
+      "step": 11574
+    },
+    {
+      "epoch": 0.9863655730720068,
+      "grad_norm": 14.75,
+      "learning_rate": 1.2852827788600309e-06,
+      "loss": 0.5743,
+      "step": 11575
+    },
+    {
+      "epoch": 0.9864507882403067,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2851148099431142e-06,
+      "loss": 0.6944,
+      "step": 11576
+    },
+    {
+      "epoch": 0.9865360034086067,
+      "grad_norm": 10.75,
+      "learning_rate": 1.284946840391638e-06,
+      "loss": 0.4466,
+      "step": 11577
+    },
+    {
+      "epoch": 0.9866212185769067,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.2847788702086369e-06,
+      "loss": 0.5045,
+      "step": 11578
+    },
+    {
+      "epoch": 0.9867064337452066,
+      "grad_norm": 17.875,
+      "learning_rate": 1.2846108993971464e-06,
+      "loss": 0.5855,
+      "step": 11579
+    },
+    {
+      "epoch": 0.9867916489135066,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.2844429279602028e-06,
+      "loss": 0.913,
+      "step": 11580
+    },
+    {
+      "epoch": 0.9868768640818065,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2842749559008405e-06,
+      "loss": 0.6198,
+      "step": 11581
+    },
+    {
+      "epoch": 0.9869620792501065,
+      "grad_norm": 15.25,
+      "learning_rate": 1.2841069832220952e-06,
+      "loss": 0.7165,
+      "step": 11582
+    },
+    {
+      "epoch": 0.9870472944184064,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.2839390099270024e-06,
+      "loss": 0.624,
+      "step": 11583
+    },
+    {
+      "epoch": 0.9871325095867064,
+      "grad_norm": 18.125,
+      "learning_rate": 1.2837710360185975e-06,
+      "loss": 1.0329,
+      "step": 11584
+    },
+    {
+      "epoch": 0.9872177247550064,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.2836030614999168e-06,
+      "loss": 0.7766,
+      "step": 11585
+    },
+    {
+      "epoch": 0.9873029399233063,
+      "grad_norm": 14.25,
+      "learning_rate": 1.2834350863739941e-06,
+      "loss": 0.319,
+      "step": 11586
+    },
+    {
+      "epoch": 0.9873881550916063,
+      "grad_norm": 18.5,
+      "learning_rate": 1.283267110643866e-06,
+      "loss": 0.7883,
+      "step": 11587
+    },
+    {
+      "epoch": 0.9874733702599062,
+      "grad_norm": 25.75,
+      "learning_rate": 1.2830991343125675e-06,
+      "loss": 0.6929,
+      "step": 11588
+    },
+    {
+      "epoch": 0.9875585854282062,
+      "grad_norm": 15.375,
+      "learning_rate": 1.2829311573831351e-06,
+      "loss": 0.7721,
+      "step": 11589
+    },
+    {
+      "epoch": 0.9876438005965061,
+      "grad_norm": 13.25,
+      "learning_rate": 1.2827631798586032e-06,
+      "loss": 0.4744,
+      "step": 11590
+    },
+    {
+      "epoch": 0.9877290157648061,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2825952017420077e-06,
+      "loss": 0.911,
+      "step": 11591
+    },
+    {
+      "epoch": 0.987814230933106,
+      "grad_norm": 12.125,
+      "learning_rate": 1.2824272230363838e-06,
+      "loss": 0.5891,
+      "step": 11592
+    },
+    {
+      "epoch": 0.987899446101406,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2822592437447678e-06,
+      "loss": 0.5944,
+      "step": 11593
+    },
+    {
+      "epoch": 0.987984661269706,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2820912638701946e-06,
+      "loss": 0.986,
+      "step": 11594
+    },
+    {
+      "epoch": 0.9880698764380059,
+      "grad_norm": 15.125,
+      "learning_rate": 1.2819232834157003e-06,
+      "loss": 0.8919,
+      "step": 11595
+    },
+    {
+      "epoch": 0.9881550916063059,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2817553023843196e-06,
+      "loss": 0.9322,
+      "step": 11596
+    },
+    {
+      "epoch": 0.9882403067746058,
+      "grad_norm": 12.875,
+      "learning_rate": 1.2815873207790891e-06,
+      "loss": 0.4923,
+      "step": 11597
+    },
+    {
+      "epoch": 0.9883255219429058,
+      "grad_norm": 21.125,
+      "learning_rate": 1.2814193386030438e-06,
+      "loss": 0.9952,
+      "step": 11598
+    },
+    {
+      "epoch": 0.9884107371112058,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2812513558592194e-06,
+      "loss": 1.0879,
+      "step": 11599
+    },
+    {
+      "epoch": 0.9884959522795057,
+      "grad_norm": 20.125,
+      "learning_rate": 1.2810833725506516e-06,
+      "loss": 0.7975,
+      "step": 11600
+    },
+    {
+      "epoch": 0.9885811674478057,
+      "grad_norm": 23.875,
+      "learning_rate": 1.2809153886803763e-06,
+      "loss": 1.2124,
+      "step": 11601
+    },
+    {
+      "epoch": 0.9886663826161056,
+      "grad_norm": 12.5,
+      "learning_rate": 1.2807474042514287e-06,
+      "loss": 0.2909,
+      "step": 11602
+    },
+    {
+      "epoch": 0.9887515977844056,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2805794192668442e-06,
+      "loss": 0.5975,
+      "step": 11603
+    },
+    {
+      "epoch": 0.9888368129527055,
+      "grad_norm": 15.375,
+      "learning_rate": 1.280411433729659e-06,
+      "loss": 0.7363,
+      "step": 11604
+    },
+    {
+      "epoch": 0.9889220281210055,
+      "grad_norm": 17.625,
+      "learning_rate": 1.280243447642909e-06,
+      "loss": 0.9198,
+      "step": 11605
+    },
+    {
+      "epoch": 0.9890072432893054,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.2800754610096293e-06,
+      "loss": 0.5472,
+      "step": 11606
+    },
+    {
+      "epoch": 0.9890924584576054,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2799074738328556e-06,
+      "loss": 0.9422,
+      "step": 11607
+    },
+    {
+      "epoch": 0.9891776736259054,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.279739486115624e-06,
+      "loss": 0.758,
+      "step": 11608
+    },
+    {
+      "epoch": 0.9892628887942053,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2795714978609699e-06,
+      "loss": 0.5117,
+      "step": 11609
+    },
+    {
+      "epoch": 0.9893481039625053,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2794035090719293e-06,
+      "loss": 0.4344,
+      "step": 11610
+    },
+    {
+      "epoch": 0.9894333191308052,
+      "grad_norm": 20.75,
+      "learning_rate": 1.2792355197515371e-06,
+      "loss": 0.8544,
+      "step": 11611
+    },
+    {
+      "epoch": 0.9895185342991052,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2790675299028302e-06,
+      "loss": 0.9502,
+      "step": 11612
+    },
+    {
+      "epoch": 0.9896037494674051,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2788995395288436e-06,
+      "loss": 0.3178,
+      "step": 11613
+    },
+    {
+      "epoch": 0.9896889646357051,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2787315486326131e-06,
+      "loss": 0.653,
+      "step": 11614
+    },
+    {
+      "epoch": 0.9897741798040051,
+      "grad_norm": 19.125,
+      "learning_rate": 1.278563557217175e-06,
+      "loss": 0.8924,
+      "step": 11615
+    },
+    {
+      "epoch": 0.989859394972305,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.2783955652855645e-06,
+      "loss": 0.6874,
+      "step": 11616
+    },
+    {
+      "epoch": 0.989944610140605,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2782275728408178e-06,
+      "loss": 0.9195,
+      "step": 11617
+    },
+    {
+      "epoch": 0.9900298253089049,
+      "grad_norm": 12.625,
+      "learning_rate": 1.27805957988597e-06,
+      "loss": 0.3683,
+      "step": 11618
+    },
+    {
+      "epoch": 0.9901150404772049,
+      "grad_norm": 14.875,
+      "learning_rate": 1.277891586424058e-06,
+      "loss": 0.895,
+      "step": 11619
+    },
+    {
+      "epoch": 0.9902002556455048,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.2777235924581164e-06,
+      "loss": 0.4849,
+      "step": 11620
+    },
+    {
+      "epoch": 0.9902854708138048,
+      "grad_norm": 13.25,
+      "learning_rate": 1.277555597991182e-06,
+      "loss": 0.5929,
+      "step": 11621
+    },
+    {
+      "epoch": 0.9903706859821048,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2773876030262899e-06,
+      "loss": 0.846,
+      "step": 11622
+    },
+    {
+      "epoch": 0.9904559011504048,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.2772196075664767e-06,
+      "loss": 0.773,
+      "step": 11623
+    },
+    {
+      "epoch": 0.9905411163187048,
+      "grad_norm": 15.375,
+      "learning_rate": 1.277051611614777e-06,
+      "loss": 0.9419,
+      "step": 11624
+    },
+    {
+      "epoch": 0.9906263314870047,
+      "grad_norm": 16.0,
+      "learning_rate": 1.2768836151742287e-06,
+      "loss": 0.9098,
+      "step": 11625
+    },
+    {
+      "epoch": 0.9907115466553047,
+      "grad_norm": 17.125,
+      "learning_rate": 1.2767156182478655e-06,
+      "loss": 0.92,
+      "step": 11626
+    },
+    {
+      "epoch": 0.9907967618236047,
+      "grad_norm": 20.5,
+      "learning_rate": 1.2765476208387247e-06,
+      "loss": 0.9697,
+      "step": 11627
+    },
+    {
+      "epoch": 0.9908819769919046,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2763796229498413e-06,
+      "loss": 0.782,
+      "step": 11628
+    },
+    {
+      "epoch": 0.9909671921602046,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.2762116245842518e-06,
+      "loss": 0.7664,
+      "step": 11629
+    },
+    {
+      "epoch": 0.9910524073285045,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.276043625744992e-06,
+      "loss": 0.64,
+      "step": 11630
+    },
+    {
+      "epoch": 0.9911376224968045,
+      "grad_norm": 23.75,
+      "learning_rate": 1.2758756264350978e-06,
+      "loss": 1.0513,
+      "step": 11631
+    },
+    {
+      "epoch": 0.9912228376651044,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2757076266576042e-06,
+      "loss": 0.7448,
+      "step": 11632
+    },
+    {
+      "epoch": 0.9913080528334044,
+      "grad_norm": 31.5,
+      "learning_rate": 1.275539626415549e-06,
+      "loss": 1.2298,
+      "step": 11633
+    },
+    {
+      "epoch": 0.9913932680017044,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.2753716257119663e-06,
+      "loss": 0.4897,
+      "step": 11634
+    },
+    {
+      "epoch": 0.9914784831700043,
+      "grad_norm": 21.25,
+      "learning_rate": 1.275203624549893e-06,
+      "loss": 0.9425,
+      "step": 11635
+    },
+    {
+      "epoch": 0.9915636983383043,
+      "grad_norm": 14.5,
+      "learning_rate": 1.275035622932365e-06,
+      "loss": 0.6666,
+      "step": 11636
+    },
+    {
+      "epoch": 0.9916489135066042,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2748676208624184e-06,
+      "loss": 1.1565,
+      "step": 11637
+    },
+    {
+      "epoch": 0.9917341286749042,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.2746996183430885e-06,
+      "loss": 0.3345,
+      "step": 11638
+    },
+    {
+      "epoch": 0.9918193438432041,
+      "grad_norm": 18.5,
+      "learning_rate": 1.2745316153774115e-06,
+      "loss": 0.8908,
+      "step": 11639
+    },
+    {
+      "epoch": 0.9919045590115041,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2743636119684238e-06,
+      "loss": 0.9383,
+      "step": 11640
+    },
+    {
+      "epoch": 0.991989774179804,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2741956081191612e-06,
+      "loss": 0.9483,
+      "step": 11641
+    },
+    {
+      "epoch": 0.992074989348104,
+      "grad_norm": 20.125,
+      "learning_rate": 1.2740276038326595e-06,
+      "loss": 0.8267,
+      "step": 11642
+    },
+    {
+      "epoch": 0.992160204516404,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.2738595991119548e-06,
+      "loss": 0.7569,
+      "step": 11643
+    },
+    {
+      "epoch": 0.9922454196847039,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.2736915939600833e-06,
+      "loss": 0.5399,
+      "step": 11644
+    },
+    {
+      "epoch": 0.9923306348530039,
+      "grad_norm": 11.375,
+      "learning_rate": 1.2735235883800808e-06,
+      "loss": 0.4138,
+      "step": 11645
+    },
+    {
+      "epoch": 0.9924158500213038,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2733555823749832e-06,
+      "loss": 0.5622,
+      "step": 11646
+    },
+    {
+      "epoch": 0.9925010651896038,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2731875759478272e-06,
+      "loss": 0.8255,
+      "step": 11647
+    },
+    {
+      "epoch": 0.9925862803579037,
+      "grad_norm": 16.375,
+      "learning_rate": 1.273019569101648e-06,
+      "loss": 0.4036,
+      "step": 11648
+    },
+    {
+      "epoch": 0.9926714955262037,
+      "grad_norm": 22.25,
+      "learning_rate": 1.2728515618394824e-06,
+      "loss": 0.6206,
+      "step": 11649
+    },
+    {
+      "epoch": 0.9927567106945037,
+      "grad_norm": 13.125,
+      "learning_rate": 1.2726835541643656e-06,
+      "loss": 0.6498,
+      "step": 11650
+    },
+    {
+      "epoch": 0.9928419258628036,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.2725155460793345e-06,
+      "loss": 0.3856,
+      "step": 11651
+    },
+    {
+      "epoch": 0.9929271410311036,
+      "grad_norm": 13.875,
+      "learning_rate": 1.2723475375874244e-06,
+      "loss": 0.4691,
+      "step": 11652
+    },
+    {
+      "epoch": 0.9930123561994035,
+      "grad_norm": 25.5,
+      "learning_rate": 1.2721795286916724e-06,
+      "loss": 0.7642,
+      "step": 11653
+    },
+    {
+      "epoch": 0.9930975713677035,
+      "grad_norm": 14.125,
+      "learning_rate": 1.2720115193951138e-06,
+      "loss": 0.5637,
+      "step": 11654
+    },
+    {
+      "epoch": 0.9931827865360034,
+      "grad_norm": 21.25,
+      "learning_rate": 1.271843509700785e-06,
+      "loss": 1.1254,
+      "step": 11655
+    },
+    {
+      "epoch": 0.9932680017043034,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.2716754996117217e-06,
+      "loss": 0.7872,
+      "step": 11656
+    },
+    {
+      "epoch": 0.9933532168726034,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.2715074891309606e-06,
+      "loss": 0.8337,
+      "step": 11657
+    },
+    {
+      "epoch": 0.9934384320409033,
+      "grad_norm": 14.375,
+      "learning_rate": 1.2713394782615376e-06,
+      "loss": 0.6906,
+      "step": 11658
+    },
+    {
+      "epoch": 0.9935236472092033,
+      "grad_norm": 19.125,
+      "learning_rate": 1.2711714670064886e-06,
+      "loss": 0.6924,
+      "step": 11659
+    },
+    {
+      "epoch": 0.9936088623775032,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.27100345536885e-06,
+      "loss": 0.4877,
+      "step": 11660
+    },
+    {
+      "epoch": 0.9936940775458032,
+      "grad_norm": 14.375,
+      "learning_rate": 1.270835443351658e-06,
+      "loss": 0.8702,
+      "step": 11661
+    },
+    {
+      "epoch": 0.9937792927141031,
+      "grad_norm": 13.875,
+      "learning_rate": 1.2706674309579483e-06,
+      "loss": 0.5319,
+      "step": 11662
+    },
+    {
+      "epoch": 0.9938645078824031,
+      "grad_norm": 14.75,
+      "learning_rate": 1.2704994181907577e-06,
+      "loss": 0.8544,
+      "step": 11663
+    },
+    {
+      "epoch": 0.9939497230507031,
+      "grad_norm": 9.5,
+      "learning_rate": 1.2703314050531213e-06,
+      "loss": 0.2873,
+      "step": 11664
+    },
+    {
+      "epoch": 0.994034938219003,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2701633915480765e-06,
+      "loss": 0.8987,
+      "step": 11665
+    },
+    {
+      "epoch": 0.994120153387303,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2699953776786589e-06,
+      "loss": 1.0721,
+      "step": 11666
+    },
+    {
+      "epoch": 0.9942053685556029,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.2698273634479044e-06,
+      "loss": 0.9995,
+      "step": 11667
+    },
+    {
+      "epoch": 0.9942905837239029,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.26965934885885e-06,
+      "loss": 1.0857,
+      "step": 11668
+    },
+    {
+      "epoch": 0.9943757988922028,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.269491333914531e-06,
+      "loss": 0.7042,
+      "step": 11669
+    },
+    {
+      "epoch": 0.9944610140605028,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2693233186179844e-06,
+      "loss": 1.1651,
+      "step": 11670
+    },
+    {
+      "epoch": 0.9945462292288028,
+      "grad_norm": 13.625,
+      "learning_rate": 1.2691553029722455e-06,
+      "loss": 0.5201,
+      "step": 11671
+    },
+    {
+      "epoch": 0.9946314443971027,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2689872869803514e-06,
+      "loss": 0.8782,
+      "step": 11672
+    },
+    {
+      "epoch": 0.9947166595654027,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2688192706453375e-06,
+      "loss": 0.7475,
+      "step": 11673
+    },
+    {
+      "epoch": 0.9948018747337026,
+      "grad_norm": 14.0,
+      "learning_rate": 1.2686512539702412e-06,
+      "loss": 0.5983,
+      "step": 11674
+    },
+    {
+      "epoch": 0.9948870899020026,
+      "grad_norm": 18.375,
+      "learning_rate": 1.2684832369580974e-06,
+      "loss": 0.536,
+      "step": 11675
+    },
+    {
+      "epoch": 0.9949723050703025,
+      "grad_norm": 12.5,
+      "learning_rate": 1.268315219611943e-06,
+      "loss": 0.5895,
+      "step": 11676
+    },
+    {
+      "epoch": 0.9950575202386025,
+      "grad_norm": 14.0,
+      "learning_rate": 1.268147201934814e-06,
+      "loss": 0.7835,
+      "step": 11677
+    },
+    {
+      "epoch": 0.9951427354069025,
+      "grad_norm": 10.125,
+      "learning_rate": 1.2679791839297476e-06,
+      "loss": 0.3089,
+      "step": 11678
+    },
+    {
+      "epoch": 0.9952279505752024,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2678111655997784e-06,
+      "loss": 0.4106,
+      "step": 11679
+    },
+    {
+      "epoch": 0.9953131657435024,
+      "grad_norm": 13.5,
+      "learning_rate": 1.267643146947944e-06,
+      "loss": 0.6568,
+      "step": 11680
+    },
+    {
+      "epoch": 0.9953983809118023,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2674751279772797e-06,
+      "loss": 0.426,
+      "step": 11681
+    },
+    {
+      "epoch": 0.9954835960801023,
+      "grad_norm": 14.125,
+      "learning_rate": 1.2673071086908226e-06,
+      "loss": 0.7429,
+      "step": 11682
+    },
+    {
+      "epoch": 0.9955688112484022,
+      "grad_norm": 13.75,
+      "learning_rate": 1.2671390890916089e-06,
+      "loss": 0.5406,
+      "step": 11683
+    },
+    {
+      "epoch": 0.9956540264167022,
+      "grad_norm": 17.625,
+      "learning_rate": 1.266971069182674e-06,
+      "loss": 1.0032,
+      "step": 11684
+    },
+    {
+      "epoch": 0.9957392415850022,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2668030489670551e-06,
+      "loss": 0.6928,
+      "step": 11685
+    },
+    {
+      "epoch": 0.9958244567533021,
+      "grad_norm": 13.875,
+      "learning_rate": 1.2666350284477885e-06,
+      "loss": 0.5017,
+      "step": 11686
+    },
+    {
+      "epoch": 0.9959096719216021,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2664670076279101e-06,
+      "loss": 0.8655,
+      "step": 11687
+    },
+    {
+      "epoch": 0.995994887089902,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2662989865104561e-06,
+      "loss": 0.7047,
+      "step": 11688
+    },
+    {
+      "epoch": 0.996080102258202,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2661309650984632e-06,
+      "loss": 0.5351,
+      "step": 11689
+    },
+    {
+      "epoch": 0.9961653174265019,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2659629433949674e-06,
+      "loss": 0.8598,
+      "step": 11690
+    },
+    {
+      "epoch": 0.9962505325948019,
+      "grad_norm": 10.75,
+      "learning_rate": 1.2657949214030055e-06,
+      "loss": 0.4937,
+      "step": 11691
+    },
+    {
+      "epoch": 0.9963357477631019,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.265626899125613e-06,
+      "loss": 0.3935,
+      "step": 11692
+    },
+    {
+      "epoch": 0.9964209629314018,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2654588765658268e-06,
+      "loss": 0.7739,
+      "step": 11693
+    },
+    {
+      "epoch": 0.9965061780997018,
+      "grad_norm": 19.75,
+      "learning_rate": 1.2652908537266833e-06,
+      "loss": 0.8091,
+      "step": 11694
+    },
+    {
+      "epoch": 0.9965913932680017,
+      "grad_norm": 9.875,
+      "learning_rate": 1.265122830611219e-06,
+      "loss": 0.3855,
+      "step": 11695
+    },
+    {
+      "epoch": 0.9966766084363017,
+      "grad_norm": 21.875,
+      "learning_rate": 1.2649548072224694e-06,
+      "loss": 1.1789,
+      "step": 11696
+    },
+    {
+      "epoch": 0.9967618236046016,
+      "grad_norm": 24.125,
+      "learning_rate": 1.2647867835634717e-06,
+      "loss": 0.8682,
+      "step": 11697
+    },
+    {
+      "epoch": 0.9968470387729016,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.264618759637262e-06,
+      "loss": 0.5722,
+      "step": 11698
+    },
+    {
+      "epoch": 0.9969322539412016,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2644507354468762e-06,
+      "loss": 0.8484,
+      "step": 11699
+    },
+    {
+      "epoch": 0.9970174691095015,
+      "grad_norm": 15.0,
+      "learning_rate": 1.2642827109953515e-06,
+      "loss": 0.9121,
+      "step": 11700
+    },
+    {
+      "epoch": 0.9971026842778015,
+      "grad_norm": 14.875,
+      "learning_rate": 1.2641146862857237e-06,
+      "loss": 1.1333,
+      "step": 11701
+    },
+    {
+      "epoch": 0.9971878994461014,
+      "grad_norm": 11.5,
+      "learning_rate": 1.2639466613210293e-06,
+      "loss": 0.5709,
+      "step": 11702
+    },
+    {
+      "epoch": 0.9972731146144014,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.2637786361043048e-06,
+      "loss": 0.6844,
+      "step": 11703
+    },
+    {
+      "epoch": 0.9973583297827013,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.2636106106385864e-06,
+      "loss": 0.6668,
+      "step": 11704
+    },
+    {
+      "epoch": 0.9974435449510013,
+      "grad_norm": 19.25,
+      "learning_rate": 1.2634425849269104e-06,
+      "loss": 0.8497,
+      "step": 11705
+    },
+    {
+      "epoch": 0.9975287601193012,
+      "grad_norm": 25.25,
+      "learning_rate": 1.2632745589723139e-06,
+      "loss": 0.8703,
+      "step": 11706
+    },
+    {
+      "epoch": 0.9976139752876012,
+      "grad_norm": 11.875,
+      "learning_rate": 1.2631065327778323e-06,
+      "loss": 0.4848,
+      "step": 11707
+    },
+    {
+      "epoch": 0.9976991904559012,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2629385063465027e-06,
+      "loss": 0.8481,
+      "step": 11708
+    },
+    {
+      "epoch": 0.9977844056242011,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.262770479681361e-06,
+      "loss": 0.5407,
+      "step": 11709
+    },
+    {
+      "epoch": 0.9978696207925011,
+      "grad_norm": 14.375,
+      "learning_rate": 1.2626024527854443e-06,
+      "loss": 0.5804,
+      "step": 11710
+    },
+    {
+      "epoch": 0.997954835960801,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.2624344256617886e-06,
+      "loss": 0.7516,
+      "step": 11711
+    },
+    {
+      "epoch": 0.998040051129101,
+      "grad_norm": 13.75,
+      "learning_rate": 1.26226639831343e-06,
+      "loss": 0.731,
+      "step": 11712
+    },
+    {
+      "epoch": 0.998125266297401,
+      "grad_norm": 20.25,
+      "learning_rate": 1.262098370743405e-06,
+      "loss": 0.8451,
+      "step": 11713
+    },
+    {
+      "epoch": 0.9982104814657009,
+      "grad_norm": 17.75,
+      "learning_rate": 1.2619303429547509e-06,
+      "loss": 0.8531,
+      "step": 11714
+    },
+    {
+      "epoch": 0.9982956966340009,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.2617623149505035e-06,
+      "loss": 0.8063,
+      "step": 11715
+    },
+    {
+      "epoch": 0.9983809118023008,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.2615942867336988e-06,
+      "loss": 0.5685,
+      "step": 11716
+    },
+    {
+      "epoch": 0.9984661269706008,
+      "grad_norm": 16.75,
+      "learning_rate": 1.261426258307374e-06,
+      "loss": 0.8099,
+      "step": 11717
+    },
+    {
+      "epoch": 0.9985513421389007,
+      "grad_norm": 13.75,
+      "learning_rate": 1.261258229674565e-06,
+      "loss": 0.7717,
+      "step": 11718
+    },
+    {
+      "epoch": 0.9986365573072007,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2610902008383086e-06,
+      "loss": 0.5017,
+      "step": 11719
+    },
+    {
+      "epoch": 0.9987217724755006,
+      "grad_norm": 16.875,
+      "learning_rate": 1.260922171801641e-06,
+      "loss": 0.5945,
+      "step": 11720
+    },
+    {
+      "epoch": 0.9988069876438006,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.260754142567599e-06,
+      "loss": 0.7152,
+      "step": 11721
+    },
+    {
+      "epoch": 0.9988922028121006,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2605861131392183e-06,
+      "loss": 1.0023,
+      "step": 11722
+    },
+    {
+      "epoch": 0.9989774179804005,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.2604180835195369e-06,
+      "loss": 0.6268,
+      "step": 11723
+    },
+    {
+      "epoch": 0.9990626331487005,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2602500537115892e-06,
+      "loss": 0.8231,
+      "step": 11724
+    },
+    {
+      "epoch": 0.9991478483170004,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2600820237184132e-06,
+      "loss": 0.8764,
+      "step": 11725
+    },
+    {
+      "epoch": 0.9992330634853004,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.2599139935430445e-06,
+      "loss": 0.5251,
+      "step": 11726
+    },
+    {
+      "epoch": 0.9993182786536003,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.259745963188521e-06,
+      "loss": 0.8632,
+      "step": 11727
+    },
+    {
+      "epoch": 0.9994034938219003,
+      "grad_norm": 12.875,
+      "learning_rate": 1.2595779326578771e-06,
+      "loss": 0.6063,
+      "step": 11728
+    },
+    {
+      "epoch": 0.9994887089902003,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.2594099019541508e-06,
+      "loss": 0.5859,
+      "step": 11729
+    },
+    {
+      "epoch": 0.9995739241585002,
+      "grad_norm": 11.25,
+      "learning_rate": 1.2592418710803778e-06,
+      "loss": 1.142,
+      "step": 11730
+    },
+    {
+      "epoch": 0.9996591393268002,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2590738400395955e-06,
+      "loss": 0.8674,
+      "step": 11731
+    },
+    {
+      "epoch": 0.9997443544951001,
+      "grad_norm": 31.875,
+      "learning_rate": 1.258905808834839e-06,
+      "loss": 0.779,
+      "step": 11732
+    },
+    {
+      "epoch": 0.9998295696634001,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2587377774691462e-06,
+      "loss": 0.5853,
+      "step": 11733
+    },
+    {
+      "epoch": 0.9999147848317,
+      "grad_norm": 19.875,
+      "learning_rate": 1.2585697459455526e-06,
+      "loss": 0.8478,
+      "step": 11734
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.258401714267095e-06,
+      "loss": 0.5969,
+      "step": 11735
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.9627697467803955,
+      "eval_runtime": 221.5463,
+      "eval_samples_per_second": 7.646,
+      "eval_steps_per_second": 7.646,
+      "step": 11735
+    },
+    {
+      "epoch": 1.0000852151683,
+      "grad_norm": 11.625,
+      "learning_rate": 1.2582336824368105e-06,
+      "loss": 0.5151,
+      "step": 11736
+    },
+    {
+      "epoch": 1.0001704303366,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2580656504577347e-06,
+      "loss": 0.5121,
+      "step": 11737
+    },
+    {
+      "epoch": 1.0002556455048999,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.2578976183329044e-06,
+      "loss": 0.7862,
+      "step": 11738
+    },
+    {
+      "epoch": 1.0003408606731998,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.2577295860653566e-06,
+      "loss": 0.638,
+      "step": 11739
+    },
+    {
+      "epoch": 1.0004260758414998,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.257561553658127e-06,
+      "loss": 0.3454,
+      "step": 11740
+    },
+    {
+      "epoch": 1.0005112910097997,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2573935211142526e-06,
+      "loss": 0.8697,
+      "step": 11741
+    },
+    {
+      "epoch": 1.0005965061780997,
+      "grad_norm": 22.5,
+      "learning_rate": 1.25722548843677e-06,
+      "loss": 0.7993,
+      "step": 11742
+    },
+    {
+      "epoch": 1.0006817213463997,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.2570574556287154e-06,
+      "loss": 0.4092,
+      "step": 11743
+    },
+    {
+      "epoch": 1.0007669365146996,
+      "grad_norm": 13.25,
+      "learning_rate": 1.2568894226931257e-06,
+      "loss": 0.5428,
+      "step": 11744
+    },
+    {
+      "epoch": 1.0008521516829996,
+      "grad_norm": 11.25,
+      "learning_rate": 1.2567213896330368e-06,
+      "loss": 0.6371,
+      "step": 11745
+    },
+    {
+      "epoch": 1.0009373668512995,
+      "grad_norm": 13.375,
+      "learning_rate": 1.256553356451486e-06,
+      "loss": 0.5713,
+      "step": 11746
+    },
+    {
+      "epoch": 1.0010225820195995,
+      "grad_norm": 10.25,
+      "learning_rate": 1.2563853231515094e-06,
+      "loss": 0.3222,
+      "step": 11747
+    },
+    {
+      "epoch": 1.0011077971878994,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.256217289736144e-06,
+      "loss": 0.3801,
+      "step": 11748
+    },
+    {
+      "epoch": 1.0011930123561994,
+      "grad_norm": 10.875,
+      "learning_rate": 1.256049256208425e-06,
+      "loss": 0.3724,
+      "step": 11749
+    },
+    {
+      "epoch": 1.0012782275244994,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.2558812225713904e-06,
+      "loss": 0.3376,
+      "step": 11750
+    },
+    {
+      "epoch": 1.0013634426927993,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.2557131888280762e-06,
+      "loss": 0.4657,
+      "step": 11751
+    },
+    {
+      "epoch": 1.0014486578610993,
+      "grad_norm": 12.125,
+      "learning_rate": 1.2555451549815187e-06,
+      "loss": 0.6529,
+      "step": 11752
+    },
+    {
+      "epoch": 1.0015338730293992,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2553771210347548e-06,
+      "loss": 0.9263,
+      "step": 11753
+    },
+    {
+      "epoch": 1.0016190881976992,
+      "grad_norm": 10.5,
+      "learning_rate": 1.2552090869908206e-06,
+      "loss": 0.3706,
+      "step": 11754
+    },
+    {
+      "epoch": 1.0017043033659991,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.2550410528527537e-06,
+      "loss": 0.4044,
+      "step": 11755
+    },
+    {
+      "epoch": 1.001789518534299,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.2548730186235895e-06,
+      "loss": 0.4461,
+      "step": 11756
+    },
+    {
+      "epoch": 1.001874733702599,
+      "grad_norm": 16.75,
+      "learning_rate": 1.254704984306365e-06,
+      "loss": 0.6447,
+      "step": 11757
+    },
+    {
+      "epoch": 1.001959948870899,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2545369499041163e-06,
+      "loss": 0.4562,
+      "step": 11758
+    },
+    {
+      "epoch": 1.002045164039199,
+      "grad_norm": 15.875,
+      "learning_rate": 1.254368915419881e-06,
+      "loss": 0.6444,
+      "step": 11759
+    },
+    {
+      "epoch": 1.002130379207499,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2542008808566946e-06,
+      "loss": 0.4757,
+      "step": 11760
+    },
+    {
+      "epoch": 1.0022155943757989,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2540328462175944e-06,
+      "loss": 0.3719,
+      "step": 11761
+    },
+    {
+      "epoch": 1.0023008095440988,
+      "grad_norm": 16.125,
+      "learning_rate": 1.253864811505616e-06,
+      "loss": 0.663,
+      "step": 11762
+    },
+    {
+      "epoch": 1.0023860247123988,
+      "grad_norm": 13.0,
+      "learning_rate": 1.2536967767237974e-06,
+      "loss": 0.6091,
+      "step": 11763
+    },
+    {
+      "epoch": 1.0024712398806988,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.2535287418751734e-06,
+      "loss": 0.3965,
+      "step": 11764
+    },
+    {
+      "epoch": 1.0025564550489987,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.2533607069627822e-06,
+      "loss": 0.4776,
+      "step": 11765
+    },
+    {
+      "epoch": 1.0026416702172987,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2531926719896592e-06,
+      "loss": 0.489,
+      "step": 11766
+    },
+    {
+      "epoch": 1.0027268853855986,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.2530246369588416e-06,
+      "loss": 0.2656,
+      "step": 11767
+    },
+    {
+      "epoch": 1.0028121005538986,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2528566018733658e-06,
+      "loss": 0.6037,
+      "step": 11768
+    },
+    {
+      "epoch": 1.0028973157221985,
+      "grad_norm": 12.875,
+      "learning_rate": 1.2526885667362685e-06,
+      "loss": 0.4394,
+      "step": 11769
+    },
+    {
+      "epoch": 1.0029825308904985,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.252520531550586e-06,
+      "loss": 0.4246,
+      "step": 11770
+    },
+    {
+      "epoch": 1.0030677460587984,
+      "grad_norm": 20.125,
+      "learning_rate": 1.2523524963193548e-06,
+      "loss": 0.9519,
+      "step": 11771
+    },
+    {
+      "epoch": 1.0031529612270984,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2521844610456118e-06,
+      "loss": 0.7885,
+      "step": 11772
+    },
+    {
+      "epoch": 1.0032381763953984,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.252016425732393e-06,
+      "loss": 0.5641,
+      "step": 11773
+    },
+    {
+      "epoch": 1.0033233915636983,
+      "grad_norm": 22.875,
+      "learning_rate": 1.2518483903827359e-06,
+      "loss": 0.3061,
+      "step": 11774
+    },
+    {
+      "epoch": 1.0034086067319983,
+      "grad_norm": 25.75,
+      "learning_rate": 1.2516803549996762e-06,
+      "loss": 1.3185,
+      "step": 11775
+    },
+    {
+      "epoch": 1.0034938219002982,
+      "grad_norm": 22.125,
+      "learning_rate": 1.251512319586251e-06,
+      "loss": 0.9861,
+      "step": 11776
+    },
+    {
+      "epoch": 1.0035790370685982,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.2513442841454962e-06,
+      "loss": 0.5692,
+      "step": 11777
+    },
+    {
+      "epoch": 1.0036642522368981,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2511762486804493e-06,
+      "loss": 0.4114,
+      "step": 11778
+    },
+    {
+      "epoch": 1.003749467405198,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.2510082131941465e-06,
+      "loss": 0.7378,
+      "step": 11779
+    },
+    {
+      "epoch": 1.003834682573498,
+      "grad_norm": 19.375,
+      "learning_rate": 1.250840177689624e-06,
+      "loss": 0.4243,
+      "step": 11780
+    },
+    {
+      "epoch": 1.003919897741798,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.2506721421699185e-06,
+      "loss": 0.5027,
+      "step": 11781
+    },
+    {
+      "epoch": 1.004005112910098,
+      "grad_norm": 14.875,
+      "learning_rate": 1.250504106638067e-06,
+      "loss": 0.6991,
+      "step": 11782
+    },
+    {
+      "epoch": 1.004090328078398,
+      "grad_norm": 13.625,
+      "learning_rate": 1.2503360710971058e-06,
+      "loss": 0.514,
+      "step": 11783
+    },
+    {
+      "epoch": 1.0041755432466979,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2501680355500711e-06,
+      "loss": 0.723,
+      "step": 11784
+    },
+    {
+      "epoch": 1.0042607584149978,
+      "grad_norm": 19.125,
+      "learning_rate": 1.25e-06,
+      "loss": 0.7626,
+      "step": 11785
+    },
+    {
+      "epoch": 1.0043459735832978,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2498319644499293e-06,
+      "loss": 0.4792,
+      "step": 11786
+    },
+    {
+      "epoch": 1.0044311887515978,
+      "grad_norm": 11.75,
+      "learning_rate": 1.2496639289028947e-06,
+      "loss": 0.36,
+      "step": 11787
+    },
+    {
+      "epoch": 1.0045164039198977,
+      "grad_norm": 21.625,
+      "learning_rate": 1.2494958933619333e-06,
+      "loss": 1.3249,
+      "step": 11788
+    },
+    {
+      "epoch": 1.0046016190881977,
+      "grad_norm": 25.25,
+      "learning_rate": 1.2493278578300817e-06,
+      "loss": 1.2345,
+      "step": 11789
+    },
+    {
+      "epoch": 1.0046868342564976,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2491598223103766e-06,
+      "loss": 0.7527,
+      "step": 11790
+    },
+    {
+      "epoch": 1.0047720494247976,
+      "grad_norm": 10.75,
+      "learning_rate": 1.248991786805854e-06,
+      "loss": 0.4175,
+      "step": 11791
+    },
+    {
+      "epoch": 1.0048572645930975,
+      "grad_norm": 12.625,
+      "learning_rate": 1.248823751319551e-06,
+      "loss": 0.3917,
+      "step": 11792
+    },
+    {
+      "epoch": 1.0049424797613975,
+      "grad_norm": 14.375,
+      "learning_rate": 1.248655715854504e-06,
+      "loss": 0.7347,
+      "step": 11793
+    },
+    {
+      "epoch": 1.0050276949296975,
+      "grad_norm": 19.25,
+      "learning_rate": 1.2484876804137493e-06,
+      "loss": 0.9084,
+      "step": 11794
+    },
+    {
+      "epoch": 1.0051129100979974,
+      "grad_norm": 13.5,
+      "learning_rate": 1.2483196450003242e-06,
+      "loss": 0.6526,
+      "step": 11795
+    },
+    {
+      "epoch": 1.0051981252662974,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2481516096172643e-06,
+      "loss": 0.8381,
+      "step": 11796
+    },
+    {
+      "epoch": 1.0052833404345973,
+      "grad_norm": 14.25,
+      "learning_rate": 1.247983574267607e-06,
+      "loss": 0.5441,
+      "step": 11797
+    },
+    {
+      "epoch": 1.0053685556028973,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2478155389543888e-06,
+      "loss": 0.4717,
+      "step": 11798
+    },
+    {
+      "epoch": 1.0054537707711972,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2476475036806456e-06,
+      "loss": 0.7018,
+      "step": 11799
+    },
+    {
+      "epoch": 1.0055389859394972,
+      "grad_norm": 10.5,
+      "learning_rate": 1.2474794684494147e-06,
+      "loss": 0.3482,
+      "step": 11800
+    },
+    {
+      "epoch": 1.0056242011077972,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2473114332637322e-06,
+      "loss": 0.7476,
+      "step": 11801
+    },
+    {
+      "epoch": 1.0057094162760971,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.2471433981266344e-06,
+      "loss": 0.7372,
+      "step": 11802
+    },
+    {
+      "epoch": 1.005794631444397,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.2469753630411586e-06,
+      "loss": 0.3906,
+      "step": 11803
+    },
+    {
+      "epoch": 1.005879846612697,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2468073280103408e-06,
+      "loss": 0.4751,
+      "step": 11804
+    },
+    {
+      "epoch": 1.005965061780997,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2466392930372182e-06,
+      "loss": 0.6205,
+      "step": 11805
+    },
+    {
+      "epoch": 1.006050276949297,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.246471258124827e-06,
+      "loss": 0.6441,
+      "step": 11806
+    },
+    {
+      "epoch": 1.006135492117597,
+      "grad_norm": 20.75,
+      "learning_rate": 1.2463032232762032e-06,
+      "loss": 0.7897,
+      "step": 11807
+    },
+    {
+      "epoch": 1.0062207072858969,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2461351884943843e-06,
+      "loss": 0.5583,
+      "step": 11808
+    },
+    {
+      "epoch": 1.0063059224541968,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.2459671537824062e-06,
+      "loss": 0.4413,
+      "step": 11809
+    },
+    {
+      "epoch": 1.0063911376224968,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.2457991191433058e-06,
+      "loss": 0.6705,
+      "step": 11810
+    },
+    {
+      "epoch": 1.0064763527907967,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2456310845801192e-06,
+      "loss": 0.8274,
+      "step": 11811
+    },
+    {
+      "epoch": 1.0065615679590967,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.2454630500958837e-06,
+      "loss": 0.3353,
+      "step": 11812
+    },
+    {
+      "epoch": 1.0066467831273966,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.2452950156936353e-06,
+      "loss": 0.4705,
+      "step": 11813
+    },
+    {
+      "epoch": 1.0067319982956966,
+      "grad_norm": 17.25,
+      "learning_rate": 1.245126981376411e-06,
+      "loss": 0.4269,
+      "step": 11814
+    },
+    {
+      "epoch": 1.0068172134639966,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.244958947147247e-06,
+      "loss": 0.4694,
+      "step": 11815
+    },
+    {
+      "epoch": 1.0069024286322965,
+      "grad_norm": 14.125,
+      "learning_rate": 1.2447909130091796e-06,
+      "loss": 0.457,
+      "step": 11816
+    },
+    {
+      "epoch": 1.0069876438005965,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.2446228789652456e-06,
+      "loss": 0.635,
+      "step": 11817
+    },
+    {
+      "epoch": 1.0070728589688964,
+      "grad_norm": 10.625,
+      "learning_rate": 1.2444548450184817e-06,
+      "loss": 0.3132,
+      "step": 11818
+    },
+    {
+      "epoch": 1.0071580741371964,
+      "grad_norm": 22.125,
+      "learning_rate": 1.2442868111719242e-06,
+      "loss": 1.086,
+      "step": 11819
+    },
+    {
+      "epoch": 1.0072432893054963,
+      "grad_norm": 20.875,
+      "learning_rate": 1.24411877742861e-06,
+      "loss": 1.3773,
+      "step": 11820
+    },
+    {
+      "epoch": 1.0073285044737963,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.2439507437915752e-06,
+      "loss": 0.6197,
+      "step": 11821
+    },
+    {
+      "epoch": 1.0074137196420963,
+      "grad_norm": 18.25,
+      "learning_rate": 1.243782710263857e-06,
+      "loss": 1.0505,
+      "step": 11822
+    },
+    {
+      "epoch": 1.0074989348103962,
+      "grad_norm": 16.25,
+      "learning_rate": 1.243614676848491e-06,
+      "loss": 0.6746,
+      "step": 11823
+    },
+    {
+      "epoch": 1.0075841499786962,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2434466435485143e-06,
+      "loss": 0.6761,
+      "step": 11824
+    },
+    {
+      "epoch": 1.0076693651469961,
+      "grad_norm": 22.75,
+      "learning_rate": 1.2432786103669634e-06,
+      "loss": 0.9316,
+      "step": 11825
+    },
+    {
+      "epoch": 1.007754580315296,
+      "grad_norm": 13.75,
+      "learning_rate": 1.2431105773068745e-06,
+      "loss": 0.6794,
+      "step": 11826
+    },
+    {
+      "epoch": 1.007839795483596,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.2429425443712848e-06,
+      "loss": 0.7683,
+      "step": 11827
+    },
+    {
+      "epoch": 1.007925010651896,
+      "grad_norm": 15.0,
+      "learning_rate": 1.24277451156323e-06,
+      "loss": 0.5239,
+      "step": 11828
+    },
+    {
+      "epoch": 1.008010225820196,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2426064788857474e-06,
+      "loss": 1.1397,
+      "step": 11829
+    },
+    {
+      "epoch": 1.008095440988496,
+      "grad_norm": 23.25,
+      "learning_rate": 1.2424384463418734e-06,
+      "loss": 0.888,
+      "step": 11830
+    },
+    {
+      "epoch": 1.0081806561567959,
+      "grad_norm": 14.5,
+      "learning_rate": 1.242270413934644e-06,
+      "loss": 0.6674,
+      "step": 11831
+    },
+    {
+      "epoch": 1.0082658713250958,
+      "grad_norm": 12.75,
+      "learning_rate": 1.2421023816670958e-06,
+      "loss": 0.4278,
+      "step": 11832
+    },
+    {
+      "epoch": 1.0083510864933958,
+      "grad_norm": 12.875,
+      "learning_rate": 1.2419343495422657e-06,
+      "loss": 0.3403,
+      "step": 11833
+    },
+    {
+      "epoch": 1.0084363016616957,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2417663175631899e-06,
+      "loss": 1.0834,
+      "step": 11834
+    },
+    {
+      "epoch": 1.0085215168299957,
+      "grad_norm": 15.875,
+      "learning_rate": 1.2415982857329051e-06,
+      "loss": 0.4746,
+      "step": 11835
+    },
+    {
+      "epoch": 1.0086067319982956,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.2414302540544474e-06,
+      "loss": 0.2932,
+      "step": 11836
+    },
+    {
+      "epoch": 1.0086919471665956,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2412622225308542e-06,
+      "loss": 0.6587,
+      "step": 11837
+    },
+    {
+      "epoch": 1.0087771623348956,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2410941911651613e-06,
+      "loss": 0.7996,
+      "step": 11838
+    },
+    {
+      "epoch": 1.0088623775031955,
+      "grad_norm": 16.5,
+      "learning_rate": 1.240926159960405e-06,
+      "loss": 0.7228,
+      "step": 11839
+    },
+    {
+      "epoch": 1.0089475926714955,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2407581289196224e-06,
+      "loss": 0.8024,
+      "step": 11840
+    },
+    {
+      "epoch": 1.0090328078397954,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2405900980458496e-06,
+      "loss": 0.5653,
+      "step": 11841
+    },
+    {
+      "epoch": 1.0091180230080954,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.240422067342123e-06,
+      "loss": 0.4692,
+      "step": 11842
+    },
+    {
+      "epoch": 1.0092032381763953,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.2402540368114795e-06,
+      "loss": 0.3533,
+      "step": 11843
+    },
+    {
+      "epoch": 1.0092884533446953,
+      "grad_norm": 14.125,
+      "learning_rate": 1.2400860064569553e-06,
+      "loss": 0.5151,
+      "step": 11844
+    },
+    {
+      "epoch": 1.0093736685129953,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.2399179762815868e-06,
+      "loss": 0.3319,
+      "step": 11845
+    },
+    {
+      "epoch": 1.0094588836812952,
+      "grad_norm": 13.25,
+      "learning_rate": 1.2397499462884114e-06,
+      "loss": 0.6069,
+      "step": 11846
+    },
+    {
+      "epoch": 1.0095440988495952,
+      "grad_norm": 13.125,
+      "learning_rate": 1.2395819164804638e-06,
+      "loss": 0.8084,
+      "step": 11847
+    },
+    {
+      "epoch": 1.0096293140178951,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2394138868607821e-06,
+      "loss": 0.6835,
+      "step": 11848
+    },
+    {
+      "epoch": 1.009714529186195,
+      "grad_norm": 16.0,
+      "learning_rate": 1.2392458574324016e-06,
+      "loss": 0.8352,
+      "step": 11849
+    },
+    {
+      "epoch": 1.009799744354495,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2390778281983594e-06,
+      "loss": 0.6048,
+      "step": 11850
+    },
+    {
+      "epoch": 1.009884959522795,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2389097991616918e-06,
+      "loss": 0.6834,
+      "step": 11851
+    },
+    {
+      "epoch": 1.009970174691095,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.2387417703254354e-06,
+      "loss": 0.5094,
+      "step": 11852
+    },
+    {
+      "epoch": 1.010055389859395,
+      "grad_norm": 19.875,
+      "learning_rate": 1.2385737416926263e-06,
+      "loss": 0.5048,
+      "step": 11853
+    },
+    {
+      "epoch": 1.0101406050276949,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2384057132663014e-06,
+      "loss": 0.7713,
+      "step": 11854
+    },
+    {
+      "epoch": 1.0102258201959948,
+      "grad_norm": 21.25,
+      "learning_rate": 1.2382376850494973e-06,
+      "loss": 0.5077,
+      "step": 11855
+    },
+    {
+      "epoch": 1.0103110353642948,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.2380696570452493e-06,
+      "loss": 0.5841,
+      "step": 11856
+    },
+    {
+      "epoch": 1.0103962505325947,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2379016292565952e-06,
+      "loss": 0.632,
+      "step": 11857
+    },
+    {
+      "epoch": 1.0104814657008947,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.2377336016865704e-06,
+      "loss": 0.5922,
+      "step": 11858
+    },
+    {
+      "epoch": 1.0105666808691947,
+      "grad_norm": 21.75,
+      "learning_rate": 1.237565574338212e-06,
+      "loss": 0.6537,
+      "step": 11859
+    },
+    {
+      "epoch": 1.0106518960374946,
+      "grad_norm": 11.5,
+      "learning_rate": 1.2373975472145559e-06,
+      "loss": 0.3533,
+      "step": 11860
+    },
+    {
+      "epoch": 1.0107371112057946,
+      "grad_norm": 11.625,
+      "learning_rate": 1.237229520318639e-06,
+      "loss": 0.3931,
+      "step": 11861
+    },
+    {
+      "epoch": 1.0108223263740945,
+      "grad_norm": 20.25,
+      "learning_rate": 1.2370614936534973e-06,
+      "loss": 0.7817,
+      "step": 11862
+    },
+    {
+      "epoch": 1.0109075415423945,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2368934672221684e-06,
+      "loss": 0.8911,
+      "step": 11863
+    },
+    {
+      "epoch": 1.0109927567106944,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2367254410276868e-06,
+      "loss": 0.855,
+      "step": 11864
+    },
+    {
+      "epoch": 1.0110779718789944,
+      "grad_norm": 11.125,
+      "learning_rate": 1.23655741507309e-06,
+      "loss": 0.2178,
+      "step": 11865
+    },
+    {
+      "epoch": 1.0111631870472944,
+      "grad_norm": 26.375,
+      "learning_rate": 1.236389389361414e-06,
+      "loss": 0.6537,
+      "step": 11866
+    },
+    {
+      "epoch": 1.0112484022155943,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.2362213638956958e-06,
+      "loss": 0.2725,
+      "step": 11867
+    },
+    {
+      "epoch": 1.0113336173838943,
+      "grad_norm": 18.5,
+      "learning_rate": 1.2360533386789709e-06,
+      "loss": 0.908,
+      "step": 11868
+    },
+    {
+      "epoch": 1.0114188325521942,
+      "grad_norm": 18.25,
+      "learning_rate": 1.2358853137142767e-06,
+      "loss": 0.6545,
+      "step": 11869
+    },
+    {
+      "epoch": 1.0115040477204942,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2357172890046487e-06,
+      "loss": 0.6677,
+      "step": 11870
+    },
+    {
+      "epoch": 1.0115892628887941,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2355492645531242e-06,
+      "loss": 0.6883,
+      "step": 11871
+    },
+    {
+      "epoch": 1.011674478057094,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.2353812403627389e-06,
+      "loss": 0.4726,
+      "step": 11872
+    },
+    {
+      "epoch": 1.011759693225394,
+      "grad_norm": 12.125,
+      "learning_rate": 1.2352132164365287e-06,
+      "loss": 0.434,
+      "step": 11873
+    },
+    {
+      "epoch": 1.011844908393694,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.235045192777531e-06,
+      "loss": 0.4478,
+      "step": 11874
+    },
+    {
+      "epoch": 1.011930123561994,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2348771693887815e-06,
+      "loss": 0.789,
+      "step": 11875
+    },
+    {
+      "epoch": 1.012015338730294,
+      "grad_norm": 15.25,
+      "learning_rate": 1.234709146273317e-06,
+      "loss": 0.6791,
+      "step": 11876
+    },
+    {
+      "epoch": 1.0121005538985939,
+      "grad_norm": 14.25,
+      "learning_rate": 1.2345411234341732e-06,
+      "loss": 0.6555,
+      "step": 11877
+    },
+    {
+      "epoch": 1.0121857690668938,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.2343731008743873e-06,
+      "loss": 0.4536,
+      "step": 11878
+    },
+    {
+      "epoch": 1.0122709842351938,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2342050785969954e-06,
+      "loss": 0.3965,
+      "step": 11879
+    },
+    {
+      "epoch": 1.0123561994034938,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.2340370566050332e-06,
+      "loss": 0.3549,
+      "step": 11880
+    },
+    {
+      "epoch": 1.0124414145717937,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2338690349015373e-06,
+      "loss": 0.6124,
+      "step": 11881
+    },
+    {
+      "epoch": 1.0125266297400937,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2337010134895445e-06,
+      "loss": 0.5992,
+      "step": 11882
+    },
+    {
+      "epoch": 1.0126118449083936,
+      "grad_norm": 14.25,
+      "learning_rate": 1.2335329923720903e-06,
+      "loss": 0.4274,
+      "step": 11883
+    },
+    {
+      "epoch": 1.0126970600766936,
+      "grad_norm": 12.875,
+      "learning_rate": 1.233364971552212e-06,
+      "loss": 0.4246,
+      "step": 11884
+    },
+    {
+      "epoch": 1.0127822752449935,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.2331969510329447e-06,
+      "loss": 0.4619,
+      "step": 11885
+    },
+    {
+      "epoch": 1.0128674904132935,
+      "grad_norm": 18.25,
+      "learning_rate": 1.233028930817326e-06,
+      "loss": 0.5692,
+      "step": 11886
+    },
+    {
+      "epoch": 1.0129527055815934,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2328609109083918e-06,
+      "loss": 0.62,
+      "step": 11887
+    },
+    {
+      "epoch": 1.0130379207498934,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.2326928913091776e-06,
+      "loss": 0.2991,
+      "step": 11888
+    },
+    {
+      "epoch": 1.0131231359181934,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2325248720227207e-06,
+      "loss": 0.6146,
+      "step": 11889
+    },
+    {
+      "epoch": 1.0132083510864933,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2323568530520564e-06,
+      "loss": 0.5334,
+      "step": 11890
+    },
+    {
+      "epoch": 1.0132935662547933,
+      "grad_norm": 14.0,
+      "learning_rate": 1.232188834400222e-06,
+      "loss": 0.4535,
+      "step": 11891
+    },
+    {
+      "epoch": 1.0133787814230932,
+      "grad_norm": 23.625,
+      "learning_rate": 1.2320208160702528e-06,
+      "loss": 0.5553,
+      "step": 11892
+    },
+    {
+      "epoch": 1.0134639965913932,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.2318527980651859e-06,
+      "loss": 0.6068,
+      "step": 11893
+    },
+    {
+      "epoch": 1.0135492117596931,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2316847803880574e-06,
+      "loss": 0.4843,
+      "step": 11894
+    },
+    {
+      "epoch": 1.013634426927993,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.2315167630419035e-06,
+      "loss": 0.5045,
+      "step": 11895
+    },
+    {
+      "epoch": 1.013719642096293,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2313487460297594e-06,
+      "loss": 0.9105,
+      "step": 11896
+    },
+    {
+      "epoch": 1.013804857264593,
+      "grad_norm": 12.0,
+      "learning_rate": 1.2311807293546627e-06,
+      "loss": 0.2828,
+      "step": 11897
+    },
+    {
+      "epoch": 1.013890072432893,
+      "grad_norm": 36.25,
+      "learning_rate": 1.2310127130196488e-06,
+      "loss": 1.1786,
+      "step": 11898
+    },
+    {
+      "epoch": 1.013975287601193,
+      "grad_norm": 22.125,
+      "learning_rate": 1.2308446970277547e-06,
+      "loss": 0.8472,
+      "step": 11899
+    },
+    {
+      "epoch": 1.0140605027694929,
+      "grad_norm": 22.5,
+      "learning_rate": 1.2306766813820158e-06,
+      "loss": 0.6123,
+      "step": 11900
+    },
+    {
+      "epoch": 1.0141457179377928,
+      "grad_norm": 25.625,
+      "learning_rate": 1.2305086660854692e-06,
+      "loss": 1.007,
+      "step": 11901
+    },
+    {
+      "epoch": 1.0142309331060928,
+      "grad_norm": 14.375,
+      "learning_rate": 1.23034065114115e-06,
+      "loss": 0.6128,
+      "step": 11902
+    },
+    {
+      "epoch": 1.0143161482743928,
+      "grad_norm": 12.75,
+      "learning_rate": 1.230172636552096e-06,
+      "loss": 0.3913,
+      "step": 11903
+    },
+    {
+      "epoch": 1.0144013634426927,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2300046223213418e-06,
+      "loss": 0.622,
+      "step": 11904
+    },
+    {
+      "epoch": 1.0144865786109927,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2298366084519237e-06,
+      "loss": 0.7441,
+      "step": 11905
+    },
+    {
+      "epoch": 1.0145717937792926,
+      "grad_norm": 15.5,
+      "learning_rate": 1.229668594946879e-06,
+      "loss": 0.822,
+      "step": 11906
+    },
+    {
+      "epoch": 1.0146570089475926,
+      "grad_norm": 12.625,
+      "learning_rate": 1.229500581809243e-06,
+      "loss": 0.4293,
+      "step": 11907
+    },
+    {
+      "epoch": 1.0147422241158925,
+      "grad_norm": 20.75,
+      "learning_rate": 1.229332569042052e-06,
+      "loss": 0.8744,
+      "step": 11908
+    },
+    {
+      "epoch": 1.0148274392841925,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2291645566483425e-06,
+      "loss": 1.002,
+      "step": 11909
+    },
+    {
+      "epoch": 1.0149126544524925,
+      "grad_norm": 22.625,
+      "learning_rate": 1.22899654463115e-06,
+      "loss": 0.524,
+      "step": 11910
+    },
+    {
+      "epoch": 1.0149978696207924,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2288285329935118e-06,
+      "loss": 0.77,
+      "step": 11911
+    },
+    {
+      "epoch": 1.0150830847890924,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.228660521738463e-06,
+      "loss": 0.3541,
+      "step": 11912
+    },
+    {
+      "epoch": 1.0151682999573923,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2284925108690398e-06,
+      "loss": 0.4796,
+      "step": 11913
+    },
+    {
+      "epoch": 1.0152535151256923,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2283245003882787e-06,
+      "loss": 0.6963,
+      "step": 11914
+    },
+    {
+      "epoch": 1.0153387302939922,
+      "grad_norm": 21.0,
+      "learning_rate": 1.2281564902992153e-06,
+      "loss": 1.0846,
+      "step": 11915
+    },
+    {
+      "epoch": 1.0154239454622922,
+      "grad_norm": 14.125,
+      "learning_rate": 1.2279884806048866e-06,
+      "loss": 0.5684,
+      "step": 11916
+    },
+    {
+      "epoch": 1.0155091606305922,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.2278204713083278e-06,
+      "loss": 0.6135,
+      "step": 11917
+    },
+    {
+      "epoch": 1.0155943757988921,
+      "grad_norm": 18.25,
+      "learning_rate": 1.2276524624125756e-06,
+      "loss": 0.6737,
+      "step": 11918
+    },
+    {
+      "epoch": 1.015679590967192,
+      "grad_norm": 19.875,
+      "learning_rate": 1.2274844539206663e-06,
+      "loss": 0.8726,
+      "step": 11919
+    },
+    {
+      "epoch": 1.015764806135492,
+      "grad_norm": 19.0,
+      "learning_rate": 1.2273164458356348e-06,
+      "loss": 0.6975,
+      "step": 11920
+    },
+    {
+      "epoch": 1.015850021303792,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.2271484381605183e-06,
+      "loss": 0.448,
+      "step": 11921
+    },
+    {
+      "epoch": 1.015935236472092,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2269804308983521e-06,
+      "loss": 0.7315,
+      "step": 11922
+    },
+    {
+      "epoch": 1.016020451640392,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.2268124240521732e-06,
+      "loss": 0.4692,
+      "step": 11923
+    },
+    {
+      "epoch": 1.0161056668086919,
+      "grad_norm": 14.875,
+      "learning_rate": 1.226644417625017e-06,
+      "loss": 0.7028,
+      "step": 11924
+    },
+    {
+      "epoch": 1.0161908819769918,
+      "grad_norm": 14.5,
+      "learning_rate": 1.2264764116199194e-06,
+      "loss": 0.3643,
+      "step": 11925
+    },
+    {
+      "epoch": 1.0162760971452918,
+      "grad_norm": 16.0,
+      "learning_rate": 1.226308406039917e-06,
+      "loss": 0.6644,
+      "step": 11926
+    },
+    {
+      "epoch": 1.0163613123135917,
+      "grad_norm": 12.875,
+      "learning_rate": 1.2261404008880452e-06,
+      "loss": 0.5277,
+      "step": 11927
+    },
+    {
+      "epoch": 1.0164465274818917,
+      "grad_norm": 16.625,
+      "learning_rate": 1.225972396167341e-06,
+      "loss": 0.5956,
+      "step": 11928
+    },
+    {
+      "epoch": 1.0165317426501916,
+      "grad_norm": 19.0,
+      "learning_rate": 1.2258043918808394e-06,
+      "loss": 0.9156,
+      "step": 11929
+    },
+    {
+      "epoch": 1.0166169578184916,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2256363880315766e-06,
+      "loss": 0.3284,
+      "step": 11930
+    },
+    {
+      "epoch": 1.0167021729867916,
+      "grad_norm": 17.125,
+      "learning_rate": 1.225468384622589e-06,
+      "loss": 0.6787,
+      "step": 11931
+    },
+    {
+      "epoch": 1.0167873881550915,
+      "grad_norm": 20.875,
+      "learning_rate": 1.225300381656912e-06,
+      "loss": 0.5145,
+      "step": 11932
+    },
+    {
+      "epoch": 1.0168726033233915,
+      "grad_norm": 13.25,
+      "learning_rate": 1.225132379137582e-06,
+      "loss": 0.5158,
+      "step": 11933
+    },
+    {
+      "epoch": 1.0169578184916914,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.224964377067635e-06,
+      "loss": 0.2742,
+      "step": 11934
+    },
+    {
+      "epoch": 1.0170430336599914,
+      "grad_norm": 14.375,
+      "learning_rate": 1.224796375450107e-06,
+      "loss": 0.5291,
+      "step": 11935
+    },
+    {
+      "epoch": 1.0171282488282913,
+      "grad_norm": 11.0,
+      "learning_rate": 1.224628374288034e-06,
+      "loss": 0.5255,
+      "step": 11936
+    },
+    {
+      "epoch": 1.0172134639965913,
+      "grad_norm": 17.125,
+      "learning_rate": 1.2244603735844515e-06,
+      "loss": 0.542,
+      "step": 11937
+    },
+    {
+      "epoch": 1.0172986791648913,
+      "grad_norm": 12.125,
+      "learning_rate": 1.224292373342396e-06,
+      "loss": 0.4438,
+      "step": 11938
+    },
+    {
+      "epoch": 1.0173838943331912,
+      "grad_norm": 10.625,
+      "learning_rate": 1.2241243735649029e-06,
+      "loss": 0.36,
+      "step": 11939
+    },
+    {
+      "epoch": 1.0174691095014912,
+      "grad_norm": 18.875,
+      "learning_rate": 1.2239563742550084e-06,
+      "loss": 0.6099,
+      "step": 11940
+    },
+    {
+      "epoch": 1.0175543246697911,
+      "grad_norm": 19.75,
+      "learning_rate": 1.2237883754157485e-06,
+      "loss": 0.6168,
+      "step": 11941
+    },
+    {
+      "epoch": 1.017639539838091,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.223620377050159e-06,
+      "loss": 0.4309,
+      "step": 11942
+    },
+    {
+      "epoch": 1.017724755006391,
+      "grad_norm": 24.25,
+      "learning_rate": 1.2234523791612757e-06,
+      "loss": 0.697,
+      "step": 11943
+    },
+    {
+      "epoch": 1.017809970174691,
+      "grad_norm": 17.75,
+      "learning_rate": 1.223284381752135e-06,
+      "loss": 0.6615,
+      "step": 11944
+    },
+    {
+      "epoch": 1.017895185342991,
+      "grad_norm": 16.5,
+      "learning_rate": 1.223116384825772e-06,
+      "loss": 0.6214,
+      "step": 11945
+    },
+    {
+      "epoch": 1.017980400511291,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2229483883852231e-06,
+      "loss": 0.685,
+      "step": 11946
+    },
+    {
+      "epoch": 1.0180656156795909,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.2227803924335237e-06,
+      "loss": 0.582,
+      "step": 11947
+    },
+    {
+      "epoch": 1.0181508308478908,
+      "grad_norm": 15.25,
+      "learning_rate": 1.2226123969737103e-06,
+      "loss": 0.4938,
+      "step": 11948
+    },
+    {
+      "epoch": 1.0182360460161908,
+      "grad_norm": 13.25,
+      "learning_rate": 1.2224444020088183e-06,
+      "loss": 0.2816,
+      "step": 11949
+    },
+    {
+      "epoch": 1.0183212611844907,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.2222764075418838e-06,
+      "loss": 0.6537,
+      "step": 11950
+    },
+    {
+      "epoch": 1.0184064763527907,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.2221084135759422e-06,
+      "loss": 0.4018,
+      "step": 11951
+    },
+    {
+      "epoch": 1.0184916915210906,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2219404201140303e-06,
+      "loss": 0.6995,
+      "step": 11952
+    },
+    {
+      "epoch": 1.0185769066893906,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.2217724271591828e-06,
+      "loss": 0.5302,
+      "step": 11953
+    },
+    {
+      "epoch": 1.0186621218576906,
+      "grad_norm": 15.125,
+      "learning_rate": 1.2216044347144357e-06,
+      "loss": 0.6748,
+      "step": 11954
+    },
+    {
+      "epoch": 1.0187473370259905,
+      "grad_norm": 21.25,
+      "learning_rate": 1.2214364427828254e-06,
+      "loss": 0.8775,
+      "step": 11955
+    },
+    {
+      "epoch": 1.0188325521942905,
+      "grad_norm": 22.875,
+      "learning_rate": 1.2212684513673873e-06,
+      "loss": 0.6695,
+      "step": 11956
+    },
+    {
+      "epoch": 1.0189177673625904,
+      "grad_norm": 17.75,
+      "learning_rate": 1.2211004604711566e-06,
+      "loss": 0.5567,
+      "step": 11957
+    },
+    {
+      "epoch": 1.0190029825308904,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2209324700971702e-06,
+      "loss": 0.5106,
+      "step": 11958
+    },
+    {
+      "epoch": 1.0190881976991903,
+      "grad_norm": 18.5,
+      "learning_rate": 1.2207644802484629e-06,
+      "loss": 0.6221,
+      "step": 11959
+    },
+    {
+      "epoch": 1.0191734128674903,
+      "grad_norm": 13.5,
+      "learning_rate": 1.2205964909280715e-06,
+      "loss": 0.4123,
+      "step": 11960
+    },
+    {
+      "epoch": 1.0192586280357903,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2204285021390308e-06,
+      "loss": 1.0866,
+      "step": 11961
+    },
+    {
+      "epoch": 1.0193438432040902,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2202605138843763e-06,
+      "loss": 0.6068,
+      "step": 11962
+    },
+    {
+      "epoch": 1.0194290583723902,
+      "grad_norm": 15.25,
+      "learning_rate": 1.2200925261671448e-06,
+      "loss": 0.5363,
+      "step": 11963
+    },
+    {
+      "epoch": 1.0195142735406901,
+      "grad_norm": 15.75,
+      "learning_rate": 1.219924538990371e-06,
+      "loss": 0.5703,
+      "step": 11964
+    },
+    {
+      "epoch": 1.0195994887089903,
+      "grad_norm": 12.125,
+      "learning_rate": 1.2197565523570914e-06,
+      "loss": 0.3931,
+      "step": 11965
+    },
+    {
+      "epoch": 1.0196847038772903,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2195885662703407e-06,
+      "loss": 0.608,
+      "step": 11966
+    },
+    {
+      "epoch": 1.0197699190455902,
+      "grad_norm": 18.625,
+      "learning_rate": 1.219420580733156e-06,
+      "loss": 0.5983,
+      "step": 11967
+    },
+    {
+      "epoch": 1.0198551342138902,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.219252595748572e-06,
+      "loss": 0.5137,
+      "step": 11968
+    },
+    {
+      "epoch": 1.0199403493821901,
+      "grad_norm": 12.125,
+      "learning_rate": 1.2190846113196242e-06,
+      "loss": 0.4368,
+      "step": 11969
+    },
+    {
+      "epoch": 1.02002556455049,
+      "grad_norm": 21.625,
+      "learning_rate": 1.2189166274493488e-06,
+      "loss": 0.7287,
+      "step": 11970
+    },
+    {
+      "epoch": 1.02011077971879,
+      "grad_norm": 25.375,
+      "learning_rate": 1.218748644140781e-06,
+      "loss": 0.7125,
+      "step": 11971
+    },
+    {
+      "epoch": 1.02019599488709,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.2185806613969564e-06,
+      "loss": 0.4505,
+      "step": 11972
+    },
+    {
+      "epoch": 1.02028121005539,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.2184126792209113e-06,
+      "loss": 0.4416,
+      "step": 11973
+    },
+    {
+      "epoch": 1.02036642522369,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.2182446976156804e-06,
+      "loss": 0.4423,
+      "step": 11974
+    },
+    {
+      "epoch": 1.0204516403919899,
+      "grad_norm": 20.75,
+      "learning_rate": 1.2180767165843001e-06,
+      "loss": 0.838,
+      "step": 11975
+    },
+    {
+      "epoch": 1.0205368555602898,
+      "grad_norm": 18.75,
+      "learning_rate": 1.217908736129806e-06,
+      "loss": 0.5747,
+      "step": 11976
+    },
+    {
+      "epoch": 1.0206220707285898,
+      "grad_norm": 21.5,
+      "learning_rate": 1.2177407562552327e-06,
+      "loss": 0.7355,
+      "step": 11977
+    },
+    {
+      "epoch": 1.0207072858968897,
+      "grad_norm": 20.875,
+      "learning_rate": 1.2175727769636166e-06,
+      "loss": 0.655,
+      "step": 11978
+    },
+    {
+      "epoch": 1.0207925010651897,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.2174047982579927e-06,
+      "loss": 0.392,
+      "step": 11979
+    },
+    {
+      "epoch": 1.0208777162334897,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2172368201413972e-06,
+      "loss": 0.605,
+      "step": 11980
+    },
+    {
+      "epoch": 1.0209629314017896,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.217068842616865e-06,
+      "loss": 0.5198,
+      "step": 11981
+    },
+    {
+      "epoch": 1.0210481465700896,
+      "grad_norm": 16.0,
+      "learning_rate": 1.2169008656874323e-06,
+      "loss": 0.5874,
+      "step": 11982
+    },
+    {
+      "epoch": 1.0211333617383895,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.216732889356134e-06,
+      "loss": 0.7363,
+      "step": 11983
+    },
+    {
+      "epoch": 1.0212185769066895,
+      "grad_norm": 19.625,
+      "learning_rate": 1.2165649136260063e-06,
+      "loss": 0.5927,
+      "step": 11984
+    },
+    {
+      "epoch": 1.0213037920749894,
+      "grad_norm": 16.25,
+      "learning_rate": 1.216396938500084e-06,
+      "loss": 0.8948,
+      "step": 11985
+    },
+    {
+      "epoch": 1.0213890072432894,
+      "grad_norm": 14.75,
+      "learning_rate": 1.2162289639814027e-06,
+      "loss": 0.7612,
+      "step": 11986
+    },
+    {
+      "epoch": 1.0214742224115894,
+      "grad_norm": 14.5,
+      "learning_rate": 1.216060990072998e-06,
+      "loss": 0.3882,
+      "step": 11987
+    },
+    {
+      "epoch": 1.0215594375798893,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.2158930167779053e-06,
+      "loss": 0.5435,
+      "step": 11988
+    },
+    {
+      "epoch": 1.0216446527481893,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.2157250440991599e-06,
+      "loss": 0.6417,
+      "step": 11989
+    },
+    {
+      "epoch": 1.0217298679164892,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2155570720397978e-06,
+      "loss": 0.4316,
+      "step": 11990
+    },
+    {
+      "epoch": 1.0218150830847892,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2153891006028536e-06,
+      "loss": 0.7241,
+      "step": 11991
+    },
+    {
+      "epoch": 1.0219002982530891,
+      "grad_norm": 14.75,
+      "learning_rate": 1.2152211297913633e-06,
+      "loss": 0.6106,
+      "step": 11992
+    },
+    {
+      "epoch": 1.021985513421389,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2150531596083628e-06,
+      "loss": 0.6398,
+      "step": 11993
+    },
+    {
+      "epoch": 1.022070728589689,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.2148851900568862e-06,
+      "loss": 0.7378,
+      "step": 11994
+    },
+    {
+      "epoch": 1.022155943757989,
+      "grad_norm": 15.0,
+      "learning_rate": 1.2147172211399698e-06,
+      "loss": 0.4501,
+      "step": 11995
+    },
+    {
+      "epoch": 1.022241158926289,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2145492528606483e-06,
+      "loss": 0.4264,
+      "step": 11996
+    },
+    {
+      "epoch": 1.022326374094589,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2143812852219577e-06,
+      "loss": 0.5939,
+      "step": 11997
+    },
+    {
+      "epoch": 1.0224115892628889,
+      "grad_norm": 15.125,
+      "learning_rate": 1.2142133182269328e-06,
+      "loss": 0.7487,
+      "step": 11998
+    },
+    {
+      "epoch": 1.0224968044311888,
+      "grad_norm": 18.75,
+      "learning_rate": 1.2140453518786097e-06,
+      "loss": 0.9839,
+      "step": 11999
+    },
+    {
+      "epoch": 1.0225820195994888,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2138773861800227e-06,
+      "loss": 0.434,
+      "step": 12000
+    },
+    {
+      "epoch": 1.0226672347677888,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.2137094211342087e-06,
+      "loss": 0.5554,
+      "step": 12001
+    },
+    {
+      "epoch": 1.0227524499360887,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.2135414567442012e-06,
+      "loss": 0.5954,
+      "step": 12002
+    },
+    {
+      "epoch": 1.0228376651043887,
+      "grad_norm": 18.875,
+      "learning_rate": 1.2133734930130365e-06,
+      "loss": 0.6646,
+      "step": 12003
+    },
+    {
+      "epoch": 1.0229228802726886,
+      "grad_norm": 9.5,
+      "learning_rate": 1.213205529943749e-06,
+      "loss": 0.4746,
+      "step": 12004
+    },
+    {
+      "epoch": 1.0230080954409886,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.2130375675393752e-06,
+      "loss": 0.8507,
+      "step": 12005
+    },
+    {
+      "epoch": 1.0230933106092885,
+      "grad_norm": 23.25,
+      "learning_rate": 1.2128696058029494e-06,
+      "loss": 0.6255,
+      "step": 12006
+    },
+    {
+      "epoch": 1.0231785257775885,
+      "grad_norm": 26.375,
+      "learning_rate": 1.2127016447375072e-06,
+      "loss": 0.7,
+      "step": 12007
+    },
+    {
+      "epoch": 1.0232637409458885,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.2125336843460837e-06,
+      "loss": 0.4802,
+      "step": 12008
+    },
+    {
+      "epoch": 1.0233489561141884,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2123657246317146e-06,
+      "loss": 0.8062,
+      "step": 12009
+    },
+    {
+      "epoch": 1.0234341712824884,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2121977655974346e-06,
+      "loss": 0.8518,
+      "step": 12010
+    },
+    {
+      "epoch": 1.0235193864507883,
+      "grad_norm": 13.25,
+      "learning_rate": 1.2120298072462787e-06,
+      "loss": 0.4189,
+      "step": 12011
+    },
+    {
+      "epoch": 1.0236046016190883,
+      "grad_norm": 11.25,
+      "learning_rate": 1.2118618495812827e-06,
+      "loss": 0.3746,
+      "step": 12012
+    },
+    {
+      "epoch": 1.0236898167873882,
+      "grad_norm": 16.25,
+      "learning_rate": 1.211693892605481e-06,
+      "loss": 0.524,
+      "step": 12013
+    },
+    {
+      "epoch": 1.0237750319556882,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2115259363219096e-06,
+      "loss": 0.6629,
+      "step": 12014
+    },
+    {
+      "epoch": 1.0238602471239882,
+      "grad_norm": 19.25,
+      "learning_rate": 1.2113579807336027e-06,
+      "loss": 0.852,
+      "step": 12015
+    },
+    {
+      "epoch": 1.0239454622922881,
+      "grad_norm": 11.5,
+      "learning_rate": 1.2111900258435964e-06,
+      "loss": 0.3769,
+      "step": 12016
+    },
+    {
+      "epoch": 1.024030677460588,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2110220716549257e-06,
+      "loss": 1.0195,
+      "step": 12017
+    },
+    {
+      "epoch": 1.024115892628888,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.210854118170625e-06,
+      "loss": 0.2603,
+      "step": 12018
+    },
+    {
+      "epoch": 1.024201107797188,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2106861653937293e-06,
+      "loss": 0.501,
+      "step": 12019
+    },
+    {
+      "epoch": 1.024286322965488,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2105182133272747e-06,
+      "loss": 0.6349,
+      "step": 12020
+    },
+    {
+      "epoch": 1.024371538133788,
+      "grad_norm": 18.125,
+      "learning_rate": 1.2103502619742954e-06,
+      "loss": 0.7254,
+      "step": 12021
+    },
+    {
+      "epoch": 1.0244567533020879,
+      "grad_norm": 18.375,
+      "learning_rate": 1.2101823113378268e-06,
+      "loss": 0.8706,
+      "step": 12022
+    },
+    {
+      "epoch": 1.0245419684703878,
+      "grad_norm": 18.0,
+      "learning_rate": 1.2100143614209038e-06,
+      "loss": 0.6695,
+      "step": 12023
+    },
+    {
+      "epoch": 1.0246271836386878,
+      "grad_norm": 20.5,
+      "learning_rate": 1.2098464122265617e-06,
+      "loss": 0.5533,
+      "step": 12024
+    },
+    {
+      "epoch": 1.0247123988069877,
+      "grad_norm": 24.0,
+      "learning_rate": 1.2096784637578355e-06,
+      "loss": 0.8595,
+      "step": 12025
+    },
+    {
+      "epoch": 1.0247976139752877,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.2095105160177594e-06,
+      "loss": 0.4157,
+      "step": 12026
+    },
+    {
+      "epoch": 1.0248828291435876,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2093425690093694e-06,
+      "loss": 0.6626,
+      "step": 12027
+    },
+    {
+      "epoch": 1.0249680443118876,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2091746227356996e-06,
+      "loss": 0.5582,
+      "step": 12028
+    },
+    {
+      "epoch": 1.0250532594801876,
+      "grad_norm": 15.25,
+      "learning_rate": 1.2090066771997858e-06,
+      "loss": 0.6928,
+      "step": 12029
+    },
+    {
+      "epoch": 1.0251384746484875,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.2088387324046624e-06,
+      "loss": 0.6567,
+      "step": 12030
+    },
+    {
+      "epoch": 1.0252236898167875,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.2086707883533649e-06,
+      "loss": 0.3253,
+      "step": 12031
+    },
+    {
+      "epoch": 1.0253089049850874,
+      "grad_norm": 22.5,
+      "learning_rate": 1.2085028450489272e-06,
+      "loss": 0.8854,
+      "step": 12032
+    },
+    {
+      "epoch": 1.0253941201533874,
+      "grad_norm": 20.875,
+      "learning_rate": 1.2083349024943857e-06,
+      "loss": 0.8109,
+      "step": 12033
+    },
+    {
+      "epoch": 1.0254793353216873,
+      "grad_norm": 13.75,
+      "learning_rate": 1.2081669606927738e-06,
+      "loss": 0.6547,
+      "step": 12034
+    },
+    {
+      "epoch": 1.0255645504899873,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.2079990196471273e-06,
+      "loss": 0.4259,
+      "step": 12035
+    },
+    {
+      "epoch": 1.0256497656582872,
+      "grad_norm": 31.875,
+      "learning_rate": 1.2078310793604802e-06,
+      "loss": 0.5704,
+      "step": 12036
+    },
+    {
+      "epoch": 1.0257349808265872,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2076631398358683e-06,
+      "loss": 0.4021,
+      "step": 12037
+    },
+    {
+      "epoch": 1.0258201959948872,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2074952010763259e-06,
+      "loss": 0.4289,
+      "step": 12038
+    },
+    {
+      "epoch": 1.0259054111631871,
+      "grad_norm": 29.625,
+      "learning_rate": 1.207327263084888e-06,
+      "loss": 0.837,
+      "step": 12039
+    },
+    {
+      "epoch": 1.025990626331487,
+      "grad_norm": 21.75,
+      "learning_rate": 1.207159325864589e-06,
+      "loss": 0.8295,
+      "step": 12040
+    },
+    {
+      "epoch": 1.026075841499787,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.2069913894184648e-06,
+      "loss": 0.5827,
+      "step": 12041
+    },
+    {
+      "epoch": 1.026161056668087,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2068234537495493e-06,
+      "loss": 0.7938,
+      "step": 12042
+    },
+    {
+      "epoch": 1.026246271836387,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2066555188608769e-06,
+      "loss": 0.6356,
+      "step": 12043
+    },
+    {
+      "epoch": 1.026331487004687,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2064875847554834e-06,
+      "loss": 0.7821,
+      "step": 12044
+    },
+    {
+      "epoch": 1.0264167021729869,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2063196514364024e-06,
+      "loss": 0.6765,
+      "step": 12045
+    },
+    {
+      "epoch": 1.0265019173412868,
+      "grad_norm": 12.625,
+      "learning_rate": 1.2061517189066695e-06,
+      "loss": 0.3162,
+      "step": 12046
+    },
+    {
+      "epoch": 1.0265871325095868,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2059837871693193e-06,
+      "loss": 0.9344,
+      "step": 12047
+    },
+    {
+      "epoch": 1.0266723476778867,
+      "grad_norm": 17.125,
+      "learning_rate": 1.205815856227386e-06,
+      "loss": 0.8398,
+      "step": 12048
+    },
+    {
+      "epoch": 1.0267575628461867,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.2056479260839053e-06,
+      "loss": 0.4291,
+      "step": 12049
+    },
+    {
+      "epoch": 1.0268427780144866,
+      "grad_norm": 25.75,
+      "learning_rate": 1.2054799967419109e-06,
+      "loss": 0.518,
+      "step": 12050
+    },
+    {
+      "epoch": 1.0269279931827866,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2053120682044372e-06,
+      "loss": 0.839,
+      "step": 12051
+    },
+    {
+      "epoch": 1.0270132083510866,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.2051441404745199e-06,
+      "loss": 0.4096,
+      "step": 12052
+    },
+    {
+      "epoch": 1.0270984235193865,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.2049762135551925e-06,
+      "loss": 0.4311,
+      "step": 12053
+    },
+    {
+      "epoch": 1.0271836386876865,
+      "grad_norm": 15.875,
+      "learning_rate": 1.204808287449491e-06,
+      "loss": 0.8613,
+      "step": 12054
+    },
+    {
+      "epoch": 1.0272688538559864,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2046403621604483e-06,
+      "loss": 0.9077,
+      "step": 12055
+    },
+    {
+      "epoch": 1.0273540690242864,
+      "grad_norm": 15.125,
+      "learning_rate": 1.2044724376911005e-06,
+      "loss": 0.5185,
+      "step": 12056
+    },
+    {
+      "epoch": 1.0274392841925863,
+      "grad_norm": 15.125,
+      "learning_rate": 1.2043045140444812e-06,
+      "loss": 0.741,
+      "step": 12057
+    },
+    {
+      "epoch": 1.0275244993608863,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2041365912236257e-06,
+      "loss": 0.5084,
+      "step": 12058
+    },
+    {
+      "epoch": 1.0276097145291863,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.203968669231568e-06,
+      "loss": 0.7012,
+      "step": 12059
+    },
+    {
+      "epoch": 1.0276949296974862,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2038007480713424e-06,
+      "loss": 0.7373,
+      "step": 12060
+    },
+    {
+      "epoch": 1.0277801448657862,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2036328277459842e-06,
+      "loss": 0.8328,
+      "step": 12061
+    },
+    {
+      "epoch": 1.0278653600340861,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.203464908258527e-06,
+      "loss": 0.5911,
+      "step": 12062
+    },
+    {
+      "epoch": 1.027950575202386,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2032969896120058e-06,
+      "loss": 1.1189,
+      "step": 12063
+    },
+    {
+      "epoch": 1.028035790370686,
+      "grad_norm": 14.75,
+      "learning_rate": 1.2031290718094552e-06,
+      "loss": 0.3555,
+      "step": 12064
+    },
+    {
+      "epoch": 1.028121005538986,
+      "grad_norm": 11.25,
+      "learning_rate": 1.202961154853909e-06,
+      "loss": 0.5411,
+      "step": 12065
+    },
+    {
+      "epoch": 1.028206220707286,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.2027932387484025e-06,
+      "loss": 0.4445,
+      "step": 12066
+    },
+    {
+      "epoch": 1.028291435875586,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2026253234959695e-06,
+      "loss": 0.4197,
+      "step": 12067
+    },
+    {
+      "epoch": 1.0283766510438859,
+      "grad_norm": 13.5,
+      "learning_rate": 1.2024574090996444e-06,
+      "loss": 0.4373,
+      "step": 12068
+    },
+    {
+      "epoch": 1.0284618662121858,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2022894955624619e-06,
+      "loss": 0.5399,
+      "step": 12069
+    },
+    {
+      "epoch": 1.0285470813804858,
+      "grad_norm": 13.25,
+      "learning_rate": 1.202121582887456e-06,
+      "loss": 0.7038,
+      "step": 12070
+    },
+    {
+      "epoch": 1.0286322965487857,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2019536710776611e-06,
+      "loss": 0.6916,
+      "step": 12071
+    },
+    {
+      "epoch": 1.0287175117170857,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.201785760136112e-06,
+      "loss": 0.2329,
+      "step": 12072
+    },
+    {
+      "epoch": 1.0288027268853857,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2016178500658426e-06,
+      "loss": 0.4703,
+      "step": 12073
+    },
+    {
+      "epoch": 1.0288879420536856,
+      "grad_norm": 33.5,
+      "learning_rate": 1.2014499408698877e-06,
+      "loss": 0.6961,
+      "step": 12074
+    },
+    {
+      "epoch": 1.0289731572219856,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.2012820325512805e-06,
+      "loss": 0.3699,
+      "step": 12075
+    },
+    {
+      "epoch": 1.0290583723902855,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2011141251130564e-06,
+      "loss": 0.7795,
+      "step": 12076
+    },
+    {
+      "epoch": 1.0291435875585855,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2009462185582489e-06,
+      "loss": 0.2929,
+      "step": 12077
+    },
+    {
+      "epoch": 1.0292288027268854,
+      "grad_norm": 12.0,
+      "learning_rate": 1.200778312889893e-06,
+      "loss": 0.5457,
+      "step": 12078
+    },
+    {
+      "epoch": 1.0293140178951854,
+      "grad_norm": 17.0,
+      "learning_rate": 1.200610408111022e-06,
+      "loss": 0.8433,
+      "step": 12079
+    },
+    {
+      "epoch": 1.0293992330634854,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2004425042246706e-06,
+      "loss": 0.5004,
+      "step": 12080
+    },
+    {
+      "epoch": 1.0294844482317853,
+      "grad_norm": 14.0,
+      "learning_rate": 1.2002746012338733e-06,
+      "loss": 0.5252,
+      "step": 12081
+    },
+    {
+      "epoch": 1.0295696634000853,
+      "grad_norm": 15.875,
+      "learning_rate": 1.2001066991416642e-06,
+      "loss": 0.3842,
+      "step": 12082
+    },
+    {
+      "epoch": 1.0296548785683852,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.1999387979510767e-06,
+      "loss": 0.3632,
+      "step": 12083
+    },
+    {
+      "epoch": 1.0297400937366852,
+      "grad_norm": 21.625,
+      "learning_rate": 1.1997708976651457e-06,
+      "loss": 0.9221,
+      "step": 12084
+    },
+    {
+      "epoch": 1.0298253089049851,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1996029982869049e-06,
+      "loss": 0.7283,
+      "step": 12085
+    },
+    {
+      "epoch": 1.029910524073285,
+      "grad_norm": 26.625,
+      "learning_rate": 1.1994350998193886e-06,
+      "loss": 0.7366,
+      "step": 12086
+    },
+    {
+      "epoch": 1.029995739241585,
+      "grad_norm": 20.625,
+      "learning_rate": 1.199267202265631e-06,
+      "loss": 0.8069,
+      "step": 12087
+    },
+    {
+      "epoch": 1.030080954409885,
+      "grad_norm": 11.625,
+      "learning_rate": 1.1990993056286662e-06,
+      "loss": 0.2936,
+      "step": 12088
+    },
+    {
+      "epoch": 1.030166169578185,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.1989314099115277e-06,
+      "loss": 0.4712,
+      "step": 12089
+    },
+    {
+      "epoch": 1.030251384746485,
+      "grad_norm": 19.625,
+      "learning_rate": 1.1987635151172507e-06,
+      "loss": 0.8653,
+      "step": 12090
+    },
+    {
+      "epoch": 1.0303365999147849,
+      "grad_norm": 11.75,
+      "learning_rate": 1.1985956212488682e-06,
+      "loss": 0.4539,
+      "step": 12091
+    },
+    {
+      "epoch": 1.0304218150830848,
+      "grad_norm": 21.0,
+      "learning_rate": 1.1984277283094142e-06,
+      "loss": 0.5299,
+      "step": 12092
+    },
+    {
+      "epoch": 1.0305070302513848,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1982598363019232e-06,
+      "loss": 0.4305,
+      "step": 12093
+    },
+    {
+      "epoch": 1.0305922454196847,
+      "grad_norm": 12.5,
+      "learning_rate": 1.198091945229429e-06,
+      "loss": 0.5654,
+      "step": 12094
+    },
+    {
+      "epoch": 1.0306774605879847,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1979240550949654e-06,
+      "loss": 0.7586,
+      "step": 12095
+    },
+    {
+      "epoch": 1.0307626757562847,
+      "grad_norm": 15.875,
+      "learning_rate": 1.1977561659015667e-06,
+      "loss": 0.6773,
+      "step": 12096
+    },
+    {
+      "epoch": 1.0308478909245846,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1975882776522663e-06,
+      "loss": 0.6426,
+      "step": 12097
+    },
+    {
+      "epoch": 1.0309331060928846,
+      "grad_norm": 46.5,
+      "learning_rate": 1.197420390350099e-06,
+      "loss": 0.7543,
+      "step": 12098
+    },
+    {
+      "epoch": 1.0310183212611845,
+      "grad_norm": 9.75,
+      "learning_rate": 1.197252503998098e-06,
+      "loss": 0.2529,
+      "step": 12099
+    },
+    {
+      "epoch": 1.0311035364294845,
+      "grad_norm": 13.75,
+      "learning_rate": 1.197084618599297e-06,
+      "loss": 0.5301,
+      "step": 12100
+    },
+    {
+      "epoch": 1.0311887515977844,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1969167341567302e-06,
+      "loss": 0.3899,
+      "step": 12101
+    },
+    {
+      "epoch": 1.0312739667660844,
+      "grad_norm": 12.5,
+      "learning_rate": 1.1967488506734313e-06,
+      "loss": 0.4672,
+      "step": 12102
+    },
+    {
+      "epoch": 1.0313591819343844,
+      "grad_norm": 14.0,
+      "learning_rate": 1.1965809681524345e-06,
+      "loss": 0.4629,
+      "step": 12103
+    },
+    {
+      "epoch": 1.0314443971026843,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.1964130865967728e-06,
+      "loss": 0.2669,
+      "step": 12104
+    },
+    {
+      "epoch": 1.0315296122709843,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.196245206009481e-06,
+      "loss": 0.5727,
+      "step": 12105
+    },
+    {
+      "epoch": 1.0316148274392842,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1960773263935927e-06,
+      "loss": 1.0707,
+      "step": 12106
+    },
+    {
+      "epoch": 1.0317000426075842,
+      "grad_norm": 13.5,
+      "learning_rate": 1.1959094477521408e-06,
+      "loss": 0.3832,
+      "step": 12107
+    },
+    {
+      "epoch": 1.0317852577758841,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.1957415700881597e-06,
+      "loss": 0.7825,
+      "step": 12108
+    },
+    {
+      "epoch": 1.031870472944184,
+      "grad_norm": 16.0,
+      "learning_rate": 1.195573693404683e-06,
+      "loss": 0.7548,
+      "step": 12109
+    },
+    {
+      "epoch": 1.031955688112484,
+      "grad_norm": 13.625,
+      "learning_rate": 1.1954058177047442e-06,
+      "loss": 0.6043,
+      "step": 12110
+    },
+    {
+      "epoch": 1.032040903280784,
+      "grad_norm": 13.875,
+      "learning_rate": 1.1952379429913773e-06,
+      "loss": 0.6998,
+      "step": 12111
+    },
+    {
+      "epoch": 1.032126118449084,
+      "grad_norm": 14.375,
+      "learning_rate": 1.1950700692676158e-06,
+      "loss": 0.7099,
+      "step": 12112
+    },
+    {
+      "epoch": 1.032211333617384,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1949021965364934e-06,
+      "loss": 0.6449,
+      "step": 12113
+    },
+    {
+      "epoch": 1.0322965487856839,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1947343248010441e-06,
+      "loss": 0.8241,
+      "step": 12114
+    },
+    {
+      "epoch": 1.0323817639539838,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1945664540643006e-06,
+      "loss": 0.6415,
+      "step": 12115
+    },
+    {
+      "epoch": 1.0324669791222838,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1943985843292973e-06,
+      "loss": 0.743,
+      "step": 12116
+    },
+    {
+      "epoch": 1.0325521942905838,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1942307155990673e-06,
+      "loss": 0.6418,
+      "step": 12117
+    },
+    {
+      "epoch": 1.0326374094588837,
+      "grad_norm": 24.25,
+      "learning_rate": 1.1940628478766445e-06,
+      "loss": 0.7129,
+      "step": 12118
+    },
+    {
+      "epoch": 1.0327226246271837,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1938949811650622e-06,
+      "loss": 0.7839,
+      "step": 12119
+    },
+    {
+      "epoch": 1.0328078397954836,
+      "grad_norm": 18.0,
+      "learning_rate": 1.193727115467354e-06,
+      "loss": 0.7728,
+      "step": 12120
+    },
+    {
+      "epoch": 1.0328930549637836,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1935592507865534e-06,
+      "loss": 0.5396,
+      "step": 12121
+    },
+    {
+      "epoch": 1.0329782701320835,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.193391387125694e-06,
+      "loss": 0.3635,
+      "step": 12122
+    },
+    {
+      "epoch": 1.0330634853003835,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1932235244878093e-06,
+      "loss": 0.5104,
+      "step": 12123
+    },
+    {
+      "epoch": 1.0331487004686835,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1930556628759326e-06,
+      "loss": 0.64,
+      "step": 12124
+    },
+    {
+      "epoch": 1.0332339156369834,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.192887802293097e-06,
+      "loss": 0.5729,
+      "step": 12125
+    },
+    {
+      "epoch": 1.0333191308052834,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1927199427423364e-06,
+      "loss": 0.777,
+      "step": 12126
+    },
+    {
+      "epoch": 1.0334043459735833,
+      "grad_norm": 18.75,
+      "learning_rate": 1.1925520842266839e-06,
+      "loss": 0.5323,
+      "step": 12127
+    },
+    {
+      "epoch": 1.0334895611418833,
+      "grad_norm": 16.875,
+      "learning_rate": 1.192384226749173e-06,
+      "loss": 0.9187,
+      "step": 12128
+    },
+    {
+      "epoch": 1.0335747763101832,
+      "grad_norm": 19.5,
+      "learning_rate": 1.192216370312837e-06,
+      "loss": 0.7986,
+      "step": 12129
+    },
+    {
+      "epoch": 1.0336599914784832,
+      "grad_norm": 24.0,
+      "learning_rate": 1.1920485149207098e-06,
+      "loss": 1.1853,
+      "step": 12130
+    },
+    {
+      "epoch": 1.0337452066467832,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.1918806605758241e-06,
+      "loss": 0.4111,
+      "step": 12131
+    },
+    {
+      "epoch": 1.0338304218150831,
+      "grad_norm": 18.125,
+      "learning_rate": 1.191712807281213e-06,
+      "loss": 0.7398,
+      "step": 12132
+    },
+    {
+      "epoch": 1.033915636983383,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.1915449550399101e-06,
+      "loss": 1.1856,
+      "step": 12133
+    },
+    {
+      "epoch": 1.034000852151683,
+      "grad_norm": 15.125,
+      "learning_rate": 1.1913771038549487e-06,
+      "loss": 0.6455,
+      "step": 12134
+    },
+    {
+      "epoch": 1.034086067319983,
+      "grad_norm": 17.75,
+      "learning_rate": 1.191209253729362e-06,
+      "loss": 0.5236,
+      "step": 12135
+    },
+    {
+      "epoch": 1.034171282488283,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1910414046661831e-06,
+      "loss": 0.7039,
+      "step": 12136
+    },
+    {
+      "epoch": 1.034256497656583,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1908735566684455e-06,
+      "loss": 0.3722,
+      "step": 12137
+    },
+    {
+      "epoch": 1.0343417128248829,
+      "grad_norm": 18.25,
+      "learning_rate": 1.1907057097391819e-06,
+      "loss": 0.441,
+      "step": 12138
+    },
+    {
+      "epoch": 1.0344269279931828,
+      "grad_norm": 16.625,
+      "learning_rate": 1.1905378638814266e-06,
+      "loss": 0.5087,
+      "step": 12139
+    },
+    {
+      "epoch": 1.0345121431614828,
+      "grad_norm": 15.0,
+      "learning_rate": 1.190370019098211e-06,
+      "loss": 0.7126,
+      "step": 12140
+    },
+    {
+      "epoch": 1.0345973583297827,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1902021753925695e-06,
+      "loss": 0.3717,
+      "step": 12141
+    },
+    {
+      "epoch": 1.0346825734980827,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1900343327675345e-06,
+      "loss": 0.8439,
+      "step": 12142
+    },
+    {
+      "epoch": 1.0347677886663826,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1898664912261397e-06,
+      "loss": 0.7514,
+      "step": 12143
+    },
+    {
+      "epoch": 1.0348530038346826,
+      "grad_norm": 14.125,
+      "learning_rate": 1.1896986507714175e-06,
+      "loss": 0.5308,
+      "step": 12144
+    },
+    {
+      "epoch": 1.0349382190029826,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1895308114064018e-06,
+      "loss": 0.6424,
+      "step": 12145
+    },
+    {
+      "epoch": 1.0350234341712825,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.1893629731341247e-06,
+      "loss": 0.6105,
+      "step": 12146
+    },
+    {
+      "epoch": 1.0351086493395825,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.1891951359576203e-06,
+      "loss": 0.4787,
+      "step": 12147
+    },
+    {
+      "epoch": 1.0351938645078824,
+      "grad_norm": 12.375,
+      "learning_rate": 1.1890272998799205e-06,
+      "loss": 0.5066,
+      "step": 12148
+    },
+    {
+      "epoch": 1.0352790796761824,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1888594649040585e-06,
+      "loss": 0.5076,
+      "step": 12149
+    },
+    {
+      "epoch": 1.0353642948444823,
+      "grad_norm": 19.125,
+      "learning_rate": 1.188691631033068e-06,
+      "loss": 0.9105,
+      "step": 12150
+    },
+    {
+      "epoch": 1.0354495100127823,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1885237982699808e-06,
+      "loss": 0.3469,
+      "step": 12151
+    },
+    {
+      "epoch": 1.0355347251810822,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1883559666178308e-06,
+      "loss": 0.6317,
+      "step": 12152
+    },
+    {
+      "epoch": 1.0356199403493822,
+      "grad_norm": 12.875,
+      "learning_rate": 1.18818813607965e-06,
+      "loss": 0.5086,
+      "step": 12153
+    },
+    {
+      "epoch": 1.0357051555176822,
+      "grad_norm": 13.5,
+      "learning_rate": 1.188020306658472e-06,
+      "loss": 0.7064,
+      "step": 12154
+    },
+    {
+      "epoch": 1.0357903706859821,
+      "grad_norm": 17.5,
+      "learning_rate": 1.1878524783573298e-06,
+      "loss": 0.7342,
+      "step": 12155
+    },
+    {
+      "epoch": 1.035875585854282,
+      "grad_norm": 13.375,
+      "learning_rate": 1.1876846511792557e-06,
+      "loss": 0.548,
+      "step": 12156
+    },
+    {
+      "epoch": 1.035960801022582,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1875168251272822e-06,
+      "loss": 0.8826,
+      "step": 12157
+    },
+    {
+      "epoch": 1.036046016190882,
+      "grad_norm": 15.75,
+      "learning_rate": 1.1873490002044428e-06,
+      "loss": 0.6361,
+      "step": 12158
+    },
+    {
+      "epoch": 1.036131231359182,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1871811764137694e-06,
+      "loss": 0.6078,
+      "step": 12159
+    },
+    {
+      "epoch": 1.036216446527482,
+      "grad_norm": 20.5,
+      "learning_rate": 1.187013353758296e-06,
+      "loss": 0.7864,
+      "step": 12160
+    },
+    {
+      "epoch": 1.0363016616957819,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1868455322410542e-06,
+      "loss": 0.8146,
+      "step": 12161
+    },
+    {
+      "epoch": 1.0363868768640818,
+      "grad_norm": 11.625,
+      "learning_rate": 1.1866777118650772e-06,
+      "loss": 0.3335,
+      "step": 12162
+    },
+    {
+      "epoch": 1.0364720920323818,
+      "grad_norm": 12.0,
+      "learning_rate": 1.1865098926333982e-06,
+      "loss": 0.5209,
+      "step": 12163
+    },
+    {
+      "epoch": 1.0365573072006817,
+      "grad_norm": 11.375,
+      "learning_rate": 1.1863420745490486e-06,
+      "loss": 0.3578,
+      "step": 12164
+    },
+    {
+      "epoch": 1.0366425223689817,
+      "grad_norm": 17.5,
+      "learning_rate": 1.186174257615062e-06,
+      "loss": 0.7336,
+      "step": 12165
+    },
+    {
+      "epoch": 1.0367277375372816,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1860064418344705e-06,
+      "loss": 0.8736,
+      "step": 12166
+    },
+    {
+      "epoch": 1.0368129527055816,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.1858386272103072e-06,
+      "loss": 0.5575,
+      "step": 12167
+    },
+    {
+      "epoch": 1.0368981678738816,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1856708137456042e-06,
+      "loss": 0.6453,
+      "step": 12168
+    },
+    {
+      "epoch": 1.0369833830421815,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1855030014433945e-06,
+      "loss": 0.7769,
+      "step": 12169
+    },
+    {
+      "epoch": 1.0370685982104815,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1853351903067103e-06,
+      "loss": 0.6374,
+      "step": 12170
+    },
+    {
+      "epoch": 1.0371538133787814,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.1851673803385848e-06,
+      "loss": 0.3573,
+      "step": 12171
+    },
+    {
+      "epoch": 1.0372390285470814,
+      "grad_norm": 20.25,
+      "learning_rate": 1.1849995715420493e-06,
+      "loss": 0.6214,
+      "step": 12172
+    },
+    {
+      "epoch": 1.0373242437153813,
+      "grad_norm": 12.625,
+      "learning_rate": 1.1848317639201372e-06,
+      "loss": 0.3044,
+      "step": 12173
+    },
+    {
+      "epoch": 1.0374094588836813,
+      "grad_norm": 13.0,
+      "learning_rate": 1.1846639574758803e-06,
+      "loss": 0.6743,
+      "step": 12174
+    },
+    {
+      "epoch": 1.0374946740519813,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1844961522123116e-06,
+      "loss": 0.5411,
+      "step": 12175
+    },
+    {
+      "epoch": 1.0375798892202812,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1843283481324632e-06,
+      "loss": 0.81,
+      "step": 12176
+    },
+    {
+      "epoch": 1.0376651043885812,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.1841605452393676e-06,
+      "loss": 0.4303,
+      "step": 12177
+    },
+    {
+      "epoch": 1.0377503195568811,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1839927435360571e-06,
+      "loss": 0.6501,
+      "step": 12178
+    },
+    {
+      "epoch": 1.037835534725181,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.1838249430255647e-06,
+      "loss": 0.38,
+      "step": 12179
+    },
+    {
+      "epoch": 1.037920749893481,
+      "grad_norm": 13.125,
+      "learning_rate": 1.1836571437109216e-06,
+      "loss": 0.5126,
+      "step": 12180
+    },
+    {
+      "epoch": 1.038005965061781,
+      "grad_norm": 20.125,
+      "learning_rate": 1.1834893455951605e-06,
+      "loss": 0.7334,
+      "step": 12181
+    },
+    {
+      "epoch": 1.038091180230081,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1833215486813141e-06,
+      "loss": 0.6822,
+      "step": 12182
+    },
+    {
+      "epoch": 1.038176395398381,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.183153752972414e-06,
+      "loss": 0.279,
+      "step": 12183
+    },
+    {
+      "epoch": 1.0382616105666809,
+      "grad_norm": 16.625,
+      "learning_rate": 1.1829859584714933e-06,
+      "loss": 0.7268,
+      "step": 12184
+    },
+    {
+      "epoch": 1.0383468257349808,
+      "grad_norm": 21.0,
+      "learning_rate": 1.1828181651815832e-06,
+      "loss": 0.9824,
+      "step": 12185
+    },
+    {
+      "epoch": 1.0384320409032808,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.1826503731057167e-06,
+      "loss": 0.6139,
+      "step": 12186
+    },
+    {
+      "epoch": 1.0385172560715807,
+      "grad_norm": 20.125,
+      "learning_rate": 1.182482582246926e-06,
+      "loss": 0.6879,
+      "step": 12187
+    },
+    {
+      "epoch": 1.0386024712398807,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.1823147926082427e-06,
+      "loss": 0.5429,
+      "step": 12188
+    },
+    {
+      "epoch": 1.0386876864081807,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.182147004192699e-06,
+      "loss": 0.402,
+      "step": 12189
+    },
+    {
+      "epoch": 1.0387729015764806,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1819792170033273e-06,
+      "loss": 0.8406,
+      "step": 12190
+    },
+    {
+      "epoch": 1.0388581167447806,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.1818114310431592e-06,
+      "loss": 0.5648,
+      "step": 12191
+    },
+    {
+      "epoch": 1.0389433319130805,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.1816436463152275e-06,
+      "loss": 0.3962,
+      "step": 12192
+    },
+    {
+      "epoch": 1.0390285470813805,
+      "grad_norm": 19.875,
+      "learning_rate": 1.1814758628225636e-06,
+      "loss": 0.6288,
+      "step": 12193
+    },
+    {
+      "epoch": 1.0391137622496804,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.1813080805682e-06,
+      "loss": 0.4618,
+      "step": 12194
+    },
+    {
+      "epoch": 1.0391989774179804,
+      "grad_norm": 15.125,
+      "learning_rate": 1.181140299555168e-06,
+      "loss": 0.3864,
+      "step": 12195
+    },
+    {
+      "epoch": 1.0392841925862804,
+      "grad_norm": 21.125,
+      "learning_rate": 1.180972519786501e-06,
+      "loss": 0.9002,
+      "step": 12196
+    },
+    {
+      "epoch": 1.0393694077545803,
+      "grad_norm": 21.25,
+      "learning_rate": 1.1808047412652292e-06,
+      "loss": 1.0315,
+      "step": 12197
+    },
+    {
+      "epoch": 1.0394546229228803,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.1806369639943855e-06,
+      "loss": 0.3812,
+      "step": 12198
+    },
+    {
+      "epoch": 1.0395398380911802,
+      "grad_norm": 14.5,
+      "learning_rate": 1.1804691879770016e-06,
+      "loss": 0.7292,
+      "step": 12199
+    },
+    {
+      "epoch": 1.0396250532594802,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.1803014132161091e-06,
+      "loss": 0.8382,
+      "step": 12200
+    },
+    {
+      "epoch": 1.0397102684277801,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1801336397147407e-06,
+      "loss": 0.5333,
+      "step": 12201
+    },
+    {
+      "epoch": 1.03979548359608,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1799658674759274e-06,
+      "loss": 0.5389,
+      "step": 12202
+    },
+    {
+      "epoch": 1.03988069876438,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.1797980965027011e-06,
+      "loss": 0.4826,
+      "step": 12203
+    },
+    {
+      "epoch": 1.03996591393268,
+      "grad_norm": 25.625,
+      "learning_rate": 1.1796303267980943e-06,
+      "loss": 0.9351,
+      "step": 12204
+    },
+    {
+      "epoch": 1.04005112910098,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1794625583651381e-06,
+      "loss": 0.4966,
+      "step": 12205
+    },
+    {
+      "epoch": 1.04013634426928,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.179294791206864e-06,
+      "loss": 0.6054,
+      "step": 12206
+    },
+    {
+      "epoch": 1.0402215594375799,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1791270253263046e-06,
+      "loss": 0.4565,
+      "step": 12207
+    },
+    {
+      "epoch": 1.0403067746058798,
+      "grad_norm": 15.5,
+      "learning_rate": 1.1789592607264907e-06,
+      "loss": 0.702,
+      "step": 12208
+    },
+    {
+      "epoch": 1.0403919897741798,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1787914974104546e-06,
+      "loss": 0.5439,
+      "step": 12209
+    },
+    {
+      "epoch": 1.0404772049424797,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1786237353812276e-06,
+      "loss": 0.3849,
+      "step": 12210
+    },
+    {
+      "epoch": 1.0405624201107797,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.1784559746418417e-06,
+      "loss": 0.6931,
+      "step": 12211
+    },
+    {
+      "epoch": 1.0406476352790797,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1782882151953285e-06,
+      "loss": 0.5247,
+      "step": 12212
+    },
+    {
+      "epoch": 1.0407328504473796,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1781204570447189e-06,
+      "loss": 0.7136,
+      "step": 12213
+    },
+    {
+      "epoch": 1.0408180656156796,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.177952700193045e-06,
+      "loss": 0.5578,
+      "step": 12214
+    },
+    {
+      "epoch": 1.0409032807839795,
+      "grad_norm": 14.125,
+      "learning_rate": 1.1777849446433382e-06,
+      "loss": 0.4661,
+      "step": 12215
+    },
+    {
+      "epoch": 1.0409884959522795,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1776171903986305e-06,
+      "loss": 0.6947,
+      "step": 12216
+    },
+    {
+      "epoch": 1.0410737111205794,
+      "grad_norm": 14.0,
+      "learning_rate": 1.1774494374619525e-06,
+      "loss": 0.5227,
+      "step": 12217
+    },
+    {
+      "epoch": 1.0411589262888794,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1772816858363363e-06,
+      "loss": 0.3847,
+      "step": 12218
+    },
+    {
+      "epoch": 1.0412441414571794,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.1771139355248131e-06,
+      "loss": 0.4344,
+      "step": 12219
+    },
+    {
+      "epoch": 1.0413293566254793,
+      "grad_norm": 16.375,
+      "learning_rate": 1.176946186530415e-06,
+      "loss": 0.9283,
+      "step": 12220
+    },
+    {
+      "epoch": 1.0414145717937793,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1767784388561718e-06,
+      "loss": 0.5963,
+      "step": 12221
+    },
+    {
+      "epoch": 1.0414997869620792,
+      "grad_norm": 19.75,
+      "learning_rate": 1.1766106925051164e-06,
+      "loss": 0.8079,
+      "step": 12222
+    },
+    {
+      "epoch": 1.0415850021303792,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1764429474802792e-06,
+      "loss": 0.6416,
+      "step": 12223
+    },
+    {
+      "epoch": 1.0416702172986791,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.1762752037846922e-06,
+      "loss": 0.615,
+      "step": 12224
+    },
+    {
+      "epoch": 1.041755432466979,
+      "grad_norm": 16.5,
+      "learning_rate": 1.176107461421386e-06,
+      "loss": 0.5623,
+      "step": 12225
+    },
+    {
+      "epoch": 1.041840647635279,
+      "grad_norm": 15.125,
+      "learning_rate": 1.175939720393393e-06,
+      "loss": 0.7099,
+      "step": 12226
+    },
+    {
+      "epoch": 1.041925862803579,
+      "grad_norm": 13.125,
+      "learning_rate": 1.1757719807037432e-06,
+      "loss": 0.4801,
+      "step": 12227
+    },
+    {
+      "epoch": 1.042011077971879,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.1756042423554688e-06,
+      "loss": 0.4208,
+      "step": 12228
+    },
+    {
+      "epoch": 1.042096293140179,
+      "grad_norm": 13.125,
+      "learning_rate": 1.1754365053516002e-06,
+      "loss": 0.2615,
+      "step": 12229
+    },
+    {
+      "epoch": 1.0421815083084789,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.1752687696951687e-06,
+      "loss": 0.5503,
+      "step": 12230
+    },
+    {
+      "epoch": 1.0422667234767788,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.1751010353892062e-06,
+      "loss": 0.3954,
+      "step": 12231
+    },
+    {
+      "epoch": 1.0423519386450788,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.1749333024367427e-06,
+      "loss": 0.4969,
+      "step": 12232
+    },
+    {
+      "epoch": 1.0424371538133788,
+      "grad_norm": 13.625,
+      "learning_rate": 1.1747655708408104e-06,
+      "loss": 0.6721,
+      "step": 12233
+    },
+    {
+      "epoch": 1.0425223689816787,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.17459784060444e-06,
+      "loss": 0.6874,
+      "step": 12234
+    },
+    {
+      "epoch": 1.0426075841499787,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.174430111730662e-06,
+      "loss": 0.3838,
+      "step": 12235
+    },
+    {
+      "epoch": 1.0426927993182786,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.1742623842225082e-06,
+      "loss": 0.4641,
+      "step": 12236
+    },
+    {
+      "epoch": 1.0427780144865786,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1740946580830094e-06,
+      "loss": 0.5698,
+      "step": 12237
+    },
+    {
+      "epoch": 1.0428632296548785,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.173926933315196e-06,
+      "loss": 0.5265,
+      "step": 12238
+    },
+    {
+      "epoch": 1.0429484448231785,
+      "grad_norm": 19.75,
+      "learning_rate": 1.1737592099220998e-06,
+      "loss": 0.6866,
+      "step": 12239
+    },
+    {
+      "epoch": 1.0430336599914785,
+      "grad_norm": 11.5,
+      "learning_rate": 1.173591487906751e-06,
+      "loss": 0.3103,
+      "step": 12240
+    },
+    {
+      "epoch": 1.0431188751597784,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.173423767272181e-06,
+      "loss": 0.5651,
+      "step": 12241
+    },
+    {
+      "epoch": 1.0432040903280784,
+      "grad_norm": 38.25,
+      "learning_rate": 1.1732560480214203e-06,
+      "loss": 0.9726,
+      "step": 12242
+    },
+    {
+      "epoch": 1.0432893054963783,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1730883301575003e-06,
+      "loss": 0.5436,
+      "step": 12243
+    },
+    {
+      "epoch": 1.0433745206646783,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1729206136834518e-06,
+      "loss": 0.639,
+      "step": 12244
+    },
+    {
+      "epoch": 1.0434597358329782,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1727528986023046e-06,
+      "loss": 0.4507,
+      "step": 12245
+    },
+    {
+      "epoch": 1.0435449510012782,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1725851849170909e-06,
+      "loss": 0.6545,
+      "step": 12246
+    },
+    {
+      "epoch": 1.0436301661695782,
+      "grad_norm": 11.25,
+      "learning_rate": 1.1724174726308401e-06,
+      "loss": 0.2853,
+      "step": 12247
+    },
+    {
+      "epoch": 1.0437153813378781,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.1722497617465841e-06,
+      "loss": 0.6409,
+      "step": 12248
+    },
+    {
+      "epoch": 1.043800596506178,
+      "grad_norm": 12.875,
+      "learning_rate": 1.172082052267353e-06,
+      "loss": 0.5506,
+      "step": 12249
+    },
+    {
+      "epoch": 1.043885811674478,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1719143441961772e-06,
+      "loss": 0.654,
+      "step": 12250
+    },
+    {
+      "epoch": 1.043971026842778,
+      "grad_norm": 17.25,
+      "learning_rate": 1.171746637536088e-06,
+      "loss": 0.6704,
+      "step": 12251
+    },
+    {
+      "epoch": 1.044056242011078,
+      "grad_norm": 13.875,
+      "learning_rate": 1.1715789322901161e-06,
+      "loss": 0.5007,
+      "step": 12252
+    },
+    {
+      "epoch": 1.044141457179378,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1714112284612912e-06,
+      "loss": 0.4469,
+      "step": 12253
+    },
+    {
+      "epoch": 1.0442266723476779,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1712435260526446e-06,
+      "loss": 0.8604,
+      "step": 12254
+    },
+    {
+      "epoch": 1.0443118875159778,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1710758250672065e-06,
+      "loss": 0.4008,
+      "step": 12255
+    },
+    {
+      "epoch": 1.0443971026842778,
+      "grad_norm": 28.75,
+      "learning_rate": 1.1709081255080078e-06,
+      "loss": 0.674,
+      "step": 12256
+    },
+    {
+      "epoch": 1.0444823178525777,
+      "grad_norm": 20.125,
+      "learning_rate": 1.1707404273780787e-06,
+      "loss": 1.0972,
+      "step": 12257
+    },
+    {
+      "epoch": 1.0445675330208777,
+      "grad_norm": 16.75,
+      "learning_rate": 1.17057273068045e-06,
+      "loss": 0.6659,
+      "step": 12258
+    },
+    {
+      "epoch": 1.0446527481891776,
+      "grad_norm": 18.25,
+      "learning_rate": 1.1704050354181516e-06,
+      "loss": 0.8329,
+      "step": 12259
+    },
+    {
+      "epoch": 1.0447379633574776,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1702373415942144e-06,
+      "loss": 0.6349,
+      "step": 12260
+    },
+    {
+      "epoch": 1.0448231785257776,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1700696492116692e-06,
+      "loss": 0.837,
+      "step": 12261
+    },
+    {
+      "epoch": 1.0449083936940775,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.169901958273545e-06,
+      "loss": 0.707,
+      "step": 12262
+    },
+    {
+      "epoch": 1.0449936088623775,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.1697342687828734e-06,
+      "loss": 0.2384,
+      "step": 12263
+    },
+    {
+      "epoch": 1.0450788240306774,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1695665807426843e-06,
+      "loss": 0.484,
+      "step": 12264
+    },
+    {
+      "epoch": 1.0451640391989774,
+      "grad_norm": 20.75,
+      "learning_rate": 1.1693988941560076e-06,
+      "loss": 0.542,
+      "step": 12265
+    },
+    {
+      "epoch": 1.0452492543672773,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1692312090258743e-06,
+      "loss": 0.9995,
+      "step": 12266
+    },
+    {
+      "epoch": 1.0453344695355773,
+      "grad_norm": 17.125,
+      "learning_rate": 1.169063525355314e-06,
+      "loss": 0.7402,
+      "step": 12267
+    },
+    {
+      "epoch": 1.0454196847038772,
+      "grad_norm": 13.125,
+      "learning_rate": 1.1688958431473573e-06,
+      "loss": 0.5055,
+      "step": 12268
+    },
+    {
+      "epoch": 1.0455048998721772,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1687281624050344e-06,
+      "loss": 0.4756,
+      "step": 12269
+    },
+    {
+      "epoch": 1.0455901150404772,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.168560483131375e-06,
+      "loss": 0.4452,
+      "step": 12270
+    },
+    {
+      "epoch": 1.0456753302087771,
+      "grad_norm": 20.75,
+      "learning_rate": 1.16839280532941e-06,
+      "loss": 0.7191,
+      "step": 12271
+    },
+    {
+      "epoch": 1.045760545377077,
+      "grad_norm": 18.75,
+      "learning_rate": 1.1682251290021687e-06,
+      "loss": 0.9346,
+      "step": 12272
+    },
+    {
+      "epoch": 1.045845760545377,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.1680574541526818e-06,
+      "loss": 0.3446,
+      "step": 12273
+    },
+    {
+      "epoch": 1.045930975713677,
+      "grad_norm": 11.625,
+      "learning_rate": 1.1678897807839786e-06,
+      "loss": 0.3199,
+      "step": 12274
+    },
+    {
+      "epoch": 1.046016190881977,
+      "grad_norm": 21.0,
+      "learning_rate": 1.1677221088990901e-06,
+      "loss": 0.7114,
+      "step": 12275
+    },
+    {
+      "epoch": 1.046101406050277,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.1675544385010456e-06,
+      "loss": 0.5859,
+      "step": 12276
+    },
+    {
+      "epoch": 1.0461866212185769,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.1673867695928758e-06,
+      "loss": 0.5342,
+      "step": 12277
+    },
+    {
+      "epoch": 1.0462718363868768,
+      "grad_norm": 18.25,
+      "learning_rate": 1.16721910217761e-06,
+      "loss": 0.8177,
+      "step": 12278
+    },
+    {
+      "epoch": 1.0463570515551768,
+      "grad_norm": 11.375,
+      "learning_rate": 1.1670514362582782e-06,
+      "loss": 0.5343,
+      "step": 12279
+    },
+    {
+      "epoch": 1.0464422667234767,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.16688377183791e-06,
+      "loss": 0.8078,
+      "step": 12280
+    },
+    {
+      "epoch": 1.0465274818917767,
+      "grad_norm": 17.125,
+      "learning_rate": 1.1667161089195358e-06,
+      "loss": 0.6942,
+      "step": 12281
+    },
+    {
+      "epoch": 1.0466126970600766,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1665484475061853e-06,
+      "loss": 0.6339,
+      "step": 12282
+    },
+    {
+      "epoch": 1.0466979122283766,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.1663807876008882e-06,
+      "loss": 0.6903,
+      "step": 12283
+    },
+    {
+      "epoch": 1.0467831273966766,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1662131292066744e-06,
+      "loss": 0.4029,
+      "step": 12284
+    },
+    {
+      "epoch": 1.0468683425649765,
+      "grad_norm": 16.5,
+      "learning_rate": 1.166045472326574e-06,
+      "loss": 0.718,
+      "step": 12285
+    },
+    {
+      "epoch": 1.0469535577332765,
+      "grad_norm": 16.875,
+      "learning_rate": 1.165877816963616e-06,
+      "loss": 0.6453,
+      "step": 12286
+    },
+    {
+      "epoch": 1.0470387729015764,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1657101631208301e-06,
+      "loss": 0.7366,
+      "step": 12287
+    },
+    {
+      "epoch": 1.0471239880698764,
+      "grad_norm": 18.25,
+      "learning_rate": 1.1655425108012467e-06,
+      "loss": 0.736,
+      "step": 12288
+    },
+    {
+      "epoch": 1.0472092032381763,
+      "grad_norm": 18.875,
+      "learning_rate": 1.165374860007895e-06,
+      "loss": 1.0333,
+      "step": 12289
+    },
+    {
+      "epoch": 1.0472944184064763,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1652072107438046e-06,
+      "loss": 0.5516,
+      "step": 12290
+    },
+    {
+      "epoch": 1.0473796335747763,
+      "grad_norm": 25.625,
+      "learning_rate": 1.1650395630120051e-06,
+      "loss": 1.1468,
+      "step": 12291
+    },
+    {
+      "epoch": 1.0474648487430762,
+      "grad_norm": 19.875,
+      "learning_rate": 1.1648719168155263e-06,
+      "loss": 0.8665,
+      "step": 12292
+    },
+    {
+      "epoch": 1.0475500639113762,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.164704272157398e-06,
+      "loss": 0.5032,
+      "step": 12293
+    },
+    {
+      "epoch": 1.0476352790796761,
+      "grad_norm": 21.625,
+      "learning_rate": 1.164536629040649e-06,
+      "loss": 0.7211,
+      "step": 12294
+    },
+    {
+      "epoch": 1.047720494247976,
+      "grad_norm": 15.75,
+      "learning_rate": 1.1643689874683085e-06,
+      "loss": 0.8516,
+      "step": 12295
+    },
+    {
+      "epoch": 1.047805709416276,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1642013474434068e-06,
+      "loss": 0.7846,
+      "step": 12296
+    },
+    {
+      "epoch": 1.047890924584576,
+      "grad_norm": 20.125,
+      "learning_rate": 1.1640337089689732e-06,
+      "loss": 0.5703,
+      "step": 12297
+    },
+    {
+      "epoch": 1.047976139752876,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1638660720480367e-06,
+      "loss": 0.6434,
+      "step": 12298
+    },
+    {
+      "epoch": 1.048061354921176,
+      "grad_norm": 12.5,
+      "learning_rate": 1.1636984366836268e-06,
+      "loss": 0.4529,
+      "step": 12299
+    },
+    {
+      "epoch": 1.0481465700894759,
+      "grad_norm": 16.625,
+      "learning_rate": 1.1635308028787733e-06,
+      "loss": 0.4214,
+      "step": 12300
+    },
+    {
+      "epoch": 1.0482317852577758,
+      "grad_norm": 11.5,
+      "learning_rate": 1.1633631706365054e-06,
+      "loss": 0.3778,
+      "step": 12301
+    },
+    {
+      "epoch": 1.0483170004260758,
+      "grad_norm": 21.125,
+      "learning_rate": 1.1631955399598513e-06,
+      "loss": 0.7536,
+      "step": 12302
+    },
+    {
+      "epoch": 1.0484022155943757,
+      "grad_norm": 20.0,
+      "learning_rate": 1.1630279108518417e-06,
+      "loss": 0.8039,
+      "step": 12303
+    },
+    {
+      "epoch": 1.0484874307626757,
+      "grad_norm": 19.75,
+      "learning_rate": 1.1628602833155047e-06,
+      "loss": 0.8201,
+      "step": 12304
+    },
+    {
+      "epoch": 1.0485726459309757,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.1626926573538705e-06,
+      "loss": 0.6493,
+      "step": 12305
+    },
+    {
+      "epoch": 1.0486578610992756,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.1625250329699673e-06,
+      "loss": 0.5935,
+      "step": 12306
+    },
+    {
+      "epoch": 1.0487430762675756,
+      "grad_norm": 16.625,
+      "learning_rate": 1.1623574101668252e-06,
+      "loss": 0.6973,
+      "step": 12307
+    },
+    {
+      "epoch": 1.0488282914358755,
+      "grad_norm": 14.5,
+      "learning_rate": 1.1621897889474725e-06,
+      "loss": 0.42,
+      "step": 12308
+    },
+    {
+      "epoch": 1.0489135066041755,
+      "grad_norm": 21.25,
+      "learning_rate": 1.1620221693149392e-06,
+      "loss": 0.8247,
+      "step": 12309
+    },
+    {
+      "epoch": 1.0489987217724754,
+      "grad_norm": 11.875,
+      "learning_rate": 1.1618545512722532e-06,
+      "loss": 0.4533,
+      "step": 12310
+    },
+    {
+      "epoch": 1.0490839369407754,
+      "grad_norm": 19.625,
+      "learning_rate": 1.1616869348224443e-06,
+      "loss": 0.8493,
+      "step": 12311
+    },
+    {
+      "epoch": 1.0491691521090754,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.161519319968541e-06,
+      "loss": 0.1809,
+      "step": 12312
+    },
+    {
+      "epoch": 1.0492543672773753,
+      "grad_norm": 20.375,
+      "learning_rate": 1.161351706713573e-06,
+      "loss": 0.841,
+      "step": 12313
+    },
+    {
+      "epoch": 1.0493395824456753,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1611840950605682e-06,
+      "loss": 0.391,
+      "step": 12314
+    },
+    {
+      "epoch": 1.0494247976139752,
+      "grad_norm": 19.125,
+      "learning_rate": 1.1610164850125566e-06,
+      "loss": 0.7404,
+      "step": 12315
+    },
+    {
+      "epoch": 1.0495100127822752,
+      "grad_norm": 20.5,
+      "learning_rate": 1.1608488765725661e-06,
+      "loss": 0.9264,
+      "step": 12316
+    },
+    {
+      "epoch": 1.0495952279505751,
+      "grad_norm": 20.375,
+      "learning_rate": 1.1606812697436268e-06,
+      "loss": 0.6435,
+      "step": 12317
+    },
+    {
+      "epoch": 1.049680443118875,
+      "grad_norm": 9.75,
+      "learning_rate": 1.1605136645287666e-06,
+      "loss": 0.3014,
+      "step": 12318
+    },
+    {
+      "epoch": 1.049765658287175,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.160346060931014e-06,
+      "loss": 0.6051,
+      "step": 12319
+    },
+    {
+      "epoch": 1.049850873455475,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1601784589533986e-06,
+      "loss": 0.4976,
+      "step": 12320
+    },
+    {
+      "epoch": 1.049936088623775,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.1600108585989485e-06,
+      "loss": 0.8374,
+      "step": 12321
+    },
+    {
+      "epoch": 1.050021303792075,
+      "grad_norm": 17.75,
+      "learning_rate": 1.159843259870693e-06,
+      "loss": 0.7385,
+      "step": 12322
+    },
+    {
+      "epoch": 1.0501065189603749,
+      "grad_norm": 14.125,
+      "learning_rate": 1.15967566277166e-06,
+      "loss": 0.8023,
+      "step": 12323
+    },
+    {
+      "epoch": 1.0501917341286748,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1595080673048791e-06,
+      "loss": 0.6686,
+      "step": 12324
+    },
+    {
+      "epoch": 1.0502769492969748,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.1593404734733784e-06,
+      "loss": 0.3566,
+      "step": 12325
+    },
+    {
+      "epoch": 1.0503621644652748,
+      "grad_norm": 13.75,
+      "learning_rate": 1.1591728812801865e-06,
+      "loss": 0.3083,
+      "step": 12326
+    },
+    {
+      "epoch": 1.0504473796335747,
+      "grad_norm": 28.5,
+      "learning_rate": 1.1590052907283316e-06,
+      "loss": 0.6044,
+      "step": 12327
+    },
+    {
+      "epoch": 1.0505325948018747,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1588377018208428e-06,
+      "loss": 0.5091,
+      "step": 12328
+    },
+    {
+      "epoch": 1.0506178099701746,
+      "grad_norm": 21.0,
+      "learning_rate": 1.1586701145607483e-06,
+      "loss": 1.0696,
+      "step": 12329
+    },
+    {
+      "epoch": 1.0507030251384746,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.1585025289510768e-06,
+      "loss": 0.63,
+      "step": 12330
+    },
+    {
+      "epoch": 1.0507882403067745,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1583349449948564e-06,
+      "loss": 0.5416,
+      "step": 12331
+    },
+    {
+      "epoch": 1.0508734554750745,
+      "grad_norm": 20.625,
+      "learning_rate": 1.158167362695116e-06,
+      "loss": 0.7185,
+      "step": 12332
+    },
+    {
+      "epoch": 1.0509586706433744,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.1579997820548835e-06,
+      "loss": 0.3981,
+      "step": 12333
+    },
+    {
+      "epoch": 1.0510438858116744,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1578322030771878e-06,
+      "loss": 0.7342,
+      "step": 12334
+    },
+    {
+      "epoch": 1.0511291009799744,
+      "grad_norm": 17.875,
+      "learning_rate": 1.1576646257650567e-06,
+      "loss": 0.7714,
+      "step": 12335
+    },
+    {
+      "epoch": 1.0512143161482743,
+      "grad_norm": 12.0,
+      "learning_rate": 1.1574970501215183e-06,
+      "loss": 0.3909,
+      "step": 12336
+    },
+    {
+      "epoch": 1.0512995313165743,
+      "grad_norm": 13.75,
+      "learning_rate": 1.1573294761496016e-06,
+      "loss": 0.6491,
+      "step": 12337
+    },
+    {
+      "epoch": 1.0513847464848742,
+      "grad_norm": 19.0,
+      "learning_rate": 1.1571619038523342e-06,
+      "loss": 0.967,
+      "step": 12338
+    },
+    {
+      "epoch": 1.0514699616531742,
+      "grad_norm": 19.75,
+      "learning_rate": 1.156994333232745e-06,
+      "loss": 0.6683,
+      "step": 12339
+    },
+    {
+      "epoch": 1.0515551768214741,
+      "grad_norm": 48.25,
+      "learning_rate": 1.1568267642938611e-06,
+      "loss": 0.7705,
+      "step": 12340
+    },
+    {
+      "epoch": 1.051640391989774,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1566591970387117e-06,
+      "loss": 0.2951,
+      "step": 12341
+    },
+    {
+      "epoch": 1.051725607158074,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1564916314703247e-06,
+      "loss": 0.3566,
+      "step": 12342
+    },
+    {
+      "epoch": 1.051810822326374,
+      "grad_norm": 14.875,
+      "learning_rate": 1.1563240675917277e-06,
+      "loss": 0.7374,
+      "step": 12343
+    },
+    {
+      "epoch": 1.051896037494674,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1561565054059487e-06,
+      "loss": 0.807,
+      "step": 12344
+    },
+    {
+      "epoch": 1.051981252662974,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1559889449160163e-06,
+      "loss": 0.7581,
+      "step": 12345
+    },
+    {
+      "epoch": 1.0520664678312739,
+      "grad_norm": 13.625,
+      "learning_rate": 1.155821386124958e-06,
+      "loss": 0.3939,
+      "step": 12346
+    },
+    {
+      "epoch": 1.0521516829995738,
+      "grad_norm": 19.375,
+      "learning_rate": 1.155653829035802e-06,
+      "loss": 0.8974,
+      "step": 12347
+    },
+    {
+      "epoch": 1.0522368981678738,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1554862736515761e-06,
+      "loss": 0.6698,
+      "step": 12348
+    },
+    {
+      "epoch": 1.0523221133361738,
+      "grad_norm": 14.375,
+      "learning_rate": 1.1553187199753087e-06,
+      "loss": 0.6102,
+      "step": 12349
+    },
+    {
+      "epoch": 1.0524073285044737,
+      "grad_norm": 15.5,
+      "learning_rate": 1.1551511680100272e-06,
+      "loss": 0.3598,
+      "step": 12350
+    },
+    {
+      "epoch": 1.0524925436727737,
+      "grad_norm": 43.5,
+      "learning_rate": 1.1549836177587591e-06,
+      "loss": 0.8307,
+      "step": 12351
+    },
+    {
+      "epoch": 1.0525777588410736,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1548160692245329e-06,
+      "loss": 0.4264,
+      "step": 12352
+    },
+    {
+      "epoch": 1.0526629740093736,
+      "grad_norm": 17.125,
+      "learning_rate": 1.1546485224103756e-06,
+      "loss": 0.6326,
+      "step": 12353
+    },
+    {
+      "epoch": 1.0527481891776735,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1544809773193158e-06,
+      "loss": 0.4014,
+      "step": 12354
+    },
+    {
+      "epoch": 1.0528334043459735,
+      "grad_norm": 19.125,
+      "learning_rate": 1.1543134339543804e-06,
+      "loss": 0.4165,
+      "step": 12355
+    },
+    {
+      "epoch": 1.0529186195142735,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1541458923185978e-06,
+      "loss": 0.9075,
+      "step": 12356
+    },
+    {
+      "epoch": 1.0530038346825734,
+      "grad_norm": 11.5,
+      "learning_rate": 1.1539783524149952e-06,
+      "loss": 0.3407,
+      "step": 12357
+    },
+    {
+      "epoch": 1.0530890498508734,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1538108142466009e-06,
+      "loss": 0.6585,
+      "step": 12358
+    },
+    {
+      "epoch": 1.0531742650191733,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1536432778164411e-06,
+      "loss": 0.647,
+      "step": 12359
+    },
+    {
+      "epoch": 1.0532594801874733,
+      "grad_norm": 14.375,
+      "learning_rate": 1.1534757431275446e-06,
+      "loss": 0.6065,
+      "step": 12360
+    },
+    {
+      "epoch": 1.0533446953557732,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1533082101829381e-06,
+      "loss": 0.8037,
+      "step": 12361
+    },
+    {
+      "epoch": 1.0534299105240732,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.1531406789856497e-06,
+      "loss": 0.7653,
+      "step": 12362
+    },
+    {
+      "epoch": 1.0535151256923732,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.1529731495387065e-06,
+      "loss": 0.4392,
+      "step": 12363
+    },
+    {
+      "epoch": 1.0536003408606731,
+      "grad_norm": 11.5,
+      "learning_rate": 1.1528056218451362e-06,
+      "loss": 0.336,
+      "step": 12364
+    },
+    {
+      "epoch": 1.053685556028973,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1526380959079657e-06,
+      "loss": 0.5924,
+      "step": 12365
+    },
+    {
+      "epoch": 1.053770771197273,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1524705717302236e-06,
+      "loss": 0.877,
+      "step": 12366
+    },
+    {
+      "epoch": 1.053855986365573,
+      "grad_norm": 13.5,
+      "learning_rate": 1.1523030493149359e-06,
+      "loss": 0.4274,
+      "step": 12367
+    },
+    {
+      "epoch": 1.053941201533873,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.15213552866513e-06,
+      "loss": 0.6436,
+      "step": 12368
+    },
+    {
+      "epoch": 1.054026416702173,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1519680097838337e-06,
+      "loss": 0.667,
+      "step": 12369
+    },
+    {
+      "epoch": 1.0541116318704729,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.151800492674074e-06,
+      "loss": 0.5561,
+      "step": 12370
+    },
+    {
+      "epoch": 1.0541968470387728,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.1516329773388782e-06,
+      "loss": 0.4434,
+      "step": 12371
+    },
+    {
+      "epoch": 1.0542820622070728,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1514654637812736e-06,
+      "loss": 0.7159,
+      "step": 12372
+    },
+    {
+      "epoch": 1.0543672773753727,
+      "grad_norm": 20.5,
+      "learning_rate": 1.1512979520042867e-06,
+      "loss": 0.654,
+      "step": 12373
+    },
+    {
+      "epoch": 1.0544524925436727,
+      "grad_norm": 20.25,
+      "learning_rate": 1.1511304420109458e-06,
+      "loss": 0.7388,
+      "step": 12374
+    },
+    {
+      "epoch": 1.0545377077119726,
+      "grad_norm": 13.25,
+      "learning_rate": 1.1509629338042768e-06,
+      "loss": 0.4155,
+      "step": 12375
+    },
+    {
+      "epoch": 1.0546229228802726,
+      "grad_norm": 15.0,
+      "learning_rate": 1.150795427387307e-06,
+      "loss": 0.7324,
+      "step": 12376
+    },
+    {
+      "epoch": 1.0547081380485726,
+      "grad_norm": 19.75,
+      "learning_rate": 1.1506279227630637e-06,
+      "loss": 0.7354,
+      "step": 12377
+    },
+    {
+      "epoch": 1.0547933532168725,
+      "grad_norm": 11.875,
+      "learning_rate": 1.1504604199345737e-06,
+      "loss": 0.4438,
+      "step": 12378
+    },
+    {
+      "epoch": 1.0548785683851725,
+      "grad_norm": 20.375,
+      "learning_rate": 1.1502929189048643e-06,
+      "loss": 0.552,
+      "step": 12379
+    },
+    {
+      "epoch": 1.0549637835534724,
+      "grad_norm": 12.25,
+      "learning_rate": 1.1501254196769616e-06,
+      "loss": 0.5927,
+      "step": 12380
+    },
+    {
+      "epoch": 1.0550489987217724,
+      "grad_norm": 15.5,
+      "learning_rate": 1.1499579222538935e-06,
+      "loss": 0.594,
+      "step": 12381
+    },
+    {
+      "epoch": 1.0551342138900723,
+      "grad_norm": 15.25,
+      "learning_rate": 1.1497904266386863e-06,
+      "loss": 0.4722,
+      "step": 12382
+    },
+    {
+      "epoch": 1.0552194290583723,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1496229328343666e-06,
+      "loss": 0.741,
+      "step": 12383
+    },
+    {
+      "epoch": 1.0553046442266723,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.1494554408439617e-06,
+      "loss": 0.5476,
+      "step": 12384
+    },
+    {
+      "epoch": 1.0553898593949722,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.1492879506704978e-06,
+      "loss": 0.3974,
+      "step": 12385
+    },
+    {
+      "epoch": 1.0554750745632722,
+      "grad_norm": 24.25,
+      "learning_rate": 1.1491204623170021e-06,
+      "loss": 0.9395,
+      "step": 12386
+    },
+    {
+      "epoch": 1.0555602897315721,
+      "grad_norm": 19.125,
+      "learning_rate": 1.1489529757865011e-06,
+      "loss": 0.6363,
+      "step": 12387
+    },
+    {
+      "epoch": 1.055645504899872,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1487854910820213e-06,
+      "loss": 0.8712,
+      "step": 12388
+    },
+    {
+      "epoch": 1.055730720068172,
+      "grad_norm": 22.125,
+      "learning_rate": 1.1486180082065896e-06,
+      "loss": 0.7786,
+      "step": 12389
+    },
+    {
+      "epoch": 1.055815935236472,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1484505271632327e-06,
+      "loss": 0.5247,
+      "step": 12390
+    },
+    {
+      "epoch": 1.055901150404772,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.1482830479549764e-06,
+      "loss": 0.4275,
+      "step": 12391
+    },
+    {
+      "epoch": 1.055986365573072,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.148115570584848e-06,
+      "loss": 0.4442,
+      "step": 12392
+    },
+    {
+      "epoch": 1.0560715807413719,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1479480950558734e-06,
+      "loss": 0.7148,
+      "step": 12393
+    },
+    {
+      "epoch": 1.0561567959096718,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.1477806213710796e-06,
+      "loss": 0.4649,
+      "step": 12394
+    },
+    {
+      "epoch": 1.0562420110779718,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.1476131495334925e-06,
+      "loss": 0.524,
+      "step": 12395
+    },
+    {
+      "epoch": 1.0563272262462717,
+      "grad_norm": 17.5,
+      "learning_rate": 1.1474456795461391e-06,
+      "loss": 0.8097,
+      "step": 12396
+    },
+    {
+      "epoch": 1.0564124414145717,
+      "grad_norm": 25.0,
+      "learning_rate": 1.147278211412045e-06,
+      "loss": 1.0126,
+      "step": 12397
+    },
+    {
+      "epoch": 1.0564976565828716,
+      "grad_norm": 11.25,
+      "learning_rate": 1.147110745134237e-06,
+      "loss": 0.374,
+      "step": 12398
+    },
+    {
+      "epoch": 1.0565828717511716,
+      "grad_norm": 14.75,
+      "learning_rate": 1.146943280715742e-06,
+      "loss": 0.5695,
+      "step": 12399
+    },
+    {
+      "epoch": 1.0566680869194716,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.1467758181595847e-06,
+      "loss": 0.4398,
+      "step": 12400
+    },
+    {
+      "epoch": 1.0567533020877715,
+      "grad_norm": 13.75,
+      "learning_rate": 1.1466083574687926e-06,
+      "loss": 0.4198,
+      "step": 12401
+    },
+    {
+      "epoch": 1.0568385172560715,
+      "grad_norm": 11.75,
+      "learning_rate": 1.1464408986463915e-06,
+      "loss": 0.4795,
+      "step": 12402
+    },
+    {
+      "epoch": 1.0569237324243714,
+      "grad_norm": 19.0,
+      "learning_rate": 1.1462734416954071e-06,
+      "loss": 0.5306,
+      "step": 12403
+    },
+    {
+      "epoch": 1.0570089475926714,
+      "grad_norm": 21.75,
+      "learning_rate": 1.146105986618866e-06,
+      "loss": 0.819,
+      "step": 12404
+    },
+    {
+      "epoch": 1.0570941627609716,
+      "grad_norm": 10.625,
+      "learning_rate": 1.1459385334197942e-06,
+      "loss": 0.3365,
+      "step": 12405
+    },
+    {
+      "epoch": 1.0571793779292715,
+      "grad_norm": 16.625,
+      "learning_rate": 1.145771082101218e-06,
+      "loss": 0.6668,
+      "step": 12406
+    },
+    {
+      "epoch": 1.0572645930975715,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.1456036326661634e-06,
+      "loss": 0.7222,
+      "step": 12407
+    },
+    {
+      "epoch": 1.0573498082658714,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1454361851176554e-06,
+      "loss": 0.4683,
+      "step": 12408
+    },
+    {
+      "epoch": 1.0574350234341714,
+      "grad_norm": 21.375,
+      "learning_rate": 1.1452687394587211e-06,
+      "loss": 0.7266,
+      "step": 12409
+    },
+    {
+      "epoch": 1.0575202386024714,
+      "grad_norm": 22.875,
+      "learning_rate": 1.1451012956923857e-06,
+      "loss": 1.0329,
+      "step": 12410
+    },
+    {
+      "epoch": 1.0576054537707713,
+      "grad_norm": 11.5,
+      "learning_rate": 1.1449338538216757e-06,
+      "loss": 0.3147,
+      "step": 12411
+    },
+    {
+      "epoch": 1.0576906689390713,
+      "grad_norm": 12.625,
+      "learning_rate": 1.1447664138496162e-06,
+      "loss": 0.4077,
+      "step": 12412
+    },
+    {
+      "epoch": 1.0577758841073712,
+      "grad_norm": 12.5,
+      "learning_rate": 1.144598975779234e-06,
+      "loss": 0.5454,
+      "step": 12413
+    },
+    {
+      "epoch": 1.0578610992756712,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1444315396135538e-06,
+      "loss": 0.6366,
+      "step": 12414
+    },
+    {
+      "epoch": 1.0579463144439711,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1442641053556023e-06,
+      "loss": 0.6623,
+      "step": 12415
+    },
+    {
+      "epoch": 1.058031529612271,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1440966730084046e-06,
+      "loss": 0.7245,
+      "step": 12416
+    },
+    {
+      "epoch": 1.058116744780571,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1439292425749867e-06,
+      "loss": 0.6474,
+      "step": 12417
+    },
+    {
+      "epoch": 1.058201959948871,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.1437618140583734e-06,
+      "loss": 0.9257,
+      "step": 12418
+    },
+    {
+      "epoch": 1.058287175117171,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1435943874615915e-06,
+      "loss": 0.7231,
+      "step": 12419
+    },
+    {
+      "epoch": 1.058372390285471,
+      "grad_norm": 17.125,
+      "learning_rate": 1.1434269627876656e-06,
+      "loss": 0.4903,
+      "step": 12420
+    },
+    {
+      "epoch": 1.0584576054537709,
+      "grad_norm": 17.375,
+      "learning_rate": 1.143259540039622e-06,
+      "loss": 0.6428,
+      "step": 12421
+    },
+    {
+      "epoch": 1.0585428206220708,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1430921192204854e-06,
+      "loss": 0.4689,
+      "step": 12422
+    },
+    {
+      "epoch": 1.0586280357903708,
+      "grad_norm": 19.375,
+      "learning_rate": 1.1429247003332824e-06,
+      "loss": 0.5604,
+      "step": 12423
+    },
+    {
+      "epoch": 1.0587132509586707,
+      "grad_norm": 17.5,
+      "learning_rate": 1.1427572833810376e-06,
+      "loss": 0.7459,
+      "step": 12424
+    },
+    {
+      "epoch": 1.0587984661269707,
+      "grad_norm": 11.5,
+      "learning_rate": 1.142589868366776e-06,
+      "loss": 0.3855,
+      "step": 12425
+    },
+    {
+      "epoch": 1.0588836812952707,
+      "grad_norm": 18.375,
+      "learning_rate": 1.142422455293524e-06,
+      "loss": 0.5321,
+      "step": 12426
+    },
+    {
+      "epoch": 1.0589688964635706,
+      "grad_norm": 12.625,
+      "learning_rate": 1.142255044164306e-06,
+      "loss": 0.3933,
+      "step": 12427
+    },
+    {
+      "epoch": 1.0590541116318706,
+      "grad_norm": 11.875,
+      "learning_rate": 1.142087634982148e-06,
+      "loss": 0.5832,
+      "step": 12428
+    },
+    {
+      "epoch": 1.0591393268001705,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1419202277500746e-06,
+      "loss": 0.8307,
+      "step": 12429
+    },
+    {
+      "epoch": 1.0592245419684705,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.1417528224711117e-06,
+      "loss": 0.3145,
+      "step": 12430
+    },
+    {
+      "epoch": 1.0593097571367704,
+      "grad_norm": 20.125,
+      "learning_rate": 1.1415854191482845e-06,
+      "loss": 0.3717,
+      "step": 12431
+    },
+    {
+      "epoch": 1.0593949723050704,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1414180177846174e-06,
+      "loss": 0.7545,
+      "step": 12432
+    },
+    {
+      "epoch": 1.0594801874733704,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.1412506183831357e-06,
+      "loss": 0.2973,
+      "step": 12433
+    },
+    {
+      "epoch": 1.0595654026416703,
+      "grad_norm": 16.75,
+      "learning_rate": 1.141083220946865e-06,
+      "loss": 0.9628,
+      "step": 12434
+    },
+    {
+      "epoch": 1.0596506178099703,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1409158254788295e-06,
+      "loss": 0.5443,
+      "step": 12435
+    },
+    {
+      "epoch": 1.0597358329782702,
+      "grad_norm": 11.625,
+      "learning_rate": 1.1407484319820553e-06,
+      "loss": 0.4232,
+      "step": 12436
+    },
+    {
+      "epoch": 1.0598210481465702,
+      "grad_norm": 10.25,
+      "learning_rate": 1.1405810404595664e-06,
+      "loss": 0.2284,
+      "step": 12437
+    },
+    {
+      "epoch": 1.0599062633148701,
+      "grad_norm": 19.375,
+      "learning_rate": 1.1404136509143883e-06,
+      "loss": 0.6412,
+      "step": 12438
+    },
+    {
+      "epoch": 1.05999147848317,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.1402462633495463e-06,
+      "loss": 0.6547,
+      "step": 12439
+    },
+    {
+      "epoch": 1.06007669365147,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1400788777680639e-06,
+      "loss": 0.5748,
+      "step": 12440
+    },
+    {
+      "epoch": 1.06016190881977,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.139911494172967e-06,
+      "loss": 0.5351,
+      "step": 12441
+    },
+    {
+      "epoch": 1.06024712398807,
+      "grad_norm": 18.75,
+      "learning_rate": 1.1397441125672798e-06,
+      "loss": 0.6128,
+      "step": 12442
+    },
+    {
+      "epoch": 1.06033233915637,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1395767329540275e-06,
+      "loss": 0.5185,
+      "step": 12443
+    },
+    {
+      "epoch": 1.0604175543246699,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1394093553362348e-06,
+      "loss": 0.877,
+      "step": 12444
+    },
+    {
+      "epoch": 1.0605027694929698,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1392419797169263e-06,
+      "loss": 0.6787,
+      "step": 12445
+    },
+    {
+      "epoch": 1.0605879846612698,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1390746060991264e-06,
+      "loss": 0.5324,
+      "step": 12446
+    },
+    {
+      "epoch": 1.0606731998295698,
+      "grad_norm": 39.25,
+      "learning_rate": 1.1389072344858606e-06,
+      "loss": 1.2895,
+      "step": 12447
+    },
+    {
+      "epoch": 1.0607584149978697,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1387398648801521e-06,
+      "loss": 0.7786,
+      "step": 12448
+    },
+    {
+      "epoch": 1.0608436301661697,
+      "grad_norm": 21.625,
+      "learning_rate": 1.1385724972850262e-06,
+      "loss": 0.6146,
+      "step": 12449
+    },
+    {
+      "epoch": 1.0609288453344696,
+      "grad_norm": 23.125,
+      "learning_rate": 1.1384051317035073e-06,
+      "loss": 0.9894,
+      "step": 12450
+    },
+    {
+      "epoch": 1.0610140605027696,
+      "grad_norm": 15.875,
+      "learning_rate": 1.13823776813862e-06,
+      "loss": 0.6607,
+      "step": 12451
+    },
+    {
+      "epoch": 1.0610992756710695,
+      "grad_norm": 16.625,
+      "learning_rate": 1.1380704065933885e-06,
+      "loss": 0.7828,
+      "step": 12452
+    },
+    {
+      "epoch": 1.0611844908393695,
+      "grad_norm": 15.0,
+      "learning_rate": 1.1379030470708375e-06,
+      "loss": 0.5605,
+      "step": 12453
+    },
+    {
+      "epoch": 1.0612697060076695,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1377356895739907e-06,
+      "loss": 0.6024,
+      "step": 12454
+    },
+    {
+      "epoch": 1.0613549211759694,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1375683341058736e-06,
+      "loss": 0.3908,
+      "step": 12455
+    },
+    {
+      "epoch": 1.0614401363442694,
+      "grad_norm": 17.125,
+      "learning_rate": 1.1374009806695095e-06,
+      "loss": 0.7532,
+      "step": 12456
+    },
+    {
+      "epoch": 1.0615253515125693,
+      "grad_norm": 14.0,
+      "learning_rate": 1.1372336292679227e-06,
+      "loss": 0.7757,
+      "step": 12457
+    },
+    {
+      "epoch": 1.0616105666808693,
+      "grad_norm": 15.0,
+      "learning_rate": 1.1370662799041378e-06,
+      "loss": 0.6946,
+      "step": 12458
+    },
+    {
+      "epoch": 1.0616957818491692,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1368989325811783e-06,
+      "loss": 0.5821,
+      "step": 12459
+    },
+    {
+      "epoch": 1.0617809970174692,
+      "grad_norm": 14.875,
+      "learning_rate": 1.1367315873020695e-06,
+      "loss": 0.4802,
+      "step": 12460
+    },
+    {
+      "epoch": 1.0618662121857692,
+      "grad_norm": 27.125,
+      "learning_rate": 1.1365642440698344e-06,
+      "loss": 1.3032,
+      "step": 12461
+    },
+    {
+      "epoch": 1.061951427354069,
+      "grad_norm": 15.0,
+      "learning_rate": 1.1363969028874976e-06,
+      "loss": 0.7396,
+      "step": 12462
+    },
+    {
+      "epoch": 1.062036642522369,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1362295637580828e-06,
+      "loss": 0.754,
+      "step": 12463
+    },
+    {
+      "epoch": 1.062121857690669,
+      "grad_norm": 14.75,
+      "learning_rate": 1.136062226684615e-06,
+      "loss": 0.7202,
+      "step": 12464
+    },
+    {
+      "epoch": 1.062207072858969,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.1358948916701166e-06,
+      "loss": 0.7034,
+      "step": 12465
+    },
+    {
+      "epoch": 1.062292288027269,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.1357275587176125e-06,
+      "loss": 0.5123,
+      "step": 12466
+    },
+    {
+      "epoch": 1.062377503195569,
+      "grad_norm": 21.5,
+      "learning_rate": 1.135560227830126e-06,
+      "loss": 0.6569,
+      "step": 12467
+    },
+    {
+      "epoch": 1.0624627183638689,
+      "grad_norm": 26.625,
+      "learning_rate": 1.1353928990106817e-06,
+      "loss": 0.7784,
+      "step": 12468
+    },
+    {
+      "epoch": 1.0625479335321688,
+      "grad_norm": 12.75,
+      "learning_rate": 1.1352255722623026e-06,
+      "loss": 0.4901,
+      "step": 12469
+    },
+    {
+      "epoch": 1.0626331487004688,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1350582475880128e-06,
+      "loss": 0.6289,
+      "step": 12470
+    },
+    {
+      "epoch": 1.0627183638687687,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.134890924990836e-06,
+      "loss": 0.4797,
+      "step": 12471
+    },
+    {
+      "epoch": 1.0628035790370687,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1347236044737965e-06,
+      "loss": 0.7833,
+      "step": 12472
+    },
+    {
+      "epoch": 1.0628887942053686,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1345562860399172e-06,
+      "loss": 0.8292,
+      "step": 12473
+    },
+    {
+      "epoch": 1.0629740093736686,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1343889696922216e-06,
+      "loss": 0.8105,
+      "step": 12474
+    },
+    {
+      "epoch": 1.0630592245419686,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.1342216554337338e-06,
+      "loss": 0.5428,
+      "step": 12475
+    },
+    {
+      "epoch": 1.0631444397102685,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1340543432674769e-06,
+      "loss": 0.7002,
+      "step": 12476
+    },
+    {
+      "epoch": 1.0632296548785685,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.1338870331964747e-06,
+      "loss": 0.7705,
+      "step": 12477
+    },
+    {
+      "epoch": 1.0633148700468684,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1337197252237504e-06,
+      "loss": 0.4863,
+      "step": 12478
+    },
+    {
+      "epoch": 1.0634000852151684,
+      "grad_norm": 12.25,
+      "learning_rate": 1.133552419352328e-06,
+      "loss": 0.3831,
+      "step": 12479
+    },
+    {
+      "epoch": 1.0634853003834683,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1333851155852307e-06,
+      "loss": 0.516,
+      "step": 12480
+    },
+    {
+      "epoch": 1.0635705155517683,
+      "grad_norm": 22.75,
+      "learning_rate": 1.133217813925481e-06,
+      "loss": 1.1684,
+      "step": 12481
+    },
+    {
+      "epoch": 1.0636557307200682,
+      "grad_norm": 57.25,
+      "learning_rate": 1.133050514376103e-06,
+      "loss": 0.7768,
+      "step": 12482
+    },
+    {
+      "epoch": 1.0637409458883682,
+      "grad_norm": 14.125,
+      "learning_rate": 1.1328832169401198e-06,
+      "loss": 0.4273,
+      "step": 12483
+    },
+    {
+      "epoch": 1.0638261610566682,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1327159216205546e-06,
+      "loss": 0.7931,
+      "step": 12484
+    },
+    {
+      "epoch": 1.0639113762249681,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.132548628420431e-06,
+      "loss": 0.801,
+      "step": 12485
+    },
+    {
+      "epoch": 1.063996591393268,
+      "grad_norm": 16.125,
+      "learning_rate": 1.132381337342771e-06,
+      "loss": 0.4861,
+      "step": 12486
+    },
+    {
+      "epoch": 1.064081806561568,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.1322140483905992e-06,
+      "loss": 0.5341,
+      "step": 12487
+    },
+    {
+      "epoch": 1.064167021729868,
+      "grad_norm": 21.375,
+      "learning_rate": 1.132046761566938e-06,
+      "loss": 0.9495,
+      "step": 12488
+    },
+    {
+      "epoch": 1.064252236898168,
+      "grad_norm": 21.625,
+      "learning_rate": 1.1318794768748103e-06,
+      "loss": 1.1944,
+      "step": 12489
+    },
+    {
+      "epoch": 1.064337452066468,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1317121943172391e-06,
+      "loss": 0.7163,
+      "step": 12490
+    },
+    {
+      "epoch": 1.0644226672347679,
+      "grad_norm": 11.75,
+      "learning_rate": 1.1315449138972475e-06,
+      "loss": 0.4529,
+      "step": 12491
+    },
+    {
+      "epoch": 1.0645078824030678,
+      "grad_norm": 15.375,
+      "learning_rate": 1.1313776356178586e-06,
+      "loss": 0.5153,
+      "step": 12492
+    },
+    {
+      "epoch": 1.0645930975713678,
+      "grad_norm": 10.875,
+      "learning_rate": 1.1312103594820948e-06,
+      "loss": 0.3603,
+      "step": 12493
+    },
+    {
+      "epoch": 1.0646783127396677,
+      "grad_norm": 25.375,
+      "learning_rate": 1.1310430854929793e-06,
+      "loss": 1.2229,
+      "step": 12494
+    },
+    {
+      "epoch": 1.0647635279079677,
+      "grad_norm": 15.75,
+      "learning_rate": 1.130875813653535e-06,
+      "loss": 0.8113,
+      "step": 12495
+    },
+    {
+      "epoch": 1.0648487430762676,
+      "grad_norm": 12.375,
+      "learning_rate": 1.1307085439667852e-06,
+      "loss": 0.5185,
+      "step": 12496
+    },
+    {
+      "epoch": 1.0649339582445676,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.130541276435751e-06,
+      "loss": 0.5143,
+      "step": 12497
+    },
+    {
+      "epoch": 1.0650191734128676,
+      "grad_norm": 11.875,
+      "learning_rate": 1.1303740110634564e-06,
+      "loss": 0.411,
+      "step": 12498
+    },
+    {
+      "epoch": 1.0651043885811675,
+      "grad_norm": 14.875,
+      "learning_rate": 1.1302067478529237e-06,
+      "loss": 0.7828,
+      "step": 12499
+    },
+    {
+      "epoch": 1.0651896037494675,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.1300394868071755e-06,
+      "loss": 0.3977,
+      "step": 12500
+    },
+    {
+      "epoch": 1.0652748189177674,
+      "grad_norm": 11.25,
+      "learning_rate": 1.1298722279292342e-06,
+      "loss": 0.3172,
+      "step": 12501
+    },
+    {
+      "epoch": 1.0653600340860674,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1297049712221228e-06,
+      "loss": 0.4312,
+      "step": 12502
+    },
+    {
+      "epoch": 1.0654452492543673,
+      "grad_norm": 19.125,
+      "learning_rate": 1.1295377166888635e-06,
+      "loss": 0.8106,
+      "step": 12503
+    },
+    {
+      "epoch": 1.0655304644226673,
+      "grad_norm": 17.0,
+      "learning_rate": 1.129370464332479e-06,
+      "loss": 0.8247,
+      "step": 12504
+    },
+    {
+      "epoch": 1.0656156795909673,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1292032141559914e-06,
+      "loss": 0.6249,
+      "step": 12505
+    },
+    {
+      "epoch": 1.0657008947592672,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.129035966162423e-06,
+      "loss": 0.2157,
+      "step": 12506
+    },
+    {
+      "epoch": 1.0657861099275672,
+      "grad_norm": 8.75,
+      "learning_rate": 1.1288687203547964e-06,
+      "loss": 0.1909,
+      "step": 12507
+    },
+    {
+      "epoch": 1.0658713250958671,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1287014767361338e-06,
+      "loss": 0.745,
+      "step": 12508
+    },
+    {
+      "epoch": 1.065956540264167,
+      "grad_norm": 20.25,
+      "learning_rate": 1.1285342353094574e-06,
+      "loss": 0.787,
+      "step": 12509
+    },
+    {
+      "epoch": 1.066041755432467,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.1283669960777896e-06,
+      "loss": 0.5329,
+      "step": 12510
+    },
+    {
+      "epoch": 1.066126970600767,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.1281997590441526e-06,
+      "loss": 0.5704,
+      "step": 12511
+    },
+    {
+      "epoch": 1.066212185769067,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1280325242115688e-06,
+      "loss": 0.5973,
+      "step": 12512
+    },
+    {
+      "epoch": 1.066297400937367,
+      "grad_norm": 12.25,
+      "learning_rate": 1.1278652915830596e-06,
+      "loss": 0.3941,
+      "step": 12513
+    },
+    {
+      "epoch": 1.0663826161056669,
+      "grad_norm": 11.125,
+      "learning_rate": 1.1276980611616472e-06,
+      "loss": 0.3744,
+      "step": 12514
+    },
+    {
+      "epoch": 1.0664678312739668,
+      "grad_norm": 14.75,
+      "learning_rate": 1.127530832950354e-06,
+      "loss": 0.6654,
+      "step": 12515
+    },
+    {
+      "epoch": 1.0665530464422668,
+      "grad_norm": 14.0,
+      "learning_rate": 1.1273636069522015e-06,
+      "loss": 0.6663,
+      "step": 12516
+    },
+    {
+      "epoch": 1.0666382616105667,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1271963831702122e-06,
+      "loss": 0.626,
+      "step": 12517
+    },
+    {
+      "epoch": 1.0667234767788667,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1270291616074075e-06,
+      "loss": 0.8603,
+      "step": 12518
+    },
+    {
+      "epoch": 1.0668086919471667,
+      "grad_norm": 13.625,
+      "learning_rate": 1.1268619422668098e-06,
+      "loss": 0.6728,
+      "step": 12519
+    },
+    {
+      "epoch": 1.0668939071154666,
+      "grad_norm": 18.0,
+      "learning_rate": 1.126694725151441e-06,
+      "loss": 0.7052,
+      "step": 12520
+    },
+    {
+      "epoch": 1.0669791222837666,
+      "grad_norm": 13.875,
+      "learning_rate": 1.126527510264322e-06,
+      "loss": 0.485,
+      "step": 12521
+    },
+    {
+      "epoch": 1.0670643374520665,
+      "grad_norm": 17.0,
+      "learning_rate": 1.126360297608475e-06,
+      "loss": 0.9502,
+      "step": 12522
+    },
+    {
+      "epoch": 1.0671495526203665,
+      "grad_norm": 13.875,
+      "learning_rate": 1.1261930871869217e-06,
+      "loss": 0.4682,
+      "step": 12523
+    },
+    {
+      "epoch": 1.0672347677886664,
+      "grad_norm": 19.875,
+      "learning_rate": 1.126025879002684e-06,
+      "loss": 0.4903,
+      "step": 12524
+    },
+    {
+      "epoch": 1.0673199829569664,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.1258586730587832e-06,
+      "loss": 0.5351,
+      "step": 12525
+    },
+    {
+      "epoch": 1.0674051981252664,
+      "grad_norm": 16.75,
+      "learning_rate": 1.125691469358241e-06,
+      "loss": 0.8411,
+      "step": 12526
+    },
+    {
+      "epoch": 1.0674904132935663,
+      "grad_norm": 22.375,
+      "learning_rate": 1.1255242679040795e-06,
+      "loss": 0.7882,
+      "step": 12527
+    },
+    {
+      "epoch": 1.0675756284618663,
+      "grad_norm": 19.75,
+      "learning_rate": 1.1253570686993188e-06,
+      "loss": 0.7725,
+      "step": 12528
+    },
+    {
+      "epoch": 1.0676608436301662,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.125189871746982e-06,
+      "loss": 0.4929,
+      "step": 12529
+    },
+    {
+      "epoch": 1.0677460587984662,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1250226770500895e-06,
+      "loss": 0.6896,
+      "step": 12530
+    },
+    {
+      "epoch": 1.0678312739667661,
+      "grad_norm": 28.0,
+      "learning_rate": 1.1248554846116625e-06,
+      "loss": 0.6349,
+      "step": 12531
+    },
+    {
+      "epoch": 1.067916489135066,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.1246882944347232e-06,
+      "loss": 0.4214,
+      "step": 12532
+    },
+    {
+      "epoch": 1.068001704303366,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.124521106522292e-06,
+      "loss": 0.4369,
+      "step": 12533
+    },
+    {
+      "epoch": 1.068086919471666,
+      "grad_norm": 17.75,
+      "learning_rate": 1.124353920877391e-06,
+      "loss": 0.5722,
+      "step": 12534
+    },
+    {
+      "epoch": 1.068172134639966,
+      "grad_norm": 12.125,
+      "learning_rate": 1.1241867375030404e-06,
+      "loss": 0.5572,
+      "step": 12535
+    },
+    {
+      "epoch": 1.068257349808266,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1240195564022625e-06,
+      "loss": 0.6553,
+      "step": 12536
+    },
+    {
+      "epoch": 1.0683425649765659,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.123852377578078e-06,
+      "loss": 0.6276,
+      "step": 12537
+    },
+    {
+      "epoch": 1.0684277801448658,
+      "grad_norm": 12.375,
+      "learning_rate": 1.1236852010335077e-06,
+      "loss": 0.4641,
+      "step": 12538
+    },
+    {
+      "epoch": 1.0685129953131658,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1235180267715728e-06,
+      "loss": 1.0508,
+      "step": 12539
+    },
+    {
+      "epoch": 1.0685982104814657,
+      "grad_norm": 12.0,
+      "learning_rate": 1.1233508547952942e-06,
+      "loss": 0.3245,
+      "step": 12540
+    },
+    {
+      "epoch": 1.0686834256497657,
+      "grad_norm": 19.375,
+      "learning_rate": 1.1231836851076931e-06,
+      "loss": 0.6954,
+      "step": 12541
+    },
+    {
+      "epoch": 1.0687686408180657,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1230165177117906e-06,
+      "loss": 0.7253,
+      "step": 12542
+    },
+    {
+      "epoch": 1.0688538559863656,
+      "grad_norm": 14.125,
+      "learning_rate": 1.122849352610607e-06,
+      "loss": 0.5273,
+      "step": 12543
+    },
+    {
+      "epoch": 1.0689390711546656,
+      "grad_norm": 22.75,
+      "learning_rate": 1.1226821898071635e-06,
+      "loss": 0.4173,
+      "step": 12544
+    },
+    {
+      "epoch": 1.0690242863229655,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1225150293044813e-06,
+      "loss": 0.7599,
+      "step": 12545
+    },
+    {
+      "epoch": 1.0691095014912655,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.12234787110558e-06,
+      "loss": 0.7337,
+      "step": 12546
+    },
+    {
+      "epoch": 1.0691947166595654,
+      "grad_norm": 21.125,
+      "learning_rate": 1.1221807152134816e-06,
+      "loss": 1.0787,
+      "step": 12547
+    },
+    {
+      "epoch": 1.0692799318278654,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.1220135616312057e-06,
+      "loss": 0.5756,
+      "step": 12548
+    },
+    {
+      "epoch": 1.0693651469961654,
+      "grad_norm": 12.75,
+      "learning_rate": 1.1218464103617737e-06,
+      "loss": 0.486,
+      "step": 12549
+    },
+    {
+      "epoch": 1.0694503621644653,
+      "grad_norm": 16.5,
+      "learning_rate": 1.121679261408206e-06,
+      "loss": 0.6116,
+      "step": 12550
+    },
+    {
+      "epoch": 1.0695355773327653,
+      "grad_norm": 33.0,
+      "learning_rate": 1.1215121147735229e-06,
+      "loss": 0.7586,
+      "step": 12551
+    },
+    {
+      "epoch": 1.0696207925010652,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.121344970460745e-06,
+      "loss": 0.3002,
+      "step": 12552
+    },
+    {
+      "epoch": 1.0697060076693652,
+      "grad_norm": 12.625,
+      "learning_rate": 1.1211778284728936e-06,
+      "loss": 0.3824,
+      "step": 12553
+    },
+    {
+      "epoch": 1.0697912228376651,
+      "grad_norm": 12.875,
+      "learning_rate": 1.121010688812988e-06,
+      "loss": 0.3992,
+      "step": 12554
+    },
+    {
+      "epoch": 1.069876438005965,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.1208435514840485e-06,
+      "loss": 0.4742,
+      "step": 12555
+    },
+    {
+      "epoch": 1.069961653174265,
+      "grad_norm": 15.0,
+      "learning_rate": 1.1206764164890962e-06,
+      "loss": 0.6216,
+      "step": 12556
+    },
+    {
+      "epoch": 1.070046868342565,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.1205092838311513e-06,
+      "loss": 0.2655,
+      "step": 12557
+    },
+    {
+      "epoch": 1.070132083510865,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1203421535132334e-06,
+      "loss": 0.7743,
+      "step": 12558
+    },
+    {
+      "epoch": 1.070217298679165,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.1201750255383636e-06,
+      "loss": 0.4042,
+      "step": 12559
+    },
+    {
+      "epoch": 1.0703025138474649,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.1200078999095612e-06,
+      "loss": 0.5265,
+      "step": 12560
+    },
+    {
+      "epoch": 1.0703877290157648,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1198407766298475e-06,
+      "loss": 0.851,
+      "step": 12561
+    },
+    {
+      "epoch": 1.0704729441840648,
+      "grad_norm": 10.75,
+      "learning_rate": 1.1196736557022414e-06,
+      "loss": 0.3177,
+      "step": 12562
+    },
+    {
+      "epoch": 1.0705581593523648,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1195065371297633e-06,
+      "loss": 0.5534,
+      "step": 12563
+    },
+    {
+      "epoch": 1.0706433745206647,
+      "grad_norm": 15.875,
+      "learning_rate": 1.1193394209154335e-06,
+      "loss": 0.5832,
+      "step": 12564
+    },
+    {
+      "epoch": 1.0707285896889647,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1191723070622714e-06,
+      "loss": 0.4795,
+      "step": 12565
+    },
+    {
+      "epoch": 1.0708138048572646,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1190051955732974e-06,
+      "loss": 0.6782,
+      "step": 12566
+    },
+    {
+      "epoch": 1.0708990200255646,
+      "grad_norm": 28.625,
+      "learning_rate": 1.1188380864515313e-06,
+      "loss": 1.1923,
+      "step": 12567
+    },
+    {
+      "epoch": 1.0709842351938645,
+      "grad_norm": 17.625,
+      "learning_rate": 1.118670979699993e-06,
+      "loss": 0.5508,
+      "step": 12568
+    },
+    {
+      "epoch": 1.0710694503621645,
+      "grad_norm": 13.375,
+      "learning_rate": 1.1185038753217024e-06,
+      "loss": 0.4533,
+      "step": 12569
+    },
+    {
+      "epoch": 1.0711546655304645,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1183367733196785e-06,
+      "loss": 0.39,
+      "step": 12570
+    },
+    {
+      "epoch": 1.0712398806987644,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1181696736969418e-06,
+      "loss": 0.7198,
+      "step": 12571
+    },
+    {
+      "epoch": 1.0713250958670644,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1180025764565117e-06,
+      "loss": 0.7945,
+      "step": 12572
+    },
+    {
+      "epoch": 1.0714103110353643,
+      "grad_norm": 21.5,
+      "learning_rate": 1.1178354816014073e-06,
+      "loss": 0.7167,
+      "step": 12573
+    },
+    {
+      "epoch": 1.0714955262036643,
+      "grad_norm": 10.875,
+      "learning_rate": 1.1176683891346491e-06,
+      "loss": 0.3108,
+      "step": 12574
+    },
+    {
+      "epoch": 1.0715807413719642,
+      "grad_norm": 17.75,
+      "learning_rate": 1.117501299059256e-06,
+      "loss": 0.4911,
+      "step": 12575
+    },
+    {
+      "epoch": 1.0716659565402642,
+      "grad_norm": 16.625,
+      "learning_rate": 1.1173342113782476e-06,
+      "loss": 0.652,
+      "step": 12576
+    },
+    {
+      "epoch": 1.0717511717085642,
+      "grad_norm": 18.0,
+      "learning_rate": 1.117167126094644e-06,
+      "loss": 0.5362,
+      "step": 12577
+    },
+    {
+      "epoch": 1.0718363868768641,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.1170000432114634e-06,
+      "loss": 0.6891,
+      "step": 12578
+    },
+    {
+      "epoch": 1.071921602045164,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.116832962731726e-06,
+      "loss": 0.5162,
+      "step": 12579
+    },
+    {
+      "epoch": 1.072006817213464,
+      "grad_norm": 12.625,
+      "learning_rate": 1.1166658846584505e-06,
+      "loss": 0.4009,
+      "step": 12580
+    },
+    {
+      "epoch": 1.072092032381764,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1164988089946568e-06,
+      "loss": 0.7076,
+      "step": 12581
+    },
+    {
+      "epoch": 1.072177247550064,
+      "grad_norm": 13.625,
+      "learning_rate": 1.1163317357433638e-06,
+      "loss": 0.5016,
+      "step": 12582
+    },
+    {
+      "epoch": 1.072262462718364,
+      "grad_norm": 12.25,
+      "learning_rate": 1.1161646649075909e-06,
+      "loss": 0.4133,
+      "step": 12583
+    },
+    {
+      "epoch": 1.0723476778866639,
+      "grad_norm": 28.875,
+      "learning_rate": 1.1159975964903566e-06,
+      "loss": 1.1796,
+      "step": 12584
+    },
+    {
+      "epoch": 1.0724328930549638,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1158305304946814e-06,
+      "loss": 0.8258,
+      "step": 12585
+    },
+    {
+      "epoch": 1.0725181082232638,
+      "grad_norm": 16.875,
+      "learning_rate": 1.115663466923583e-06,
+      "loss": 0.5508,
+      "step": 12586
+    },
+    {
+      "epoch": 1.0726033233915637,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.1154964057800806e-06,
+      "loss": 0.8667,
+      "step": 12587
+    },
+    {
+      "epoch": 1.0726885385598637,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.1153293470671936e-06,
+      "loss": 0.4107,
+      "step": 12588
+    },
+    {
+      "epoch": 1.0727737537281636,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1151622907879407e-06,
+      "loss": 0.6532,
+      "step": 12589
+    },
+    {
+      "epoch": 1.0728589688964636,
+      "grad_norm": 20.375,
+      "learning_rate": 1.1149952369453404e-06,
+      "loss": 0.7967,
+      "step": 12590
+    },
+    {
+      "epoch": 1.0729441840647636,
+      "grad_norm": 13.875,
+      "learning_rate": 1.1148281855424125e-06,
+      "loss": 0.795,
+      "step": 12591
+    },
+    {
+      "epoch": 1.0730293992330635,
+      "grad_norm": 24.75,
+      "learning_rate": 1.114661136582175e-06,
+      "loss": 1.0112,
+      "step": 12592
+    },
+    {
+      "epoch": 1.0731146144013635,
+      "grad_norm": 19.375,
+      "learning_rate": 1.114494090067647e-06,
+      "loss": 0.7698,
+      "step": 12593
+    },
+    {
+      "epoch": 1.0731998295696634,
+      "grad_norm": 13.625,
+      "learning_rate": 1.114327046001847e-06,
+      "loss": 0.3142,
+      "step": 12594
+    },
+    {
+      "epoch": 1.0732850447379634,
+      "grad_norm": 14.5,
+      "learning_rate": 1.1141600043877935e-06,
+      "loss": 0.6812,
+      "step": 12595
+    },
+    {
+      "epoch": 1.0733702599062633,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1139929652285056e-06,
+      "loss": 0.4265,
+      "step": 12596
+    },
+    {
+      "epoch": 1.0734554750745633,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.113825928527001e-06,
+      "loss": 0.4827,
+      "step": 12597
+    },
+    {
+      "epoch": 1.0735406902428632,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1136588942862995e-06,
+      "loss": 0.5722,
+      "step": 12598
+    },
+    {
+      "epoch": 1.0736259054111632,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1134918625094185e-06,
+      "loss": 0.522,
+      "step": 12599
+    },
+    {
+      "epoch": 1.0737111205794632,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1133248331993771e-06,
+      "loss": 0.898,
+      "step": 12600
+    },
+    {
+      "epoch": 1.0737963357477631,
+      "grad_norm": 16.5,
+      "learning_rate": 1.113157806359193e-06,
+      "loss": 0.6983,
+      "step": 12601
+    },
+    {
+      "epoch": 1.073881550916063,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1129907819918856e-06,
+      "loss": 0.7055,
+      "step": 12602
+    },
+    {
+      "epoch": 1.073966766084363,
+      "grad_norm": 17.125,
+      "learning_rate": 1.112823760100472e-06,
+      "loss": 0.8291,
+      "step": 12603
+    },
+    {
+      "epoch": 1.074051981252663,
+      "grad_norm": 23.375,
+      "learning_rate": 1.1126567406879713e-06,
+      "loss": 1.2439,
+      "step": 12604
+    },
+    {
+      "epoch": 1.074137196420963,
+      "grad_norm": 17.25,
+      "learning_rate": 1.112489723757401e-06,
+      "loss": 0.5468,
+      "step": 12605
+    },
+    {
+      "epoch": 1.074222411589263,
+      "grad_norm": 15.625,
+      "learning_rate": 1.11232270931178e-06,
+      "loss": 0.6623,
+      "step": 12606
+    },
+    {
+      "epoch": 1.0743076267575629,
+      "grad_norm": 19.0,
+      "learning_rate": 1.1121556973541259e-06,
+      "loss": 0.7655,
+      "step": 12607
+    },
+    {
+      "epoch": 1.0743928419258628,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.111988687887457e-06,
+      "loss": 0.7381,
+      "step": 12608
+    },
+    {
+      "epoch": 1.0744780570941628,
+      "grad_norm": 21.375,
+      "learning_rate": 1.1118216809147913e-06,
+      "loss": 0.6058,
+      "step": 12609
+    },
+    {
+      "epoch": 1.0745632722624627,
+      "grad_norm": 16.25,
+      "learning_rate": 1.111654676439147e-06,
+      "loss": 0.8145,
+      "step": 12610
+    },
+    {
+      "epoch": 1.0746484874307627,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1114876744635416e-06,
+      "loss": 0.3852,
+      "step": 12611
+    },
+    {
+      "epoch": 1.0747337025990626,
+      "grad_norm": 16.875,
+      "learning_rate": 1.111320674990993e-06,
+      "loss": 0.5855,
+      "step": 12612
+    },
+    {
+      "epoch": 1.0748189177673626,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1111536780245196e-06,
+      "loss": 0.4772,
+      "step": 12613
+    },
+    {
+      "epoch": 1.0749041329356626,
+      "grad_norm": 10.875,
+      "learning_rate": 1.1109866835671386e-06,
+      "loss": 0.2645,
+      "step": 12614
+    },
+    {
+      "epoch": 1.0749893481039625,
+      "grad_norm": 15.0,
+      "learning_rate": 1.1108196916218683e-06,
+      "loss": 0.5453,
+      "step": 12615
+    },
+    {
+      "epoch": 1.0750745632722625,
+      "grad_norm": 15.375,
+      "learning_rate": 1.1106527021917259e-06,
+      "loss": 0.6427,
+      "step": 12616
+    },
+    {
+      "epoch": 1.0751597784405624,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1104857152797294e-06,
+      "loss": 0.7207,
+      "step": 12617
+    },
+    {
+      "epoch": 1.0752449936088624,
+      "grad_norm": 19.875,
+      "learning_rate": 1.1103187308888967e-06,
+      "loss": 0.5573,
+      "step": 12618
+    },
+    {
+      "epoch": 1.0753302087771623,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.1101517490222447e-06,
+      "loss": 0.3856,
+      "step": 12619
+    },
+    {
+      "epoch": 1.0754154239454623,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1099847696827908e-06,
+      "loss": 0.4324,
+      "step": 12620
+    },
+    {
+      "epoch": 1.0755006391137623,
+      "grad_norm": 10.375,
+      "learning_rate": 1.1098177928735534e-06,
+      "loss": 0.2635,
+      "step": 12621
+    },
+    {
+      "epoch": 1.0755858542820622,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.1096508185975492e-06,
+      "loss": 0.2998,
+      "step": 12622
+    },
+    {
+      "epoch": 1.0756710694503622,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.109483846857796e-06,
+      "loss": 0.5418,
+      "step": 12623
+    },
+    {
+      "epoch": 1.0757562846186621,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1093168776573105e-06,
+      "loss": 0.6295,
+      "step": 12624
+    },
+    {
+      "epoch": 1.075841499786962,
+      "grad_norm": 20.375,
+      "learning_rate": 1.109149910999111e-06,
+      "loss": 0.7588,
+      "step": 12625
+    },
+    {
+      "epoch": 1.075926714955262,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.1089829468862146e-06,
+      "loss": 0.5768,
+      "step": 12626
+    },
+    {
+      "epoch": 1.076011930123562,
+      "grad_norm": 14.875,
+      "learning_rate": 1.1088159853216374e-06,
+      "loss": 0.9127,
+      "step": 12627
+    },
+    {
+      "epoch": 1.076097145291862,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1086490263083977e-06,
+      "loss": 0.5684,
+      "step": 12628
+    },
+    {
+      "epoch": 1.076182360460162,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1084820698495121e-06,
+      "loss": 0.7395,
+      "step": 12629
+    },
+    {
+      "epoch": 1.0762675756284619,
+      "grad_norm": 21.375,
+      "learning_rate": 1.1083151159479977e-06,
+      "loss": 0.5173,
+      "step": 12630
+    },
+    {
+      "epoch": 1.0763527907967618,
+      "grad_norm": 14.125,
+      "learning_rate": 1.1081481646068718e-06,
+      "loss": 0.4354,
+      "step": 12631
+    },
+    {
+      "epoch": 1.0764380059650618,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1079812158291512e-06,
+      "loss": 0.5181,
+      "step": 12632
+    },
+    {
+      "epoch": 1.0765232211333617,
+      "grad_norm": 13.125,
+      "learning_rate": 1.1078142696178525e-06,
+      "loss": 0.557,
+      "step": 12633
+    },
+    {
+      "epoch": 1.0766084363016617,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.107647325975994e-06,
+      "loss": 0.4769,
+      "step": 12634
+    },
+    {
+      "epoch": 1.0766936514699617,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1074803849065906e-06,
+      "loss": 0.8167,
+      "step": 12635
+    },
+    {
+      "epoch": 1.0767788666382616,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.10731344641266e-06,
+      "loss": 0.4167,
+      "step": 12636
+    },
+    {
+      "epoch": 1.0768640818065616,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1071465104972188e-06,
+      "loss": 0.3861,
+      "step": 12637
+    },
+    {
+      "epoch": 1.0769492969748615,
+      "grad_norm": 14.375,
+      "learning_rate": 1.1069795771632841e-06,
+      "loss": 0.4314,
+      "step": 12638
+    },
+    {
+      "epoch": 1.0770345121431615,
+      "grad_norm": 11.5,
+      "learning_rate": 1.106812646413872e-06,
+      "loss": 0.3936,
+      "step": 12639
+    },
+    {
+      "epoch": 1.0771197273114614,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1066457182519997e-06,
+      "loss": 0.6304,
+      "step": 12640
+    },
+    {
+      "epoch": 1.0772049424797614,
+      "grad_norm": 12.5,
+      "learning_rate": 1.106478792680683e-06,
+      "loss": 0.4781,
+      "step": 12641
+    },
+    {
+      "epoch": 1.0772901576480614,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.1063118697029394e-06,
+      "loss": 0.6092,
+      "step": 12642
+    },
+    {
+      "epoch": 1.0773753728163613,
+      "grad_norm": 41.75,
+      "learning_rate": 1.1061449493217844e-06,
+      "loss": 0.6386,
+      "step": 12643
+    },
+    {
+      "epoch": 1.0774605879846613,
+      "grad_norm": 22.375,
+      "learning_rate": 1.1059780315402348e-06,
+      "loss": 0.8606,
+      "step": 12644
+    },
+    {
+      "epoch": 1.0775458031529612,
+      "grad_norm": 10.5,
+      "learning_rate": 1.1058111163613071e-06,
+      "loss": 0.2788,
+      "step": 12645
+    },
+    {
+      "epoch": 1.0776310183212612,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1056442037880174e-06,
+      "loss": 0.6694,
+      "step": 12646
+    },
+    {
+      "epoch": 1.0777162334895611,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.1054772938233825e-06,
+      "loss": 0.6951,
+      "step": 12647
+    },
+    {
+      "epoch": 1.077801448657861,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1053103864704175e-06,
+      "loss": 0.7417,
+      "step": 12648
+    },
+    {
+      "epoch": 1.077886663826161,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.10514348173214e-06,
+      "loss": 0.4677,
+      "step": 12649
+    },
+    {
+      "epoch": 1.077971878994461,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1049765796115655e-06,
+      "loss": 0.8182,
+      "step": 12650
+    },
+    {
+      "epoch": 1.078057094162761,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1048096801117098e-06,
+      "loss": 0.7342,
+      "step": 12651
+    },
+    {
+      "epoch": 1.078142309331061,
+      "grad_norm": 19.875,
+      "learning_rate": 1.104642783235589e-06,
+      "loss": 0.5371,
+      "step": 12652
+    },
+    {
+      "epoch": 1.0782275244993609,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.1044758889862195e-06,
+      "loss": 0.5905,
+      "step": 12653
+    },
+    {
+      "epoch": 1.0783127396676608,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.1043089973666167e-06,
+      "loss": 0.4751,
+      "step": 12654
+    },
+    {
+      "epoch": 1.0783979548359608,
+      "grad_norm": 14.625,
+      "learning_rate": 1.104142108379797e-06,
+      "loss": 0.5421,
+      "step": 12655
+    },
+    {
+      "epoch": 1.0784831700042607,
+      "grad_norm": 15.625,
+      "learning_rate": 1.103975222028776e-06,
+      "loss": 0.7648,
+      "step": 12656
+    },
+    {
+      "epoch": 1.0785683851725607,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1038083383165698e-06,
+      "loss": 0.635,
+      "step": 12657
+    },
+    {
+      "epoch": 1.0786536003408607,
+      "grad_norm": 19.625,
+      "learning_rate": 1.1036414572461942e-06,
+      "loss": 1.1928,
+      "step": 12658
+    },
+    {
+      "epoch": 1.0787388155091606,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.1034745788206642e-06,
+      "loss": 0.6254,
+      "step": 12659
+    },
+    {
+      "epoch": 1.0788240306774606,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1033077030429958e-06,
+      "loss": 0.4726,
+      "step": 12660
+    },
+    {
+      "epoch": 1.0789092458457605,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1031408299162049e-06,
+      "loss": 0.8441,
+      "step": 12661
+    },
+    {
+      "epoch": 1.0789944610140605,
+      "grad_norm": 37.0,
+      "learning_rate": 1.102973959443307e-06,
+      "loss": 0.841,
+      "step": 12662
+    },
+    {
+      "epoch": 1.0790796761823604,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.102807091627317e-06,
+      "loss": 0.5267,
+      "step": 12663
+    },
+    {
+      "epoch": 1.0791648913506604,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1026402264712515e-06,
+      "loss": 0.8202,
+      "step": 12664
+    },
+    {
+      "epoch": 1.0792501065189604,
+      "grad_norm": 14.75,
+      "learning_rate": 1.102473363978125e-06,
+      "loss": 0.3713,
+      "step": 12665
+    },
+    {
+      "epoch": 1.0793353216872603,
+      "grad_norm": 15.0,
+      "learning_rate": 1.102306504150953e-06,
+      "loss": 0.7345,
+      "step": 12666
+    },
+    {
+      "epoch": 1.0794205368555603,
+      "grad_norm": 15.25,
+      "learning_rate": 1.1021396469927517e-06,
+      "loss": 0.8014,
+      "step": 12667
+    },
+    {
+      "epoch": 1.0795057520238602,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1019727925065352e-06,
+      "loss": 0.4512,
+      "step": 12668
+    },
+    {
+      "epoch": 1.0795909671921602,
+      "grad_norm": 16.875,
+      "learning_rate": 1.101805940695319e-06,
+      "loss": 0.7712,
+      "step": 12669
+    },
+    {
+      "epoch": 1.0796761823604601,
+      "grad_norm": 12.125,
+      "learning_rate": 1.1016390915621186e-06,
+      "loss": 0.4908,
+      "step": 12670
+    },
+    {
+      "epoch": 1.07976139752876,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.1014722451099489e-06,
+      "loss": 0.6974,
+      "step": 12671
+    },
+    {
+      "epoch": 1.07984661269706,
+      "grad_norm": 15.125,
+      "learning_rate": 1.101305401341825e-06,
+      "loss": 0.529,
+      "step": 12672
+    },
+    {
+      "epoch": 1.07993182786536,
+      "grad_norm": 19.75,
+      "learning_rate": 1.101138560260762e-06,
+      "loss": 0.5152,
+      "step": 12673
+    },
+    {
+      "epoch": 1.08001704303366,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1009717218697753e-06,
+      "loss": 0.799,
+      "step": 12674
+    },
+    {
+      "epoch": 1.08010225820196,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.1008048861718798e-06,
+      "loss": 0.3283,
+      "step": 12675
+    },
+    {
+      "epoch": 1.0801874733702599,
+      "grad_norm": 15.75,
+      "learning_rate": 1.100638053170089e-06,
+      "loss": 0.7773,
+      "step": 12676
+    },
+    {
+      "epoch": 1.0802726885385598,
+      "grad_norm": 20.125,
+      "learning_rate": 1.1004712228674196e-06,
+      "loss": 0.6056,
+      "step": 12677
+    },
+    {
+      "epoch": 1.0803579037068598,
+      "grad_norm": 22.125,
+      "learning_rate": 1.100304395266885e-06,
+      "loss": 1.0682,
+      "step": 12678
+    },
+    {
+      "epoch": 1.0804431188751598,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.1001375703715008e-06,
+      "loss": 0.2952,
+      "step": 12679
+    },
+    {
+      "epoch": 1.0805283340434597,
+      "grad_norm": 45.0,
+      "learning_rate": 1.0999707481842816e-06,
+      "loss": 0.8158,
+      "step": 12680
+    },
+    {
+      "epoch": 1.0806135492117597,
+      "grad_norm": 15.375,
+      "learning_rate": 1.0998039287082412e-06,
+      "loss": 0.5468,
+      "step": 12681
+    },
+    {
+      "epoch": 1.0806987643800596,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.0996371119463955e-06,
+      "loss": 0.3753,
+      "step": 12682
+    },
+    {
+      "epoch": 1.0807839795483596,
+      "grad_norm": 15.125,
+      "learning_rate": 1.0994702979017584e-06,
+      "loss": 0.5479,
+      "step": 12683
+    },
+    {
+      "epoch": 1.0808691947166595,
+      "grad_norm": 14.25,
+      "learning_rate": 1.099303486577344e-06,
+      "loss": 0.3836,
+      "step": 12684
+    },
+    {
+      "epoch": 1.0809544098849595,
+      "grad_norm": 20.125,
+      "learning_rate": 1.0991366779761673e-06,
+      "loss": 0.8661,
+      "step": 12685
+    },
+    {
+      "epoch": 1.0810396250532595,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0989698721012424e-06,
+      "loss": 0.3521,
+      "step": 12686
+    },
+    {
+      "epoch": 1.0811248402215594,
+      "grad_norm": 27.0,
+      "learning_rate": 1.098803068955584e-06,
+      "loss": 0.7299,
+      "step": 12687
+    },
+    {
+      "epoch": 1.0812100553898594,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.0986362685422057e-06,
+      "loss": 0.5865,
+      "step": 12688
+    },
+    {
+      "epoch": 1.0812952705581593,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.0984694708641228e-06,
+      "loss": 0.7113,
+      "step": 12689
+    },
+    {
+      "epoch": 1.0813804857264593,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.0983026759243484e-06,
+      "loss": 0.4307,
+      "step": 12690
+    },
+    {
+      "epoch": 1.0814657008947592,
+      "grad_norm": 14.75,
+      "learning_rate": 1.0981358837258979e-06,
+      "loss": 0.5553,
+      "step": 12691
+    },
+    {
+      "epoch": 1.0815509160630592,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0979690942717843e-06,
+      "loss": 0.3945,
+      "step": 12692
+    },
+    {
+      "epoch": 1.0816361312313592,
+      "grad_norm": 16.5,
+      "learning_rate": 1.097802307565022e-06,
+      "loss": 0.6793,
+      "step": 12693
+    },
+    {
+      "epoch": 1.0817213463996591,
+      "grad_norm": 34.0,
+      "learning_rate": 1.0976355236086252e-06,
+      "loss": 0.7968,
+      "step": 12694
+    },
+    {
+      "epoch": 1.081806561567959,
+      "grad_norm": 20.25,
+      "learning_rate": 1.0974687424056076e-06,
+      "loss": 0.6531,
+      "step": 12695
+    },
+    {
+      "epoch": 1.081891776736259,
+      "grad_norm": 16.875,
+      "learning_rate": 1.097301963958983e-06,
+      "loss": 0.7813,
+      "step": 12696
+    },
+    {
+      "epoch": 1.081976991904559,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0971351882717659e-06,
+      "loss": 0.5741,
+      "step": 12697
+    },
+    {
+      "epoch": 1.082062207072859,
+      "grad_norm": 34.75,
+      "learning_rate": 1.0969684153469692e-06,
+      "loss": 0.7351,
+      "step": 12698
+    },
+    {
+      "epoch": 1.082147422241159,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0968016451876078e-06,
+      "loss": 0.6071,
+      "step": 12699
+    },
+    {
+      "epoch": 1.0822326374094589,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.0966348777966942e-06,
+      "loss": 0.5334,
+      "step": 12700
+    },
+    {
+      "epoch": 1.0823178525777588,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0964681131772426e-06,
+      "loss": 0.4378,
+      "step": 12701
+    },
+    {
+      "epoch": 1.0824030677460588,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0963013513322668e-06,
+      "loss": 0.5015,
+      "step": 12702
+    },
+    {
+      "epoch": 1.0824882829143587,
+      "grad_norm": 15.375,
+      "learning_rate": 1.0961345922647796e-06,
+      "loss": 0.7422,
+      "step": 12703
+    },
+    {
+      "epoch": 1.0825734980826587,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0959678359777956e-06,
+      "loss": 0.8333,
+      "step": 12704
+    },
+    {
+      "epoch": 1.0826587132509586,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0958010824743273e-06,
+      "loss": 0.5575,
+      "step": 12705
+    },
+    {
+      "epoch": 1.0827439284192586,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.095634331757389e-06,
+      "loss": 0.508,
+      "step": 12706
+    },
+    {
+      "epoch": 1.0828291435875586,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0954675838299936e-06,
+      "loss": 0.8037,
+      "step": 12707
+    },
+    {
+      "epoch": 1.0829143587558585,
+      "grad_norm": 18.125,
+      "learning_rate": 1.0953008386951538e-06,
+      "loss": 0.7792,
+      "step": 12708
+    },
+    {
+      "epoch": 1.0829995739241585,
+      "grad_norm": 39.25,
+      "learning_rate": 1.0951340963558836e-06,
+      "loss": 1.1254,
+      "step": 12709
+    },
+    {
+      "epoch": 1.0830847890924584,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.0949673568151962e-06,
+      "loss": 0.4919,
+      "step": 12710
+    },
+    {
+      "epoch": 1.0831700042607584,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0948006200761042e-06,
+      "loss": 0.3435,
+      "step": 12711
+    },
+    {
+      "epoch": 1.0832552194290583,
+      "grad_norm": 18.25,
+      "learning_rate": 1.0946338861416213e-06,
+      "loss": 0.2558,
+      "step": 12712
+    },
+    {
+      "epoch": 1.0833404345973583,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.09446715501476e-06,
+      "loss": 0.5498,
+      "step": 12713
+    },
+    {
+      "epoch": 1.0834256497656582,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.094300426698534e-06,
+      "loss": 0.5399,
+      "step": 12714
+    },
+    {
+      "epoch": 1.0835108649339582,
+      "grad_norm": 14.375,
+      "learning_rate": 1.0941337011959562e-06,
+      "loss": 0.83,
+      "step": 12715
+    },
+    {
+      "epoch": 1.0835960801022582,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0939669785100384e-06,
+      "loss": 0.2942,
+      "step": 12716
+    },
+    {
+      "epoch": 1.0836812952705581,
+      "grad_norm": 13.0,
+      "learning_rate": 1.0938002586437947e-06,
+      "loss": 0.5229,
+      "step": 12717
+    },
+    {
+      "epoch": 1.083766510438858,
+      "grad_norm": 19.875,
+      "learning_rate": 1.093633541600237e-06,
+      "loss": 0.9032,
+      "step": 12718
+    },
+    {
+      "epoch": 1.083851725607158,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0934668273823787e-06,
+      "loss": 0.7821,
+      "step": 12719
+    },
+    {
+      "epoch": 1.083936940775458,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.093300115993232e-06,
+      "loss": 0.6005,
+      "step": 12720
+    },
+    {
+      "epoch": 1.084022155943758,
+      "grad_norm": 15.875,
+      "learning_rate": 1.09313340743581e-06,
+      "loss": 0.1398,
+      "step": 12721
+    },
+    {
+      "epoch": 1.084107371112058,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0929667017131248e-06,
+      "loss": 0.8508,
+      "step": 12722
+    },
+    {
+      "epoch": 1.0841925862803579,
+      "grad_norm": 16.75,
+      "learning_rate": 1.0927999988281898e-06,
+      "loss": 0.7227,
+      "step": 12723
+    },
+    {
+      "epoch": 1.0842778014486578,
+      "grad_norm": 20.375,
+      "learning_rate": 1.0926332987840167e-06,
+      "loss": 0.8254,
+      "step": 12724
+    },
+    {
+      "epoch": 1.0843630166169578,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.092466601583618e-06,
+      "loss": 0.7405,
+      "step": 12725
+    },
+    {
+      "epoch": 1.0844482317852577,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.092299907230006e-06,
+      "loss": 0.5781,
+      "step": 12726
+    },
+    {
+      "epoch": 1.0845334469535577,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.0921332157261935e-06,
+      "loss": 0.54,
+      "step": 12727
+    },
+    {
+      "epoch": 1.0846186621218576,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0919665270751923e-06,
+      "loss": 0.2435,
+      "step": 12728
+    },
+    {
+      "epoch": 1.0847038772901576,
+      "grad_norm": 23.625,
+      "learning_rate": 1.0917998412800151e-06,
+      "loss": 0.7783,
+      "step": 12729
+    },
+    {
+      "epoch": 1.0847890924584576,
+      "grad_norm": 17.625,
+      "learning_rate": 1.0916331583436735e-06,
+      "loss": 0.9214,
+      "step": 12730
+    },
+    {
+      "epoch": 1.0848743076267575,
+      "grad_norm": 21.125,
+      "learning_rate": 1.0914664782691805e-06,
+      "loss": 0.5946,
+      "step": 12731
+    },
+    {
+      "epoch": 1.0849595227950575,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0912998010595477e-06,
+      "loss": 0.6533,
+      "step": 12732
+    },
+    {
+      "epoch": 1.0850447379633574,
+      "grad_norm": 15.25,
+      "learning_rate": 1.0911331267177866e-06,
+      "loss": 0.6576,
+      "step": 12733
+    },
+    {
+      "epoch": 1.0851299531316574,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.09096645524691e-06,
+      "loss": 0.3513,
+      "step": 12734
+    },
+    {
+      "epoch": 1.0852151682999573,
+      "grad_norm": 17.25,
+      "learning_rate": 1.0907997866499293e-06,
+      "loss": 0.7094,
+      "step": 12735
+    },
+    {
+      "epoch": 1.0853003834682573,
+      "grad_norm": 12.25,
+      "learning_rate": 1.0906331209298565e-06,
+      "loss": 0.4623,
+      "step": 12736
+    },
+    {
+      "epoch": 1.0853855986365573,
+      "grad_norm": 14.75,
+      "learning_rate": 1.0904664580897034e-06,
+      "loss": 0.6455,
+      "step": 12737
+    },
+    {
+      "epoch": 1.0854708138048572,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.090299798132482e-06,
+      "loss": 0.5296,
+      "step": 12738
+    },
+    {
+      "epoch": 1.0855560289731572,
+      "grad_norm": 25.25,
+      "learning_rate": 1.0901331410612036e-06,
+      "loss": 0.8654,
+      "step": 12739
+    },
+    {
+      "epoch": 1.0856412441414571,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.089966486878881e-06,
+      "loss": 0.5028,
+      "step": 12740
+    },
+    {
+      "epoch": 1.085726459309757,
+      "grad_norm": 16.0,
+      "learning_rate": 1.089799835588524e-06,
+      "loss": 0.3618,
+      "step": 12741
+    },
+    {
+      "epoch": 1.085811674478057,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0896331871931454e-06,
+      "loss": 0.4043,
+      "step": 12742
+    },
+    {
+      "epoch": 1.085896889646357,
+      "grad_norm": 14.75,
+      "learning_rate": 1.089466541695756e-06,
+      "loss": 0.679,
+      "step": 12743
+    },
+    {
+      "epoch": 1.085982104814657,
+      "grad_norm": 21.0,
+      "learning_rate": 1.089299899099368e-06,
+      "loss": 0.7628,
+      "step": 12744
+    },
+    {
+      "epoch": 1.086067319982957,
+      "grad_norm": 17.625,
+      "learning_rate": 1.089133259406992e-06,
+      "loss": 0.9788,
+      "step": 12745
+    },
+    {
+      "epoch": 1.0861525351512569,
+      "grad_norm": 26.625,
+      "learning_rate": 1.0889666226216404e-06,
+      "loss": 0.7057,
+      "step": 12746
+    },
+    {
+      "epoch": 1.0862377503195568,
+      "grad_norm": 33.75,
+      "learning_rate": 1.0887999887463233e-06,
+      "loss": 0.8494,
+      "step": 12747
+    },
+    {
+      "epoch": 1.0863229654878568,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.088633357784053e-06,
+      "loss": 0.6497,
+      "step": 12748
+    },
+    {
+      "epoch": 1.0864081806561567,
+      "grad_norm": 18.875,
+      "learning_rate": 1.08846672973784e-06,
+      "loss": 0.8629,
+      "step": 12749
+    },
+    {
+      "epoch": 1.0864933958244567,
+      "grad_norm": 19.125,
+      "learning_rate": 1.0883001046106953e-06,
+      "loss": 0.6444,
+      "step": 12750
+    },
+    {
+      "epoch": 1.0865786109927567,
+      "grad_norm": 22.625,
+      "learning_rate": 1.0881334824056307e-06,
+      "loss": 0.9823,
+      "step": 12751
+    },
+    {
+      "epoch": 1.0866638261610566,
+      "grad_norm": 15.0,
+      "learning_rate": 1.0879668631256565e-06,
+      "loss": 0.6912,
+      "step": 12752
+    },
+    {
+      "epoch": 1.0867490413293566,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0878002467737843e-06,
+      "loss": 0.5737,
+      "step": 12753
+    },
+    {
+      "epoch": 1.0868342564976565,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0876336333530244e-06,
+      "loss": 0.5326,
+      "step": 12754
+    },
+    {
+      "epoch": 1.0869194716659565,
+      "grad_norm": 15.5,
+      "learning_rate": 1.0874670228663883e-06,
+      "loss": 0.7217,
+      "step": 12755
+    },
+    {
+      "epoch": 1.0870046868342564,
+      "grad_norm": 16.25,
+      "learning_rate": 1.087300415316887e-06,
+      "loss": 0.6236,
+      "step": 12756
+    },
+    {
+      "epoch": 1.0870899020025564,
+      "grad_norm": 15.125,
+      "learning_rate": 1.0871338107075304e-06,
+      "loss": 0.6636,
+      "step": 12757
+    },
+    {
+      "epoch": 1.0871751171708564,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0869672090413294e-06,
+      "loss": 0.6618,
+      "step": 12758
+    },
+    {
+      "epoch": 1.0872603323391563,
+      "grad_norm": 13.875,
+      "learning_rate": 1.086800610321295e-06,
+      "loss": 0.6011,
+      "step": 12759
+    },
+    {
+      "epoch": 1.0873455475074563,
+      "grad_norm": 19.75,
+      "learning_rate": 1.0866340145504376e-06,
+      "loss": 0.7407,
+      "step": 12760
+    },
+    {
+      "epoch": 1.0874307626757562,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.0864674217317678e-06,
+      "loss": 0.7223,
+      "step": 12761
+    },
+    {
+      "epoch": 1.0875159778440562,
+      "grad_norm": 16.125,
+      "learning_rate": 1.086300831868296e-06,
+      "loss": 0.8548,
+      "step": 12762
+    },
+    {
+      "epoch": 1.0876011930123561,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.0861342449630333e-06,
+      "loss": 0.4573,
+      "step": 12763
+    },
+    {
+      "epoch": 1.087686408180656,
+      "grad_norm": 14.875,
+      "learning_rate": 1.0859676610189896e-06,
+      "loss": 0.5312,
+      "step": 12764
+    },
+    {
+      "epoch": 1.087771623348956,
+      "grad_norm": 14.875,
+      "learning_rate": 1.0858010800391746e-06,
+      "loss": 0.6443,
+      "step": 12765
+    },
+    {
+      "epoch": 1.087856838517256,
+      "grad_norm": 20.5,
+      "learning_rate": 1.0856345020265996e-06,
+      "loss": 0.9176,
+      "step": 12766
+    },
+    {
+      "epoch": 1.087942053685556,
+      "grad_norm": 14.75,
+      "learning_rate": 1.0854679269842745e-06,
+      "loss": 0.5941,
+      "step": 12767
+    },
+    {
+      "epoch": 1.088027268853856,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0853013549152092e-06,
+      "loss": 0.5012,
+      "step": 12768
+    },
+    {
+      "epoch": 1.0881124840221559,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.0851347858224138e-06,
+      "loss": 0.474,
+      "step": 12769
+    },
+    {
+      "epoch": 1.0881976991904558,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.084968219708899e-06,
+      "loss": 0.1327,
+      "step": 12770
+    },
+    {
+      "epoch": 1.0882829143587558,
+      "grad_norm": 13.625,
+      "learning_rate": 1.0848016565776742e-06,
+      "loss": 0.4218,
+      "step": 12771
+    },
+    {
+      "epoch": 1.0883681295270557,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0846350964317503e-06,
+      "loss": 0.6918,
+      "step": 12772
+    },
+    {
+      "epoch": 1.0884533446953557,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0844685392741357e-06,
+      "loss": 0.6078,
+      "step": 12773
+    },
+    {
+      "epoch": 1.0885385598636557,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0843019851078413e-06,
+      "loss": 0.9734,
+      "step": 12774
+    },
+    {
+      "epoch": 1.0886237750319556,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0841354339358763e-06,
+      "loss": 0.5221,
+      "step": 12775
+    },
+    {
+      "epoch": 1.0887089902002556,
+      "grad_norm": 13.25,
+      "learning_rate": 1.0839688857612512e-06,
+      "loss": 0.5736,
+      "step": 12776
+    },
+    {
+      "epoch": 1.0887942053685555,
+      "grad_norm": 34.75,
+      "learning_rate": 1.0838023405869751e-06,
+      "loss": 0.7536,
+      "step": 12777
+    },
+    {
+      "epoch": 1.0888794205368555,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.083635798416058e-06,
+      "loss": 0.5081,
+      "step": 12778
+    },
+    {
+      "epoch": 1.0889646357051554,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0834692592515091e-06,
+      "loss": 0.8246,
+      "step": 12779
+    },
+    {
+      "epoch": 1.0890498508734554,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0833027230963387e-06,
+      "loss": 0.7068,
+      "step": 12780
+    },
+    {
+      "epoch": 1.0891350660417554,
+      "grad_norm": 15.375,
+      "learning_rate": 1.0831361899535553e-06,
+      "loss": 0.4467,
+      "step": 12781
+    },
+    {
+      "epoch": 1.0892202812100553,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.0829696598261686e-06,
+      "loss": 0.4083,
+      "step": 12782
+    },
+    {
+      "epoch": 1.0893054963783553,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.0828031327171883e-06,
+      "loss": 0.416,
+      "step": 12783
+    },
+    {
+      "epoch": 1.0893907115466552,
+      "grad_norm": 21.875,
+      "learning_rate": 1.0826366086296233e-06,
+      "loss": 0.7159,
+      "step": 12784
+    },
+    {
+      "epoch": 1.0894759267149552,
+      "grad_norm": 15.0,
+      "learning_rate": 1.0824700875664834e-06,
+      "loss": 0.6925,
+      "step": 12785
+    },
+    {
+      "epoch": 1.0895611418832551,
+      "grad_norm": 18.625,
+      "learning_rate": 1.082303569530777e-06,
+      "loss": 0.5471,
+      "step": 12786
+    },
+    {
+      "epoch": 1.089646357051555,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.0821370545255142e-06,
+      "loss": 0.4998,
+      "step": 12787
+    },
+    {
+      "epoch": 1.089731572219855,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0819705425537038e-06,
+      "loss": 0.8055,
+      "step": 12788
+    },
+    {
+      "epoch": 1.089816787388155,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.0818040336183545e-06,
+      "loss": 0.5433,
+      "step": 12789
+    },
+    {
+      "epoch": 1.089902002556455,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.081637527722475e-06,
+      "loss": 0.6502,
+      "step": 12790
+    },
+    {
+      "epoch": 1.089987217724755,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0814710248690751e-06,
+      "loss": 0.555,
+      "step": 12791
+    },
+    {
+      "epoch": 1.0900724328930549,
+      "grad_norm": 17.0,
+      "learning_rate": 1.081304525061163e-06,
+      "loss": 0.6878,
+      "step": 12792
+    },
+    {
+      "epoch": 1.0901576480613548,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.0811380283017479e-06,
+      "loss": 0.4999,
+      "step": 12793
+    },
+    {
+      "epoch": 1.0902428632296548,
+      "grad_norm": 24.75,
+      "learning_rate": 1.0809715345938383e-06,
+      "loss": 0.8888,
+      "step": 12794
+    },
+    {
+      "epoch": 1.0903280783979548,
+      "grad_norm": 31.0,
+      "learning_rate": 1.0808050439404432e-06,
+      "loss": 0.6933,
+      "step": 12795
+    },
+    {
+      "epoch": 1.0904132935662547,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0806385563445715e-06,
+      "loss": 0.8258,
+      "step": 12796
+    },
+    {
+      "epoch": 1.0904985087345547,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0804720718092306e-06,
+      "loss": 0.8295,
+      "step": 12797
+    },
+    {
+      "epoch": 1.0905837239028546,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0803055903374304e-06,
+      "loss": 0.5162,
+      "step": 12798
+    },
+    {
+      "epoch": 1.0906689390711546,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0801391119321785e-06,
+      "loss": 0.6569,
+      "step": 12799
+    },
+    {
+      "epoch": 1.0907541542394545,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0799726365964839e-06,
+      "loss": 0.7901,
+      "step": 12800
+    },
+    {
+      "epoch": 1.0908393694077545,
+      "grad_norm": 14.25,
+      "learning_rate": 1.0798061643333544e-06,
+      "loss": 0.3801,
+      "step": 12801
+    },
+    {
+      "epoch": 1.0909245845760545,
+      "grad_norm": 14.875,
+      "learning_rate": 1.0796396951457988e-06,
+      "loss": 0.498,
+      "step": 12802
+    },
+    {
+      "epoch": 1.0910097997443544,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0794732290368252e-06,
+      "loss": 0.7334,
+      "step": 12803
+    },
+    {
+      "epoch": 1.0910950149126544,
+      "grad_norm": 18.75,
+      "learning_rate": 1.079306766009442e-06,
+      "loss": 0.7772,
+      "step": 12804
+    },
+    {
+      "epoch": 1.0911802300809543,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0791403060666573e-06,
+      "loss": 0.7102,
+      "step": 12805
+    },
+    {
+      "epoch": 1.0912654452492543,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.078973849211479e-06,
+      "loss": 0.5594,
+      "step": 12806
+    },
+    {
+      "epoch": 1.0913506604175542,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.078807395446915e-06,
+      "loss": 0.4549,
+      "step": 12807
+    },
+    {
+      "epoch": 1.0914358755858542,
+      "grad_norm": 15.25,
+      "learning_rate": 1.0786409447759737e-06,
+      "loss": 0.4186,
+      "step": 12808
+    },
+    {
+      "epoch": 1.0915210907541542,
+      "grad_norm": 17.625,
+      "learning_rate": 1.0784744972016625e-06,
+      "loss": 1.0159,
+      "step": 12809
+    },
+    {
+      "epoch": 1.0916063059224541,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.0783080527269898e-06,
+      "loss": 0.3958,
+      "step": 12810
+    },
+    {
+      "epoch": 1.091691521090754,
+      "grad_norm": 20.875,
+      "learning_rate": 1.0781416113549633e-06,
+      "loss": 0.7264,
+      "step": 12811
+    },
+    {
+      "epoch": 1.091776736259054,
+      "grad_norm": 22.5,
+      "learning_rate": 1.0779751730885907e-06,
+      "loss": 0.8182,
+      "step": 12812
+    },
+    {
+      "epoch": 1.091861951427354,
+      "grad_norm": 20.875,
+      "learning_rate": 1.07780873793088e-06,
+      "loss": 0.8166,
+      "step": 12813
+    },
+    {
+      "epoch": 1.091947166595654,
+      "grad_norm": 12.25,
+      "learning_rate": 1.077642305884838e-06,
+      "loss": 0.4081,
+      "step": 12814
+    },
+    {
+      "epoch": 1.092032381763954,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0774758769534734e-06,
+      "loss": 0.7799,
+      "step": 12815
+    },
+    {
+      "epoch": 1.0921175969322539,
+      "grad_norm": 13.5,
+      "learning_rate": 1.0773094511397928e-06,
+      "loss": 0.5278,
+      "step": 12816
+    },
+    {
+      "epoch": 1.0922028121005538,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.0771430284468042e-06,
+      "loss": 0.5195,
+      "step": 12817
+    },
+    {
+      "epoch": 1.0922880272688538,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0769766088775146e-06,
+      "loss": 0.6012,
+      "step": 12818
+    },
+    {
+      "epoch": 1.0923732424371537,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.076810192434932e-06,
+      "loss": 0.6268,
+      "step": 12819
+    },
+    {
+      "epoch": 1.0924584576054537,
+      "grad_norm": 17.125,
+      "learning_rate": 1.0766437791220635e-06,
+      "loss": 0.7731,
+      "step": 12820
+    },
+    {
+      "epoch": 1.0925436727737536,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.0764773689419164e-06,
+      "loss": 0.4953,
+      "step": 12821
+    },
+    {
+      "epoch": 1.0926288879420536,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0763109618974973e-06,
+      "loss": 0.6061,
+      "step": 12822
+    },
+    {
+      "epoch": 1.0927141031103536,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.0761445579918139e-06,
+      "loss": 0.7843,
+      "step": 12823
+    },
+    {
+      "epoch": 1.0927993182786535,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.075978157227873e-06,
+      "loss": 0.4908,
+      "step": 12824
+    },
+    {
+      "epoch": 1.0928845334469535,
+      "grad_norm": 15.125,
+      "learning_rate": 1.0758117596086818e-06,
+      "loss": 0.5381,
+      "step": 12825
+    },
+    {
+      "epoch": 1.0929697486152534,
+      "grad_norm": 18.5,
+      "learning_rate": 1.0756453651372473e-06,
+      "loss": 0.8341,
+      "step": 12826
+    },
+    {
+      "epoch": 1.0930549637835534,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.0754789738165766e-06,
+      "loss": 0.4798,
+      "step": 12827
+    },
+    {
+      "epoch": 1.0931401789518533,
+      "grad_norm": 11.75,
+      "learning_rate": 1.075312585649676e-06,
+      "loss": 0.4415,
+      "step": 12828
+    },
+    {
+      "epoch": 1.0932253941201533,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.075146200639553e-06,
+      "loss": 0.3664,
+      "step": 12829
+    },
+    {
+      "epoch": 1.0933106092884533,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.0749798187892142e-06,
+      "loss": 0.4136,
+      "step": 12830
+    },
+    {
+      "epoch": 1.0933958244567532,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0748134401016655e-06,
+      "loss": 0.5164,
+      "step": 12831
+    },
+    {
+      "epoch": 1.0934810396250532,
+      "grad_norm": 24.5,
+      "learning_rate": 1.0746470645799142e-06,
+      "loss": 0.6794,
+      "step": 12832
+    },
+    {
+      "epoch": 1.0935662547933531,
+      "grad_norm": 16.625,
+      "learning_rate": 1.0744806922269666e-06,
+      "loss": 0.6106,
+      "step": 12833
+    },
+    {
+      "epoch": 1.093651469961653,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0743143230458298e-06,
+      "loss": 0.6231,
+      "step": 12834
+    },
+    {
+      "epoch": 1.093736685129953,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0741479570395099e-06,
+      "loss": 0.6513,
+      "step": 12835
+    },
+    {
+      "epoch": 1.093821900298253,
+      "grad_norm": 9.75,
+      "learning_rate": 1.0739815942110128e-06,
+      "loss": 0.5876,
+      "step": 12836
+    },
+    {
+      "epoch": 1.093907115466553,
+      "grad_norm": 16.25,
+      "learning_rate": 1.073815234563346e-06,
+      "loss": 0.5734,
+      "step": 12837
+    },
+    {
+      "epoch": 1.093992330634853,
+      "grad_norm": 17.875,
+      "learning_rate": 1.0736488780995148e-06,
+      "loss": 0.5672,
+      "step": 12838
+    },
+    {
+      "epoch": 1.0940775458031529,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0734825248225252e-06,
+      "loss": 0.5338,
+      "step": 12839
+    },
+    {
+      "epoch": 1.0941627609714528,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0733161747353842e-06,
+      "loss": 0.3231,
+      "step": 12840
+    },
+    {
+      "epoch": 1.0942479761397528,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0731498278410975e-06,
+      "loss": 0.8591,
+      "step": 12841
+    },
+    {
+      "epoch": 1.0943331913080527,
+      "grad_norm": 19.875,
+      "learning_rate": 1.0729834841426714e-06,
+      "loss": 0.7308,
+      "step": 12842
+    },
+    {
+      "epoch": 1.0944184064763527,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0728171436431114e-06,
+      "loss": 0.8915,
+      "step": 12843
+    },
+    {
+      "epoch": 1.0945036216446526,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0726508063454242e-06,
+      "loss": 0.556,
+      "step": 12844
+    },
+    {
+      "epoch": 1.0945888368129526,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.0724844722526156e-06,
+      "loss": 0.5479,
+      "step": 12845
+    },
+    {
+      "epoch": 1.0946740519812526,
+      "grad_norm": 10.125,
+      "learning_rate": 1.0723181413676904e-06,
+      "loss": 0.2236,
+      "step": 12846
+    },
+    {
+      "epoch": 1.0947592671495525,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0721518136936555e-06,
+      "loss": 0.3286,
+      "step": 12847
+    },
+    {
+      "epoch": 1.0948444823178525,
+      "grad_norm": 14.375,
+      "learning_rate": 1.0719854892335158e-06,
+      "loss": 0.6719,
+      "step": 12848
+    },
+    {
+      "epoch": 1.0949296974861524,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.0718191679902777e-06,
+      "loss": 0.6264,
+      "step": 12849
+    },
+    {
+      "epoch": 1.0950149126544524,
+      "grad_norm": 12.5,
+      "learning_rate": 1.0716528499669461e-06,
+      "loss": 0.297,
+      "step": 12850
+    },
+    {
+      "epoch": 1.0951001278227523,
+      "grad_norm": 13.375,
+      "learning_rate": 1.0714865351665267e-06,
+      "loss": 0.6676,
+      "step": 12851
+    },
+    {
+      "epoch": 1.0951853429910523,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0713202235920254e-06,
+      "loss": 0.7653,
+      "step": 12852
+    },
+    {
+      "epoch": 1.0952705581593523,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0711539152464476e-06,
+      "loss": 0.7831,
+      "step": 12853
+    },
+    {
+      "epoch": 1.0953557733276522,
+      "grad_norm": 21.875,
+      "learning_rate": 1.070987610132798e-06,
+      "loss": 0.5187,
+      "step": 12854
+    },
+    {
+      "epoch": 1.0954409884959522,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0708213082540826e-06,
+      "loss": 0.7532,
+      "step": 12855
+    },
+    {
+      "epoch": 1.0955262036642521,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0706550096133058e-06,
+      "loss": 0.7686,
+      "step": 12856
+    },
+    {
+      "epoch": 1.095611418832552,
+      "grad_norm": 12.25,
+      "learning_rate": 1.0704887142134738e-06,
+      "loss": 0.4692,
+      "step": 12857
+    },
+    {
+      "epoch": 1.095696634000852,
+      "grad_norm": 9.875,
+      "learning_rate": 1.0703224220575908e-06,
+      "loss": 0.1641,
+      "step": 12858
+    },
+    {
+      "epoch": 1.095781849169152,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0701561331486627e-06,
+      "loss": 0.495,
+      "step": 12859
+    },
+    {
+      "epoch": 1.095867064337452,
+      "grad_norm": 16.125,
+      "learning_rate": 1.069989847489694e-06,
+      "loss": 0.4269,
+      "step": 12860
+    },
+    {
+      "epoch": 1.095952279505752,
+      "grad_norm": 19.625,
+      "learning_rate": 1.06982356508369e-06,
+      "loss": 0.8992,
+      "step": 12861
+    },
+    {
+      "epoch": 1.0960374946740519,
+      "grad_norm": 18.125,
+      "learning_rate": 1.0696572859336553e-06,
+      "loss": 0.5528,
+      "step": 12862
+    },
+    {
+      "epoch": 1.0961227098423518,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.0694910100425944e-06,
+      "loss": 0.4664,
+      "step": 12863
+    },
+    {
+      "epoch": 1.0962079250106518,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0693247374135127e-06,
+      "loss": 0.5167,
+      "step": 12864
+    },
+    {
+      "epoch": 1.0962931401789517,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.0691584680494148e-06,
+      "loss": 0.6103,
+      "step": 12865
+    },
+    {
+      "epoch": 1.0963783553472517,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.068992201953305e-06,
+      "loss": 0.493,
+      "step": 12866
+    },
+    {
+      "epoch": 1.0964635705155517,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0688259391281883e-06,
+      "loss": 0.749,
+      "step": 12867
+    },
+    {
+      "epoch": 1.0965487856838516,
+      "grad_norm": 11.25,
+      "learning_rate": 1.068659679577069e-06,
+      "loss": 0.2643,
+      "step": 12868
+    },
+    {
+      "epoch": 1.0966340008521516,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0684934233029517e-06,
+      "loss": 0.6015,
+      "step": 12869
+    },
+    {
+      "epoch": 1.0967192160204515,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.068327170308841e-06,
+      "loss": 0.7019,
+      "step": 12870
+    },
+    {
+      "epoch": 1.0968044311887515,
+      "grad_norm": 18.625,
+      "learning_rate": 1.0681609205977407e-06,
+      "loss": 0.9605,
+      "step": 12871
+    },
+    {
+      "epoch": 1.0968896463570514,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0679946741726553e-06,
+      "loss": 0.8498,
+      "step": 12872
+    },
+    {
+      "epoch": 1.0969748615253514,
+      "grad_norm": 15.25,
+      "learning_rate": 1.0678284310365891e-06,
+      "loss": 0.4518,
+      "step": 12873
+    },
+    {
+      "epoch": 1.0970600766936514,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0676621911925466e-06,
+      "loss": 0.6102,
+      "step": 12874
+    },
+    {
+      "epoch": 1.0971452918619513,
+      "grad_norm": 19.125,
+      "learning_rate": 1.0674959546435315e-06,
+      "loss": 0.9691,
+      "step": 12875
+    },
+    {
+      "epoch": 1.0972305070302513,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0673297213925482e-06,
+      "loss": 0.55,
+      "step": 12876
+    },
+    {
+      "epoch": 1.0973157221985512,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.0671634914426003e-06,
+      "loss": 0.2295,
+      "step": 12877
+    },
+    {
+      "epoch": 1.0974009373668514,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0669972647966924e-06,
+      "loss": 0.596,
+      "step": 12878
+    },
+    {
+      "epoch": 1.0974861525351514,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0668310414578278e-06,
+      "loss": 0.4683,
+      "step": 12879
+    },
+    {
+      "epoch": 1.0975713677034513,
+      "grad_norm": 11.125,
+      "learning_rate": 1.0666648214290104e-06,
+      "loss": 0.4774,
+      "step": 12880
+    },
+    {
+      "epoch": 1.0976565828717513,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.066498604713244e-06,
+      "loss": 0.4846,
+      "step": 12881
+    },
+    {
+      "epoch": 1.0977417980400512,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0663323913135324e-06,
+      "loss": 0.6343,
+      "step": 12882
+    },
+    {
+      "epoch": 1.0978270132083512,
+      "grad_norm": 24.875,
+      "learning_rate": 1.0661661812328787e-06,
+      "loss": 1.0339,
+      "step": 12883
+    },
+    {
+      "epoch": 1.0979122283766511,
+      "grad_norm": 12.875,
+      "learning_rate": 1.0659999744742874e-06,
+      "loss": 0.5148,
+      "step": 12884
+    },
+    {
+      "epoch": 1.097997443544951,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.0658337710407616e-06,
+      "loss": 0.5663,
+      "step": 12885
+    },
+    {
+      "epoch": 1.098082658713251,
+      "grad_norm": 12.0,
+      "learning_rate": 1.065667570935305e-06,
+      "loss": 0.2614,
+      "step": 12886
+    },
+    {
+      "epoch": 1.098167873881551,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0655013741609205e-06,
+      "loss": 0.8164,
+      "step": 12887
+    },
+    {
+      "epoch": 1.098253089049851,
+      "grad_norm": 23.125,
+      "learning_rate": 1.0653351807206113e-06,
+      "loss": 0.5434,
+      "step": 12888
+    },
+    {
+      "epoch": 1.098338304218151,
+      "grad_norm": 19.75,
+      "learning_rate": 1.0651689906173817e-06,
+      "loss": 0.6547,
+      "step": 12889
+    },
+    {
+      "epoch": 1.0984235193864509,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0650028038542337e-06,
+      "loss": 0.859,
+      "step": 12890
+    },
+    {
+      "epoch": 1.0985087345547508,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0648366204341715e-06,
+      "loss": 0.6401,
+      "step": 12891
+    },
+    {
+      "epoch": 1.0985939497230508,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0646704403601975e-06,
+      "loss": 0.5541,
+      "step": 12892
+    },
+    {
+      "epoch": 1.0986791648913508,
+      "grad_norm": 13.625,
+      "learning_rate": 1.0645042636353152e-06,
+      "loss": 0.5803,
+      "step": 12893
+    },
+    {
+      "epoch": 1.0987643800596507,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0643380902625277e-06,
+      "loss": 0.8313,
+      "step": 12894
+    },
+    {
+      "epoch": 1.0988495952279507,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0641719202448373e-06,
+      "loss": 0.5836,
+      "step": 12895
+    },
+    {
+      "epoch": 1.0989348103962506,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.064005753585247e-06,
+      "loss": 0.4099,
+      "step": 12896
+    },
+    {
+      "epoch": 1.0990200255645506,
+      "grad_norm": 28.0,
+      "learning_rate": 1.06383959028676e-06,
+      "loss": 0.9255,
+      "step": 12897
+    },
+    {
+      "epoch": 1.0991052407328505,
+      "grad_norm": 37.5,
+      "learning_rate": 1.0636734303523785e-06,
+      "loss": 0.8101,
+      "step": 12898
+    },
+    {
+      "epoch": 1.0991904559011505,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0635072737851056e-06,
+      "loss": 0.5173,
+      "step": 12899
+    },
+    {
+      "epoch": 1.0992756710694505,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0633411205879437e-06,
+      "loss": 0.7854,
+      "step": 12900
+    },
+    {
+      "epoch": 1.0993608862377504,
+      "grad_norm": 24.125,
+      "learning_rate": 1.063174970763896e-06,
+      "loss": 0.9447,
+      "step": 12901
+    },
+    {
+      "epoch": 1.0994461014060504,
+      "grad_norm": 22.0,
+      "learning_rate": 1.063008824315964e-06,
+      "loss": 1.0113,
+      "step": 12902
+    },
+    {
+      "epoch": 1.0995313165743503,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0628426812471506e-06,
+      "loss": 0.6508,
+      "step": 12903
+    },
+    {
+      "epoch": 1.0996165317426503,
+      "grad_norm": 15.25,
+      "learning_rate": 1.0626765415604583e-06,
+      "loss": 0.5525,
+      "step": 12904
+    },
+    {
+      "epoch": 1.0997017469109502,
+      "grad_norm": 14.5,
+      "learning_rate": 1.062510405258889e-06,
+      "loss": 0.6913,
+      "step": 12905
+    },
+    {
+      "epoch": 1.0997869620792502,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0623442723454455e-06,
+      "loss": 0.6473,
+      "step": 12906
+    },
+    {
+      "epoch": 1.0998721772475502,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0621781428231293e-06,
+      "loss": 0.8004,
+      "step": 12907
+    },
+    {
+      "epoch": 1.09995739241585,
+      "grad_norm": 18.375,
+      "learning_rate": 1.0620120166949432e-06,
+      "loss": 0.8201,
+      "step": 12908
+    },
+    {
+      "epoch": 1.10004260758415,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0618458939638888e-06,
+      "loss": 0.4078,
+      "step": 12909
+    },
+    {
+      "epoch": 1.10012782275245,
+      "grad_norm": 18.625,
+      "learning_rate": 1.0616797746329691e-06,
+      "loss": 0.8423,
+      "step": 12910
+    },
+    {
+      "epoch": 1.10021303792075,
+      "grad_norm": 18.375,
+      "learning_rate": 1.0615136587051845e-06,
+      "loss": 0.705,
+      "step": 12911
+    },
+    {
+      "epoch": 1.10029825308905,
+      "grad_norm": 12.375,
+      "learning_rate": 1.0613475461835379e-06,
+      "loss": 0.2477,
+      "step": 12912
+    },
+    {
+      "epoch": 1.10038346825735,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0611814370710305e-06,
+      "loss": 0.7207,
+      "step": 12913
+    },
+    {
+      "epoch": 1.1004686834256499,
+      "grad_norm": 13.5,
+      "learning_rate": 1.0610153313706647e-06,
+      "loss": 0.4509,
+      "step": 12914
+    },
+    {
+      "epoch": 1.1005538985939498,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0608492290854417e-06,
+      "loss": 0.5664,
+      "step": 12915
+    },
+    {
+      "epoch": 1.1006391137622498,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0606831302183635e-06,
+      "loss": 0.4848,
+      "step": 12916
+    },
+    {
+      "epoch": 1.1007243289305497,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0605170347724312e-06,
+      "loss": 0.4664,
+      "step": 12917
+    },
+    {
+      "epoch": 1.1008095440988497,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0603509427506473e-06,
+      "loss": 0.4241,
+      "step": 12918
+    },
+    {
+      "epoch": 1.1008947592671496,
+      "grad_norm": 18.625,
+      "learning_rate": 1.0601848541560122e-06,
+      "loss": 0.8123,
+      "step": 12919
+    },
+    {
+      "epoch": 1.1009799744354496,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.0600187689915274e-06,
+      "loss": 0.4523,
+      "step": 12920
+    },
+    {
+      "epoch": 1.1010651896037496,
+      "grad_norm": 35.75,
+      "learning_rate": 1.0598526872601948e-06,
+      "loss": 1.0185,
+      "step": 12921
+    },
+    {
+      "epoch": 1.1011504047720495,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.059686608965015e-06,
+      "loss": 0.4586,
+      "step": 12922
+    },
+    {
+      "epoch": 1.1012356199403495,
+      "grad_norm": 14.875,
+      "learning_rate": 1.0595205341089898e-06,
+      "loss": 0.4927,
+      "step": 12923
+    },
+    {
+      "epoch": 1.1013208351086494,
+      "grad_norm": 17.25,
+      "learning_rate": 1.0593544626951199e-06,
+      "loss": 0.747,
+      "step": 12924
+    },
+    {
+      "epoch": 1.1014060502769494,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0591883947264068e-06,
+      "loss": 0.7902,
+      "step": 12925
+    },
+    {
+      "epoch": 1.1014912654452493,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0590223302058517e-06,
+      "loss": 0.5812,
+      "step": 12926
+    },
+    {
+      "epoch": 1.1015764806135493,
+      "grad_norm": 24.375,
+      "learning_rate": 1.0588562691364547e-06,
+      "loss": 0.8328,
+      "step": 12927
+    },
+    {
+      "epoch": 1.1016616957818492,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.058690211521217e-06,
+      "loss": 0.4655,
+      "step": 12928
+    },
+    {
+      "epoch": 1.1017469109501492,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0585241573631397e-06,
+      "loss": 0.6824,
+      "step": 12929
+    },
+    {
+      "epoch": 1.1018321261184492,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0583581066652232e-06,
+      "loss": 0.3801,
+      "step": 12930
+    },
+    {
+      "epoch": 1.1019173412867491,
+      "grad_norm": 10.875,
+      "learning_rate": 1.0581920594304687e-06,
+      "loss": 0.1891,
+      "step": 12931
+    },
+    {
+      "epoch": 1.102002556455049,
+      "grad_norm": 14.625,
+      "learning_rate": 1.058026015661876e-06,
+      "loss": 0.6136,
+      "step": 12932
+    },
+    {
+      "epoch": 1.102087771623349,
+      "grad_norm": 14.875,
+      "learning_rate": 1.057859975362447e-06,
+      "loss": 0.6085,
+      "step": 12933
+    },
+    {
+      "epoch": 1.102172986791649,
+      "grad_norm": 17.75,
+      "learning_rate": 1.057693938535181e-06,
+      "loss": 0.6933,
+      "step": 12934
+    },
+    {
+      "epoch": 1.102258201959949,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0575279051830793e-06,
+      "loss": 0.7987,
+      "step": 12935
+    },
+    {
+      "epoch": 1.102343417128249,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0573618753091417e-06,
+      "loss": 0.3716,
+      "step": 12936
+    },
+    {
+      "epoch": 1.1024286322965489,
+      "grad_norm": 16.75,
+      "learning_rate": 1.0571958489163685e-06,
+      "loss": 1.1064,
+      "step": 12937
+    },
+    {
+      "epoch": 1.1025138474648488,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.0570298260077605e-06,
+      "loss": 0.5074,
+      "step": 12938
+    },
+    {
+      "epoch": 1.1025990626331488,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.0568638065863175e-06,
+      "loss": 0.4701,
+      "step": 12939
+    },
+    {
+      "epoch": 1.1026842778014487,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.0566977906550396e-06,
+      "loss": 0.4696,
+      "step": 12940
+    },
+    {
+      "epoch": 1.1027694929697487,
+      "grad_norm": 14.125,
+      "learning_rate": 1.0565317782169269e-06,
+      "loss": 0.9267,
+      "step": 12941
+    },
+    {
+      "epoch": 1.1028547081380486,
+      "grad_norm": 13.875,
+      "learning_rate": 1.05636576927498e-06,
+      "loss": 0.5396,
+      "step": 12942
+    },
+    {
+      "epoch": 1.1029399233063486,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0561997638321983e-06,
+      "loss": 0.6771,
+      "step": 12943
+    },
+    {
+      "epoch": 1.1030251384746486,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.0560337618915815e-06,
+      "loss": 0.586,
+      "step": 12944
+    },
+    {
+      "epoch": 1.1031103536429485,
+      "grad_norm": 17.625,
+      "learning_rate": 1.0558677634561295e-06,
+      "loss": 0.5079,
+      "step": 12945
+    },
+    {
+      "epoch": 1.1031955688112485,
+      "grad_norm": 12.875,
+      "learning_rate": 1.0557017685288427e-06,
+      "loss": 0.2925,
+      "step": 12946
+    },
+    {
+      "epoch": 1.1032807839795484,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.0555357771127197e-06,
+      "loss": 0.7335,
+      "step": 12947
+    },
+    {
+      "epoch": 1.1033659991478484,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0553697892107614e-06,
+      "loss": 0.9047,
+      "step": 12948
+    },
+    {
+      "epoch": 1.1034512143161483,
+      "grad_norm": 16.0,
+      "learning_rate": 1.055203804825966e-06,
+      "loss": 0.4865,
+      "step": 12949
+    },
+    {
+      "epoch": 1.1035364294844483,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0550378239613343e-06,
+      "loss": 0.4186,
+      "step": 12950
+    },
+    {
+      "epoch": 1.1036216446527483,
+      "grad_norm": 13.5,
+      "learning_rate": 1.0548718466198655e-06,
+      "loss": 0.5239,
+      "step": 12951
+    },
+    {
+      "epoch": 1.1037068598210482,
+      "grad_norm": 18.5,
+      "learning_rate": 1.054705872804558e-06,
+      "loss": 0.6996,
+      "step": 12952
+    },
+    {
+      "epoch": 1.1037920749893482,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0545399025184118e-06,
+      "loss": 0.4461,
+      "step": 12953
+    },
+    {
+      "epoch": 1.1038772901576481,
+      "grad_norm": 13.875,
+      "learning_rate": 1.054373935764426e-06,
+      "loss": 0.4231,
+      "step": 12954
+    },
+    {
+      "epoch": 1.103962505325948,
+      "grad_norm": 14.75,
+      "learning_rate": 1.0542079725456005e-06,
+      "loss": 0.6483,
+      "step": 12955
+    },
+    {
+      "epoch": 1.104047720494248,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0540420128649332e-06,
+      "loss": 1.0375,
+      "step": 12956
+    },
+    {
+      "epoch": 1.104132935662548,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0538760567254242e-06,
+      "loss": 0.6338,
+      "step": 12957
+    },
+    {
+      "epoch": 1.104218150830848,
+      "grad_norm": 15.625,
+      "learning_rate": 1.053710104130072e-06,
+      "loss": 0.5914,
+      "step": 12958
+    },
+    {
+      "epoch": 1.104303365999148,
+      "grad_norm": 12.125,
+      "learning_rate": 1.0535441550818758e-06,
+      "loss": 0.4574,
+      "step": 12959
+    },
+    {
+      "epoch": 1.1043885811674479,
+      "grad_norm": 21.375,
+      "learning_rate": 1.053378209583834e-06,
+      "loss": 0.7857,
+      "step": 12960
+    },
+    {
+      "epoch": 1.1044737963357478,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0532122676389456e-06,
+      "loss": 0.905,
+      "step": 12961
+    },
+    {
+      "epoch": 1.1045590115040478,
+      "grad_norm": 21.0,
+      "learning_rate": 1.0530463292502095e-06,
+      "loss": 0.8267,
+      "step": 12962
+    },
+    {
+      "epoch": 1.1046442266723477,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.0528803944206242e-06,
+      "loss": 0.4209,
+      "step": 12963
+    },
+    {
+      "epoch": 1.1047294418406477,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0527144631531882e-06,
+      "loss": 1.0379,
+      "step": 12964
+    },
+    {
+      "epoch": 1.1048146570089477,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0525485354509007e-06,
+      "loss": 0.2531,
+      "step": 12965
+    },
+    {
+      "epoch": 1.1048998721772476,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.0523826113167592e-06,
+      "loss": 0.4102,
+      "step": 12966
+    },
+    {
+      "epoch": 1.1049850873455476,
+      "grad_norm": 12.875,
+      "learning_rate": 1.0522166907537632e-06,
+      "loss": 0.4423,
+      "step": 12967
+    },
+    {
+      "epoch": 1.1050703025138475,
+      "grad_norm": 13.375,
+      "learning_rate": 1.0520507737649105e-06,
+      "loss": 0.3806,
+      "step": 12968
+    },
+    {
+      "epoch": 1.1051555176821475,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.051884860353199e-06,
+      "loss": 0.4255,
+      "step": 12969
+    },
+    {
+      "epoch": 1.1052407328504474,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0517189505216272e-06,
+      "loss": 0.6645,
+      "step": 12970
+    },
+    {
+      "epoch": 1.1053259480187474,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0515530442731935e-06,
+      "loss": 0.4232,
+      "step": 12971
+    },
+    {
+      "epoch": 1.1054111631870474,
+      "grad_norm": 36.25,
+      "learning_rate": 1.0513871416108959e-06,
+      "loss": 0.8509,
+      "step": 12972
+    },
+    {
+      "epoch": 1.1054963783553473,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0512212425377325e-06,
+      "loss": 0.5004,
+      "step": 12973
+    },
+    {
+      "epoch": 1.1055815935236473,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0510553470567007e-06,
+      "loss": 0.8302,
+      "step": 12974
+    },
+    {
+      "epoch": 1.1056668086919472,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0508894551707997e-06,
+      "loss": 0.3675,
+      "step": 12975
+    },
+    {
+      "epoch": 1.1057520238602472,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0507235668830258e-06,
+      "loss": 0.5378,
+      "step": 12976
+    },
+    {
+      "epoch": 1.1058372390285471,
+      "grad_norm": 15.5,
+      "learning_rate": 1.0505576821963774e-06,
+      "loss": 0.7414,
+      "step": 12977
+    },
+    {
+      "epoch": 1.105922454196847,
+      "grad_norm": 9.125,
+      "learning_rate": 1.0503918011138524e-06,
+      "loss": 0.2718,
+      "step": 12978
+    },
+    {
+      "epoch": 1.106007669365147,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0502259236384482e-06,
+      "loss": 1.2556,
+      "step": 12979
+    },
+    {
+      "epoch": 1.106092884533447,
+      "grad_norm": 13.375,
+      "learning_rate": 1.0500600497731625e-06,
+      "loss": 0.5361,
+      "step": 12980
+    },
+    {
+      "epoch": 1.106178099701747,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0498941795209924e-06,
+      "loss": 0.5595,
+      "step": 12981
+    },
+    {
+      "epoch": 1.106263314870047,
+      "grad_norm": 18.125,
+      "learning_rate": 1.049728312884936e-06,
+      "loss": 0.5345,
+      "step": 12982
+    },
+    {
+      "epoch": 1.1063485300383469,
+      "grad_norm": 13.25,
+      "learning_rate": 1.049562449867991e-06,
+      "loss": 0.3807,
+      "step": 12983
+    },
+    {
+      "epoch": 1.1064337452066468,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0493965904731532e-06,
+      "loss": 0.4706,
+      "step": 12984
+    },
+    {
+      "epoch": 1.1065189603749468,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0492307347034212e-06,
+      "loss": 0.3667,
+      "step": 12985
+    },
+    {
+      "epoch": 1.1066041755432467,
+      "grad_norm": 14.5,
+      "learning_rate": 1.0490648825617913e-06,
+      "loss": 0.606,
+      "step": 12986
+    },
+    {
+      "epoch": 1.1066893907115467,
+      "grad_norm": 23.875,
+      "learning_rate": 1.0488990340512615e-06,
+      "loss": 1.1199,
+      "step": 12987
+    },
+    {
+      "epoch": 1.1067746058798467,
+      "grad_norm": 16.25,
+      "learning_rate": 1.048733189174828e-06,
+      "loss": 0.7381,
+      "step": 12988
+    },
+    {
+      "epoch": 1.1068598210481466,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0485673479354882e-06,
+      "loss": 0.4409,
+      "step": 12989
+    },
+    {
+      "epoch": 1.1069450362164466,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0484015103362392e-06,
+      "loss": 0.4177,
+      "step": 12990
+    },
+    {
+      "epoch": 1.1070302513847465,
+      "grad_norm": 16.375,
+      "learning_rate": 1.048235676380078e-06,
+      "loss": 0.683,
+      "step": 12991
+    },
+    {
+      "epoch": 1.1071154665530465,
+      "grad_norm": 20.25,
+      "learning_rate": 1.0480698460700003e-06,
+      "loss": 1.0075,
+      "step": 12992
+    },
+    {
+      "epoch": 1.1072006817213464,
+      "grad_norm": 18.625,
+      "learning_rate": 1.0479040194090039e-06,
+      "loss": 0.767,
+      "step": 12993
+    },
+    {
+      "epoch": 1.1072858968896464,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0477381964000848e-06,
+      "loss": 0.5861,
+      "step": 12994
+    },
+    {
+      "epoch": 1.1073711120579464,
+      "grad_norm": 15.25,
+      "learning_rate": 1.0475723770462402e-06,
+      "loss": 0.7565,
+      "step": 12995
+    },
+    {
+      "epoch": 1.1074563272262463,
+      "grad_norm": 15.75,
+      "learning_rate": 1.047406561350466e-06,
+      "loss": 0.7408,
+      "step": 12996
+    },
+    {
+      "epoch": 1.1075415423945463,
+      "grad_norm": 15.375,
+      "learning_rate": 1.047240749315759e-06,
+      "loss": 0.534,
+      "step": 12997
+    },
+    {
+      "epoch": 1.1076267575628462,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0470749409451156e-06,
+      "loss": 0.7307,
+      "step": 12998
+    },
+    {
+      "epoch": 1.1077119727311462,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0469091362415323e-06,
+      "loss": 0.4105,
+      "step": 12999
+    },
+    {
+      "epoch": 1.1077971878994461,
+      "grad_norm": 16.375,
+      "learning_rate": 1.046743335208005e-06,
+      "loss": 0.6593,
+      "step": 13000
+    },
+    {
+      "epoch": 1.107882403067746,
+      "grad_norm": 16.5,
+      "learning_rate": 1.04657753784753e-06,
+      "loss": 0.5729,
+      "step": 13001
+    },
+    {
+      "epoch": 1.107967618236046,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.0464117441631035e-06,
+      "loss": 0.5392,
+      "step": 13002
+    },
+    {
+      "epoch": 1.108052833404346,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0462459541577214e-06,
+      "loss": 0.8327,
+      "step": 13003
+    },
+    {
+      "epoch": 1.108138048572646,
+      "grad_norm": 29.625,
+      "learning_rate": 1.0460801678343797e-06,
+      "loss": 1.6851,
+      "step": 13004
+    },
+    {
+      "epoch": 1.108223263740946,
+      "grad_norm": 20.625,
+      "learning_rate": 1.0459143851960748e-06,
+      "loss": 0.6165,
+      "step": 13005
+    },
+    {
+      "epoch": 1.1083084789092459,
+      "grad_norm": 28.5,
+      "learning_rate": 1.0457486062458017e-06,
+      "loss": 0.9852,
+      "step": 13006
+    },
+    {
+      "epoch": 1.1083936940775458,
+      "grad_norm": 20.375,
+      "learning_rate": 1.045582830986557e-06,
+      "loss": 0.9211,
+      "step": 13007
+    },
+    {
+      "epoch": 1.1084789092458458,
+      "grad_norm": 17.125,
+      "learning_rate": 1.0454170594213363e-06,
+      "loss": 0.8565,
+      "step": 13008
+    },
+    {
+      "epoch": 1.1085641244141458,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0452512915531347e-06,
+      "loss": 0.3924,
+      "step": 13009
+    },
+    {
+      "epoch": 1.1086493395824457,
+      "grad_norm": 17.25,
+      "learning_rate": 1.0450855273849484e-06,
+      "loss": 0.6035,
+      "step": 13010
+    },
+    {
+      "epoch": 1.1087345547507457,
+      "grad_norm": 21.625,
+      "learning_rate": 1.0449197669197722e-06,
+      "loss": 0.7959,
+      "step": 13011
+    },
+    {
+      "epoch": 1.1088197699190456,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0447540101606025e-06,
+      "loss": 0.7922,
+      "step": 13012
+    },
+    {
+      "epoch": 1.1089049850873456,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.0445882571104338e-06,
+      "loss": 0.6423,
+      "step": 13013
+    },
+    {
+      "epoch": 1.1089902002556455,
+      "grad_norm": 25.25,
+      "learning_rate": 1.0444225077722623e-06,
+      "loss": 0.9106,
+      "step": 13014
+    },
+    {
+      "epoch": 1.1090754154239455,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0442567621490824e-06,
+      "loss": 0.6706,
+      "step": 13015
+    },
+    {
+      "epoch": 1.1091606305922455,
+      "grad_norm": 19.125,
+      "learning_rate": 1.0440910202438903e-06,
+      "loss": 0.5476,
+      "step": 13016
+    },
+    {
+      "epoch": 1.1092458457605454,
+      "grad_norm": 11.875,
+      "learning_rate": 1.0439252820596803e-06,
+      "loss": 0.3312,
+      "step": 13017
+    },
+    {
+      "epoch": 1.1093310609288454,
+      "grad_norm": 18.625,
+      "learning_rate": 1.0437595475994477e-06,
+      "loss": 0.4511,
+      "step": 13018
+    },
+    {
+      "epoch": 1.1094162760971453,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.0435938168661872e-06,
+      "loss": 0.3428,
+      "step": 13019
+    },
+    {
+      "epoch": 1.1095014912654453,
+      "grad_norm": 16.0,
+      "learning_rate": 1.043428089862894e-06,
+      "loss": 0.7999,
+      "step": 13020
+    },
+    {
+      "epoch": 1.1095867064337452,
+      "grad_norm": 20.125,
+      "learning_rate": 1.043262366592563e-06,
+      "loss": 0.3464,
+      "step": 13021
+    },
+    {
+      "epoch": 1.1096719216020452,
+      "grad_norm": 12.875,
+      "learning_rate": 1.0430966470581889e-06,
+      "loss": 0.3667,
+      "step": 13022
+    },
+    {
+      "epoch": 1.1097571367703452,
+      "grad_norm": 19.625,
+      "learning_rate": 1.0429309312627664e-06,
+      "loss": 0.6985,
+      "step": 13023
+    },
+    {
+      "epoch": 1.109842351938645,
+      "grad_norm": 26.0,
+      "learning_rate": 1.0427652192092908e-06,
+      "loss": 1.0088,
+      "step": 13024
+    },
+    {
+      "epoch": 1.109927567106945,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0425995109007555e-06,
+      "loss": 0.6296,
+      "step": 13025
+    },
+    {
+      "epoch": 1.110012782275245,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0424338063401555e-06,
+      "loss": 0.6085,
+      "step": 13026
+    },
+    {
+      "epoch": 1.110097997443545,
+      "grad_norm": 12.375,
+      "learning_rate": 1.0422681055304857e-06,
+      "loss": 0.4145,
+      "step": 13027
+    },
+    {
+      "epoch": 1.110183212611845,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0421024084747397e-06,
+      "loss": 0.5407,
+      "step": 13028
+    },
+    {
+      "epoch": 1.110268427780145,
+      "grad_norm": 19.75,
+      "learning_rate": 1.0419367151759127e-06,
+      "loss": 0.8379,
+      "step": 13029
+    },
+    {
+      "epoch": 1.1103536429484449,
+      "grad_norm": 21.125,
+      "learning_rate": 1.041771025636998e-06,
+      "loss": 0.8537,
+      "step": 13030
+    },
+    {
+      "epoch": 1.1104388581167448,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0416053398609907e-06,
+      "loss": 0.4359,
+      "step": 13031
+    },
+    {
+      "epoch": 1.1105240732850448,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0414396578508845e-06,
+      "loss": 0.5,
+      "step": 13032
+    },
+    {
+      "epoch": 1.1106092884533447,
+      "grad_norm": 20.375,
+      "learning_rate": 1.0412739796096732e-06,
+      "loss": 0.5673,
+      "step": 13033
+    },
+    {
+      "epoch": 1.1106945036216447,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0411083051403508e-06,
+      "loss": 0.7234,
+      "step": 13034
+    },
+    {
+      "epoch": 1.1107797187899446,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0409426344459117e-06,
+      "loss": 0.5188,
+      "step": 13035
+    },
+    {
+      "epoch": 1.1108649339582446,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.040776967529349e-06,
+      "loss": 0.5039,
+      "step": 13036
+    },
+    {
+      "epoch": 1.1109501491265446,
+      "grad_norm": 19.125,
+      "learning_rate": 1.0406113043936573e-06,
+      "loss": 1.0194,
+      "step": 13037
+    },
+    {
+      "epoch": 1.1110353642948445,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.0404456450418296e-06,
+      "loss": 0.4799,
+      "step": 13038
+    },
+    {
+      "epoch": 1.1111205794631445,
+      "grad_norm": 11.875,
+      "learning_rate": 1.04027998947686e-06,
+      "loss": 0.3237,
+      "step": 13039
+    },
+    {
+      "epoch": 1.1112057946314444,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0401143377017423e-06,
+      "loss": 0.1946,
+      "step": 13040
+    },
+    {
+      "epoch": 1.1112910097997444,
+      "grad_norm": 11.375,
+      "learning_rate": 1.0399486897194688e-06,
+      "loss": 0.4274,
+      "step": 13041
+    },
+    {
+      "epoch": 1.1113762249680443,
+      "grad_norm": 13.625,
+      "learning_rate": 1.039783045533034e-06,
+      "loss": 0.442,
+      "step": 13042
+    },
+    {
+      "epoch": 1.1114614401363443,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.039617405145431e-06,
+      "loss": 0.3231,
+      "step": 13043
+    },
+    {
+      "epoch": 1.1115466553046442,
+      "grad_norm": 18.0,
+      "learning_rate": 1.039451768559653e-06,
+      "loss": 0.786,
+      "step": 13044
+    },
+    {
+      "epoch": 1.1116318704729442,
+      "grad_norm": 28.5,
+      "learning_rate": 1.0392861357786932e-06,
+      "loss": 0.4373,
+      "step": 13045
+    },
+    {
+      "epoch": 1.1117170856412442,
+      "grad_norm": 13.25,
+      "learning_rate": 1.0391205068055448e-06,
+      "loss": 0.6301,
+      "step": 13046
+    },
+    {
+      "epoch": 1.1118023008095441,
+      "grad_norm": 18.5,
+      "learning_rate": 1.0389548816432009e-06,
+      "loss": 0.513,
+      "step": 13047
+    },
+    {
+      "epoch": 1.111887515977844,
+      "grad_norm": 18.25,
+      "learning_rate": 1.0387892602946551e-06,
+      "loss": 0.9109,
+      "step": 13048
+    },
+    {
+      "epoch": 1.111972731146144,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0386236427628994e-06,
+      "loss": 1.0801,
+      "step": 13049
+    },
+    {
+      "epoch": 1.112057946314444,
+      "grad_norm": 14.25,
+      "learning_rate": 1.0384580290509268e-06,
+      "loss": 0.461,
+      "step": 13050
+    },
+    {
+      "epoch": 1.112143161482744,
+      "grad_norm": 23.25,
+      "learning_rate": 1.0382924191617301e-06,
+      "loss": 0.5671,
+      "step": 13051
+    },
+    {
+      "epoch": 1.112228376651044,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.0381268130983029e-06,
+      "loss": 0.3289,
+      "step": 13052
+    },
+    {
+      "epoch": 1.1123135918193439,
+      "grad_norm": 16.75,
+      "learning_rate": 1.0379612108636367e-06,
+      "loss": 0.6262,
+      "step": 13053
+    },
+    {
+      "epoch": 1.1123988069876438,
+      "grad_norm": 17.375,
+      "learning_rate": 1.037795612460725e-06,
+      "loss": 0.7017,
+      "step": 13054
+    },
+    {
+      "epoch": 1.1124840221559438,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0376300178925597e-06,
+      "loss": 0.4279,
+      "step": 13055
+    },
+    {
+      "epoch": 1.1125692373242437,
+      "grad_norm": 18.5,
+      "learning_rate": 1.037464427162134e-06,
+      "loss": 0.5531,
+      "step": 13056
+    },
+    {
+      "epoch": 1.1126544524925437,
+      "grad_norm": 13.0,
+      "learning_rate": 1.0372988402724395e-06,
+      "loss": 0.5188,
+      "step": 13057
+    },
+    {
+      "epoch": 1.1127396676608436,
+      "grad_norm": 11.375,
+      "learning_rate": 1.0371332572264687e-06,
+      "loss": 0.2751,
+      "step": 13058
+    },
+    {
+      "epoch": 1.1128248828291436,
+      "grad_norm": 12.375,
+      "learning_rate": 1.0369676780272142e-06,
+      "loss": 0.4019,
+      "step": 13059
+    },
+    {
+      "epoch": 1.1129100979974436,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0368021026776678e-06,
+      "loss": 0.5013,
+      "step": 13060
+    },
+    {
+      "epoch": 1.1129953131657435,
+      "grad_norm": 21.0,
+      "learning_rate": 1.036636531180822e-06,
+      "loss": 1.0437,
+      "step": 13061
+    },
+    {
+      "epoch": 1.1130805283340435,
+      "grad_norm": 22.0,
+      "learning_rate": 1.0364709635396684e-06,
+      "loss": 0.5727,
+      "step": 13062
+    },
+    {
+      "epoch": 1.1131657435023434,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0363053997571995e-06,
+      "loss": 0.6494,
+      "step": 13063
+    },
+    {
+      "epoch": 1.1132509586706434,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.036139839836407e-06,
+      "loss": 0.6923,
+      "step": 13064
+    },
+    {
+      "epoch": 1.1133361738389433,
+      "grad_norm": 13.375,
+      "learning_rate": 1.0359742837802826e-06,
+      "loss": 0.3933,
+      "step": 13065
+    },
+    {
+      "epoch": 1.1134213890072433,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0358087315918178e-06,
+      "loss": 0.5149,
+      "step": 13066
+    },
+    {
+      "epoch": 1.1135066041755433,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.0356431832740047e-06,
+      "loss": 0.5284,
+      "step": 13067
+    },
+    {
+      "epoch": 1.1135918193438432,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0354776388298346e-06,
+      "loss": 0.4593,
+      "step": 13068
+    },
+    {
+      "epoch": 1.1136770345121432,
+      "grad_norm": 11.5,
+      "learning_rate": 1.0353120982622998e-06,
+      "loss": 0.3664,
+      "step": 13069
+    },
+    {
+      "epoch": 1.1137622496804431,
+      "grad_norm": 16.125,
+      "learning_rate": 1.035146561574391e-06,
+      "loss": 0.569,
+      "step": 13070
+    },
+    {
+      "epoch": 1.113847464848743,
+      "grad_norm": 15.5,
+      "learning_rate": 1.0349810287690998e-06,
+      "loss": 0.4902,
+      "step": 13071
+    },
+    {
+      "epoch": 1.113932680017043,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0348154998494176e-06,
+      "loss": 0.6578,
+      "step": 13072
+    },
+    {
+      "epoch": 1.114017895185343,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0346499748183361e-06,
+      "loss": 0.4136,
+      "step": 13073
+    },
+    {
+      "epoch": 1.114103110353643,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0344844536788461e-06,
+      "loss": 0.4552,
+      "step": 13074
+    },
+    {
+      "epoch": 1.114188325521943,
+      "grad_norm": 22.75,
+      "learning_rate": 1.0343189364339382e-06,
+      "loss": 0.7652,
+      "step": 13075
+    },
+    {
+      "epoch": 1.1142735406902429,
+      "grad_norm": 59.0,
+      "learning_rate": 1.0341534230866044e-06,
+      "loss": 0.6078,
+      "step": 13076
+    },
+    {
+      "epoch": 1.1143587558585428,
+      "grad_norm": 15.0,
+      "learning_rate": 1.033987913639835e-06,
+      "loss": 0.4455,
+      "step": 13077
+    },
+    {
+      "epoch": 1.1144439710268428,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0338224080966216e-06,
+      "loss": 0.4727,
+      "step": 13078
+    },
+    {
+      "epoch": 1.1145291861951427,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0336569064599542e-06,
+      "loss": 0.6407,
+      "step": 13079
+    },
+    {
+      "epoch": 1.1146144013634427,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0334914087328244e-06,
+      "loss": 0.616,
+      "step": 13080
+    },
+    {
+      "epoch": 1.1146996165317427,
+      "grad_norm": 18.875,
+      "learning_rate": 1.0333259149182228e-06,
+      "loss": 0.887,
+      "step": 13081
+    },
+    {
+      "epoch": 1.1147848317000426,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0331604250191396e-06,
+      "loss": 0.5509,
+      "step": 13082
+    },
+    {
+      "epoch": 1.1148700468683426,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0329949390385653e-06,
+      "loss": 0.5918,
+      "step": 13083
+    },
+    {
+      "epoch": 1.1149552620366425,
+      "grad_norm": 24.0,
+      "learning_rate": 1.032829456979491e-06,
+      "loss": 1.0502,
+      "step": 13084
+    },
+    {
+      "epoch": 1.1150404772049425,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0326639788449063e-06,
+      "loss": 0.7153,
+      "step": 13085
+    },
+    {
+      "epoch": 1.1151256923732424,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0324985046378024e-06,
+      "loss": 0.5648,
+      "step": 13086
+    },
+    {
+      "epoch": 1.1152109075415424,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0323330343611692e-06,
+      "loss": 0.3868,
+      "step": 13087
+    },
+    {
+      "epoch": 1.1152961227098424,
+      "grad_norm": 17.25,
+      "learning_rate": 1.032167568017997e-06,
+      "loss": 0.598,
+      "step": 13088
+    },
+    {
+      "epoch": 1.1153813378781423,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.032002105611276e-06,
+      "loss": 0.5334,
+      "step": 13089
+    },
+    {
+      "epoch": 1.1154665530464423,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0318366471439959e-06,
+      "loss": 0.7744,
+      "step": 13090
+    },
+    {
+      "epoch": 1.1155517682147422,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.0316711926191473e-06,
+      "loss": 0.3883,
+      "step": 13091
+    },
+    {
+      "epoch": 1.1156369833830422,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.0315057420397193e-06,
+      "loss": 0.8604,
+      "step": 13092
+    },
+    {
+      "epoch": 1.1157221985513421,
+      "grad_norm": 20.75,
+      "learning_rate": 1.0313402954087026e-06,
+      "loss": 1.0109,
+      "step": 13093
+    },
+    {
+      "epoch": 1.115807413719642,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.0311748527290864e-06,
+      "loss": 0.3822,
+      "step": 13094
+    },
+    {
+      "epoch": 1.115892628887942,
+      "grad_norm": 16.0,
+      "learning_rate": 1.031009414003861e-06,
+      "loss": 0.4772,
+      "step": 13095
+    },
+    {
+      "epoch": 1.115977844056242,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.0308439792360153e-06,
+      "loss": 0.9222,
+      "step": 13096
+    },
+    {
+      "epoch": 1.116063059224542,
+      "grad_norm": 21.0,
+      "learning_rate": 1.03067854842854e-06,
+      "loss": 0.353,
+      "step": 13097
+    },
+    {
+      "epoch": 1.116148274392842,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.0305131215844233e-06,
+      "loss": 0.2562,
+      "step": 13098
+    },
+    {
+      "epoch": 1.1162334895611419,
+      "grad_norm": 12.875,
+      "learning_rate": 1.0303476987066554e-06,
+      "loss": 0.4775,
+      "step": 13099
+    },
+    {
+      "epoch": 1.1163187047294418,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.030182279798225e-06,
+      "loss": 0.734,
+      "step": 13100
+    },
+    {
+      "epoch": 1.1164039198977418,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0300168648621223e-06,
+      "loss": 0.319,
+      "step": 13101
+    },
+    {
+      "epoch": 1.1164891350660417,
+      "grad_norm": 25.125,
+      "learning_rate": 1.0298514539013357e-06,
+      "loss": 0.6406,
+      "step": 13102
+    },
+    {
+      "epoch": 1.1165743502343417,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0296860469188553e-06,
+      "loss": 0.6202,
+      "step": 13103
+    },
+    {
+      "epoch": 1.1166595654026417,
+      "grad_norm": 14.875,
+      "learning_rate": 1.0295206439176688e-06,
+      "loss": 0.6371,
+      "step": 13104
+    },
+    {
+      "epoch": 1.1167447805709416,
+      "grad_norm": 13.625,
+      "learning_rate": 1.029355244900767e-06,
+      "loss": 0.3767,
+      "step": 13105
+    },
+    {
+      "epoch": 1.1168299957392416,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0291898498711373e-06,
+      "loss": 0.7552,
+      "step": 13106
+    },
+    {
+      "epoch": 1.1169152109075415,
+      "grad_norm": 20.625,
+      "learning_rate": 1.0290244588317686e-06,
+      "loss": 0.5777,
+      "step": 13107
+    },
+    {
+      "epoch": 1.1170004260758415,
+      "grad_norm": 14.75,
+      "learning_rate": 1.0288590717856507e-06,
+      "loss": 0.4842,
+      "step": 13108
+    },
+    {
+      "epoch": 1.1170856412441414,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.0286936887357713e-06,
+      "loss": 0.5881,
+      "step": 13109
+    },
+    {
+      "epoch": 1.1171708564124414,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.0285283096851197e-06,
+      "loss": 0.3175,
+      "step": 13110
+    },
+    {
+      "epoch": 1.1172560715807414,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0283629346366842e-06,
+      "loss": 0.6931,
+      "step": 13111
+    },
+    {
+      "epoch": 1.1173412867490413,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0281975635934534e-06,
+      "loss": 0.4789,
+      "step": 13112
+    },
+    {
+      "epoch": 1.1174265019173413,
+      "grad_norm": 14.625,
+      "learning_rate": 1.028032196558416e-06,
+      "loss": 0.5222,
+      "step": 13113
+    },
+    {
+      "epoch": 1.1175117170856412,
+      "grad_norm": 28.875,
+      "learning_rate": 1.0278668335345597e-06,
+      "loss": 0.8614,
+      "step": 13114
+    },
+    {
+      "epoch": 1.1175969322539412,
+      "grad_norm": 15.125,
+      "learning_rate": 1.027701474524873e-06,
+      "loss": 0.8517,
+      "step": 13115
+    },
+    {
+      "epoch": 1.1176821474222411,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0275361195323445e-06,
+      "loss": 0.3669,
+      "step": 13116
+    },
+    {
+      "epoch": 1.117767362590541,
+      "grad_norm": 14.125,
+      "learning_rate": 1.0273707685599616e-06,
+      "loss": 0.4826,
+      "step": 13117
+    },
+    {
+      "epoch": 1.117852577758841,
+      "grad_norm": 16.75,
+      "learning_rate": 1.027205421610713e-06,
+      "loss": 0.833,
+      "step": 13118
+    },
+    {
+      "epoch": 1.117937792927141,
+      "grad_norm": 12.5,
+      "learning_rate": 1.0270400786875864e-06,
+      "loss": 0.3049,
+      "step": 13119
+    },
+    {
+      "epoch": 1.118023008095441,
+      "grad_norm": 17.5,
+      "learning_rate": 1.02687473979357e-06,
+      "loss": 0.6963,
+      "step": 13120
+    },
+    {
+      "epoch": 1.118108223263741,
+      "grad_norm": 12.5,
+      "learning_rate": 1.0267094049316514e-06,
+      "loss": 0.5541,
+      "step": 13121
+    },
+    {
+      "epoch": 1.1181934384320409,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0265440741048183e-06,
+      "loss": 0.4425,
+      "step": 13122
+    },
+    {
+      "epoch": 1.1182786536003408,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0263787473160584e-06,
+      "loss": 0.8104,
+      "step": 13123
+    },
+    {
+      "epoch": 1.1183638687686408,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0262134245683593e-06,
+      "loss": 0.476,
+      "step": 13124
+    },
+    {
+      "epoch": 1.1184490839369408,
+      "grad_norm": 16.125,
+      "learning_rate": 1.026048105864709e-06,
+      "loss": 0.6981,
+      "step": 13125
+    },
+    {
+      "epoch": 1.1185342991052407,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0258827912080941e-06,
+      "loss": 0.385,
+      "step": 13126
+    },
+    {
+      "epoch": 1.1186195142735407,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.025717480601503e-06,
+      "loss": 0.4115,
+      "step": 13127
+    },
+    {
+      "epoch": 1.1187047294418406,
+      "grad_norm": 12.5,
+      "learning_rate": 1.0255521740479225e-06,
+      "loss": 0.237,
+      "step": 13128
+    },
+    {
+      "epoch": 1.1187899446101406,
+      "grad_norm": 19.875,
+      "learning_rate": 1.02538687155034e-06,
+      "loss": 1.0516,
+      "step": 13129
+    },
+    {
+      "epoch": 1.1188751597784405,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.0252215731117424e-06,
+      "loss": 0.3883,
+      "step": 13130
+    },
+    {
+      "epoch": 1.1189603749467405,
+      "grad_norm": 17.5,
+      "learning_rate": 1.025056278735117e-06,
+      "loss": 0.9291,
+      "step": 13131
+    },
+    {
+      "epoch": 1.1190455901150405,
+      "grad_norm": 14.875,
+      "learning_rate": 1.0248909884234509e-06,
+      "loss": 0.5778,
+      "step": 13132
+    },
+    {
+      "epoch": 1.1191308052833404,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0247257021797308e-06,
+      "loss": 0.7954,
+      "step": 13133
+    },
+    {
+      "epoch": 1.1192160204516404,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0245604200069439e-06,
+      "loss": 0.6463,
+      "step": 13134
+    },
+    {
+      "epoch": 1.1193012356199403,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.024395141908077e-06,
+      "loss": 0.6824,
+      "step": 13135
+    },
+    {
+      "epoch": 1.1193864507882403,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.0242298678861166e-06,
+      "loss": 0.4984,
+      "step": 13136
+    },
+    {
+      "epoch": 1.1194716659565402,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.0240645979440496e-06,
+      "loss": 0.5623,
+      "step": 13137
+    },
+    {
+      "epoch": 1.1195568811248402,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0238993320848631e-06,
+      "loss": 0.73,
+      "step": 13138
+    },
+    {
+      "epoch": 1.1196420962931402,
+      "grad_norm": 15.625,
+      "learning_rate": 1.0237340703115423e-06,
+      "loss": 0.8047,
+      "step": 13139
+    },
+    {
+      "epoch": 1.1197273114614401,
+      "grad_norm": 19.375,
+      "learning_rate": 1.0235688126270746e-06,
+      "loss": 1.034,
+      "step": 13140
+    },
+    {
+      "epoch": 1.11981252662974,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.023403559034446e-06,
+      "loss": 0.3668,
+      "step": 13141
+    },
+    {
+      "epoch": 1.11989774179804,
+      "grad_norm": 19.125,
+      "learning_rate": 1.023238309536643e-06,
+      "loss": 0.5025,
+      "step": 13142
+    },
+    {
+      "epoch": 1.11998295696634,
+      "grad_norm": 17.75,
+      "learning_rate": 1.023073064136652e-06,
+      "loss": 0.8449,
+      "step": 13143
+    },
+    {
+      "epoch": 1.12006817213464,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0229078228374585e-06,
+      "loss": 0.4999,
+      "step": 13144
+    },
+    {
+      "epoch": 1.12015338730294,
+      "grad_norm": 12.0,
+      "learning_rate": 1.0227425856420492e-06,
+      "loss": 0.4149,
+      "step": 13145
+    },
+    {
+      "epoch": 1.1202386024712399,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0225773525534104e-06,
+      "loss": 0.5834,
+      "step": 13146
+    },
+    {
+      "epoch": 1.1203238176395398,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.022412123574527e-06,
+      "loss": 0.857,
+      "step": 13147
+    },
+    {
+      "epoch": 1.1204090328078398,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0222468987083855e-06,
+      "loss": 0.6731,
+      "step": 13148
+    },
+    {
+      "epoch": 1.1204942479761397,
+      "grad_norm": 15.0,
+      "learning_rate": 1.0220816779579714e-06,
+      "loss": 0.4319,
+      "step": 13149
+    },
+    {
+      "epoch": 1.1205794631444397,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.0219164613262707e-06,
+      "loss": 0.5113,
+      "step": 13150
+    },
+    {
+      "epoch": 1.1206646783127396,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0217512488162688e-06,
+      "loss": 0.6501,
+      "step": 13151
+    },
+    {
+      "epoch": 1.1207498934810396,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0215860404309516e-06,
+      "loss": 0.3189,
+      "step": 13152
+    },
+    {
+      "epoch": 1.1208351086493396,
+      "grad_norm": 17.875,
+      "learning_rate": 1.021420836173304e-06,
+      "loss": 0.8867,
+      "step": 13153
+    },
+    {
+      "epoch": 1.1209203238176395,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0212556360463122e-06,
+      "loss": 0.7039,
+      "step": 13154
+    },
+    {
+      "epoch": 1.1210055389859395,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.021090440052961e-06,
+      "loss": 0.7019,
+      "step": 13155
+    },
+    {
+      "epoch": 1.1210907541542394,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0209252481962352e-06,
+      "loss": 0.7521,
+      "step": 13156
+    },
+    {
+      "epoch": 1.1211759693225394,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.020760060479121e-06,
+      "loss": 0.6964,
+      "step": 13157
+    },
+    {
+      "epoch": 1.1212611844908393,
+      "grad_norm": 19.125,
+      "learning_rate": 1.020594876904603e-06,
+      "loss": 0.7076,
+      "step": 13158
+    },
+    {
+      "epoch": 1.1213463996591393,
+      "grad_norm": 15.5,
+      "learning_rate": 1.020429697475666e-06,
+      "loss": 0.4491,
+      "step": 13159
+    },
+    {
+      "epoch": 1.1214316148274392,
+      "grad_norm": 22.25,
+      "learning_rate": 1.0202645221952956e-06,
+      "loss": 0.8533,
+      "step": 13160
+    },
+    {
+      "epoch": 1.1215168299957392,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0200993510664758e-06,
+      "loss": 0.3223,
+      "step": 13161
+    },
+    {
+      "epoch": 1.1216020451640392,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0199341840921924e-06,
+      "loss": 0.5168,
+      "step": 13162
+    },
+    {
+      "epoch": 1.1216872603323391,
+      "grad_norm": 21.0,
+      "learning_rate": 1.0197690212754296e-06,
+      "loss": 0.7427,
+      "step": 13163
+    },
+    {
+      "epoch": 1.121772475500639,
+      "grad_norm": 24.375,
+      "learning_rate": 1.0196038626191717e-06,
+      "loss": 0.9318,
+      "step": 13164
+    },
+    {
+      "epoch": 1.121857690668939,
+      "grad_norm": 32.25,
+      "learning_rate": 1.0194387081264042e-06,
+      "loss": 1.1205,
+      "step": 13165
+    },
+    {
+      "epoch": 1.121942905837239,
+      "grad_norm": 12.375,
+      "learning_rate": 1.0192735578001105e-06,
+      "loss": 0.4297,
+      "step": 13166
+    },
+    {
+      "epoch": 1.122028121005539,
+      "grad_norm": 19.875,
+      "learning_rate": 1.019108411643276e-06,
+      "loss": 0.5472,
+      "step": 13167
+    },
+    {
+      "epoch": 1.122113336173839,
+      "grad_norm": 13.375,
+      "learning_rate": 1.0189432696588843e-06,
+      "loss": 0.483,
+      "step": 13168
+    },
+    {
+      "epoch": 1.1221985513421389,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0187781318499208e-06,
+      "loss": 0.6847,
+      "step": 13169
+    },
+    {
+      "epoch": 1.1222837665104388,
+      "grad_norm": 22.5,
+      "learning_rate": 1.0186129982193688e-06,
+      "loss": 0.9738,
+      "step": 13170
+    },
+    {
+      "epoch": 1.1223689816787388,
+      "grad_norm": 29.0,
+      "learning_rate": 1.018447868770212e-06,
+      "loss": 0.8715,
+      "step": 13171
+    },
+    {
+      "epoch": 1.1224541968470387,
+      "grad_norm": 19.625,
+      "learning_rate": 1.0182827435054356e-06,
+      "loss": 0.5456,
+      "step": 13172
+    },
+    {
+      "epoch": 1.1225394120153387,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.018117622428023e-06,
+      "loss": 0.3682,
+      "step": 13173
+    },
+    {
+      "epoch": 1.1226246271836386,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0179525055409576e-06,
+      "loss": 1.0837,
+      "step": 13174
+    },
+    {
+      "epoch": 1.1227098423519386,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0177873928472242e-06,
+      "loss": 0.2664,
+      "step": 13175
+    },
+    {
+      "epoch": 1.1227950575202386,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.017622284349806e-06,
+      "loss": 0.3701,
+      "step": 13176
+    },
+    {
+      "epoch": 1.1228802726885385,
+      "grad_norm": 19.5,
+      "learning_rate": 1.017457180051687e-06,
+      "loss": 0.859,
+      "step": 13177
+    },
+    {
+      "epoch": 1.1229654878568385,
+      "grad_norm": 17.875,
+      "learning_rate": 1.0172920799558508e-06,
+      "loss": 0.4577,
+      "step": 13178
+    },
+    {
+      "epoch": 1.1230507030251384,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.01712698406528e-06,
+      "loss": 0.3891,
+      "step": 13179
+    },
+    {
+      "epoch": 1.1231359181934384,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0169618923829593e-06,
+      "loss": 0.7796,
+      "step": 13180
+    },
+    {
+      "epoch": 1.1232211333617383,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0167968049118712e-06,
+      "loss": 0.3676,
+      "step": 13181
+    },
+    {
+      "epoch": 1.1233063485300383,
+      "grad_norm": 12.125,
+      "learning_rate": 1.0166317216549995e-06,
+      "loss": 0.2856,
+      "step": 13182
+    },
+    {
+      "epoch": 1.1233915636983383,
+      "grad_norm": 17.5,
+      "learning_rate": 1.0164666426153271e-06,
+      "loss": 0.8112,
+      "step": 13183
+    },
+    {
+      "epoch": 1.1234767788666382,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0163015677958376e-06,
+      "loss": 0.716,
+      "step": 13184
+    },
+    {
+      "epoch": 1.1235619940349382,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0161364971995136e-06,
+      "loss": 0.509,
+      "step": 13185
+    },
+    {
+      "epoch": 1.1236472092032381,
+      "grad_norm": 17.875,
+      "learning_rate": 1.0159714308293387e-06,
+      "loss": 0.756,
+      "step": 13186
+    },
+    {
+      "epoch": 1.123732424371538,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.0158063686882948e-06,
+      "loss": 0.4773,
+      "step": 13187
+    },
+    {
+      "epoch": 1.123817639539838,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.0156413107793658e-06,
+      "loss": 0.4807,
+      "step": 13188
+    },
+    {
+      "epoch": 1.123902854708138,
+      "grad_norm": 13.0,
+      "learning_rate": 1.0154762571055335e-06,
+      "loss": 0.5748,
+      "step": 13189
+    },
+    {
+      "epoch": 1.123988069876438,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0153112076697813e-06,
+      "loss": 0.399,
+      "step": 13190
+    },
+    {
+      "epoch": 1.124073285044738,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.0151461624750914e-06,
+      "loss": 0.4311,
+      "step": 13191
+    },
+    {
+      "epoch": 1.1241585002130379,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0149811215244468e-06,
+      "loss": 0.8434,
+      "step": 13192
+    },
+    {
+      "epoch": 1.1242437153813378,
+      "grad_norm": 17.625,
+      "learning_rate": 1.0148160848208293e-06,
+      "loss": 0.6584,
+      "step": 13193
+    },
+    {
+      "epoch": 1.1243289305496378,
+      "grad_norm": 20.75,
+      "learning_rate": 1.0146510523672221e-06,
+      "loss": 0.7541,
+      "step": 13194
+    },
+    {
+      "epoch": 1.1244141457179377,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0144860241666068e-06,
+      "loss": 0.2725,
+      "step": 13195
+    },
+    {
+      "epoch": 1.1244993608862377,
+      "grad_norm": 23.0,
+      "learning_rate": 1.0143210002219657e-06,
+      "loss": 0.8792,
+      "step": 13196
+    },
+    {
+      "epoch": 1.1245845760545377,
+      "grad_norm": 15.625,
+      "learning_rate": 1.0141559805362815e-06,
+      "loss": 0.6599,
+      "step": 13197
+    },
+    {
+      "epoch": 1.1246697912228376,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.0139909651125354e-06,
+      "loss": 0.4025,
+      "step": 13198
+    },
+    {
+      "epoch": 1.1247550063911376,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.0138259539537103e-06,
+      "loss": 0.4674,
+      "step": 13199
+    },
+    {
+      "epoch": 1.1248402215594375,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0136609470627874e-06,
+      "loss": 0.3598,
+      "step": 13200
+    },
+    {
+      "epoch": 1.1249254367277375,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0134959444427492e-06,
+      "loss": 0.6847,
+      "step": 13201
+    },
+    {
+      "epoch": 1.1250106518960374,
+      "grad_norm": 11.75,
+      "learning_rate": 1.0133309460965771e-06,
+      "loss": 0.3975,
+      "step": 13202
+    },
+    {
+      "epoch": 1.1250958670643374,
+      "grad_norm": 15.5,
+      "learning_rate": 1.0131659520272527e-06,
+      "loss": 0.3462,
+      "step": 13203
+    },
+    {
+      "epoch": 1.1251810822326374,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0130009622377575e-06,
+      "loss": 0.4185,
+      "step": 13204
+    },
+    {
+      "epoch": 1.1252662974009373,
+      "grad_norm": 16.75,
+      "learning_rate": 1.0128359767310734e-06,
+      "loss": 0.8716,
+      "step": 13205
+    },
+    {
+      "epoch": 1.1253515125692373,
+      "grad_norm": 15.25,
+      "learning_rate": 1.0126709955101815e-06,
+      "loss": 0.7207,
+      "step": 13206
+    },
+    {
+      "epoch": 1.1254367277375372,
+      "grad_norm": 16.625,
+      "learning_rate": 1.0125060185780636e-06,
+      "loss": 0.6605,
+      "step": 13207
+    },
+    {
+      "epoch": 1.1255219429058372,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0123410459377004e-06,
+      "loss": 0.7494,
+      "step": 13208
+    },
+    {
+      "epoch": 1.1256071580741371,
+      "grad_norm": 18.125,
+      "learning_rate": 1.0121760775920739e-06,
+      "loss": 0.6069,
+      "step": 13209
+    },
+    {
+      "epoch": 1.125692373242437,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0120111135441644e-06,
+      "loss": 0.5261,
+      "step": 13210
+    },
+    {
+      "epoch": 1.125777588410737,
+      "grad_norm": 9.625,
+      "learning_rate": 1.011846153796954e-06,
+      "loss": 0.2247,
+      "step": 13211
+    },
+    {
+      "epoch": 1.125862803579037,
+      "grad_norm": 18.0,
+      "learning_rate": 1.011681198353423e-06,
+      "loss": 0.5614,
+      "step": 13212
+    },
+    {
+      "epoch": 1.125948018747337,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0115162472165517e-06,
+      "loss": 0.6167,
+      "step": 13213
+    },
+    {
+      "epoch": 1.126033233915637,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0113513003893223e-06,
+      "loss": 0.4604,
+      "step": 13214
+    },
+    {
+      "epoch": 1.1261184490839369,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0111863578747145e-06,
+      "loss": 0.5409,
+      "step": 13215
+    },
+    {
+      "epoch": 1.1262036642522368,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0110214196757095e-06,
+      "loss": 0.4521,
+      "step": 13216
+    },
+    {
+      "epoch": 1.1262888794205368,
+      "grad_norm": 17.25,
+      "learning_rate": 1.0108564857952876e-06,
+      "loss": 0.4512,
+      "step": 13217
+    },
+    {
+      "epoch": 1.1263740945888367,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0106915562364298e-06,
+      "loss": 0.7895,
+      "step": 13218
+    },
+    {
+      "epoch": 1.1264593097571367,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0105266310021164e-06,
+      "loss": 0.6925,
+      "step": 13219
+    },
+    {
+      "epoch": 1.1265445249254367,
+      "grad_norm": 17.25,
+      "learning_rate": 1.0103617100953273e-06,
+      "loss": 0.5922,
+      "step": 13220
+    },
+    {
+      "epoch": 1.1266297400937366,
+      "grad_norm": 16.875,
+      "learning_rate": 1.010196793519043e-06,
+      "loss": 0.7036,
+      "step": 13221
+    },
+    {
+      "epoch": 1.1267149552620366,
+      "grad_norm": 24.625,
+      "learning_rate": 1.010031881276244e-06,
+      "loss": 0.8022,
+      "step": 13222
+    },
+    {
+      "epoch": 1.1268001704303365,
+      "grad_norm": 14.125,
+      "learning_rate": 1.0098669733699097e-06,
+      "loss": 0.4868,
+      "step": 13223
+    },
+    {
+      "epoch": 1.1268853855986365,
+      "grad_norm": 12.875,
+      "learning_rate": 1.009702069803021e-06,
+      "loss": 0.4106,
+      "step": 13224
+    },
+    {
+      "epoch": 1.1269706007669364,
+      "grad_norm": 18.875,
+      "learning_rate": 1.0095371705785574e-06,
+      "loss": 0.5967,
+      "step": 13225
+    },
+    {
+      "epoch": 1.1270558159352364,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0093722756994992e-06,
+      "loss": 0.8156,
+      "step": 13226
+    },
+    {
+      "epoch": 1.1271410311035364,
+      "grad_norm": 13.25,
+      "learning_rate": 1.0092073851688262e-06,
+      "loss": 0.2918,
+      "step": 13227
+    },
+    {
+      "epoch": 1.1272262462718363,
+      "grad_norm": 21.25,
+      "learning_rate": 1.0090424989895172e-06,
+      "loss": 0.7232,
+      "step": 13228
+    },
+    {
+      "epoch": 1.1273114614401363,
+      "grad_norm": 17.5,
+      "learning_rate": 1.0088776171645529e-06,
+      "loss": 0.828,
+      "step": 13229
+    },
+    {
+      "epoch": 1.1273966766084362,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.0087127396969124e-06,
+      "loss": 0.5333,
+      "step": 13230
+    },
+    {
+      "epoch": 1.1274818917767362,
+      "grad_norm": 20.25,
+      "learning_rate": 1.0085478665895753e-06,
+      "loss": 0.8569,
+      "step": 13231
+    },
+    {
+      "epoch": 1.1275671069450361,
+      "grad_norm": 16.75,
+      "learning_rate": 1.008382997845521e-06,
+      "loss": 0.6145,
+      "step": 13232
+    },
+    {
+      "epoch": 1.127652322113336,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.008218133467729e-06,
+      "loss": 0.2534,
+      "step": 13233
+    },
+    {
+      "epoch": 1.127737537281636,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0080532734591784e-06,
+      "loss": 0.616,
+      "step": 13234
+    },
+    {
+      "epoch": 1.127822752449936,
+      "grad_norm": 20.125,
+      "learning_rate": 1.0078884178228492e-06,
+      "loss": 0.481,
+      "step": 13235
+    },
+    {
+      "epoch": 1.127907967618236,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.0077235665617188e-06,
+      "loss": 0.5209,
+      "step": 13236
+    },
+    {
+      "epoch": 1.127993182786536,
+      "grad_norm": 10.375,
+      "learning_rate": 1.0075587196787677e-06,
+      "loss": 0.2546,
+      "step": 13237
+    },
+    {
+      "epoch": 1.1280783979548359,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0073938771769738e-06,
+      "loss": 0.7662,
+      "step": 13238
+    },
+    {
+      "epoch": 1.1281636131231358,
+      "grad_norm": 18.5,
+      "learning_rate": 1.007229039059317e-06,
+      "loss": 0.7725,
+      "step": 13239
+    },
+    {
+      "epoch": 1.1282488282914358,
+      "grad_norm": 16.625,
+      "learning_rate": 1.0070642053287751e-06,
+      "loss": 0.7769,
+      "step": 13240
+    },
+    {
+      "epoch": 1.1283340434597358,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0068993759883278e-06,
+      "loss": 0.4144,
+      "step": 13241
+    },
+    {
+      "epoch": 1.1284192586280357,
+      "grad_norm": 9.75,
+      "learning_rate": 1.0067345510409527e-06,
+      "loss": 0.2127,
+      "step": 13242
+    },
+    {
+      "epoch": 1.1285044737963357,
+      "grad_norm": 16.625,
+      "learning_rate": 1.0065697304896296e-06,
+      "loss": 0.9401,
+      "step": 13243
+    },
+    {
+      "epoch": 1.1285896889646356,
+      "grad_norm": 13.625,
+      "learning_rate": 1.006404914337336e-06,
+      "loss": 0.4444,
+      "step": 13244
+    },
+    {
+      "epoch": 1.1286749041329356,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0062401025870501e-06,
+      "loss": 0.7788,
+      "step": 13245
+    },
+    {
+      "epoch": 1.1287601193012355,
+      "grad_norm": 18.25,
+      "learning_rate": 1.006075295241751e-06,
+      "loss": 0.8345,
+      "step": 13246
+    },
+    {
+      "epoch": 1.1288453344695355,
+      "grad_norm": 29.25,
+      "learning_rate": 1.0059104923044166e-06,
+      "loss": 0.7864,
+      "step": 13247
+    },
+    {
+      "epoch": 1.1289305496378355,
+      "grad_norm": 10.625,
+      "learning_rate": 1.0057456937780252e-06,
+      "loss": 0.2497,
+      "step": 13248
+    },
+    {
+      "epoch": 1.1290157648061354,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.0055808996655542e-06,
+      "loss": 0.5187,
+      "step": 13249
+    },
+    {
+      "epoch": 1.1291009799744354,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0054161099699826e-06,
+      "loss": 0.745,
+      "step": 13250
+    },
+    {
+      "epoch": 1.1291861951427353,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0052513246942879e-06,
+      "loss": 0.6369,
+      "step": 13251
+    },
+    {
+      "epoch": 1.1292714103110353,
+      "grad_norm": 25.75,
+      "learning_rate": 1.005086543841448e-06,
+      "loss": 0.4194,
+      "step": 13252
+    },
+    {
+      "epoch": 1.1293566254793352,
+      "grad_norm": 10.375,
+      "learning_rate": 1.00492176741444e-06,
+      "loss": 0.37,
+      "step": 13253
+    },
+    {
+      "epoch": 1.1294418406476352,
+      "grad_norm": 17.75,
+      "learning_rate": 1.0047569954162424e-06,
+      "loss": 0.7184,
+      "step": 13254
+    },
+    {
+      "epoch": 1.1295270558159352,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0045922278498323e-06,
+      "loss": 0.5522,
+      "step": 13255
+    },
+    {
+      "epoch": 1.1296122709842351,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.0044274647181876e-06,
+      "loss": 0.24,
+      "step": 13256
+    },
+    {
+      "epoch": 1.129697486152535,
+      "grad_norm": 22.875,
+      "learning_rate": 1.0042627060242854e-06,
+      "loss": 0.8819,
+      "step": 13257
+    },
+    {
+      "epoch": 1.129782701320835,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0040979517711034e-06,
+      "loss": 0.6125,
+      "step": 13258
+    },
+    {
+      "epoch": 1.129867916489135,
+      "grad_norm": 19.0,
+      "learning_rate": 1.003933201961619e-06,
+      "loss": 0.5613,
+      "step": 13259
+    },
+    {
+      "epoch": 1.129953131657435,
+      "grad_norm": 27.5,
+      "learning_rate": 1.0037684565988084e-06,
+      "loss": 0.707,
+      "step": 13260
+    },
+    {
+      "epoch": 1.130038346825735,
+      "grad_norm": 16.875,
+      "learning_rate": 1.00360371568565e-06,
+      "loss": 0.7815,
+      "step": 13261
+    },
+    {
+      "epoch": 1.1301235619940349,
+      "grad_norm": 18.375,
+      "learning_rate": 1.0034389792251196e-06,
+      "loss": 0.7995,
+      "step": 13262
+    },
+    {
+      "epoch": 1.1302087771623348,
+      "grad_norm": 20.625,
+      "learning_rate": 1.0032742472201952e-06,
+      "loss": 0.5258,
+      "step": 13263
+    },
+    {
+      "epoch": 1.1302939923306348,
+      "grad_norm": 16.0,
+      "learning_rate": 1.003109519673853e-06,
+      "loss": 0.5172,
+      "step": 13264
+    },
+    {
+      "epoch": 1.1303792074989347,
+      "grad_norm": 31.0,
+      "learning_rate": 1.0029447965890702e-06,
+      "loss": 0.5182,
+      "step": 13265
+    },
+    {
+      "epoch": 1.1304644226672347,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0027800779688236e-06,
+      "loss": 0.498,
+      "step": 13266
+    },
+    {
+      "epoch": 1.1305496378355346,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0026153638160896e-06,
+      "loss": 0.513,
+      "step": 13267
+    },
+    {
+      "epoch": 1.1306348530038346,
+      "grad_norm": 20.625,
+      "learning_rate": 1.0024506541338444e-06,
+      "loss": 0.8987,
+      "step": 13268
+    },
+    {
+      "epoch": 1.1307200681721346,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.002285948925065e-06,
+      "loss": 0.5029,
+      "step": 13269
+    },
+    {
+      "epoch": 1.1308052833404345,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0021212481927272e-06,
+      "loss": 0.7434,
+      "step": 13270
+    },
+    {
+      "epoch": 1.1308904985087345,
+      "grad_norm": 12.5,
+      "learning_rate": 1.001956551939808e-06,
+      "loss": 0.5046,
+      "step": 13271
+    },
+    {
+      "epoch": 1.1309757136770344,
+      "grad_norm": 13.5,
+      "learning_rate": 1.0017918601692833e-06,
+      "loss": 0.5205,
+      "step": 13272
+    },
+    {
+      "epoch": 1.1310609288453344,
+      "grad_norm": 14.0,
+      "learning_rate": 1.0016271728841294e-06,
+      "loss": 0.6635,
+      "step": 13273
+    },
+    {
+      "epoch": 1.1311461440136343,
+      "grad_norm": 22.0,
+      "learning_rate": 1.001462490087322e-06,
+      "loss": 0.3674,
+      "step": 13274
+    },
+    {
+      "epoch": 1.1312313591819343,
+      "grad_norm": 11.875,
+      "learning_rate": 1.0012978117818375e-06,
+      "loss": 0.4243,
+      "step": 13275
+    },
+    {
+      "epoch": 1.1313165743502342,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.0011331379706522e-06,
+      "loss": 0.6768,
+      "step": 13276
+    },
+    {
+      "epoch": 1.1314017895185342,
+      "grad_norm": 15.625,
+      "learning_rate": 1.0009684686567404e-06,
+      "loss": 0.9479,
+      "step": 13277
+    },
+    {
+      "epoch": 1.1314870046868342,
+      "grad_norm": 29.125,
+      "learning_rate": 1.0008038038430791e-06,
+      "loss": 0.3934,
+      "step": 13278
+    },
+    {
+      "epoch": 1.1315722198551341,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0006391435326438e-06,
+      "loss": 0.7388,
+      "step": 13279
+    },
+    {
+      "epoch": 1.131657435023434,
+      "grad_norm": 20.875,
+      "learning_rate": 1.0004744877284099e-06,
+      "loss": 1.0583,
+      "step": 13280
+    },
+    {
+      "epoch": 1.131742650191734,
+      "grad_norm": 17.25,
+      "learning_rate": 1.000309836433353e-06,
+      "loss": 0.8705,
+      "step": 13281
+    },
+    {
+      "epoch": 1.131827865360034,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.000145189650448e-06,
+      "loss": 0.3678,
+      "step": 13282
+    },
+    {
+      "epoch": 1.131913080528334,
+      "grad_norm": 15.875,
+      "learning_rate": 9.99980547382671e-07,
+      "loss": 0.6978,
+      "step": 13283
+    },
+    {
+      "epoch": 1.131998295696634,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.998159096329974e-07,
+      "loss": 0.6857,
+      "step": 13284
+    },
+    {
+      "epoch": 1.1320835108649339,
+      "grad_norm": 10.6875,
+      "learning_rate": 9.99651276404401e-07,
+      "loss": 0.2527,
+      "step": 13285
+    },
+    {
+      "epoch": 1.1321687260332338,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.994866476998583e-07,
+      "loss": 0.5626,
+      "step": 13286
+    },
+    {
+      "epoch": 1.132253941201534,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.993220235223437e-07,
+      "loss": 0.6575,
+      "step": 13287
+    },
+    {
+      "epoch": 1.132339156369834,
+      "grad_norm": 13.75,
+      "learning_rate": 9.991574038748322e-07,
+      "loss": 0.6853,
+      "step": 13288
+    },
+    {
+      "epoch": 1.132424371538134,
+      "grad_norm": 13.75,
+      "learning_rate": 9.989927887602984e-07,
+      "loss": 0.3997,
+      "step": 13289
+    },
+    {
+      "epoch": 1.1325095867064339,
+      "grad_norm": 15.625,
+      "learning_rate": 9.988281781817178e-07,
+      "loss": 0.7059,
+      "step": 13290
+    },
+    {
+      "epoch": 1.1325948018747338,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.98663572142064e-07,
+      "loss": 0.4542,
+      "step": 13291
+    },
+    {
+      "epoch": 1.1326800170430338,
+      "grad_norm": 36.0,
+      "learning_rate": 9.984989706443129e-07,
+      "loss": 0.7358,
+      "step": 13292
+    },
+    {
+      "epoch": 1.1327652322113337,
+      "grad_norm": 57.5,
+      "learning_rate": 9.98334373691438e-07,
+      "loss": 0.7715,
+      "step": 13293
+    },
+    {
+      "epoch": 1.1328504473796337,
+      "grad_norm": 21.75,
+      "learning_rate": 9.981697812864138e-07,
+      "loss": 0.3737,
+      "step": 13294
+    },
+    {
+      "epoch": 1.1329356625479337,
+      "grad_norm": 22.5,
+      "learning_rate": 9.980051934322152e-07,
+      "loss": 0.5305,
+      "step": 13295
+    },
+    {
+      "epoch": 1.1330208777162336,
+      "grad_norm": 16.75,
+      "learning_rate": 9.97840610131816e-07,
+      "loss": 0.7419,
+      "step": 13296
+    },
+    {
+      "epoch": 1.1331060928845336,
+      "grad_norm": 19.25,
+      "learning_rate": 9.976760313881903e-07,
+      "loss": 0.5595,
+      "step": 13297
+    },
+    {
+      "epoch": 1.1331913080528335,
+      "grad_norm": 17.25,
+      "learning_rate": 9.97511457204313e-07,
+      "loss": 0.7737,
+      "step": 13298
+    },
+    {
+      "epoch": 1.1332765232211335,
+      "grad_norm": 19.625,
+      "learning_rate": 9.973468875831569e-07,
+      "loss": 0.7084,
+      "step": 13299
+    },
+    {
+      "epoch": 1.1333617383894334,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.971823225276973e-07,
+      "loss": 0.4992,
+      "step": 13300
+    },
+    {
+      "epoch": 1.1334469535577334,
+      "grad_norm": 18.875,
+      "learning_rate": 9.970177620409072e-07,
+      "loss": 1.0615,
+      "step": 13301
+    },
+    {
+      "epoch": 1.1335321687260334,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.968532061257603e-07,
+      "loss": 0.7005,
+      "step": 13302
+    },
+    {
+      "epoch": 1.1336173838943333,
+      "grad_norm": 21.875,
+      "learning_rate": 9.966886547852305e-07,
+      "loss": 0.8372,
+      "step": 13303
+    },
+    {
+      "epoch": 1.1337025990626333,
+      "grad_norm": 18.75,
+      "learning_rate": 9.965241080222913e-07,
+      "loss": 1.094,
+      "step": 13304
+    },
+    {
+      "epoch": 1.1337878142309332,
+      "grad_norm": 22.125,
+      "learning_rate": 9.963595658399167e-07,
+      "loss": 0.8261,
+      "step": 13305
+    },
+    {
+      "epoch": 1.1338730293992332,
+      "grad_norm": 18.375,
+      "learning_rate": 9.961950282410794e-07,
+      "loss": 0.4777,
+      "step": 13306
+    },
+    {
+      "epoch": 1.1339582445675331,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.960304952287536e-07,
+      "loss": 0.3664,
+      "step": 13307
+    },
+    {
+      "epoch": 1.134043459735833,
+      "grad_norm": 14.75,
+      "learning_rate": 9.958659668059123e-07,
+      "loss": 0.5838,
+      "step": 13308
+    },
+    {
+      "epoch": 1.134128674904133,
+      "grad_norm": 17.875,
+      "learning_rate": 9.95701442975528e-07,
+      "loss": 0.7497,
+      "step": 13309
+    },
+    {
+      "epoch": 1.134213890072433,
+      "grad_norm": 11.0,
+      "learning_rate": 9.955369237405745e-07,
+      "loss": 0.2642,
+      "step": 13310
+    },
+    {
+      "epoch": 1.134299105240733,
+      "grad_norm": 17.125,
+      "learning_rate": 9.95372409104025e-07,
+      "loss": 0.5166,
+      "step": 13311
+    },
+    {
+      "epoch": 1.134384320409033,
+      "grad_norm": 16.5,
+      "learning_rate": 9.952078990688515e-07,
+      "loss": 0.7807,
+      "step": 13312
+    },
+    {
+      "epoch": 1.1344695355773329,
+      "grad_norm": 13.375,
+      "learning_rate": 9.95043393638028e-07,
+      "loss": 0.6093,
+      "step": 13313
+    },
+    {
+      "epoch": 1.1345547507456328,
+      "grad_norm": 18.375,
+      "learning_rate": 9.948788928145262e-07,
+      "loss": 1.1307,
+      "step": 13314
+    },
+    {
+      "epoch": 1.1346399659139328,
+      "grad_norm": 14.125,
+      "learning_rate": 9.9471439660132e-07,
+      "loss": 0.5128,
+      "step": 13315
+    },
+    {
+      "epoch": 1.1347251810822327,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.945499050013813e-07,
+      "loss": 0.6648,
+      "step": 13316
+    },
+    {
+      "epoch": 1.1348103962505327,
+      "grad_norm": 15.0,
+      "learning_rate": 9.943854180176824e-07,
+      "loss": 0.5517,
+      "step": 13317
+    },
+    {
+      "epoch": 1.1348956114188327,
+      "grad_norm": 18.0,
+      "learning_rate": 9.942209356531962e-07,
+      "loss": 0.904,
+      "step": 13318
+    },
+    {
+      "epoch": 1.1349808265871326,
+      "grad_norm": 22.25,
+      "learning_rate": 9.940564579108946e-07,
+      "loss": 0.7452,
+      "step": 13319
+    },
+    {
+      "epoch": 1.1350660417554326,
+      "grad_norm": 17.5,
+      "learning_rate": 9.938919847937504e-07,
+      "loss": 0.6283,
+      "step": 13320
+    },
+    {
+      "epoch": 1.1351512569237325,
+      "grad_norm": 17.0,
+      "learning_rate": 9.937275163047353e-07,
+      "loss": 0.589,
+      "step": 13321
+    },
+    {
+      "epoch": 1.1352364720920325,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.93563052446822e-07,
+      "loss": 0.5597,
+      "step": 13322
+    },
+    {
+      "epoch": 1.1353216872603324,
+      "grad_norm": 15.75,
+      "learning_rate": 9.93398593222982e-07,
+      "loss": 0.6667,
+      "step": 13323
+    },
+    {
+      "epoch": 1.1354069024286324,
+      "grad_norm": 18.75,
+      "learning_rate": 9.932341386361878e-07,
+      "loss": 0.7381,
+      "step": 13324
+    },
+    {
+      "epoch": 1.1354921175969324,
+      "grad_norm": 18.375,
+      "learning_rate": 9.93069688689411e-07,
+      "loss": 0.7696,
+      "step": 13325
+    },
+    {
+      "epoch": 1.1355773327652323,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.929052433856227e-07,
+      "loss": 0.6384,
+      "step": 13326
+    },
+    {
+      "epoch": 1.1356625479335323,
+      "grad_norm": 15.125,
+      "learning_rate": 9.927408027277957e-07,
+      "loss": 0.8175,
+      "step": 13327
+    },
+    {
+      "epoch": 1.1357477631018322,
+      "grad_norm": 15.625,
+      "learning_rate": 9.925763667189009e-07,
+      "loss": 0.6364,
+      "step": 13328
+    },
+    {
+      "epoch": 1.1358329782701322,
+      "grad_norm": 24.0,
+      "learning_rate": 9.924119353619099e-07,
+      "loss": 0.9324,
+      "step": 13329
+    },
+    {
+      "epoch": 1.1359181934384321,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.922475086597944e-07,
+      "loss": 0.6118,
+      "step": 13330
+    },
+    {
+      "epoch": 1.136003408606732,
+      "grad_norm": 14.875,
+      "learning_rate": 9.920830866155253e-07,
+      "loss": 0.5916,
+      "step": 13331
+    },
+    {
+      "epoch": 1.136088623775032,
+      "grad_norm": 18.875,
+      "learning_rate": 9.919186692320748e-07,
+      "loss": 0.4644,
+      "step": 13332
+    },
+    {
+      "epoch": 1.136173838943332,
+      "grad_norm": 20.75,
+      "learning_rate": 9.91754256512413e-07,
+      "loss": 0.5872,
+      "step": 13333
+    },
+    {
+      "epoch": 1.136259054111632,
+      "grad_norm": 16.875,
+      "learning_rate": 9.915898484595115e-07,
+      "loss": 1.0306,
+      "step": 13334
+    },
+    {
+      "epoch": 1.136344269279932,
+      "grad_norm": 19.375,
+      "learning_rate": 9.914254450763414e-07,
+      "loss": 0.7943,
+      "step": 13335
+    },
+    {
+      "epoch": 1.1364294844482319,
+      "grad_norm": 41.0,
+      "learning_rate": 9.912610463658734e-07,
+      "loss": 1.1428,
+      "step": 13336
+    },
+    {
+      "epoch": 1.1365146996165318,
+      "grad_norm": 11.0,
+      "learning_rate": 9.910966523310786e-07,
+      "loss": 0.3458,
+      "step": 13337
+    },
+    {
+      "epoch": 1.1365999147848318,
+      "grad_norm": 24.875,
+      "learning_rate": 9.909322629749273e-07,
+      "loss": 0.7317,
+      "step": 13338
+    },
+    {
+      "epoch": 1.1366851299531318,
+      "grad_norm": 16.375,
+      "learning_rate": 9.90767878300391e-07,
+      "loss": 0.5539,
+      "step": 13339
+    },
+    {
+      "epoch": 1.1367703451214317,
+      "grad_norm": 13.25,
+      "learning_rate": 9.906034983104393e-07,
+      "loss": 0.5132,
+      "step": 13340
+    },
+    {
+      "epoch": 1.1368555602897317,
+      "grad_norm": 17.5,
+      "learning_rate": 9.90439123008044e-07,
+      "loss": 0.7343,
+      "step": 13341
+    },
+    {
+      "epoch": 1.1369407754580316,
+      "grad_norm": 19.125,
+      "learning_rate": 9.902747523961744e-07,
+      "loss": 0.5739,
+      "step": 13342
+    },
+    {
+      "epoch": 1.1370259906263316,
+      "grad_norm": 14.5,
+      "learning_rate": 9.901103864778012e-07,
+      "loss": 0.2709,
+      "step": 13343
+    },
+    {
+      "epoch": 1.1371112057946315,
+      "grad_norm": 14.375,
+      "learning_rate": 9.899460252558942e-07,
+      "loss": 0.2251,
+      "step": 13344
+    },
+    {
+      "epoch": 1.1371964209629315,
+      "grad_norm": 18.75,
+      "learning_rate": 9.897816687334247e-07,
+      "loss": 0.932,
+      "step": 13345
+    },
+    {
+      "epoch": 1.1372816361312315,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.896173169133615e-07,
+      "loss": 0.591,
+      "step": 13346
+    },
+    {
+      "epoch": 1.1373668512995314,
+      "grad_norm": 16.375,
+      "learning_rate": 9.894529697986758e-07,
+      "loss": 0.755,
+      "step": 13347
+    },
+    {
+      "epoch": 1.1374520664678314,
+      "grad_norm": 15.875,
+      "learning_rate": 9.892886273923363e-07,
+      "loss": 0.8592,
+      "step": 13348
+    },
+    {
+      "epoch": 1.1375372816361313,
+      "grad_norm": 21.75,
+      "learning_rate": 9.891242896973143e-07,
+      "loss": 0.5365,
+      "step": 13349
+    },
+    {
+      "epoch": 1.1376224968044313,
+      "grad_norm": 15.625,
+      "learning_rate": 9.889599567165782e-07,
+      "loss": 0.7757,
+      "step": 13350
+    },
+    {
+      "epoch": 1.1377077119727312,
+      "grad_norm": 18.375,
+      "learning_rate": 9.88795628453098e-07,
+      "loss": 0.7902,
+      "step": 13351
+    },
+    {
+      "epoch": 1.1377929271410312,
+      "grad_norm": 13.5,
+      "learning_rate": 9.88631304909844e-07,
+      "loss": 0.6232,
+      "step": 13352
+    },
+    {
+      "epoch": 1.1378781423093312,
+      "grad_norm": 19.625,
+      "learning_rate": 9.884669860897845e-07,
+      "loss": 0.5116,
+      "step": 13353
+    },
+    {
+      "epoch": 1.137963357477631,
+      "grad_norm": 22.25,
+      "learning_rate": 9.883026719958899e-07,
+      "loss": 1.0277,
+      "step": 13354
+    },
+    {
+      "epoch": 1.138048572645931,
+      "grad_norm": 21.75,
+      "learning_rate": 9.88138362631129e-07,
+      "loss": 0.7742,
+      "step": 13355
+    },
+    {
+      "epoch": 1.138133787814231,
+      "grad_norm": 17.125,
+      "learning_rate": 9.879740579984716e-07,
+      "loss": 0.7077,
+      "step": 13356
+    },
+    {
+      "epoch": 1.138219002982531,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.878097581008864e-07,
+      "loss": 0.4034,
+      "step": 13357
+    },
+    {
+      "epoch": 1.138304218150831,
+      "grad_norm": 12.3125,
+      "learning_rate": 9.876454629413425e-07,
+      "loss": 0.3284,
+      "step": 13358
+    },
+    {
+      "epoch": 1.138389433319131,
+      "grad_norm": 16.875,
+      "learning_rate": 9.874811725228086e-07,
+      "loss": 0.7156,
+      "step": 13359
+    },
+    {
+      "epoch": 1.1384746484874309,
+      "grad_norm": 16.0,
+      "learning_rate": 9.87316886848254e-07,
+      "loss": 0.7129,
+      "step": 13360
+    },
+    {
+      "epoch": 1.1385598636557308,
+      "grad_norm": 19.75,
+      "learning_rate": 9.871526059206473e-07,
+      "loss": 0.6555,
+      "step": 13361
+    },
+    {
+      "epoch": 1.1386450788240308,
+      "grad_norm": 11.75,
+      "learning_rate": 9.869883297429575e-07,
+      "loss": 0.249,
+      "step": 13362
+    },
+    {
+      "epoch": 1.1387302939923307,
+      "grad_norm": 12.25,
+      "learning_rate": 9.868240583181528e-07,
+      "loss": 0.447,
+      "step": 13363
+    },
+    {
+      "epoch": 1.1388155091606307,
+      "grad_norm": 13.125,
+      "learning_rate": 9.866597916492024e-07,
+      "loss": 0.4346,
+      "step": 13364
+    },
+    {
+      "epoch": 1.1389007243289306,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.864955297390746e-07,
+      "loss": 0.8351,
+      "step": 13365
+    },
+    {
+      "epoch": 1.1389859394972306,
+      "grad_norm": 17.5,
+      "learning_rate": 9.86331272590737e-07,
+      "loss": 0.7371,
+      "step": 13366
+    },
+    {
+      "epoch": 1.1390711546655305,
+      "grad_norm": 18.875,
+      "learning_rate": 9.861670202071586e-07,
+      "loss": 1.1333,
+      "step": 13367
+    },
+    {
+      "epoch": 1.1391563698338305,
+      "grad_norm": 18.875,
+      "learning_rate": 9.860027725913074e-07,
+      "loss": 0.8659,
+      "step": 13368
+    },
+    {
+      "epoch": 1.1392415850021305,
+      "grad_norm": 15.375,
+      "learning_rate": 9.858385297461518e-07,
+      "loss": 0.6953,
+      "step": 13369
+    },
+    {
+      "epoch": 1.1393268001704304,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.856742916746592e-07,
+      "loss": 0.3334,
+      "step": 13370
+    },
+    {
+      "epoch": 1.1394120153387304,
+      "grad_norm": 23.75,
+      "learning_rate": 9.855100583797983e-07,
+      "loss": 1.0496,
+      "step": 13371
+    },
+    {
+      "epoch": 1.1394972305070303,
+      "grad_norm": 16.875,
+      "learning_rate": 9.853458298645363e-07,
+      "loss": 0.6059,
+      "step": 13372
+    },
+    {
+      "epoch": 1.1395824456753303,
+      "grad_norm": 18.375,
+      "learning_rate": 9.85181606131842e-07,
+      "loss": 0.4935,
+      "step": 13373
+    },
+    {
+      "epoch": 1.1396676608436302,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.850173871846817e-07,
+      "loss": 0.5685,
+      "step": 13374
+    },
+    {
+      "epoch": 1.1397528760119302,
+      "grad_norm": 14.375,
+      "learning_rate": 9.848531730260238e-07,
+      "loss": 0.5107,
+      "step": 13375
+    },
+    {
+      "epoch": 1.1398380911802302,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.846889636588357e-07,
+      "loss": 0.4051,
+      "step": 13376
+    },
+    {
+      "epoch": 1.1399233063485301,
+      "grad_norm": 12.375,
+      "learning_rate": 9.845247590860848e-07,
+      "loss": 0.4834,
+      "step": 13377
+    },
+    {
+      "epoch": 1.14000852151683,
+      "grad_norm": 13.5,
+      "learning_rate": 9.843605593107383e-07,
+      "loss": 0.5974,
+      "step": 13378
+    },
+    {
+      "epoch": 1.14009373668513,
+      "grad_norm": 18.75,
+      "learning_rate": 9.84196364335764e-07,
+      "loss": 0.7099,
+      "step": 13379
+    },
+    {
+      "epoch": 1.14017895185343,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.840321741641281e-07,
+      "loss": 0.6355,
+      "step": 13380
+    },
+    {
+      "epoch": 1.14026416702173,
+      "grad_norm": 16.375,
+      "learning_rate": 9.83867988798799e-07,
+      "loss": 0.9901,
+      "step": 13381
+    },
+    {
+      "epoch": 1.14034938219003,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.83703808242743e-07,
+      "loss": 0.354,
+      "step": 13382
+    },
+    {
+      "epoch": 1.1404345973583299,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.835396324989263e-07,
+      "loss": 0.4641,
+      "step": 13383
+    },
+    {
+      "epoch": 1.1405198125266298,
+      "grad_norm": 12.5,
+      "learning_rate": 9.833754615703166e-07,
+      "loss": 0.3003,
+      "step": 13384
+    },
+    {
+      "epoch": 1.1406050276949298,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.832112954598804e-07,
+      "loss": 0.5548,
+      "step": 13385
+    },
+    {
+      "epoch": 1.1406902428632297,
+      "grad_norm": 9.75,
+      "learning_rate": 9.830471341705846e-07,
+      "loss": 0.2483,
+      "step": 13386
+    },
+    {
+      "epoch": 1.1407754580315297,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.828829777053952e-07,
+      "loss": 0.5521,
+      "step": 13387
+    },
+    {
+      "epoch": 1.1408606731998296,
+      "grad_norm": 19.75,
+      "learning_rate": 9.827188260672792e-07,
+      "loss": 0.6579,
+      "step": 13388
+    },
+    {
+      "epoch": 1.1409458883681296,
+      "grad_norm": 18.625,
+      "learning_rate": 9.82554679259203e-07,
+      "loss": 0.7812,
+      "step": 13389
+    },
+    {
+      "epoch": 1.1410311035364296,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.823905372841325e-07,
+      "loss": 0.3598,
+      "step": 13390
+    },
+    {
+      "epoch": 1.1411163187047295,
+      "grad_norm": 22.5,
+      "learning_rate": 9.82226400145034e-07,
+      "loss": 0.5544,
+      "step": 13391
+    },
+    {
+      "epoch": 1.1412015338730295,
+      "grad_norm": 21.25,
+      "learning_rate": 9.820622678448738e-07,
+      "loss": 0.7871,
+      "step": 13392
+    },
+    {
+      "epoch": 1.1412867490413294,
+      "grad_norm": 17.125,
+      "learning_rate": 9.818981403866177e-07,
+      "loss": 0.6372,
+      "step": 13393
+    },
+    {
+      "epoch": 1.1413719642096294,
+      "grad_norm": 16.125,
+      "learning_rate": 9.81734017773232e-07,
+      "loss": 0.6488,
+      "step": 13394
+    },
+    {
+      "epoch": 1.1414571793779293,
+      "grad_norm": 13.75,
+      "learning_rate": 9.81569900007682e-07,
+      "loss": 0.461,
+      "step": 13395
+    },
+    {
+      "epoch": 1.1415423945462293,
+      "grad_norm": 34.0,
+      "learning_rate": 9.81405787092934e-07,
+      "loss": 0.6902,
+      "step": 13396
+    },
+    {
+      "epoch": 1.1416276097145293,
+      "grad_norm": 14.625,
+      "learning_rate": 9.812416790319542e-07,
+      "loss": 0.272,
+      "step": 13397
+    },
+    {
+      "epoch": 1.1417128248828292,
+      "grad_norm": 21.75,
+      "learning_rate": 9.810775758277067e-07,
+      "loss": 0.5177,
+      "step": 13398
+    },
+    {
+      "epoch": 1.1417980400511292,
+      "grad_norm": 14.125,
+      "learning_rate": 9.809134774831583e-07,
+      "loss": 0.5513,
+      "step": 13399
+    },
+    {
+      "epoch": 1.1418832552194291,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.807493840012735e-07,
+      "loss": 0.5496,
+      "step": 13400
+    },
+    {
+      "epoch": 1.141968470387729,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.805852953850183e-07,
+      "loss": 0.327,
+      "step": 13401
+    },
+    {
+      "epoch": 1.142053685556029,
+      "grad_norm": 20.25,
+      "learning_rate": 9.804212116373577e-07,
+      "loss": 0.5069,
+      "step": 13402
+    },
+    {
+      "epoch": 1.142138900724329,
+      "grad_norm": 13.5,
+      "learning_rate": 9.802571327612568e-07,
+      "loss": 0.5047,
+      "step": 13403
+    },
+    {
+      "epoch": 1.142224115892629,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.800930587596807e-07,
+      "loss": 0.5939,
+      "step": 13404
+    },
+    {
+      "epoch": 1.142309331060929,
+      "grad_norm": 18.375,
+      "learning_rate": 9.799289896355955e-07,
+      "loss": 0.6073,
+      "step": 13405
+    },
+    {
+      "epoch": 1.1423945462292289,
+      "grad_norm": 17.375,
+      "learning_rate": 9.79764925391964e-07,
+      "loss": 0.8255,
+      "step": 13406
+    },
+    {
+      "epoch": 1.1424797613975288,
+      "grad_norm": 13.375,
+      "learning_rate": 9.796008660317522e-07,
+      "loss": 0.637,
+      "step": 13407
+    },
+    {
+      "epoch": 1.1425649765658288,
+      "grad_norm": 18.25,
+      "learning_rate": 9.794368115579244e-07,
+      "loss": 0.6506,
+      "step": 13408
+    },
+    {
+      "epoch": 1.1426501917341287,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.792727619734457e-07,
+      "loss": 0.7287,
+      "step": 13409
+    },
+    {
+      "epoch": 1.1427354069024287,
+      "grad_norm": 10.4375,
+      "learning_rate": 9.791087172812804e-07,
+      "loss": 0.2484,
+      "step": 13410
+    },
+    {
+      "epoch": 1.1428206220707287,
+      "grad_norm": 16.25,
+      "learning_rate": 9.78944677484393e-07,
+      "loss": 0.6125,
+      "step": 13411
+    },
+    {
+      "epoch": 1.1429058372390286,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.787806425857478e-07,
+      "loss": 0.2827,
+      "step": 13412
+    },
+    {
+      "epoch": 1.1429910524073286,
+      "grad_norm": 19.625,
+      "learning_rate": 9.786166125883092e-07,
+      "loss": 1.1507,
+      "step": 13413
+    },
+    {
+      "epoch": 1.1430762675756285,
+      "grad_norm": 16.25,
+      "learning_rate": 9.784525874950417e-07,
+      "loss": 0.6389,
+      "step": 13414
+    },
+    {
+      "epoch": 1.1431614827439285,
+      "grad_norm": 16.875,
+      "learning_rate": 9.782885673089084e-07,
+      "loss": 0.969,
+      "step": 13415
+    },
+    {
+      "epoch": 1.1432466979122284,
+      "grad_norm": 20.75,
+      "learning_rate": 9.781245520328745e-07,
+      "loss": 1.0094,
+      "step": 13416
+    },
+    {
+      "epoch": 1.1433319130805284,
+      "grad_norm": 42.0,
+      "learning_rate": 9.779605416699028e-07,
+      "loss": 1.3504,
+      "step": 13417
+    },
+    {
+      "epoch": 1.1434171282488284,
+      "grad_norm": 18.25,
+      "learning_rate": 9.777965362229578e-07,
+      "loss": 0.7894,
+      "step": 13418
+    },
+    {
+      "epoch": 1.1435023434171283,
+      "grad_norm": 16.375,
+      "learning_rate": 9.77632535695003e-07,
+      "loss": 0.5772,
+      "step": 13419
+    },
+    {
+      "epoch": 1.1435875585854283,
+      "grad_norm": 12.3125,
+      "learning_rate": 9.774685400890024e-07,
+      "loss": 0.4071,
+      "step": 13420
+    },
+    {
+      "epoch": 1.1436727737537282,
+      "grad_norm": 13.0,
+      "learning_rate": 9.773045494079192e-07,
+      "loss": 0.4914,
+      "step": 13421
+    },
+    {
+      "epoch": 1.1437579889220282,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.771405636547175e-07,
+      "loss": 0.5116,
+      "step": 13422
+    },
+    {
+      "epoch": 1.1438432040903281,
+      "grad_norm": 15.625,
+      "learning_rate": 9.769765828323597e-07,
+      "loss": 0.535,
+      "step": 13423
+    },
+    {
+      "epoch": 1.143928419258628,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.768126069438097e-07,
+      "loss": 0.7335,
+      "step": 13424
+    },
+    {
+      "epoch": 1.144013634426928,
+      "grad_norm": 18.0,
+      "learning_rate": 9.766486359920304e-07,
+      "loss": 1.0532,
+      "step": 13425
+    },
+    {
+      "epoch": 1.144098849595228,
+      "grad_norm": 14.125,
+      "learning_rate": 9.764846699799854e-07,
+      "loss": 0.7496,
+      "step": 13426
+    },
+    {
+      "epoch": 1.144184064763528,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.76320708910637e-07,
+      "loss": 0.5166,
+      "step": 13427
+    },
+    {
+      "epoch": 1.144269279931828,
+      "grad_norm": 16.375,
+      "learning_rate": 9.76156752786949e-07,
+      "loss": 0.3215,
+      "step": 13428
+    },
+    {
+      "epoch": 1.1443544951001279,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.759928016118838e-07,
+      "loss": 0.8494,
+      "step": 13429
+    },
+    {
+      "epoch": 1.1444397102684278,
+      "grad_norm": 16.625,
+      "learning_rate": 9.758288553884044e-07,
+      "loss": 0.6422,
+      "step": 13430
+    },
+    {
+      "epoch": 1.1445249254367278,
+      "grad_norm": 19.25,
+      "learning_rate": 9.756649141194732e-07,
+      "loss": 0.5523,
+      "step": 13431
+    },
+    {
+      "epoch": 1.1446101406050277,
+      "grad_norm": 23.875,
+      "learning_rate": 9.755009778080526e-07,
+      "loss": 0.771,
+      "step": 13432
+    },
+    {
+      "epoch": 1.1446953557733277,
+      "grad_norm": 14.25,
+      "learning_rate": 9.753370464571057e-07,
+      "loss": 0.5322,
+      "step": 13433
+    },
+    {
+      "epoch": 1.1447805709416277,
+      "grad_norm": 13.625,
+      "learning_rate": 9.75173120069594e-07,
+      "loss": 0.652,
+      "step": 13434
+    },
+    {
+      "epoch": 1.1448657861099276,
+      "grad_norm": 18.25,
+      "learning_rate": 9.75009198648481e-07,
+      "loss": 0.4819,
+      "step": 13435
+    },
+    {
+      "epoch": 1.1449510012782276,
+      "grad_norm": 16.625,
+      "learning_rate": 9.74845282196728e-07,
+      "loss": 0.5884,
+      "step": 13436
+    },
+    {
+      "epoch": 1.1450362164465275,
+      "grad_norm": 13.25,
+      "learning_rate": 9.746813707172973e-07,
+      "loss": 0.5908,
+      "step": 13437
+    },
+    {
+      "epoch": 1.1451214316148275,
+      "grad_norm": 20.125,
+      "learning_rate": 9.745174642131518e-07,
+      "loss": 0.5963,
+      "step": 13438
+    },
+    {
+      "epoch": 1.1452066467831274,
+      "grad_norm": 19.0,
+      "learning_rate": 9.74353562687252e-07,
+      "loss": 0.6282,
+      "step": 13439
+    },
+    {
+      "epoch": 1.1452918619514274,
+      "grad_norm": 21.875,
+      "learning_rate": 9.741896661425609e-07,
+      "loss": 0.7913,
+      "step": 13440
+    },
+    {
+      "epoch": 1.1453770771197274,
+      "grad_norm": 23.375,
+      "learning_rate": 9.740257745820398e-07,
+      "loss": 1.0568,
+      "step": 13441
+    },
+    {
+      "epoch": 1.1454622922880273,
+      "grad_norm": 18.125,
+      "learning_rate": 9.7386188800865e-07,
+      "loss": 0.8974,
+      "step": 13442
+    },
+    {
+      "epoch": 1.1455475074563273,
+      "grad_norm": 18.5,
+      "learning_rate": 9.736980064253542e-07,
+      "loss": 0.8557,
+      "step": 13443
+    },
+    {
+      "epoch": 1.1456327226246272,
+      "grad_norm": 17.25,
+      "learning_rate": 9.735341298351128e-07,
+      "loss": 0.5335,
+      "step": 13444
+    },
+    {
+      "epoch": 1.1457179377929272,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.733702582408879e-07,
+      "loss": 0.692,
+      "step": 13445
+    },
+    {
+      "epoch": 1.1458031529612271,
+      "grad_norm": 19.5,
+      "learning_rate": 9.73206391645641e-07,
+      "loss": 0.3934,
+      "step": 13446
+    },
+    {
+      "epoch": 1.145888368129527,
+      "grad_norm": 16.375,
+      "learning_rate": 9.730425300523324e-07,
+      "loss": 0.8496,
+      "step": 13447
+    },
+    {
+      "epoch": 1.145973583297827,
+      "grad_norm": 16.875,
+      "learning_rate": 9.72878673463924e-07,
+      "loss": 0.6115,
+      "step": 13448
+    },
+    {
+      "epoch": 1.146058798466127,
+      "grad_norm": 17.125,
+      "learning_rate": 9.727148218833763e-07,
+      "loss": 0.6767,
+      "step": 13449
+    },
+    {
+      "epoch": 1.146144013634427,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.72550975313651e-07,
+      "loss": 0.9049,
+      "step": 13450
+    },
+    {
+      "epoch": 1.146229228802727,
+      "grad_norm": 12.0,
+      "learning_rate": 9.723871337577085e-07,
+      "loss": 0.4095,
+      "step": 13451
+    },
+    {
+      "epoch": 1.1463144439710269,
+      "grad_norm": 18.25,
+      "learning_rate": 9.722232972185094e-07,
+      "loss": 0.6787,
+      "step": 13452
+    },
+    {
+      "epoch": 1.1463996591393268,
+      "grad_norm": 20.625,
+      "learning_rate": 9.720594656990148e-07,
+      "loss": 0.6588,
+      "step": 13453
+    },
+    {
+      "epoch": 1.1464848743076268,
+      "grad_norm": 21.125,
+      "learning_rate": 9.718956392021856e-07,
+      "loss": 0.8133,
+      "step": 13454
+    },
+    {
+      "epoch": 1.1465700894759268,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.717318177309812e-07,
+      "loss": 0.7094,
+      "step": 13455
+    },
+    {
+      "epoch": 1.1466553046442267,
+      "grad_norm": 15.75,
+      "learning_rate": 9.71568001288363e-07,
+      "loss": 0.7029,
+      "step": 13456
+    },
+    {
+      "epoch": 1.1467405198125267,
+      "grad_norm": 19.625,
+      "learning_rate": 9.714041898772908e-07,
+      "loss": 0.396,
+      "step": 13457
+    },
+    {
+      "epoch": 1.1468257349808266,
+      "grad_norm": 14.25,
+      "learning_rate": 9.712403835007252e-07,
+      "loss": 0.3976,
+      "step": 13458
+    },
+    {
+      "epoch": 1.1469109501491266,
+      "grad_norm": 24.0,
+      "learning_rate": 9.71076582161626e-07,
+      "loss": 0.598,
+      "step": 13459
+    },
+    {
+      "epoch": 1.1469961653174265,
+      "grad_norm": 16.75,
+      "learning_rate": 9.709127858629536e-07,
+      "loss": 0.4045,
+      "step": 13460
+    },
+    {
+      "epoch": 1.1470813804857265,
+      "grad_norm": 17.75,
+      "learning_rate": 9.707489946076675e-07,
+      "loss": 0.4125,
+      "step": 13461
+    },
+    {
+      "epoch": 1.1471665956540265,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.705852083987285e-07,
+      "loss": 0.5206,
+      "step": 13462
+    },
+    {
+      "epoch": 1.1472518108223264,
+      "grad_norm": 16.125,
+      "learning_rate": 9.704214272390956e-07,
+      "loss": 0.5586,
+      "step": 13463
+    },
+    {
+      "epoch": 1.1473370259906264,
+      "grad_norm": 17.5,
+      "learning_rate": 9.702576511317283e-07,
+      "loss": 0.682,
+      "step": 13464
+    },
+    {
+      "epoch": 1.1474222411589263,
+      "grad_norm": 17.625,
+      "learning_rate": 9.700938800795867e-07,
+      "loss": 0.8917,
+      "step": 13465
+    },
+    {
+      "epoch": 1.1475074563272263,
+      "grad_norm": 19.875,
+      "learning_rate": 9.699301140856304e-07,
+      "loss": 0.5282,
+      "step": 13466
+    },
+    {
+      "epoch": 1.1475926714955262,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.69766353152818e-07,
+      "loss": 0.489,
+      "step": 13467
+    },
+    {
+      "epoch": 1.1476778866638262,
+      "grad_norm": 12.875,
+      "learning_rate": 9.696025972841099e-07,
+      "loss": 0.338,
+      "step": 13468
+    },
+    {
+      "epoch": 1.1477631018321262,
+      "grad_norm": 19.0,
+      "learning_rate": 9.694388464824645e-07,
+      "loss": 0.7455,
+      "step": 13469
+    },
+    {
+      "epoch": 1.147848317000426,
+      "grad_norm": 12.375,
+      "learning_rate": 9.692751007508416e-07,
+      "loss": 0.4396,
+      "step": 13470
+    },
+    {
+      "epoch": 1.147933532168726,
+      "grad_norm": 16.25,
+      "learning_rate": 9.691113600921998e-07,
+      "loss": 0.8635,
+      "step": 13471
+    },
+    {
+      "epoch": 1.148018747337026,
+      "grad_norm": 18.125,
+      "learning_rate": 9.68947624509498e-07,
+      "loss": 0.4474,
+      "step": 13472
+    },
+    {
+      "epoch": 1.148103962505326,
+      "grad_norm": 18.75,
+      "learning_rate": 9.687838940056956e-07,
+      "loss": 0.7303,
+      "step": 13473
+    },
+    {
+      "epoch": 1.148189177673626,
+      "grad_norm": 21.0,
+      "learning_rate": 9.686201685837504e-07,
+      "loss": 0.8052,
+      "step": 13474
+    },
+    {
+      "epoch": 1.148274392841926,
+      "grad_norm": 17.0,
+      "learning_rate": 9.684564482466222e-07,
+      "loss": 0.7654,
+      "step": 13475
+    },
+    {
+      "epoch": 1.1483596080102259,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.682927329972689e-07,
+      "loss": 0.4802,
+      "step": 13476
+    },
+    {
+      "epoch": 1.1484448231785258,
+      "grad_norm": 17.125,
+      "learning_rate": 9.681290228386494e-07,
+      "loss": 0.7754,
+      "step": 13477
+    },
+    {
+      "epoch": 1.1485300383468258,
+      "grad_norm": 12.75,
+      "learning_rate": 9.679653177737216e-07,
+      "loss": 0.362,
+      "step": 13478
+    },
+    {
+      "epoch": 1.1486152535151257,
+      "grad_norm": 14.625,
+      "learning_rate": 9.678016178054446e-07,
+      "loss": 0.7624,
+      "step": 13479
+    },
+    {
+      "epoch": 1.1487004686834257,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.676379229367759e-07,
+      "loss": 0.3636,
+      "step": 13480
+    },
+    {
+      "epoch": 1.1487856838517256,
+      "grad_norm": 16.25,
+      "learning_rate": 9.674742331706738e-07,
+      "loss": 0.5063,
+      "step": 13481
+    },
+    {
+      "epoch": 1.1488708990200256,
+      "grad_norm": 18.5,
+      "learning_rate": 9.673105485100963e-07,
+      "loss": 0.442,
+      "step": 13482
+    },
+    {
+      "epoch": 1.1489561141883256,
+      "grad_norm": 12.625,
+      "learning_rate": 9.671468689580017e-07,
+      "loss": 0.3088,
+      "step": 13483
+    },
+    {
+      "epoch": 1.1490413293566255,
+      "grad_norm": 14.125,
+      "learning_rate": 9.669831945173472e-07,
+      "loss": 0.5611,
+      "step": 13484
+    },
+    {
+      "epoch": 1.1491265445249255,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.668195251910915e-07,
+      "loss": 0.5355,
+      "step": 13485
+    },
+    {
+      "epoch": 1.1492117596932254,
+      "grad_norm": 13.0,
+      "learning_rate": 9.66655860982191e-07,
+      "loss": 0.6047,
+      "step": 13486
+    },
+    {
+      "epoch": 1.1492969748615254,
+      "grad_norm": 17.875,
+      "learning_rate": 9.664922018936051e-07,
+      "loss": 0.7379,
+      "step": 13487
+    },
+    {
+      "epoch": 1.1493821900298253,
+      "grad_norm": 14.875,
+      "learning_rate": 9.663285479282895e-07,
+      "loss": 0.7082,
+      "step": 13488
+    },
+    {
+      "epoch": 1.1494674051981253,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.661648990892023e-07,
+      "loss": 0.514,
+      "step": 13489
+    },
+    {
+      "epoch": 1.1495526203664252,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.660012553793009e-07,
+      "loss": 0.5388,
+      "step": 13490
+    },
+    {
+      "epoch": 1.1496378355347252,
+      "grad_norm": 16.75,
+      "learning_rate": 9.658376168015422e-07,
+      "loss": 0.6193,
+      "step": 13491
+    },
+    {
+      "epoch": 1.1497230507030252,
+      "grad_norm": 16.125,
+      "learning_rate": 9.656739833588837e-07,
+      "loss": 0.8265,
+      "step": 13492
+    },
+    {
+      "epoch": 1.1498082658713251,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.655103550542822e-07,
+      "loss": 0.3289,
+      "step": 13493
+    },
+    {
+      "epoch": 1.149893481039625,
+      "grad_norm": 20.25,
+      "learning_rate": 9.653467318906948e-07,
+      "loss": 0.8454,
+      "step": 13494
+    },
+    {
+      "epoch": 1.149978696207925,
+      "grad_norm": 14.75,
+      "learning_rate": 9.651831138710784e-07,
+      "loss": 0.3903,
+      "step": 13495
+    },
+    {
+      "epoch": 1.150063911376225,
+      "grad_norm": 25.25,
+      "learning_rate": 9.650195009983894e-07,
+      "loss": 0.6945,
+      "step": 13496
+    },
+    {
+      "epoch": 1.150149126544525,
+      "grad_norm": 24.625,
+      "learning_rate": 9.648558932755841e-07,
+      "loss": 0.9499,
+      "step": 13497
+    },
+    {
+      "epoch": 1.150234341712825,
+      "grad_norm": 24.625,
+      "learning_rate": 9.6469229070562e-07,
+      "loss": 0.6027,
+      "step": 13498
+    },
+    {
+      "epoch": 1.1503195568811249,
+      "grad_norm": 37.0,
+      "learning_rate": 9.645286932914526e-07,
+      "loss": 0.5165,
+      "step": 13499
+    },
+    {
+      "epoch": 1.1504047720494248,
+      "grad_norm": 24.25,
+      "learning_rate": 9.643651010360393e-07,
+      "loss": 0.8691,
+      "step": 13500
+    },
+    {
+      "epoch": 1.1504899872177248,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.642015139423355e-07,
+      "loss": 0.6577,
+      "step": 13501
+    },
+    {
+      "epoch": 1.1505752023860247,
+      "grad_norm": 14.875,
+      "learning_rate": 9.640379320132982e-07,
+      "loss": 0.3295,
+      "step": 13502
+    },
+    {
+      "epoch": 1.1506604175543247,
+      "grad_norm": 15.0,
+      "learning_rate": 9.63874355251883e-07,
+      "loss": 0.6709,
+      "step": 13503
+    },
+    {
+      "epoch": 1.1507456327226246,
+      "grad_norm": 21.125,
+      "learning_rate": 9.637107836610455e-07,
+      "loss": 0.7441,
+      "step": 13504
+    },
+    {
+      "epoch": 1.1508308478909246,
+      "grad_norm": 20.0,
+      "learning_rate": 9.63547217243742e-07,
+      "loss": 0.7832,
+      "step": 13505
+    },
+    {
+      "epoch": 1.1509160630592246,
+      "grad_norm": 17.25,
+      "learning_rate": 9.633836560029286e-07,
+      "loss": 0.7449,
+      "step": 13506
+    },
+    {
+      "epoch": 1.1510012782275245,
+      "grad_norm": 32.5,
+      "learning_rate": 9.632200999415607e-07,
+      "loss": 0.544,
+      "step": 13507
+    },
+    {
+      "epoch": 1.1510864933958245,
+      "grad_norm": 23.25,
+      "learning_rate": 9.630565490625938e-07,
+      "loss": 0.9638,
+      "step": 13508
+    },
+    {
+      "epoch": 1.1511717085641244,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.628930033689837e-07,
+      "loss": 0.3941,
+      "step": 13509
+    },
+    {
+      "epoch": 1.1512569237324244,
+      "grad_norm": 17.625,
+      "learning_rate": 9.627294628636857e-07,
+      "loss": 0.7794,
+      "step": 13510
+    },
+    {
+      "epoch": 1.1513421389007243,
+      "grad_norm": 20.125,
+      "learning_rate": 9.625659275496558e-07,
+      "loss": 0.8737,
+      "step": 13511
+    },
+    {
+      "epoch": 1.1514273540690243,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.62402397429848e-07,
+      "loss": 0.2886,
+      "step": 13512
+    },
+    {
+      "epoch": 1.1515125692373243,
+      "grad_norm": 17.125,
+      "learning_rate": 9.622388725072182e-07,
+      "loss": 0.6164,
+      "step": 13513
+    },
+    {
+      "epoch": 1.1515977844056242,
+      "grad_norm": 14.625,
+      "learning_rate": 9.620753527847212e-07,
+      "loss": 0.5109,
+      "step": 13514
+    },
+    {
+      "epoch": 1.1516829995739242,
+      "grad_norm": 26.125,
+      "learning_rate": 9.619118382653123e-07,
+      "loss": 0.93,
+      "step": 13515
+    },
+    {
+      "epoch": 1.1517682147422241,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.61748328951946e-07,
+      "loss": 0.633,
+      "step": 13516
+    },
+    {
+      "epoch": 1.151853429910524,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.615848248475774e-07,
+      "loss": 0.7057,
+      "step": 13517
+    },
+    {
+      "epoch": 1.151938645078824,
+      "grad_norm": 19.75,
+      "learning_rate": 9.614213259551607e-07,
+      "loss": 0.7244,
+      "step": 13518
+    },
+    {
+      "epoch": 1.152023860247124,
+      "grad_norm": 11.5625,
+      "learning_rate": 9.612578322776516e-07,
+      "loss": 0.2486,
+      "step": 13519
+    },
+    {
+      "epoch": 1.152109075415424,
+      "grad_norm": 15.375,
+      "learning_rate": 9.610943438180033e-07,
+      "loss": 0.956,
+      "step": 13520
+    },
+    {
+      "epoch": 1.152194290583724,
+      "grad_norm": 17.0,
+      "learning_rate": 9.609308605791705e-07,
+      "loss": 0.6803,
+      "step": 13521
+    },
+    {
+      "epoch": 1.1522795057520239,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.607673825641083e-07,
+      "loss": 0.4474,
+      "step": 13522
+    },
+    {
+      "epoch": 1.1523647209203238,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.606039097757698e-07,
+      "loss": 0.3392,
+      "step": 13523
+    },
+    {
+      "epoch": 1.1524499360886238,
+      "grad_norm": 16.875,
+      "learning_rate": 9.6044044221711e-07,
+      "loss": 0.8876,
+      "step": 13524
+    },
+    {
+      "epoch": 1.1525351512569237,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.602769798910824e-07,
+      "loss": 0.7512,
+      "step": 13525
+    },
+    {
+      "epoch": 1.1526203664252237,
+      "grad_norm": 18.5,
+      "learning_rate": 9.601135228006412e-07,
+      "loss": 0.3645,
+      "step": 13526
+    },
+    {
+      "epoch": 1.1527055815935237,
+      "grad_norm": 15.5,
+      "learning_rate": 9.599500709487405e-07,
+      "loss": 0.7833,
+      "step": 13527
+    },
+    {
+      "epoch": 1.1527907967618236,
+      "grad_norm": 20.625,
+      "learning_rate": 9.597866243383334e-07,
+      "loss": 0.6381,
+      "step": 13528
+    },
+    {
+      "epoch": 1.1528760119301236,
+      "grad_norm": 12.875,
+      "learning_rate": 9.596231829723736e-07,
+      "loss": 0.7029,
+      "step": 13529
+    },
+    {
+      "epoch": 1.1529612270984235,
+      "grad_norm": 22.75,
+      "learning_rate": 9.594597468538151e-07,
+      "loss": 0.9073,
+      "step": 13530
+    },
+    {
+      "epoch": 1.1530464422667235,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.59296315985611e-07,
+      "loss": 0.5385,
+      "step": 13531
+    },
+    {
+      "epoch": 1.1531316574350234,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.59132890370715e-07,
+      "loss": 0.505,
+      "step": 13532
+    },
+    {
+      "epoch": 1.1532168726033234,
+      "grad_norm": 14.625,
+      "learning_rate": 9.589694700120798e-07,
+      "loss": 0.7409,
+      "step": 13533
+    },
+    {
+      "epoch": 1.1533020877716234,
+      "grad_norm": 14.625,
+      "learning_rate": 9.588060549126591e-07,
+      "loss": 0.5704,
+      "step": 13534
+    },
+    {
+      "epoch": 1.1533873029399233,
+      "grad_norm": 10.25,
+      "learning_rate": 9.586426450754063e-07,
+      "loss": 0.2538,
+      "step": 13535
+    },
+    {
+      "epoch": 1.1534725181082233,
+      "grad_norm": 13.0,
+      "learning_rate": 9.584792405032734e-07,
+      "loss": 0.3588,
+      "step": 13536
+    },
+    {
+      "epoch": 1.1535577332765232,
+      "grad_norm": 17.5,
+      "learning_rate": 9.583158411992139e-07,
+      "loss": 0.8585,
+      "step": 13537
+    },
+    {
+      "epoch": 1.1536429484448232,
+      "grad_norm": 18.125,
+      "learning_rate": 9.581524471661802e-07,
+      "loss": 0.6913,
+      "step": 13538
+    },
+    {
+      "epoch": 1.1537281636131231,
+      "grad_norm": 16.875,
+      "learning_rate": 9.579890584071254e-07,
+      "loss": 0.6071,
+      "step": 13539
+    },
+    {
+      "epoch": 1.153813378781423,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.578256749250018e-07,
+      "loss": 0.7357,
+      "step": 13540
+    },
+    {
+      "epoch": 1.153898593949723,
+      "grad_norm": 16.625,
+      "learning_rate": 9.576622967227626e-07,
+      "loss": 0.7746,
+      "step": 13541
+    },
+    {
+      "epoch": 1.153983809118023,
+      "grad_norm": 18.0,
+      "learning_rate": 9.57498923803359e-07,
+      "loss": 0.7199,
+      "step": 13542
+    },
+    {
+      "epoch": 1.154069024286323,
+      "grad_norm": 13.375,
+      "learning_rate": 9.573355561697443e-07,
+      "loss": 0.4421,
+      "step": 13543
+    },
+    {
+      "epoch": 1.154154239454623,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.57172193824871e-07,
+      "loss": 0.6116,
+      "step": 13544
+    },
+    {
+      "epoch": 1.1542394546229229,
+      "grad_norm": 10.875,
+      "learning_rate": 9.5700883677169e-07,
+      "loss": 0.3635,
+      "step": 13545
+    },
+    {
+      "epoch": 1.1543246697912228,
+      "grad_norm": 20.625,
+      "learning_rate": 9.568454850131536e-07,
+      "loss": 0.7502,
+      "step": 13546
+    },
+    {
+      "epoch": 1.1544098849595228,
+      "grad_norm": 9.6875,
+      "learning_rate": 9.566821385522145e-07,
+      "loss": 0.2851,
+      "step": 13547
+    },
+    {
+      "epoch": 1.1544951001278227,
+      "grad_norm": 18.5,
+      "learning_rate": 9.56518797391824e-07,
+      "loss": 0.8178,
+      "step": 13548
+    },
+    {
+      "epoch": 1.1545803152961227,
+      "grad_norm": 20.75,
+      "learning_rate": 9.56355461534934e-07,
+      "loss": 0.7996,
+      "step": 13549
+    },
+    {
+      "epoch": 1.1546655304644227,
+      "grad_norm": 22.75,
+      "learning_rate": 9.56192130984496e-07,
+      "loss": 0.5011,
+      "step": 13550
+    },
+    {
+      "epoch": 1.1547507456327226,
+      "grad_norm": 16.25,
+      "learning_rate": 9.560288057434618e-07,
+      "loss": 0.7961,
+      "step": 13551
+    },
+    {
+      "epoch": 1.1548359608010226,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.558654858147829e-07,
+      "loss": 0.5697,
+      "step": 13552
+    },
+    {
+      "epoch": 1.1549211759693225,
+      "grad_norm": 18.875,
+      "learning_rate": 9.5570217120141e-07,
+      "loss": 0.5314,
+      "step": 13553
+    },
+    {
+      "epoch": 1.1550063911376225,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.55538861906295e-07,
+      "loss": 0.6523,
+      "step": 13554
+    },
+    {
+      "epoch": 1.1550916063059224,
+      "grad_norm": 19.5,
+      "learning_rate": 9.553755579323888e-07,
+      "loss": 0.799,
+      "step": 13555
+    },
+    {
+      "epoch": 1.1551768214742224,
+      "grad_norm": 10.0,
+      "learning_rate": 9.552122592826427e-07,
+      "loss": 0.2523,
+      "step": 13556
+    },
+    {
+      "epoch": 1.1552620366425224,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.55048965960007e-07,
+      "loss": 0.5398,
+      "step": 13557
+    },
+    {
+      "epoch": 1.1553472518108223,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.548856779674338e-07,
+      "loss": 0.4667,
+      "step": 13558
+    },
+    {
+      "epoch": 1.1554324669791223,
+      "grad_norm": 15.125,
+      "learning_rate": 9.54722395307873e-07,
+      "loss": 0.5802,
+      "step": 13559
+    },
+    {
+      "epoch": 1.1555176821474222,
+      "grad_norm": 20.125,
+      "learning_rate": 9.545591179842755e-07,
+      "loss": 0.6534,
+      "step": 13560
+    },
+    {
+      "epoch": 1.1556028973157222,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.543958459995914e-07,
+      "loss": 0.3055,
+      "step": 13561
+    },
+    {
+      "epoch": 1.1556881124840221,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.542325793567718e-07,
+      "loss": 0.6445,
+      "step": 13562
+    },
+    {
+      "epoch": 1.155773327652322,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.540693180587667e-07,
+      "loss": 0.361,
+      "step": 13563
+    },
+    {
+      "epoch": 1.155858542820622,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.53906062108527e-07,
+      "loss": 0.3159,
+      "step": 13564
+    },
+    {
+      "epoch": 1.155943757988922,
+      "grad_norm": 23.5,
+      "learning_rate": 9.53742811509002e-07,
+      "loss": 0.8819,
+      "step": 13565
+    },
+    {
+      "epoch": 1.156028973157222,
+      "grad_norm": 13.25,
+      "learning_rate": 9.535795662631428e-07,
+      "loss": 0.5447,
+      "step": 13566
+    },
+    {
+      "epoch": 1.156114188325522,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.534163263738985e-07,
+      "loss": 0.3405,
+      "step": 13567
+    },
+    {
+      "epoch": 1.1561994034938219,
+      "grad_norm": 15.375,
+      "learning_rate": 9.532530918442201e-07,
+      "loss": 0.6456,
+      "step": 13568
+    },
+    {
+      "epoch": 1.1562846186621218,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.530898626770564e-07,
+      "loss": 0.5433,
+      "step": 13569
+    },
+    {
+      "epoch": 1.1563698338304218,
+      "grad_norm": 15.125,
+      "learning_rate": 9.529266388753572e-07,
+      "loss": 0.6616,
+      "step": 13570
+    },
+    {
+      "epoch": 1.1564550489987218,
+      "grad_norm": 49.0,
+      "learning_rate": 9.527634204420726e-07,
+      "loss": 0.9666,
+      "step": 13571
+    },
+    {
+      "epoch": 1.1565402641670217,
+      "grad_norm": 11.6875,
+      "learning_rate": 9.526002073801516e-07,
+      "loss": 0.4294,
+      "step": 13572
+    },
+    {
+      "epoch": 1.1566254793353217,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.524369996925442e-07,
+      "loss": 0.5046,
+      "step": 13573
+    },
+    {
+      "epoch": 1.1567106945036216,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.522737973821994e-07,
+      "loss": 0.3816,
+      "step": 13574
+    },
+    {
+      "epoch": 1.1567959096719216,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.521106004520662e-07,
+      "loss": 0.7074,
+      "step": 13575
+    },
+    {
+      "epoch": 1.1568811248402215,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.519474089050945e-07,
+      "loss": 0.5165,
+      "step": 13576
+    },
+    {
+      "epoch": 1.1569663400085215,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.517842227442327e-07,
+      "loss": 0.8808,
+      "step": 13577
+    },
+    {
+      "epoch": 1.1570515551768215,
+      "grad_norm": 16.125,
+      "learning_rate": 9.516210419724295e-07,
+      "loss": 0.5519,
+      "step": 13578
+    },
+    {
+      "epoch": 1.1571367703451214,
+      "grad_norm": 16.625,
+      "learning_rate": 9.514578665926345e-07,
+      "loss": 0.7716,
+      "step": 13579
+    },
+    {
+      "epoch": 1.1572219855134214,
+      "grad_norm": 20.25,
+      "learning_rate": 9.512946966077957e-07,
+      "loss": 0.6582,
+      "step": 13580
+    },
+    {
+      "epoch": 1.1573072006817213,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.511315320208623e-07,
+      "loss": 0.5491,
+      "step": 13581
+    },
+    {
+      "epoch": 1.1573924158500213,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.509683728347825e-07,
+      "loss": 0.326,
+      "step": 13582
+    },
+    {
+      "epoch": 1.1574776310183212,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.508052190525052e-07,
+      "loss": 0.5705,
+      "step": 13583
+    },
+    {
+      "epoch": 1.1575628461866212,
+      "grad_norm": 13.25,
+      "learning_rate": 9.506420706769787e-07,
+      "loss": 0.627,
+      "step": 13584
+    },
+    {
+      "epoch": 1.1576480613549212,
+      "grad_norm": 13.875,
+      "learning_rate": 9.504789277111504e-07,
+      "loss": 0.4842,
+      "step": 13585
+    },
+    {
+      "epoch": 1.157733276523221,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.503157901579693e-07,
+      "loss": 0.7308,
+      "step": 13586
+    },
+    {
+      "epoch": 1.157818491691521,
+      "grad_norm": 13.25,
+      "learning_rate": 9.501526580203832e-07,
+      "loss": 0.5224,
+      "step": 13587
+    },
+    {
+      "epoch": 1.157903706859821,
+      "grad_norm": 13.875,
+      "learning_rate": 9.499895313013401e-07,
+      "loss": 0.6051,
+      "step": 13588
+    },
+    {
+      "epoch": 1.157988922028121,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.498264100037879e-07,
+      "loss": 0.552,
+      "step": 13589
+    },
+    {
+      "epoch": 1.158074137196421,
+      "grad_norm": 22.125,
+      "learning_rate": 9.496632941306741e-07,
+      "loss": 0.5993,
+      "step": 13590
+    },
+    {
+      "epoch": 1.158159352364721,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.495001836849468e-07,
+      "loss": 0.4013,
+      "step": 13591
+    },
+    {
+      "epoch": 1.1582445675330209,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.493370786695538e-07,
+      "loss": 0.5326,
+      "step": 13592
+    },
+    {
+      "epoch": 1.1583297827013208,
+      "grad_norm": 15.0,
+      "learning_rate": 9.491739790874414e-07,
+      "loss": 0.598,
+      "step": 13593
+    },
+    {
+      "epoch": 1.1584149978696208,
+      "grad_norm": 16.5,
+      "learning_rate": 9.49010884941558e-07,
+      "loss": 0.6714,
+      "step": 13594
+    },
+    {
+      "epoch": 1.1585002130379207,
+      "grad_norm": 11.0,
+      "learning_rate": 9.488477962348506e-07,
+      "loss": 0.3743,
+      "step": 13595
+    },
+    {
+      "epoch": 1.1585854282062207,
+      "grad_norm": 20.5,
+      "learning_rate": 9.486847129702661e-07,
+      "loss": 0.8852,
+      "step": 13596
+    },
+    {
+      "epoch": 1.1586706433745206,
+      "grad_norm": 10.875,
+      "learning_rate": 9.48521635150752e-07,
+      "loss": 0.3323,
+      "step": 13597
+    },
+    {
+      "epoch": 1.1587558585428206,
+      "grad_norm": 17.25,
+      "learning_rate": 9.483585627792552e-07,
+      "loss": 0.9642,
+      "step": 13598
+    },
+    {
+      "epoch": 1.1588410737111206,
+      "grad_norm": 29.0,
+      "learning_rate": 9.481954958587223e-07,
+      "loss": 0.4036,
+      "step": 13599
+    },
+    {
+      "epoch": 1.1589262888794205,
+      "grad_norm": 16.0,
+      "learning_rate": 9.480324343921006e-07,
+      "loss": 0.614,
+      "step": 13600
+    },
+    {
+      "epoch": 1.1590115040477205,
+      "grad_norm": 21.125,
+      "learning_rate": 9.478693783823364e-07,
+      "loss": 0.7071,
+      "step": 13601
+    },
+    {
+      "epoch": 1.1590967192160204,
+      "grad_norm": 15.75,
+      "learning_rate": 9.47706327832376e-07,
+      "loss": 0.5266,
+      "step": 13602
+    },
+    {
+      "epoch": 1.1591819343843204,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.475432827451666e-07,
+      "loss": 0.5305,
+      "step": 13603
+    },
+    {
+      "epoch": 1.1592671495526203,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.473802431236538e-07,
+      "loss": 0.3812,
+      "step": 13604
+    },
+    {
+      "epoch": 1.1593523647209203,
+      "grad_norm": 13.875,
+      "learning_rate": 9.472172089707847e-07,
+      "loss": 0.4425,
+      "step": 13605
+    },
+    {
+      "epoch": 1.1594375798892202,
+      "grad_norm": 17.5,
+      "learning_rate": 9.47054180289505e-07,
+      "loss": 0.4946,
+      "step": 13606
+    },
+    {
+      "epoch": 1.1595227950575202,
+      "grad_norm": 17.0,
+      "learning_rate": 9.468911570827607e-07,
+      "loss": 0.7057,
+      "step": 13607
+    },
+    {
+      "epoch": 1.1596080102258202,
+      "grad_norm": 22.25,
+      "learning_rate": 9.467281393534986e-07,
+      "loss": 1.0526,
+      "step": 13608
+    },
+    {
+      "epoch": 1.1596932253941201,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.465651271046637e-07,
+      "loss": 0.6051,
+      "step": 13609
+    },
+    {
+      "epoch": 1.15977844056242,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.464021203392018e-07,
+      "loss": 0.5853,
+      "step": 13610
+    },
+    {
+      "epoch": 1.15986365573072,
+      "grad_norm": 17.5,
+      "learning_rate": 9.46239119060059e-07,
+      "loss": 0.7617,
+      "step": 13611
+    },
+    {
+      "epoch": 1.15994887089902,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.460761232701807e-07,
+      "loss": 0.2305,
+      "step": 13612
+    },
+    {
+      "epoch": 1.16003408606732,
+      "grad_norm": 15.125,
+      "learning_rate": 9.459131329725127e-07,
+      "loss": 0.7347,
+      "step": 13613
+    },
+    {
+      "epoch": 1.16011930123562,
+      "grad_norm": 17.625,
+      "learning_rate": 9.457501481699997e-07,
+      "loss": 0.5301,
+      "step": 13614
+    },
+    {
+      "epoch": 1.1602045164039199,
+      "grad_norm": 14.125,
+      "learning_rate": 9.45587168865588e-07,
+      "loss": 0.6893,
+      "step": 13615
+    },
+    {
+      "epoch": 1.1602897315722198,
+      "grad_norm": 17.625,
+      "learning_rate": 9.454241950622219e-07,
+      "loss": 0.7127,
+      "step": 13616
+    },
+    {
+      "epoch": 1.1603749467405198,
+      "grad_norm": 20.25,
+      "learning_rate": 9.452612267628473e-07,
+      "loss": 0.9187,
+      "step": 13617
+    },
+    {
+      "epoch": 1.1604601619088197,
+      "grad_norm": 14.75,
+      "learning_rate": 9.450982639704087e-07,
+      "loss": 0.6128,
+      "step": 13618
+    },
+    {
+      "epoch": 1.1605453770771197,
+      "grad_norm": 12.125,
+      "learning_rate": 9.449353066878507e-07,
+      "loss": 0.4894,
+      "step": 13619
+    },
+    {
+      "epoch": 1.1606305922454196,
+      "grad_norm": 17.75,
+      "learning_rate": 9.447723549181187e-07,
+      "loss": 0.9162,
+      "step": 13620
+    },
+    {
+      "epoch": 1.1607158074137196,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.446094086641573e-07,
+      "loss": 0.989,
+      "step": 13621
+    },
+    {
+      "epoch": 1.1608010225820196,
+      "grad_norm": 17.0,
+      "learning_rate": 9.444464679289106e-07,
+      "loss": 0.6169,
+      "step": 13622
+    },
+    {
+      "epoch": 1.1608862377503195,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.442835327153238e-07,
+      "loss": 0.4222,
+      "step": 13623
+    },
+    {
+      "epoch": 1.1609714529186195,
+      "grad_norm": 16.5,
+      "learning_rate": 9.441206030263407e-07,
+      "loss": 0.6476,
+      "step": 13624
+    },
+    {
+      "epoch": 1.1610566680869194,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.439576788649065e-07,
+      "loss": 0.4112,
+      "step": 13625
+    },
+    {
+      "epoch": 1.1611418832552194,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.437947602339644e-07,
+      "loss": 0.6339,
+      "step": 13626
+    },
+    {
+      "epoch": 1.1612270984235193,
+      "grad_norm": 16.5,
+      "learning_rate": 9.436318471364586e-07,
+      "loss": 0.6932,
+      "step": 13627
+    },
+    {
+      "epoch": 1.1613123135918193,
+      "grad_norm": 18.75,
+      "learning_rate": 9.434689395753338e-07,
+      "loss": 0.9164,
+      "step": 13628
+    },
+    {
+      "epoch": 1.1613975287601193,
+      "grad_norm": 23.125,
+      "learning_rate": 9.433060375535333e-07,
+      "loss": 0.5634,
+      "step": 13629
+    },
+    {
+      "epoch": 1.1614827439284192,
+      "grad_norm": 18.75,
+      "learning_rate": 9.431431410740012e-07,
+      "loss": 0.7379,
+      "step": 13630
+    },
+    {
+      "epoch": 1.1615679590967192,
+      "grad_norm": 16.5,
+      "learning_rate": 9.42980250139681e-07,
+      "loss": 0.6562,
+      "step": 13631
+    },
+    {
+      "epoch": 1.1616531742650191,
+      "grad_norm": 12.25,
+      "learning_rate": 9.428173647535165e-07,
+      "loss": 0.3048,
+      "step": 13632
+    },
+    {
+      "epoch": 1.161738389433319,
+      "grad_norm": 16.625,
+      "learning_rate": 9.426544849184514e-07,
+      "loss": 0.6302,
+      "step": 13633
+    },
+    {
+      "epoch": 1.161823604601619,
+      "grad_norm": 17.875,
+      "learning_rate": 9.424916106374283e-07,
+      "loss": 0.4418,
+      "step": 13634
+    },
+    {
+      "epoch": 1.161908819769919,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.423287419133914e-07,
+      "loss": 0.3419,
+      "step": 13635
+    },
+    {
+      "epoch": 1.161994034938219,
+      "grad_norm": 15.875,
+      "learning_rate": 9.421658787492835e-07,
+      "loss": 0.4799,
+      "step": 13636
+    },
+    {
+      "epoch": 1.162079250106519,
+      "grad_norm": 28.0,
+      "learning_rate": 9.420030211480474e-07,
+      "loss": 0.6386,
+      "step": 13637
+    },
+    {
+      "epoch": 1.1621644652748189,
+      "grad_norm": 11.125,
+      "learning_rate": 9.418401691126266e-07,
+      "loss": 0.1991,
+      "step": 13638
+    },
+    {
+      "epoch": 1.1622496804431188,
+      "grad_norm": 18.375,
+      "learning_rate": 9.416773226459637e-07,
+      "loss": 0.7864,
+      "step": 13639
+    },
+    {
+      "epoch": 1.1623348956114188,
+      "grad_norm": 18.375,
+      "learning_rate": 9.415144817510019e-07,
+      "loss": 0.9134,
+      "step": 13640
+    },
+    {
+      "epoch": 1.1624201107797187,
+      "grad_norm": 18.75,
+      "learning_rate": 9.413516464306838e-07,
+      "loss": 0.3923,
+      "step": 13641
+    },
+    {
+      "epoch": 1.1625053259480187,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.411888166879512e-07,
+      "loss": 0.3891,
+      "step": 13642
+    },
+    {
+      "epoch": 1.1625905411163187,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.410259925257477e-07,
+      "loss": 0.7715,
+      "step": 13643
+    },
+    {
+      "epoch": 1.1626757562846186,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.408631739470149e-07,
+      "loss": 0.5817,
+      "step": 13644
+    },
+    {
+      "epoch": 1.1627609714529186,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.407003609546955e-07,
+      "loss": 0.2861,
+      "step": 13645
+    },
+    {
+      "epoch": 1.1628461866212185,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.405375535517315e-07,
+      "loss": 0.3416,
+      "step": 13646
+    },
+    {
+      "epoch": 1.1629314017895185,
+      "grad_norm": 13.375,
+      "learning_rate": 9.403747517410653e-07,
+      "loss": 0.3693,
+      "step": 13647
+    },
+    {
+      "epoch": 1.1630166169578184,
+      "grad_norm": 16.0,
+      "learning_rate": 9.402119555256385e-07,
+      "loss": 0.6877,
+      "step": 13648
+    },
+    {
+      "epoch": 1.1631018321261184,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.400491649083937e-07,
+      "loss": 0.4036,
+      "step": 13649
+    },
+    {
+      "epoch": 1.1631870472944184,
+      "grad_norm": 18.375,
+      "learning_rate": 9.398863798922718e-07,
+      "loss": 0.5185,
+      "step": 13650
+    },
+    {
+      "epoch": 1.1632722624627183,
+      "grad_norm": 10.9375,
+      "learning_rate": 9.397236004802149e-07,
+      "loss": 0.2909,
+      "step": 13651
+    },
+    {
+      "epoch": 1.1633574776310183,
+      "grad_norm": 14.375,
+      "learning_rate": 9.395608266751643e-07,
+      "loss": 0.6689,
+      "step": 13652
+    },
+    {
+      "epoch": 1.1634426927993182,
+      "grad_norm": 16.125,
+      "learning_rate": 9.393980584800621e-07,
+      "loss": 0.7906,
+      "step": 13653
+    },
+    {
+      "epoch": 1.1635279079676182,
+      "grad_norm": 16.125,
+      "learning_rate": 9.392352958978489e-07,
+      "loss": 0.9041,
+      "step": 13654
+    },
+    {
+      "epoch": 1.1636131231359181,
+      "grad_norm": 16.0,
+      "learning_rate": 9.390725389314667e-07,
+      "loss": 0.6811,
+      "step": 13655
+    },
+    {
+      "epoch": 1.163698338304218,
+      "grad_norm": 17.625,
+      "learning_rate": 9.389097875838562e-07,
+      "loss": 0.8108,
+      "step": 13656
+    },
+    {
+      "epoch": 1.163783553472518,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.38747041857959e-07,
+      "loss": 0.5247,
+      "step": 13657
+    },
+    {
+      "epoch": 1.163868768640818,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.385843017567156e-07,
+      "loss": 0.6277,
+      "step": 13658
+    },
+    {
+      "epoch": 1.163953983809118,
+      "grad_norm": 13.875,
+      "learning_rate": 9.384215672830668e-07,
+      "loss": 0.5695,
+      "step": 13659
+    },
+    {
+      "epoch": 1.164039198977418,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.382588384399538e-07,
+      "loss": 0.8171,
+      "step": 13660
+    },
+    {
+      "epoch": 1.1641244141457179,
+      "grad_norm": 21.5,
+      "learning_rate": 9.380961152303168e-07,
+      "loss": 0.9488,
+      "step": 13661
+    },
+    {
+      "epoch": 1.1642096293140178,
+      "grad_norm": 12.625,
+      "learning_rate": 9.37933397657097e-07,
+      "loss": 0.4288,
+      "step": 13662
+    },
+    {
+      "epoch": 1.1642948444823178,
+      "grad_norm": 17.75,
+      "learning_rate": 9.377706857232341e-07,
+      "loss": 0.4675,
+      "step": 13663
+    },
+    {
+      "epoch": 1.1643800596506177,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.376079794316692e-07,
+      "loss": 0.4697,
+      "step": 13664
+    },
+    {
+      "epoch": 1.1644652748189177,
+      "grad_norm": 13.5,
+      "learning_rate": 9.374452787853426e-07,
+      "loss": 0.602,
+      "step": 13665
+    },
+    {
+      "epoch": 1.1645504899872177,
+      "grad_norm": 13.375,
+      "learning_rate": 9.372825837871938e-07,
+      "loss": 0.5384,
+      "step": 13666
+    },
+    {
+      "epoch": 1.1646357051555176,
+      "grad_norm": 8.5,
+      "learning_rate": 9.371198944401628e-07,
+      "loss": 0.1984,
+      "step": 13667
+    },
+    {
+      "epoch": 1.1647209203238176,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.369572107471903e-07,
+      "loss": 0.6151,
+      "step": 13668
+    },
+    {
+      "epoch": 1.1648061354921175,
+      "grad_norm": 17.25,
+      "learning_rate": 9.367945327112155e-07,
+      "loss": 0.5947,
+      "step": 13669
+    },
+    {
+      "epoch": 1.1648913506604175,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.366318603351787e-07,
+      "loss": 0.5175,
+      "step": 13670
+    },
+    {
+      "epoch": 1.1649765658287174,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.364691936220191e-07,
+      "loss": 0.6977,
+      "step": 13671
+    },
+    {
+      "epoch": 1.1650617809970174,
+      "grad_norm": 20.375,
+      "learning_rate": 9.363065325746767e-07,
+      "loss": 0.5001,
+      "step": 13672
+    },
+    {
+      "epoch": 1.1651469961653174,
+      "grad_norm": 22.125,
+      "learning_rate": 9.361438771960909e-07,
+      "loss": 0.8938,
+      "step": 13673
+    },
+    {
+      "epoch": 1.1652322113336173,
+      "grad_norm": 20.5,
+      "learning_rate": 9.359812274892004e-07,
+      "loss": 1.0476,
+      "step": 13674
+    },
+    {
+      "epoch": 1.1653174265019173,
+      "grad_norm": 14.0,
+      "learning_rate": 9.35818583456945e-07,
+      "loss": 0.3981,
+      "step": 13675
+    },
+    {
+      "epoch": 1.1654026416702172,
+      "grad_norm": 16.0,
+      "learning_rate": 9.356559451022637e-07,
+      "loss": 0.6305,
+      "step": 13676
+    },
+    {
+      "epoch": 1.1654878568385172,
+      "grad_norm": 15.25,
+      "learning_rate": 9.354933124280957e-07,
+      "loss": 0.7307,
+      "step": 13677
+    },
+    {
+      "epoch": 1.1655730720068171,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.353306854373794e-07,
+      "loss": 0.3523,
+      "step": 13678
+    },
+    {
+      "epoch": 1.165658287175117,
+      "grad_norm": 14.25,
+      "learning_rate": 9.351680641330546e-07,
+      "loss": 0.4692,
+      "step": 13679
+    },
+    {
+      "epoch": 1.165743502343417,
+      "grad_norm": 16.5,
+      "learning_rate": 9.35005448518059e-07,
+      "loss": 0.7912,
+      "step": 13680
+    },
+    {
+      "epoch": 1.165828717511717,
+      "grad_norm": 14.125,
+      "learning_rate": 9.34842838595332e-07,
+      "loss": 0.4384,
+      "step": 13681
+    },
+    {
+      "epoch": 1.165913932680017,
+      "grad_norm": 15.0,
+      "learning_rate": 9.346802343678121e-07,
+      "loss": 0.7061,
+      "step": 13682
+    },
+    {
+      "epoch": 1.165999147848317,
+      "grad_norm": 16.875,
+      "learning_rate": 9.345176358384373e-07,
+      "loss": 1.0105,
+      "step": 13683
+    },
+    {
+      "epoch": 1.1660843630166169,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.343550430101455e-07,
+      "loss": 0.2593,
+      "step": 13684
+    },
+    {
+      "epoch": 1.1661695781849168,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.341924558858759e-07,
+      "loss": 0.4889,
+      "step": 13685
+    },
+    {
+      "epoch": 1.1662547933532168,
+      "grad_norm": 17.875,
+      "learning_rate": 9.34029874468566e-07,
+      "loss": 0.6194,
+      "step": 13686
+    },
+    {
+      "epoch": 1.1663400085215168,
+      "grad_norm": 13.5,
+      "learning_rate": 9.338672987611542e-07,
+      "loss": 0.6946,
+      "step": 13687
+    },
+    {
+      "epoch": 1.1664252236898167,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.337047287665779e-07,
+      "loss": 0.592,
+      "step": 13688
+    },
+    {
+      "epoch": 1.1665104388581167,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.335421644877757e-07,
+      "loss": 0.6409,
+      "step": 13689
+    },
+    {
+      "epoch": 1.1665956540264166,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.333796059276847e-07,
+      "loss": 0.3282,
+      "step": 13690
+    },
+    {
+      "epoch": 1.1666808691947166,
+      "grad_norm": 15.5,
+      "learning_rate": 9.332170530892423e-07,
+      "loss": 0.6501,
+      "step": 13691
+    },
+    {
+      "epoch": 1.1667660843630165,
+      "grad_norm": 25.375,
+      "learning_rate": 9.330545059753866e-07,
+      "loss": 0.5105,
+      "step": 13692
+    },
+    {
+      "epoch": 1.1668512995313165,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.328919645890544e-07,
+      "loss": 0.7121,
+      "step": 13693
+    },
+    {
+      "epoch": 1.1669365146996165,
+      "grad_norm": 22.125,
+      "learning_rate": 9.327294289331835e-07,
+      "loss": 0.8792,
+      "step": 13694
+    },
+    {
+      "epoch": 1.1670217298679164,
+      "grad_norm": 17.25,
+      "learning_rate": 9.325668990107106e-07,
+      "loss": 0.9586,
+      "step": 13695
+    },
+    {
+      "epoch": 1.1671069450362164,
+      "grad_norm": 14.875,
+      "learning_rate": 9.324043748245734e-07,
+      "loss": 0.5544,
+      "step": 13696
+    },
+    {
+      "epoch": 1.1671921602045163,
+      "grad_norm": 15.375,
+      "learning_rate": 9.32241856377708e-07,
+      "loss": 0.5766,
+      "step": 13697
+    },
+    {
+      "epoch": 1.1672773753728163,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.32079343673053e-07,
+      "loss": 0.4521,
+      "step": 13698
+    },
+    {
+      "epoch": 1.1673625905411162,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.319168367135428e-07,
+      "loss": 0.5398,
+      "step": 13699
+    },
+    {
+      "epoch": 1.1674478057094162,
+      "grad_norm": 14.5,
+      "learning_rate": 9.317543355021155e-07,
+      "loss": 0.6367,
+      "step": 13700
+    },
+    {
+      "epoch": 1.1675330208777162,
+      "grad_norm": 14.75,
+      "learning_rate": 9.315918400417073e-07,
+      "loss": 0.5078,
+      "step": 13701
+    },
+    {
+      "epoch": 1.1676182360460161,
+      "grad_norm": 18.0,
+      "learning_rate": 9.314293503352547e-07,
+      "loss": 0.7722,
+      "step": 13702
+    },
+    {
+      "epoch": 1.167703451214316,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.312668663856941e-07,
+      "loss": 0.3216,
+      "step": 13703
+    },
+    {
+      "epoch": 1.167788666382616,
+      "grad_norm": 18.5,
+      "learning_rate": 9.311043881959619e-07,
+      "loss": 0.7287,
+      "step": 13704
+    },
+    {
+      "epoch": 1.167873881550916,
+      "grad_norm": 18.625,
+      "learning_rate": 9.309419157689939e-07,
+      "loss": 0.8367,
+      "step": 13705
+    },
+    {
+      "epoch": 1.167959096719216,
+      "grad_norm": 12.625,
+      "learning_rate": 9.307794491077269e-07,
+      "loss": 0.4761,
+      "step": 13706
+    },
+    {
+      "epoch": 1.168044311887516,
+      "grad_norm": 25.875,
+      "learning_rate": 9.306169882150959e-07,
+      "loss": 0.9498,
+      "step": 13707
+    },
+    {
+      "epoch": 1.1681295270558159,
+      "grad_norm": 15.125,
+      "learning_rate": 9.304545330940368e-07,
+      "loss": 0.7091,
+      "step": 13708
+    },
+    {
+      "epoch": 1.1682147422241158,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.302920837474857e-07,
+      "loss": 0.7179,
+      "step": 13709
+    },
+    {
+      "epoch": 1.1682999573924158,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.301296401783781e-07,
+      "loss": 0.7572,
+      "step": 13710
+    },
+    {
+      "epoch": 1.1683851725607157,
+      "grad_norm": 17.375,
+      "learning_rate": 9.299672023896497e-07,
+      "loss": 0.7923,
+      "step": 13711
+    },
+    {
+      "epoch": 1.1684703877290157,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.298047703842356e-07,
+      "loss": 0.3043,
+      "step": 13712
+    },
+    {
+      "epoch": 1.1685556028973156,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.296423441650715e-07,
+      "loss": 0.643,
+      "step": 13713
+    },
+    {
+      "epoch": 1.1686408180656156,
+      "grad_norm": 13.625,
+      "learning_rate": 9.294799237350926e-07,
+      "loss": 0.4616,
+      "step": 13714
+    },
+    {
+      "epoch": 1.1687260332339156,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.293175090972337e-07,
+      "loss": 0.3855,
+      "step": 13715
+    },
+    {
+      "epoch": 1.1688112484022155,
+      "grad_norm": 16.25,
+      "learning_rate": 9.291551002544295e-07,
+      "loss": 0.6121,
+      "step": 13716
+    },
+    {
+      "epoch": 1.1688964635705155,
+      "grad_norm": 20.125,
+      "learning_rate": 9.289926972096156e-07,
+      "loss": 0.6761,
+      "step": 13717
+    },
+    {
+      "epoch": 1.1689816787388154,
+      "grad_norm": 13.125,
+      "learning_rate": 9.288302999657262e-07,
+      "loss": 0.5429,
+      "step": 13718
+    },
+    {
+      "epoch": 1.1690668939071154,
+      "grad_norm": 21.0,
+      "learning_rate": 9.286679085256963e-07,
+      "loss": 0.6478,
+      "step": 13719
+    },
+    {
+      "epoch": 1.1691521090754153,
+      "grad_norm": 16.5,
+      "learning_rate": 9.285055228924604e-07,
+      "loss": 0.639,
+      "step": 13720
+    },
+    {
+      "epoch": 1.1692373242437153,
+      "grad_norm": 18.875,
+      "learning_rate": 9.283431430689532e-07,
+      "loss": 0.4565,
+      "step": 13721
+    },
+    {
+      "epoch": 1.1693225394120152,
+      "grad_norm": 16.25,
+      "learning_rate": 9.281807690581091e-07,
+      "loss": 0.5764,
+      "step": 13722
+    },
+    {
+      "epoch": 1.1694077545803152,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.280184008628614e-07,
+      "loss": 0.5861,
+      "step": 13723
+    },
+    {
+      "epoch": 1.1694929697486152,
+      "grad_norm": 18.75,
+      "learning_rate": 9.278560384861454e-07,
+      "loss": 1.1764,
+      "step": 13724
+    },
+    {
+      "epoch": 1.1695781849169151,
+      "grad_norm": 17.0,
+      "learning_rate": 9.276936819308946e-07,
+      "loss": 0.4114,
+      "step": 13725
+    },
+    {
+      "epoch": 1.169663400085215,
+      "grad_norm": 16.25,
+      "learning_rate": 9.275313312000431e-07,
+      "loss": 0.5187,
+      "step": 13726
+    },
+    {
+      "epoch": 1.169748615253515,
+      "grad_norm": 15.0,
+      "learning_rate": 9.273689862965246e-07,
+      "loss": 0.6566,
+      "step": 13727
+    },
+    {
+      "epoch": 1.169833830421815,
+      "grad_norm": 17.0,
+      "learning_rate": 9.272066472232733e-07,
+      "loss": 0.5981,
+      "step": 13728
+    },
+    {
+      "epoch": 1.169919045590115,
+      "grad_norm": 14.125,
+      "learning_rate": 9.270443139832222e-07,
+      "loss": 0.5107,
+      "step": 13729
+    },
+    {
+      "epoch": 1.170004260758415,
+      "grad_norm": 16.375,
+      "learning_rate": 9.268819865793055e-07,
+      "loss": 0.484,
+      "step": 13730
+    },
+    {
+      "epoch": 1.1700894759267149,
+      "grad_norm": 20.875,
+      "learning_rate": 9.267196650144558e-07,
+      "loss": 0.5193,
+      "step": 13731
+    },
+    {
+      "epoch": 1.1701746910950148,
+      "grad_norm": 11.5625,
+      "learning_rate": 9.265573492916069e-07,
+      "loss": 0.4721,
+      "step": 13732
+    },
+    {
+      "epoch": 1.1702599062633148,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.263950394136918e-07,
+      "loss": 0.5427,
+      "step": 13733
+    },
+    {
+      "epoch": 1.1703451214316147,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.26232735383644e-07,
+      "loss": 0.4442,
+      "step": 13734
+    },
+    {
+      "epoch": 1.1704303365999147,
+      "grad_norm": 19.875,
+      "learning_rate": 9.260704372043961e-07,
+      "loss": 0.6256,
+      "step": 13735
+    },
+    {
+      "epoch": 1.1705155517682146,
+      "grad_norm": 26.875,
+      "learning_rate": 9.259081448788812e-07,
+      "loss": 0.7938,
+      "step": 13736
+    },
+    {
+      "epoch": 1.1706007669365146,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.257458584100318e-07,
+      "loss": 0.4748,
+      "step": 13737
+    },
+    {
+      "epoch": 1.1706859821048146,
+      "grad_norm": 18.375,
+      "learning_rate": 9.255835778007813e-07,
+      "loss": 0.4346,
+      "step": 13738
+    },
+    {
+      "epoch": 1.1707711972731145,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.254213030540615e-07,
+      "loss": 0.5094,
+      "step": 13739
+    },
+    {
+      "epoch": 1.1708564124414145,
+      "grad_norm": 17.625,
+      "learning_rate": 9.252590341728052e-07,
+      "loss": 0.7029,
+      "step": 13740
+    },
+    {
+      "epoch": 1.1709416276097144,
+      "grad_norm": 17.0,
+      "learning_rate": 9.250967711599446e-07,
+      "loss": 0.8056,
+      "step": 13741
+    },
+    {
+      "epoch": 1.1710268427780144,
+      "grad_norm": 18.5,
+      "learning_rate": 9.249345140184118e-07,
+      "loss": 0.8193,
+      "step": 13742
+    },
+    {
+      "epoch": 1.1711120579463143,
+      "grad_norm": 15.875,
+      "learning_rate": 9.247722627511396e-07,
+      "loss": 0.7369,
+      "step": 13743
+    },
+    {
+      "epoch": 1.1711972731146143,
+      "grad_norm": 16.125,
+      "learning_rate": 9.246100173610598e-07,
+      "loss": 0.3672,
+      "step": 13744
+    },
+    {
+      "epoch": 1.1712824882829143,
+      "grad_norm": 17.875,
+      "learning_rate": 9.244477778511038e-07,
+      "loss": 0.2977,
+      "step": 13745
+    },
+    {
+      "epoch": 1.1713677034512142,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.24285544224204e-07,
+      "loss": 0.9983,
+      "step": 13746
+    },
+    {
+      "epoch": 1.1714529186195142,
+      "grad_norm": 17.0,
+      "learning_rate": 9.241233164832922e-07,
+      "loss": 0.8154,
+      "step": 13747
+    },
+    {
+      "epoch": 1.1715381337878141,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.239610946312993e-07,
+      "loss": 0.4281,
+      "step": 13748
+    },
+    {
+      "epoch": 1.171623348956114,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.237988786711577e-07,
+      "loss": 0.3146,
+      "step": 13749
+    },
+    {
+      "epoch": 1.171708564124414,
+      "grad_norm": 16.25,
+      "learning_rate": 9.236366686057979e-07,
+      "loss": 0.7633,
+      "step": 13750
+    },
+    {
+      "epoch": 1.171793779292714,
+      "grad_norm": 13.75,
+      "learning_rate": 9.23474464438152e-07,
+      "loss": 0.4556,
+      "step": 13751
+    },
+    {
+      "epoch": 1.171878994461014,
+      "grad_norm": 19.5,
+      "learning_rate": 9.233122661711507e-07,
+      "loss": 0.9847,
+      "step": 13752
+    },
+    {
+      "epoch": 1.171964209629314,
+      "grad_norm": 16.5,
+      "learning_rate": 9.231500738077254e-07,
+      "loss": 0.484,
+      "step": 13753
+    },
+    {
+      "epoch": 1.1720494247976139,
+      "grad_norm": 19.0,
+      "learning_rate": 9.229878873508066e-07,
+      "loss": 0.804,
+      "step": 13754
+    },
+    {
+      "epoch": 1.1721346399659138,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.228257068033262e-07,
+      "loss": 0.351,
+      "step": 13755
+    },
+    {
+      "epoch": 1.1722198551342138,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.22663532168214e-07,
+      "loss": 0.3316,
+      "step": 13756
+    },
+    {
+      "epoch": 1.1723050703025137,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.225013634484006e-07,
+      "loss": 0.505,
+      "step": 13757
+    },
+    {
+      "epoch": 1.1723902854708137,
+      "grad_norm": 20.75,
+      "learning_rate": 9.223392006468174e-07,
+      "loss": 0.9544,
+      "step": 13758
+    },
+    {
+      "epoch": 1.1724755006391137,
+      "grad_norm": 13.375,
+      "learning_rate": 9.22177043766394e-07,
+      "loss": 0.4804,
+      "step": 13759
+    },
+    {
+      "epoch": 1.1725607158074136,
+      "grad_norm": 10.6875,
+      "learning_rate": 9.220148928100609e-07,
+      "loss": 0.3642,
+      "step": 13760
+    },
+    {
+      "epoch": 1.1726459309757136,
+      "grad_norm": 16.125,
+      "learning_rate": 9.218527477807488e-07,
+      "loss": 1.0494,
+      "step": 13761
+    },
+    {
+      "epoch": 1.1727311461440135,
+      "grad_norm": 9.6875,
+      "learning_rate": 9.216906086813873e-07,
+      "loss": 0.2008,
+      "step": 13762
+    },
+    {
+      "epoch": 1.1728163613123135,
+      "grad_norm": 15.375,
+      "learning_rate": 9.215284755149071e-07,
+      "loss": 0.5116,
+      "step": 13763
+    },
+    {
+      "epoch": 1.1729015764806134,
+      "grad_norm": 13.625,
+      "learning_rate": 9.213663482842375e-07,
+      "loss": 0.4188,
+      "step": 13764
+    },
+    {
+      "epoch": 1.1729867916489134,
+      "grad_norm": 18.75,
+      "learning_rate": 9.21204226992308e-07,
+      "loss": 0.6175,
+      "step": 13765
+    },
+    {
+      "epoch": 1.1730720068172134,
+      "grad_norm": 22.375,
+      "learning_rate": 9.210421116420493e-07,
+      "loss": 0.7288,
+      "step": 13766
+    },
+    {
+      "epoch": 1.1731572219855133,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.208800022363899e-07,
+      "loss": 0.53,
+      "step": 13767
+    },
+    {
+      "epoch": 1.1732424371538133,
+      "grad_norm": 12.875,
+      "learning_rate": 9.207178987782602e-07,
+      "loss": 0.4425,
+      "step": 13768
+    },
+    {
+      "epoch": 1.1733276523221132,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.20555801270589e-07,
+      "loss": 0.6485,
+      "step": 13769
+    },
+    {
+      "epoch": 1.1734128674904132,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.203937097163058e-07,
+      "loss": 0.4844,
+      "step": 13770
+    },
+    {
+      "epoch": 1.1734980826587131,
+      "grad_norm": 14.5,
+      "learning_rate": 9.202316241183401e-07,
+      "loss": 0.4934,
+      "step": 13771
+    },
+    {
+      "epoch": 1.173583297827013,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.200695444796201e-07,
+      "loss": 0.6111,
+      "step": 13772
+    },
+    {
+      "epoch": 1.173668512995313,
+      "grad_norm": 11.5,
+      "learning_rate": 9.199074708030753e-07,
+      "loss": 0.406,
+      "step": 13773
+    },
+    {
+      "epoch": 1.173753728163613,
+      "grad_norm": 34.5,
+      "learning_rate": 9.197454030916345e-07,
+      "loss": 0.8554,
+      "step": 13774
+    },
+    {
+      "epoch": 1.173838943331913,
+      "grad_norm": 18.875,
+      "learning_rate": 9.195833413482261e-07,
+      "loss": 0.5937,
+      "step": 13775
+    },
+    {
+      "epoch": 1.173924158500213,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.194212855757793e-07,
+      "loss": 0.4867,
+      "step": 13776
+    },
+    {
+      "epoch": 1.1740093736685129,
+      "grad_norm": 23.0,
+      "learning_rate": 9.19259235777222e-07,
+      "loss": 0.93,
+      "step": 13777
+    },
+    {
+      "epoch": 1.1740945888368128,
+      "grad_norm": 17.125,
+      "learning_rate": 9.190971919554831e-07,
+      "loss": 0.5663,
+      "step": 13778
+    },
+    {
+      "epoch": 1.1741798040051128,
+      "grad_norm": 19.875,
+      "learning_rate": 9.189351541134911e-07,
+      "loss": 0.9228,
+      "step": 13779
+    },
+    {
+      "epoch": 1.1742650191734127,
+      "grad_norm": 15.875,
+      "learning_rate": 9.187731222541732e-07,
+      "loss": 0.918,
+      "step": 13780
+    },
+    {
+      "epoch": 1.1743502343417127,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.186110963804582e-07,
+      "loss": 0.646,
+      "step": 13781
+    },
+    {
+      "epoch": 1.1744354495100127,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.184490764952737e-07,
+      "loss": 0.4019,
+      "step": 13782
+    },
+    {
+      "epoch": 1.1745206646783126,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.18287062601548e-07,
+      "loss": 0.3562,
+      "step": 13783
+    },
+    {
+      "epoch": 1.1746058798466126,
+      "grad_norm": 22.375,
+      "learning_rate": 9.181250547022086e-07,
+      "loss": 0.6302,
+      "step": 13784
+    },
+    {
+      "epoch": 1.1746910950149125,
+      "grad_norm": 16.875,
+      "learning_rate": 9.179630528001833e-07,
+      "loss": 0.485,
+      "step": 13785
+    },
+    {
+      "epoch": 1.1747763101832125,
+      "grad_norm": 22.5,
+      "learning_rate": 9.178010568983993e-07,
+      "loss": 0.9022,
+      "step": 13786
+    },
+    {
+      "epoch": 1.1748615253515124,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.176390669997848e-07,
+      "loss": 0.4142,
+      "step": 13787
+    },
+    {
+      "epoch": 1.1749467405198124,
+      "grad_norm": 19.875,
+      "learning_rate": 9.174770831072663e-07,
+      "loss": 0.6718,
+      "step": 13788
+    },
+    {
+      "epoch": 1.1750319556881124,
+      "grad_norm": 27.125,
+      "learning_rate": 9.173151052237712e-07,
+      "loss": 0.6082,
+      "step": 13789
+    },
+    {
+      "epoch": 1.1751171708564123,
+      "grad_norm": 19.75,
+      "learning_rate": 9.171531333522266e-07,
+      "loss": 0.7176,
+      "step": 13790
+    },
+    {
+      "epoch": 1.1752023860247123,
+      "grad_norm": 21.5,
+      "learning_rate": 9.169911674955596e-07,
+      "loss": 0.4735,
+      "step": 13791
+    },
+    {
+      "epoch": 1.1752876011930122,
+      "grad_norm": 16.25,
+      "learning_rate": 9.16829207656697e-07,
+      "loss": 0.7935,
+      "step": 13792
+    },
+    {
+      "epoch": 1.1753728163613122,
+      "grad_norm": 13.625,
+      "learning_rate": 9.166672538385659e-07,
+      "loss": 0.6668,
+      "step": 13793
+    },
+    {
+      "epoch": 1.1754580315296121,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.165053060440923e-07,
+      "loss": 0.6673,
+      "step": 13794
+    },
+    {
+      "epoch": 1.1755432466979123,
+      "grad_norm": 21.75,
+      "learning_rate": 9.163433642762038e-07,
+      "loss": 0.7526,
+      "step": 13795
+    },
+    {
+      "epoch": 1.1756284618662123,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.16181428537826e-07,
+      "loss": 0.5234,
+      "step": 13796
+    },
+    {
+      "epoch": 1.1757136770345122,
+      "grad_norm": 20.625,
+      "learning_rate": 9.160194988318851e-07,
+      "loss": 0.3877,
+      "step": 13797
+    },
+    {
+      "epoch": 1.1757988922028122,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.15857575161308e-07,
+      "loss": 0.6685,
+      "step": 13798
+    },
+    {
+      "epoch": 1.1758841073711122,
+      "grad_norm": 14.75,
+      "learning_rate": 9.156956575290205e-07,
+      "loss": 0.4307,
+      "step": 13799
+    },
+    {
+      "epoch": 1.175969322539412,
+      "grad_norm": 18.875,
+      "learning_rate": 9.155337459379485e-07,
+      "loss": 0.8261,
+      "step": 13800
+    },
+    {
+      "epoch": 1.176054537707712,
+      "grad_norm": 12.25,
+      "learning_rate": 9.15371840391018e-07,
+      "loss": 0.4362,
+      "step": 13801
+    },
+    {
+      "epoch": 1.176139752876012,
+      "grad_norm": 16.625,
+      "learning_rate": 9.152099408911551e-07,
+      "loss": 0.7719,
+      "step": 13802
+    },
+    {
+      "epoch": 1.176224968044312,
+      "grad_norm": 12.875,
+      "learning_rate": 9.150480474412854e-07,
+      "loss": 0.4996,
+      "step": 13803
+    },
+    {
+      "epoch": 1.176310183212612,
+      "grad_norm": 17.375,
+      "learning_rate": 9.148861600443341e-07,
+      "loss": 0.9614,
+      "step": 13804
+    },
+    {
+      "epoch": 1.176395398380912,
+      "grad_norm": 20.25,
+      "learning_rate": 9.147242787032266e-07,
+      "loss": 0.5946,
+      "step": 13805
+    },
+    {
+      "epoch": 1.1764806135492119,
+      "grad_norm": 16.25,
+      "learning_rate": 9.145624034208889e-07,
+      "loss": 0.7093,
+      "step": 13806
+    },
+    {
+      "epoch": 1.1765658287175118,
+      "grad_norm": 14.125,
+      "learning_rate": 9.144005342002455e-07,
+      "loss": 0.3679,
+      "step": 13807
+    },
+    {
+      "epoch": 1.1766510438858118,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.142386710442222e-07,
+      "loss": 0.6171,
+      "step": 13808
+    },
+    {
+      "epoch": 1.1767362590541117,
+      "grad_norm": 17.375,
+      "learning_rate": 9.140768139557435e-07,
+      "loss": 0.5618,
+      "step": 13809
+    },
+    {
+      "epoch": 1.1768214742224117,
+      "grad_norm": 18.0,
+      "learning_rate": 9.139149629377347e-07,
+      "loss": 0.7244,
+      "step": 13810
+    },
+    {
+      "epoch": 1.1769066893907116,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.137531179931208e-07,
+      "loss": 0.6206,
+      "step": 13811
+    },
+    {
+      "epoch": 1.1769919045590116,
+      "grad_norm": 14.75,
+      "learning_rate": 9.135912791248258e-07,
+      "loss": 0.4439,
+      "step": 13812
+    },
+    {
+      "epoch": 1.1770771197273115,
+      "grad_norm": 15.875,
+      "learning_rate": 9.134294463357749e-07,
+      "loss": 0.4244,
+      "step": 13813
+    },
+    {
+      "epoch": 1.1771623348956115,
+      "grad_norm": 13.125,
+      "learning_rate": 9.13267619628892e-07,
+      "loss": 0.4025,
+      "step": 13814
+    },
+    {
+      "epoch": 1.1772475500639115,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.131057990071021e-07,
+      "loss": 0.3839,
+      "step": 13815
+    },
+    {
+      "epoch": 1.1773327652322114,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.12943984473329e-07,
+      "loss": 0.6203,
+      "step": 13816
+    },
+    {
+      "epoch": 1.1774179804005114,
+      "grad_norm": 14.5,
+      "learning_rate": 9.127821760304972e-07,
+      "loss": 0.4257,
+      "step": 13817
+    },
+    {
+      "epoch": 1.1775031955688113,
+      "grad_norm": 11.5,
+      "learning_rate": 9.126203736815303e-07,
+      "loss": 0.4317,
+      "step": 13818
+    },
+    {
+      "epoch": 1.1775884107371113,
+      "grad_norm": 17.625,
+      "learning_rate": 9.124585774293529e-07,
+      "loss": 0.5322,
+      "step": 13819
+    },
+    {
+      "epoch": 1.1776736259054112,
+      "grad_norm": 40.0,
+      "learning_rate": 9.122967872768887e-07,
+      "loss": 0.4118,
+      "step": 13820
+    },
+    {
+      "epoch": 1.1777588410737112,
+      "grad_norm": 15.625,
+      "learning_rate": 9.121350032270607e-07,
+      "loss": 0.7193,
+      "step": 13821
+    },
+    {
+      "epoch": 1.1778440562420112,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.119732252827928e-07,
+      "loss": 0.4431,
+      "step": 13822
+    },
+    {
+      "epoch": 1.1779292714103111,
+      "grad_norm": 18.625,
+      "learning_rate": 9.118114534470089e-07,
+      "loss": 0.5925,
+      "step": 13823
+    },
+    {
+      "epoch": 1.178014486578611,
+      "grad_norm": 14.875,
+      "learning_rate": 9.116496877226319e-07,
+      "loss": 0.7285,
+      "step": 13824
+    },
+    {
+      "epoch": 1.178099701746911,
+      "grad_norm": 17.25,
+      "learning_rate": 9.114879281125854e-07,
+      "loss": 0.4615,
+      "step": 13825
+    },
+    {
+      "epoch": 1.178184916915211,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.113261746197923e-07,
+      "loss": 0.3311,
+      "step": 13826
+    },
+    {
+      "epoch": 1.178270132083511,
+      "grad_norm": 14.375,
+      "learning_rate": 9.111644272471759e-07,
+      "loss": 0.5541,
+      "step": 13827
+    },
+    {
+      "epoch": 1.178355347251811,
+      "grad_norm": 20.875,
+      "learning_rate": 9.110026859976595e-07,
+      "loss": 0.8298,
+      "step": 13828
+    },
+    {
+      "epoch": 1.1784405624201109,
+      "grad_norm": 16.375,
+      "learning_rate": 9.108409508741647e-07,
+      "loss": 0.6907,
+      "step": 13829
+    },
+    {
+      "epoch": 1.1785257775884108,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.106792218796154e-07,
+      "loss": 0.5155,
+      "step": 13830
+    },
+    {
+      "epoch": 1.1786109927567108,
+      "grad_norm": 18.25,
+      "learning_rate": 9.105174990169334e-07,
+      "loss": 0.56,
+      "step": 13831
+    },
+    {
+      "epoch": 1.1786962079250107,
+      "grad_norm": 15.75,
+      "learning_rate": 9.103557822890418e-07,
+      "loss": 0.641,
+      "step": 13832
+    },
+    {
+      "epoch": 1.1787814230933107,
+      "grad_norm": 12.75,
+      "learning_rate": 9.101940716988626e-07,
+      "loss": 0.5763,
+      "step": 13833
+    },
+    {
+      "epoch": 1.1788666382616106,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.100323672493186e-07,
+      "loss": 0.2797,
+      "step": 13834
+    },
+    {
+      "epoch": 1.1789518534299106,
+      "grad_norm": 17.0,
+      "learning_rate": 9.09870668943331e-07,
+      "loss": 0.6225,
+      "step": 13835
+    },
+    {
+      "epoch": 1.1790370685982106,
+      "grad_norm": 39.5,
+      "learning_rate": 9.097089767838235e-07,
+      "loss": 1.3705,
+      "step": 13836
+    },
+    {
+      "epoch": 1.1791222837665105,
+      "grad_norm": 17.375,
+      "learning_rate": 9.095472907737162e-07,
+      "loss": 0.6613,
+      "step": 13837
+    },
+    {
+      "epoch": 1.1792074989348105,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.093856109159319e-07,
+      "loss": 0.7647,
+      "step": 13838
+    },
+    {
+      "epoch": 1.1792927141031104,
+      "grad_norm": 16.0,
+      "learning_rate": 9.092239372133918e-07,
+      "loss": 0.5336,
+      "step": 13839
+    },
+    {
+      "epoch": 1.1793779292714104,
+      "grad_norm": 12.0625,
+      "learning_rate": 9.090622696690179e-07,
+      "loss": 0.4046,
+      "step": 13840
+    },
+    {
+      "epoch": 1.1794631444397103,
+      "grad_norm": 37.75,
+      "learning_rate": 9.089006082857317e-07,
+      "loss": 1.3158,
+      "step": 13841
+    },
+    {
+      "epoch": 1.1795483596080103,
+      "grad_norm": 14.75,
+      "learning_rate": 9.087389530664547e-07,
+      "loss": 1.0166,
+      "step": 13842
+    },
+    {
+      "epoch": 1.1796335747763103,
+      "grad_norm": 17.625,
+      "learning_rate": 9.085773040141077e-07,
+      "loss": 0.5921,
+      "step": 13843
+    },
+    {
+      "epoch": 1.1797187899446102,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.084156611316126e-07,
+      "loss": 0.4077,
+      "step": 13844
+    },
+    {
+      "epoch": 1.1798040051129102,
+      "grad_norm": 18.75,
+      "learning_rate": 9.082540244218899e-07,
+      "loss": 0.734,
+      "step": 13845
+    },
+    {
+      "epoch": 1.1798892202812101,
+      "grad_norm": 23.75,
+      "learning_rate": 9.080923938878602e-07,
+      "loss": 1.0257,
+      "step": 13846
+    },
+    {
+      "epoch": 1.17997443544951,
+      "grad_norm": 16.0,
+      "learning_rate": 9.079307695324451e-07,
+      "loss": 0.7709,
+      "step": 13847
+    },
+    {
+      "epoch": 1.18005965061781,
+      "grad_norm": 16.875,
+      "learning_rate": 9.077691513585645e-07,
+      "loss": 0.632,
+      "step": 13848
+    },
+    {
+      "epoch": 1.18014486578611,
+      "grad_norm": 17.875,
+      "learning_rate": 9.0760753936914e-07,
+      "loss": 0.814,
+      "step": 13849
+    },
+    {
+      "epoch": 1.18023008095441,
+      "grad_norm": 13.75,
+      "learning_rate": 9.074459335670913e-07,
+      "loss": 0.3345,
+      "step": 13850
+    },
+    {
+      "epoch": 1.18031529612271,
+      "grad_norm": 31.5,
+      "learning_rate": 9.072843339553391e-07,
+      "loss": 0.5699,
+      "step": 13851
+    },
+    {
+      "epoch": 1.1804005112910099,
+      "grad_norm": 15.125,
+      "learning_rate": 9.071227405368041e-07,
+      "loss": 0.5589,
+      "step": 13852
+    },
+    {
+      "epoch": 1.1804857264593098,
+      "grad_norm": 29.875,
+      "learning_rate": 9.069611533144057e-07,
+      "loss": 0.3087,
+      "step": 13853
+    },
+    {
+      "epoch": 1.1805709416276098,
+      "grad_norm": 16.125,
+      "learning_rate": 9.067995722910639e-07,
+      "loss": 0.5764,
+      "step": 13854
+    },
+    {
+      "epoch": 1.1806561567959097,
+      "grad_norm": 18.75,
+      "learning_rate": 9.066379974696989e-07,
+      "loss": 0.6613,
+      "step": 13855
+    },
+    {
+      "epoch": 1.1807413719642097,
+      "grad_norm": 16.0,
+      "learning_rate": 9.064764288532308e-07,
+      "loss": 0.7291,
+      "step": 13856
+    },
+    {
+      "epoch": 1.1808265871325097,
+      "grad_norm": 19.25,
+      "learning_rate": 9.063148664445789e-07,
+      "loss": 0.6896,
+      "step": 13857
+    },
+    {
+      "epoch": 1.1809118023008096,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.06153310246663e-07,
+      "loss": 0.4896,
+      "step": 13858
+    },
+    {
+      "epoch": 1.1809970174691096,
+      "grad_norm": 13.625,
+      "learning_rate": 9.059917602624027e-07,
+      "loss": 0.525,
+      "step": 13859
+    },
+    {
+      "epoch": 1.1810822326374095,
+      "grad_norm": 19.25,
+      "learning_rate": 9.058302164947174e-07,
+      "loss": 0.7842,
+      "step": 13860
+    },
+    {
+      "epoch": 1.1811674478057095,
+      "grad_norm": 19.125,
+      "learning_rate": 9.056686789465258e-07,
+      "loss": 0.3967,
+      "step": 13861
+    },
+    {
+      "epoch": 1.1812526629740094,
+      "grad_norm": 15.0,
+      "learning_rate": 9.055071476207477e-07,
+      "loss": 0.7483,
+      "step": 13862
+    },
+    {
+      "epoch": 1.1813378781423094,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.053456225203016e-07,
+      "loss": 0.6002,
+      "step": 13863
+    },
+    {
+      "epoch": 1.1814230933106094,
+      "grad_norm": 17.5,
+      "learning_rate": 9.051841036481068e-07,
+      "loss": 0.7455,
+      "step": 13864
+    },
+    {
+      "epoch": 1.1815083084789093,
+      "grad_norm": 13.125,
+      "learning_rate": 9.050225910070817e-07,
+      "loss": 0.4873,
+      "step": 13865
+    },
+    {
+      "epoch": 1.1815935236472093,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.048610846001457e-07,
+      "loss": 0.5632,
+      "step": 13866
+    },
+    {
+      "epoch": 1.1816787388155092,
+      "grad_norm": 14.5,
+      "learning_rate": 9.04699584430217e-07,
+      "loss": 0.6127,
+      "step": 13867
+    },
+    {
+      "epoch": 1.1817639539838092,
+      "grad_norm": 15.375,
+      "learning_rate": 9.045380905002141e-07,
+      "loss": 0.7485,
+      "step": 13868
+    },
+    {
+      "epoch": 1.1818491691521091,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.043766028130549e-07,
+      "loss": 0.8294,
+      "step": 13869
+    },
+    {
+      "epoch": 1.181934384320409,
+      "grad_norm": 17.5,
+      "learning_rate": 9.042151213716582e-07,
+      "loss": 0.548,
+      "step": 13870
+    },
+    {
+      "epoch": 1.182019599488709,
+      "grad_norm": 17.625,
+      "learning_rate": 9.040536461789418e-07,
+      "loss": 0.7102,
+      "step": 13871
+    },
+    {
+      "epoch": 1.182104814657009,
+      "grad_norm": 17.0,
+      "learning_rate": 9.038921772378242e-07,
+      "loss": 0.4423,
+      "step": 13872
+    },
+    {
+      "epoch": 1.182190029825309,
+      "grad_norm": 17.125,
+      "learning_rate": 9.037307145512225e-07,
+      "loss": 0.8903,
+      "step": 13873
+    },
+    {
+      "epoch": 1.182275244993609,
+      "grad_norm": 17.5,
+      "learning_rate": 9.035692581220555e-07,
+      "loss": 0.5376,
+      "step": 13874
+    },
+    {
+      "epoch": 1.1823604601619089,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.0340780795324e-07,
+      "loss": 0.4616,
+      "step": 13875
+    },
+    {
+      "epoch": 1.1824456753302088,
+      "grad_norm": 15.375,
+      "learning_rate": 9.032463640476947e-07,
+      "loss": 0.7267,
+      "step": 13876
+    },
+    {
+      "epoch": 1.1825308904985088,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.030849264083357e-07,
+      "loss": 0.6071,
+      "step": 13877
+    },
+    {
+      "epoch": 1.1826161056668087,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.029234950380808e-07,
+      "loss": 0.4414,
+      "step": 13878
+    },
+    {
+      "epoch": 1.1827013208351087,
+      "grad_norm": 19.0,
+      "learning_rate": 9.027620699398474e-07,
+      "loss": 0.7998,
+      "step": 13879
+    },
+    {
+      "epoch": 1.1827865360034087,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.026006511165527e-07,
+      "loss": 0.4086,
+      "step": 13880
+    },
+    {
+      "epoch": 1.1828717511717086,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.024392385711135e-07,
+      "loss": 0.5875,
+      "step": 13881
+    },
+    {
+      "epoch": 1.1829569663400086,
+      "grad_norm": 19.0,
+      "learning_rate": 9.022778323064466e-07,
+      "loss": 0.8234,
+      "step": 13882
+    },
+    {
+      "epoch": 1.1830421815083085,
+      "grad_norm": 15.5,
+      "learning_rate": 9.021164323254692e-07,
+      "loss": 0.5214,
+      "step": 13883
+    },
+    {
+      "epoch": 1.1831273966766085,
+      "grad_norm": 15.75,
+      "learning_rate": 9.019550386310978e-07,
+      "loss": 0.7611,
+      "step": 13884
+    },
+    {
+      "epoch": 1.1832126118449084,
+      "grad_norm": 19.75,
+      "learning_rate": 9.01793651226249e-07,
+      "loss": 0.7538,
+      "step": 13885
+    },
+    {
+      "epoch": 1.1832978270132084,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.016322701138386e-07,
+      "loss": 0.6235,
+      "step": 13886
+    },
+    {
+      "epoch": 1.1833830421815084,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.014708952967837e-07,
+      "loss": 0.5189,
+      "step": 13887
+    },
+    {
+      "epoch": 1.1834682573498083,
+      "grad_norm": 18.0,
+      "learning_rate": 9.013095267779998e-07,
+      "loss": 1.1996,
+      "step": 13888
+    },
+    {
+      "epoch": 1.1835534725181083,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.011481645604036e-07,
+      "loss": 0.8322,
+      "step": 13889
+    },
+    {
+      "epoch": 1.1836386876864082,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.009868086469108e-07,
+      "loss": 0.5539,
+      "step": 13890
+    },
+    {
+      "epoch": 1.1837239028547082,
+      "grad_norm": 14.375,
+      "learning_rate": 9.008254590404375e-07,
+      "loss": 0.6563,
+      "step": 13891
+    },
+    {
+      "epoch": 1.1838091180230081,
+      "grad_norm": 21.125,
+      "learning_rate": 9.006641157438989e-07,
+      "loss": 0.6192,
+      "step": 13892
+    },
+    {
+      "epoch": 1.183894333191308,
+      "grad_norm": 16.625,
+      "learning_rate": 9.005027787602117e-07,
+      "loss": 0.585,
+      "step": 13893
+    },
+    {
+      "epoch": 1.183979548359608,
+      "grad_norm": 14.0,
+      "learning_rate": 9.003414480922904e-07,
+      "loss": 0.544,
+      "step": 13894
+    },
+    {
+      "epoch": 1.184064763527908,
+      "grad_norm": 17.125,
+      "learning_rate": 9.001801237430505e-07,
+      "loss": 0.7996,
+      "step": 13895
+    },
+    {
+      "epoch": 1.184149978696208,
+      "grad_norm": 17.875,
+      "learning_rate": 9.000188057154077e-07,
+      "loss": 0.7155,
+      "step": 13896
+    },
+    {
+      "epoch": 1.184235193864508,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.99857494012277e-07,
+      "loss": 0.368,
+      "step": 13897
+    },
+    {
+      "epoch": 1.1843204090328079,
+      "grad_norm": 14.0,
+      "learning_rate": 8.996961886365735e-07,
+      "loss": 0.5922,
+      "step": 13898
+    },
+    {
+      "epoch": 1.1844056242011078,
+      "grad_norm": 20.0,
+      "learning_rate": 8.995348895912123e-07,
+      "loss": 0.6272,
+      "step": 13899
+    },
+    {
+      "epoch": 1.1844908393694078,
+      "grad_norm": 20.625,
+      "learning_rate": 8.993735968791076e-07,
+      "loss": 0.8241,
+      "step": 13900
+    },
+    {
+      "epoch": 1.1845760545377078,
+      "grad_norm": 20.0,
+      "learning_rate": 8.992123105031752e-07,
+      "loss": 0.796,
+      "step": 13901
+    },
+    {
+      "epoch": 1.1846612697060077,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.990510304663288e-07,
+      "loss": 0.6819,
+      "step": 13902
+    },
+    {
+      "epoch": 1.1847464848743077,
+      "grad_norm": 17.375,
+      "learning_rate": 8.988897567714831e-07,
+      "loss": 0.6611,
+      "step": 13903
+    },
+    {
+      "epoch": 1.1848317000426076,
+      "grad_norm": 16.375,
+      "learning_rate": 8.987284894215528e-07,
+      "loss": 0.4065,
+      "step": 13904
+    },
+    {
+      "epoch": 1.1849169152109076,
+      "grad_norm": 39.5,
+      "learning_rate": 8.985672284194517e-07,
+      "loss": 0.6687,
+      "step": 13905
+    },
+    {
+      "epoch": 1.1850021303792075,
+      "grad_norm": 10.0,
+      "learning_rate": 8.984059737680941e-07,
+      "loss": 0.2108,
+      "step": 13906
+    },
+    {
+      "epoch": 1.1850873455475075,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.982447254703941e-07,
+      "loss": 0.6192,
+      "step": 13907
+    },
+    {
+      "epoch": 1.1851725607158075,
+      "grad_norm": 16.125,
+      "learning_rate": 8.980834835292658e-07,
+      "loss": 0.7532,
+      "step": 13908
+    },
+    {
+      "epoch": 1.1852577758841074,
+      "grad_norm": 16.5,
+      "learning_rate": 8.979222479476232e-07,
+      "loss": 0.7998,
+      "step": 13909
+    },
+    {
+      "epoch": 1.1853429910524074,
+      "grad_norm": 19.25,
+      "learning_rate": 8.97761018728379e-07,
+      "loss": 0.9819,
+      "step": 13910
+    },
+    {
+      "epoch": 1.1854282062207073,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.975997958744477e-07,
+      "loss": 0.6064,
+      "step": 13911
+    },
+    {
+      "epoch": 1.1855134213890073,
+      "grad_norm": 18.5,
+      "learning_rate": 8.974385793887422e-07,
+      "loss": 0.6795,
+      "step": 13912
+    },
+    {
+      "epoch": 1.1855986365573072,
+      "grad_norm": 14.0,
+      "learning_rate": 8.972773692741762e-07,
+      "loss": 0.5897,
+      "step": 13913
+    },
+    {
+      "epoch": 1.1856838517256072,
+      "grad_norm": 15.125,
+      "learning_rate": 8.971161655336631e-07,
+      "loss": 0.4291,
+      "step": 13914
+    },
+    {
+      "epoch": 1.1857690668939072,
+      "grad_norm": 14.0,
+      "learning_rate": 8.969549681701153e-07,
+      "loss": 0.6136,
+      "step": 13915
+    },
+    {
+      "epoch": 1.185854282062207,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.967937771864466e-07,
+      "loss": 0.5899,
+      "step": 13916
+    },
+    {
+      "epoch": 1.185939497230507,
+      "grad_norm": 15.625,
+      "learning_rate": 8.966325925855698e-07,
+      "loss": 0.6131,
+      "step": 13917
+    },
+    {
+      "epoch": 1.186024712398807,
+      "grad_norm": 16.5,
+      "learning_rate": 8.964714143703968e-07,
+      "loss": 0.3838,
+      "step": 13918
+    },
+    {
+      "epoch": 1.186109927567107,
+      "grad_norm": 12.375,
+      "learning_rate": 8.963102425438411e-07,
+      "loss": 0.3785,
+      "step": 13919
+    },
+    {
+      "epoch": 1.186195142735407,
+      "grad_norm": 15.625,
+      "learning_rate": 8.961490771088146e-07,
+      "loss": 0.7403,
+      "step": 13920
+    },
+    {
+      "epoch": 1.186280357903707,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.959879180682307e-07,
+      "loss": 0.3227,
+      "step": 13921
+    },
+    {
+      "epoch": 1.1863655730720069,
+      "grad_norm": 19.0,
+      "learning_rate": 8.958267654250005e-07,
+      "loss": 0.942,
+      "step": 13922
+    },
+    {
+      "epoch": 1.1864507882403068,
+      "grad_norm": 20.625,
+      "learning_rate": 8.956656191820373e-07,
+      "loss": 0.6239,
+      "step": 13923
+    },
+    {
+      "epoch": 1.1865360034086068,
+      "grad_norm": 10.375,
+      "learning_rate": 8.955044793422522e-07,
+      "loss": 0.2948,
+      "step": 13924
+    },
+    {
+      "epoch": 1.1866212185769067,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.953433459085584e-07,
+      "loss": 0.5747,
+      "step": 13925
+    },
+    {
+      "epoch": 1.1867064337452067,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.951822188838666e-07,
+      "loss": 0.235,
+      "step": 13926
+    },
+    {
+      "epoch": 1.1867916489135066,
+      "grad_norm": 16.125,
+      "learning_rate": 8.950210982710885e-07,
+      "loss": 0.7468,
+      "step": 13927
+    },
+    {
+      "epoch": 1.1868768640818066,
+      "grad_norm": 19.0,
+      "learning_rate": 8.948599840731364e-07,
+      "loss": 0.8483,
+      "step": 13928
+    },
+    {
+      "epoch": 1.1869620792501065,
+      "grad_norm": 17.375,
+      "learning_rate": 8.946988762929218e-07,
+      "loss": 0.7305,
+      "step": 13929
+    },
+    {
+      "epoch": 1.1870472944184065,
+      "grad_norm": 25.75,
+      "learning_rate": 8.945377749333553e-07,
+      "loss": 0.6006,
+      "step": 13930
+    },
+    {
+      "epoch": 1.1871325095867065,
+      "grad_norm": 17.75,
+      "learning_rate": 8.943766799973489e-07,
+      "loss": 0.5573,
+      "step": 13931
+    },
+    {
+      "epoch": 1.1872177247550064,
+      "grad_norm": 19.125,
+      "learning_rate": 8.942155914878132e-07,
+      "loss": 0.7629,
+      "step": 13932
+    },
+    {
+      "epoch": 1.1873029399233064,
+      "grad_norm": 21.0,
+      "learning_rate": 8.940545094076602e-07,
+      "loss": 0.6928,
+      "step": 13933
+    },
+    {
+      "epoch": 1.1873881550916063,
+      "grad_norm": 18.0,
+      "learning_rate": 8.938934337597999e-07,
+      "loss": 0.5207,
+      "step": 13934
+    },
+    {
+      "epoch": 1.1874733702599063,
+      "grad_norm": 10.75,
+      "learning_rate": 8.937323645471429e-07,
+      "loss": 0.2774,
+      "step": 13935
+    },
+    {
+      "epoch": 1.1875585854282062,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.935713017726006e-07,
+      "loss": 0.3826,
+      "step": 13936
+    },
+    {
+      "epoch": 1.1876438005965062,
+      "grad_norm": 11.3125,
+      "learning_rate": 8.934102454390832e-07,
+      "loss": 0.5029,
+      "step": 13937
+    },
+    {
+      "epoch": 1.1877290157648062,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.932491955495015e-07,
+      "loss": 0.3429,
+      "step": 13938
+    },
+    {
+      "epoch": 1.1878142309331061,
+      "grad_norm": 18.0,
+      "learning_rate": 8.930881521067652e-07,
+      "loss": 0.5205,
+      "step": 13939
+    },
+    {
+      "epoch": 1.187899446101406,
+      "grad_norm": 16.375,
+      "learning_rate": 8.929271151137852e-07,
+      "loss": 0.6538,
+      "step": 13940
+    },
+    {
+      "epoch": 1.187984661269706,
+      "grad_norm": 20.25,
+      "learning_rate": 8.927660845734714e-07,
+      "loss": 0.6821,
+      "step": 13941
+    },
+    {
+      "epoch": 1.188069876438006,
+      "grad_norm": 17.75,
+      "learning_rate": 8.926050604887334e-07,
+      "loss": 0.7899,
+      "step": 13942
+    },
+    {
+      "epoch": 1.188155091606306,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.924440428624815e-07,
+      "loss": 0.491,
+      "step": 13943
+    },
+    {
+      "epoch": 1.188240306774606,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.922830316976252e-07,
+      "loss": 0.3275,
+      "step": 13944
+    },
+    {
+      "epoch": 1.1883255219429059,
+      "grad_norm": 16.75,
+      "learning_rate": 8.921220269970743e-07,
+      "loss": 0.6647,
+      "step": 13945
+    },
+    {
+      "epoch": 1.1884107371112058,
+      "grad_norm": 16.375,
+      "learning_rate": 8.919610287637381e-07,
+      "loss": 0.7892,
+      "step": 13946
+    },
+    {
+      "epoch": 1.1884959522795058,
+      "grad_norm": 16.75,
+      "learning_rate": 8.918000370005261e-07,
+      "loss": 0.8509,
+      "step": 13947
+    },
+    {
+      "epoch": 1.1885811674478057,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.916390517103479e-07,
+      "loss": 0.3842,
+      "step": 13948
+    },
+    {
+      "epoch": 1.1886663826161057,
+      "grad_norm": 12.625,
+      "learning_rate": 8.914780728961121e-07,
+      "loss": 0.3759,
+      "step": 13949
+    },
+    {
+      "epoch": 1.1887515977844056,
+      "grad_norm": 15.8125,
+      "learning_rate": 8.913171005607285e-07,
+      "loss": 0.6726,
+      "step": 13950
+    },
+    {
+      "epoch": 1.1888368129527056,
+      "grad_norm": 20.375,
+      "learning_rate": 8.911561347071055e-07,
+      "loss": 0.6583,
+      "step": 13951
+    },
+    {
+      "epoch": 1.1889220281210056,
+      "grad_norm": 16.125,
+      "learning_rate": 8.909951753381515e-07,
+      "loss": 0.7319,
+      "step": 13952
+    },
+    {
+      "epoch": 1.1890072432893055,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.908342224567762e-07,
+      "loss": 0.2693,
+      "step": 13953
+    },
+    {
+      "epoch": 1.1890924584576055,
+      "grad_norm": 18.75,
+      "learning_rate": 8.906732760658871e-07,
+      "loss": 0.4593,
+      "step": 13954
+    },
+    {
+      "epoch": 1.1891776736259054,
+      "grad_norm": 12.875,
+      "learning_rate": 8.905123361683937e-07,
+      "loss": 0.3662,
+      "step": 13955
+    },
+    {
+      "epoch": 1.1892628887942054,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.903514027672036e-07,
+      "loss": 0.6233,
+      "step": 13956
+    },
+    {
+      "epoch": 1.1893481039625053,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.901904758652254e-07,
+      "loss": 0.7191,
+      "step": 13957
+    },
+    {
+      "epoch": 1.1894333191308053,
+      "grad_norm": 13.375,
+      "learning_rate": 8.900295554653676e-07,
+      "loss": 0.4753,
+      "step": 13958
+    },
+    {
+      "epoch": 1.1895185342991053,
+      "grad_norm": 21.5,
+      "learning_rate": 8.898686415705374e-07,
+      "loss": 0.5589,
+      "step": 13959
+    },
+    {
+      "epoch": 1.1896037494674052,
+      "grad_norm": 17.75,
+      "learning_rate": 8.897077341836427e-07,
+      "loss": 0.9451,
+      "step": 13960
+    },
+    {
+      "epoch": 1.1896889646357052,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.895468333075918e-07,
+      "loss": 0.7401,
+      "step": 13961
+    },
+    {
+      "epoch": 1.1897741798040051,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.893859389452919e-07,
+      "loss": 0.2874,
+      "step": 13962
+    },
+    {
+      "epoch": 1.189859394972305,
+      "grad_norm": 16.625,
+      "learning_rate": 8.892250510996508e-07,
+      "loss": 0.7872,
+      "step": 13963
+    },
+    {
+      "epoch": 1.189944610140605,
+      "grad_norm": 12.8125,
+      "learning_rate": 8.890641697735757e-07,
+      "loss": 0.3925,
+      "step": 13964
+    },
+    {
+      "epoch": 1.190029825308905,
+      "grad_norm": 12.375,
+      "learning_rate": 8.889032949699742e-07,
+      "loss": 0.4832,
+      "step": 13965
+    },
+    {
+      "epoch": 1.190115040477205,
+      "grad_norm": 17.75,
+      "learning_rate": 8.887424266917535e-07,
+      "loss": 0.5644,
+      "step": 13966
+    },
+    {
+      "epoch": 1.190200255645505,
+      "grad_norm": 11.0,
+      "learning_rate": 8.8858156494182e-07,
+      "loss": 0.3026,
+      "step": 13967
+    },
+    {
+      "epoch": 1.1902854708138049,
+      "grad_norm": 12.8125,
+      "learning_rate": 8.884207097230812e-07,
+      "loss": 0.3956,
+      "step": 13968
+    },
+    {
+      "epoch": 1.1903706859821048,
+      "grad_norm": 14.9375,
+      "learning_rate": 8.882598610384437e-07,
+      "loss": 0.5994,
+      "step": 13969
+    },
+    {
+      "epoch": 1.1904559011504048,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.880990188908143e-07,
+      "loss": 0.6867,
+      "step": 13970
+    },
+    {
+      "epoch": 1.1905411163187047,
+      "grad_norm": 12.25,
+      "learning_rate": 8.879381832830994e-07,
+      "loss": 0.3718,
+      "step": 13971
+    },
+    {
+      "epoch": 1.1906263314870047,
+      "grad_norm": 17.125,
+      "learning_rate": 8.87777354218206e-07,
+      "loss": 0.4537,
+      "step": 13972
+    },
+    {
+      "epoch": 1.1907115466553047,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.876165316990396e-07,
+      "loss": 0.7113,
+      "step": 13973
+    },
+    {
+      "epoch": 1.1907967618236046,
+      "grad_norm": 14.0,
+      "learning_rate": 8.874557157285077e-07,
+      "loss": 0.418,
+      "step": 13974
+    },
+    {
+      "epoch": 1.1908819769919046,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.872949063095148e-07,
+      "loss": 0.6324,
+      "step": 13975
+    },
+    {
+      "epoch": 1.1909671921602045,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.871341034449679e-07,
+      "loss": 0.5445,
+      "step": 13976
+    },
+    {
+      "epoch": 1.1910524073285045,
+      "grad_norm": 12.75,
+      "learning_rate": 8.869733071377725e-07,
+      "loss": 0.3347,
+      "step": 13977
+    },
+    {
+      "epoch": 1.1911376224968044,
+      "grad_norm": 18.375,
+      "learning_rate": 8.868125173908346e-07,
+      "loss": 0.9493,
+      "step": 13978
+    },
+    {
+      "epoch": 1.1912228376651044,
+      "grad_norm": 17.875,
+      "learning_rate": 8.866517342070596e-07,
+      "loss": 0.7775,
+      "step": 13979
+    },
+    {
+      "epoch": 1.1913080528334044,
+      "grad_norm": 11.0625,
+      "learning_rate": 8.864909575893534e-07,
+      "loss": 0.4234,
+      "step": 13980
+    },
+    {
+      "epoch": 1.1913932680017043,
+      "grad_norm": 15.25,
+      "learning_rate": 8.863301875406208e-07,
+      "loss": 0.5289,
+      "step": 13981
+    },
+    {
+      "epoch": 1.1914784831700043,
+      "grad_norm": 17.625,
+      "learning_rate": 8.86169424063768e-07,
+      "loss": 0.7465,
+      "step": 13982
+    },
+    {
+      "epoch": 1.1915636983383042,
+      "grad_norm": 12.875,
+      "learning_rate": 8.860086671616991e-07,
+      "loss": 0.3823,
+      "step": 13983
+    },
+    {
+      "epoch": 1.1916489135066042,
+      "grad_norm": 23.0,
+      "learning_rate": 8.858479168373196e-07,
+      "loss": 0.7136,
+      "step": 13984
+    },
+    {
+      "epoch": 1.1917341286749041,
+      "grad_norm": 14.125,
+      "learning_rate": 8.856871730935346e-07,
+      "loss": 0.605,
+      "step": 13985
+    },
+    {
+      "epoch": 1.191819343843204,
+      "grad_norm": 9.6875,
+      "learning_rate": 8.855264359332483e-07,
+      "loss": 0.1831,
+      "step": 13986
+    },
+    {
+      "epoch": 1.191904559011504,
+      "grad_norm": 12.75,
+      "learning_rate": 8.853657053593664e-07,
+      "loss": 0.2646,
+      "step": 13987
+    },
+    {
+      "epoch": 1.191989774179804,
+      "grad_norm": 15.75,
+      "learning_rate": 8.852049813747924e-07,
+      "loss": 0.3613,
+      "step": 13988
+    },
+    {
+      "epoch": 1.192074989348104,
+      "grad_norm": 10.375,
+      "learning_rate": 8.850442639824314e-07,
+      "loss": 0.3577,
+      "step": 13989
+    },
+    {
+      "epoch": 1.192160204516404,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.848835531851879e-07,
+      "loss": 0.5365,
+      "step": 13990
+    },
+    {
+      "epoch": 1.1922454196847039,
+      "grad_norm": 13.125,
+      "learning_rate": 8.847228489859655e-07,
+      "loss": 0.5703,
+      "step": 13991
+    },
+    {
+      "epoch": 1.1923306348530038,
+      "grad_norm": 14.875,
+      "learning_rate": 8.845621513876683e-07,
+      "loss": 0.4117,
+      "step": 13992
+    },
+    {
+      "epoch": 1.1924158500213038,
+      "grad_norm": 16.75,
+      "learning_rate": 8.844014603932008e-07,
+      "loss": 0.8187,
+      "step": 13993
+    },
+    {
+      "epoch": 1.1925010651896037,
+      "grad_norm": 13.375,
+      "learning_rate": 8.842407760054661e-07,
+      "loss": 0.414,
+      "step": 13994
+    },
+    {
+      "epoch": 1.1925862803579037,
+      "grad_norm": 16.75,
+      "learning_rate": 8.840800982273687e-07,
+      "loss": 0.7227,
+      "step": 13995
+    },
+    {
+      "epoch": 1.1926714955262037,
+      "grad_norm": 12.5,
+      "learning_rate": 8.839194270618118e-07,
+      "loss": 0.2721,
+      "step": 13996
+    },
+    {
+      "epoch": 1.1927567106945036,
+      "grad_norm": 20.625,
+      "learning_rate": 8.83758762511699e-07,
+      "loss": 0.6246,
+      "step": 13997
+    },
+    {
+      "epoch": 1.1928419258628036,
+      "grad_norm": 16.5,
+      "learning_rate": 8.835981045799341e-07,
+      "loss": 0.7627,
+      "step": 13998
+    },
+    {
+      "epoch": 1.1929271410311035,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.834374532694193e-07,
+      "loss": 0.4493,
+      "step": 13999
+    },
+    {
+      "epoch": 1.1930123561994035,
+      "grad_norm": 15.6875,
+      "learning_rate": 8.832768085830585e-07,
+      "loss": 0.6372,
+      "step": 14000
+    },
+    {
+      "epoch": 1.1930975713677034,
+      "grad_norm": 25.5,
+      "learning_rate": 8.831161705237544e-07,
+      "loss": 0.6108,
+      "step": 14001
+    },
+    {
+      "epoch": 1.1931827865360034,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.829555390944103e-07,
+      "loss": 0.6974,
+      "step": 14002
+    },
+    {
+      "epoch": 1.1932680017043034,
+      "grad_norm": 21.0,
+      "learning_rate": 8.827949142979284e-07,
+      "loss": 1.0006,
+      "step": 14003
+    },
+    {
+      "epoch": 1.1933532168726033,
+      "grad_norm": 16.375,
+      "learning_rate": 8.826342961372118e-07,
+      "loss": 0.6715,
+      "step": 14004
+    },
+    {
+      "epoch": 1.1934384320409033,
+      "grad_norm": 15.0,
+      "learning_rate": 8.824736846151627e-07,
+      "loss": 0.5677,
+      "step": 14005
+    },
+    {
+      "epoch": 1.1935236472092032,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.823130797346846e-07,
+      "loss": 0.5664,
+      "step": 14006
+    },
+    {
+      "epoch": 1.1936088623775032,
+      "grad_norm": 18.25,
+      "learning_rate": 8.821524814986778e-07,
+      "loss": 0.973,
+      "step": 14007
+    },
+    {
+      "epoch": 1.1936940775458031,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.819918899100462e-07,
+      "loss": 0.6489,
+      "step": 14008
+    },
+    {
+      "epoch": 1.193779292714103,
+      "grad_norm": 17.625,
+      "learning_rate": 8.818313049716908e-07,
+      "loss": 0.7236,
+      "step": 14009
+    },
+    {
+      "epoch": 1.193864507882403,
+      "grad_norm": 17.75,
+      "learning_rate": 8.81670726686514e-07,
+      "loss": 0.7463,
+      "step": 14010
+    },
+    {
+      "epoch": 1.193949723050703,
+      "grad_norm": 23.25,
+      "learning_rate": 8.815101550574177e-07,
+      "loss": 1.1033,
+      "step": 14011
+    },
+    {
+      "epoch": 1.194034938219003,
+      "grad_norm": 11.125,
+      "learning_rate": 8.813495900873034e-07,
+      "loss": 0.3491,
+      "step": 14012
+    },
+    {
+      "epoch": 1.194120153387303,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.811890317790725e-07,
+      "loss": 0.306,
+      "step": 14013
+    },
+    {
+      "epoch": 1.1942053685556029,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.81028480135627e-07,
+      "loss": 0.4726,
+      "step": 14014
+    },
+    {
+      "epoch": 1.1942905837239028,
+      "grad_norm": 20.25,
+      "learning_rate": 8.808679351598681e-07,
+      "loss": 0.5591,
+      "step": 14015
+    },
+    {
+      "epoch": 1.1943757988922028,
+      "grad_norm": 17.0,
+      "learning_rate": 8.807073968546963e-07,
+      "loss": 0.6262,
+      "step": 14016
+    },
+    {
+      "epoch": 1.1944610140605028,
+      "grad_norm": 16.25,
+      "learning_rate": 8.805468652230134e-07,
+      "loss": 0.5463,
+      "step": 14017
+    },
+    {
+      "epoch": 1.1945462292288027,
+      "grad_norm": 18.375,
+      "learning_rate": 8.803863402677199e-07,
+      "loss": 0.9624,
+      "step": 14018
+    },
+    {
+      "epoch": 1.1946314443971027,
+      "grad_norm": 19.0,
+      "learning_rate": 8.802258219917173e-07,
+      "loss": 0.4853,
+      "step": 14019
+    },
+    {
+      "epoch": 1.1947166595654026,
+      "grad_norm": 29.625,
+      "learning_rate": 8.800653103979055e-07,
+      "loss": 0.911,
+      "step": 14020
+    },
+    {
+      "epoch": 1.1948018747337026,
+      "grad_norm": 17.25,
+      "learning_rate": 8.79904805489186e-07,
+      "loss": 0.6465,
+      "step": 14021
+    },
+    {
+      "epoch": 1.1948870899020025,
+      "grad_norm": 16.875,
+      "learning_rate": 8.797443072684586e-07,
+      "loss": 0.6253,
+      "step": 14022
+    },
+    {
+      "epoch": 1.1949723050703025,
+      "grad_norm": 33.75,
+      "learning_rate": 8.795838157386242e-07,
+      "loss": 0.8825,
+      "step": 14023
+    },
+    {
+      "epoch": 1.1950575202386025,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.794233309025824e-07,
+      "loss": 0.5183,
+      "step": 14024
+    },
+    {
+      "epoch": 1.1951427354069024,
+      "grad_norm": 22.625,
+      "learning_rate": 8.792628527632337e-07,
+      "loss": 0.8591,
+      "step": 14025
+    },
+    {
+      "epoch": 1.1952279505752024,
+      "grad_norm": 16.5,
+      "learning_rate": 8.791023813234778e-07,
+      "loss": 0.4932,
+      "step": 14026
+    },
+    {
+      "epoch": 1.1953131657435023,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.789419165862153e-07,
+      "loss": 0.4498,
+      "step": 14027
+    },
+    {
+      "epoch": 1.1953983809118023,
+      "grad_norm": 16.0,
+      "learning_rate": 8.78781458554345e-07,
+      "loss": 0.7703,
+      "step": 14028
+    },
+    {
+      "epoch": 1.1954835960801022,
+      "grad_norm": 24.0,
+      "learning_rate": 8.786210072307675e-07,
+      "loss": 0.7145,
+      "step": 14029
+    },
+    {
+      "epoch": 1.1955688112484022,
+      "grad_norm": 20.625,
+      "learning_rate": 8.784605626183813e-07,
+      "loss": 0.6528,
+      "step": 14030
+    },
+    {
+      "epoch": 1.1956540264167022,
+      "grad_norm": 15.3125,
+      "learning_rate": 8.78300124720087e-07,
+      "loss": 0.5606,
+      "step": 14031
+    },
+    {
+      "epoch": 1.195739241585002,
+      "grad_norm": 14.625,
+      "learning_rate": 8.781396935387832e-07,
+      "loss": 0.5712,
+      "step": 14032
+    },
+    {
+      "epoch": 1.195824456753302,
+      "grad_norm": 11.25,
+      "learning_rate": 8.779792690773686e-07,
+      "loss": 0.3576,
+      "step": 14033
+    },
+    {
+      "epoch": 1.195909671921602,
+      "grad_norm": 17.875,
+      "learning_rate": 8.778188513387428e-07,
+      "loss": 0.8496,
+      "step": 14034
+    },
+    {
+      "epoch": 1.195994887089902,
+      "grad_norm": 15.5,
+      "learning_rate": 8.776584403258046e-07,
+      "loss": 0.8143,
+      "step": 14035
+    },
+    {
+      "epoch": 1.196080102258202,
+      "grad_norm": 15.3125,
+      "learning_rate": 8.774980360414531e-07,
+      "loss": 0.3802,
+      "step": 14036
+    },
+    {
+      "epoch": 1.196165317426502,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.773376384885865e-07,
+      "loss": 0.3343,
+      "step": 14037
+    },
+    {
+      "epoch": 1.1962505325948019,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.771772476701033e-07,
+      "loss": 0.5611,
+      "step": 14038
+    },
+    {
+      "epoch": 1.1963357477631018,
+      "grad_norm": 14.875,
+      "learning_rate": 8.770168635889027e-07,
+      "loss": 0.7927,
+      "step": 14039
+    },
+    {
+      "epoch": 1.1964209629314018,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.768564862478822e-07,
+      "loss": 0.5856,
+      "step": 14040
+    },
+    {
+      "epoch": 1.1965061780997017,
+      "grad_norm": 17.5,
+      "learning_rate": 8.766961156499398e-07,
+      "loss": 0.5329,
+      "step": 14041
+    },
+    {
+      "epoch": 1.1965913932680017,
+      "grad_norm": 12.125,
+      "learning_rate": 8.765357517979745e-07,
+      "loss": 0.5797,
+      "step": 14042
+    },
+    {
+      "epoch": 1.1966766084363016,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.763753946948833e-07,
+      "loss": 0.3884,
+      "step": 14043
+    },
+    {
+      "epoch": 1.1967618236046016,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.762150443435646e-07,
+      "loss": 0.4425,
+      "step": 14044
+    },
+    {
+      "epoch": 1.1968470387729016,
+      "grad_norm": 16.25,
+      "learning_rate": 8.760547007469158e-07,
+      "loss": 0.4221,
+      "step": 14045
+    },
+    {
+      "epoch": 1.1969322539412015,
+      "grad_norm": 12.5,
+      "learning_rate": 8.758943639078347e-07,
+      "loss": 0.4596,
+      "step": 14046
+    },
+    {
+      "epoch": 1.1970174691095015,
+      "grad_norm": 17.5,
+      "learning_rate": 8.757340338292191e-07,
+      "loss": 0.5352,
+      "step": 14047
+    },
+    {
+      "epoch": 1.1971026842778014,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.755737105139653e-07,
+      "loss": 0.5569,
+      "step": 14048
+    },
+    {
+      "epoch": 1.1971878994461014,
+      "grad_norm": 16.375,
+      "learning_rate": 8.754133939649712e-07,
+      "loss": 0.6973,
+      "step": 14049
+    },
+    {
+      "epoch": 1.1972731146144013,
+      "grad_norm": 19.875,
+      "learning_rate": 8.752530841851335e-07,
+      "loss": 0.7481,
+      "step": 14050
+    },
+    {
+      "epoch": 1.1973583297827013,
+      "grad_norm": 13.0,
+      "learning_rate": 8.750927811773496e-07,
+      "loss": 0.467,
+      "step": 14051
+    },
+    {
+      "epoch": 1.1974435449510012,
+      "grad_norm": 18.75,
+      "learning_rate": 8.749324849445165e-07,
+      "loss": 0.5897,
+      "step": 14052
+    },
+    {
+      "epoch": 1.1975287601193012,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.7477219548953e-07,
+      "loss": 0.8792,
+      "step": 14053
+    },
+    {
+      "epoch": 1.1976139752876012,
+      "grad_norm": 10.25,
+      "learning_rate": 8.746119128152875e-07,
+      "loss": 0.3574,
+      "step": 14054
+    },
+    {
+      "epoch": 1.1976991904559011,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.744516369246855e-07,
+      "loss": 0.4556,
+      "step": 14055
+    },
+    {
+      "epoch": 1.197784405624201,
+      "grad_norm": 12.875,
+      "learning_rate": 8.742913678206197e-07,
+      "loss": 0.5211,
+      "step": 14056
+    },
+    {
+      "epoch": 1.197869620792501,
+      "grad_norm": 18.375,
+      "learning_rate": 8.741311055059868e-07,
+      "loss": 0.75,
+      "step": 14057
+    },
+    {
+      "epoch": 1.197954835960801,
+      "grad_norm": 14.0,
+      "learning_rate": 8.739708499836824e-07,
+      "loss": 0.4655,
+      "step": 14058
+    },
+    {
+      "epoch": 1.198040051129101,
+      "grad_norm": 17.25,
+      "learning_rate": 8.738106012566034e-07,
+      "loss": 1.029,
+      "step": 14059
+    },
+    {
+      "epoch": 1.198125266297401,
+      "grad_norm": 16.375,
+      "learning_rate": 8.736503593276448e-07,
+      "loss": 0.8588,
+      "step": 14060
+    },
+    {
+      "epoch": 1.1982104814657009,
+      "grad_norm": 14.125,
+      "learning_rate": 8.734901241997029e-07,
+      "loss": 0.5043,
+      "step": 14061
+    },
+    {
+      "epoch": 1.1982956966340008,
+      "grad_norm": 16.25,
+      "learning_rate": 8.733298958756727e-07,
+      "loss": 0.7785,
+      "step": 14062
+    },
+    {
+      "epoch": 1.1983809118023008,
+      "grad_norm": 18.625,
+      "learning_rate": 8.731696743584505e-07,
+      "loss": 0.6924,
+      "step": 14063
+    },
+    {
+      "epoch": 1.1984661269706007,
+      "grad_norm": 19.375,
+      "learning_rate": 8.730094596509311e-07,
+      "loss": 0.5844,
+      "step": 14064
+    },
+    {
+      "epoch": 1.1985513421389007,
+      "grad_norm": 14.0,
+      "learning_rate": 8.728492517560097e-07,
+      "loss": 0.5937,
+      "step": 14065
+    },
+    {
+      "epoch": 1.1986365573072006,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.726890506765818e-07,
+      "loss": 0.4447,
+      "step": 14066
+    },
+    {
+      "epoch": 1.1987217724755006,
+      "grad_norm": 13.5,
+      "learning_rate": 8.725288564155421e-07,
+      "loss": 0.6119,
+      "step": 14067
+    },
+    {
+      "epoch": 1.1988069876438006,
+      "grad_norm": 26.75,
+      "learning_rate": 8.723686689757852e-07,
+      "loss": 0.8273,
+      "step": 14068
+    },
+    {
+      "epoch": 1.1988922028121005,
+      "grad_norm": 13.625,
+      "learning_rate": 8.722084883602067e-07,
+      "loss": 0.5634,
+      "step": 14069
+    },
+    {
+      "epoch": 1.1989774179804005,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.720483145717001e-07,
+      "loss": 0.7647,
+      "step": 14070
+    },
+    {
+      "epoch": 1.1990626331487004,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.718881476131613e-07,
+      "loss": 0.7546,
+      "step": 14071
+    },
+    {
+      "epoch": 1.1991478483170004,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.717279874874836e-07,
+      "loss": 0.72,
+      "step": 14072
+    },
+    {
+      "epoch": 1.1992330634853003,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.715678341975612e-07,
+      "loss": 0.4732,
+      "step": 14073
+    },
+    {
+      "epoch": 1.1993182786536003,
+      "grad_norm": 19.375,
+      "learning_rate": 8.714076877462889e-07,
+      "loss": 0.6179,
+      "step": 14074
+    },
+    {
+      "epoch": 1.1994034938219003,
+      "grad_norm": 30.125,
+      "learning_rate": 8.7124754813656e-07,
+      "loss": 1.0587,
+      "step": 14075
+    },
+    {
+      "epoch": 1.1994887089902002,
+      "grad_norm": 19.25,
+      "learning_rate": 8.710874153712693e-07,
+      "loss": 0.9087,
+      "step": 14076
+    },
+    {
+      "epoch": 1.1995739241585002,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.709272894533096e-07,
+      "loss": 0.5339,
+      "step": 14077
+    },
+    {
+      "epoch": 1.1996591393268001,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.707671703855752e-07,
+      "loss": 0.492,
+      "step": 14078
+    },
+    {
+      "epoch": 1.1997443544951,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.706070581709597e-07,
+      "loss": 0.8418,
+      "step": 14079
+    },
+    {
+      "epoch": 1.1998295696634,
+      "grad_norm": 17.5,
+      "learning_rate": 8.704469528123554e-07,
+      "loss": 0.6859,
+      "step": 14080
+    },
+    {
+      "epoch": 1.1999147848317,
+      "grad_norm": 17.375,
+      "learning_rate": 8.702868543126569e-07,
+      "loss": 0.5632,
+      "step": 14081
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 13.375,
+      "learning_rate": 8.701267626747567e-07,
+      "loss": 0.5121,
+      "step": 14082
+    },
+    {
+      "epoch": 1.2000852151683,
+      "grad_norm": 17.75,
+      "learning_rate": 8.699666779015476e-07,
+      "loss": 0.3943,
+      "step": 14083
+    },
+    {
+      "epoch": 1.2001704303365999,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.698065999959232e-07,
+      "loss": 0.431,
+      "step": 14084
+    },
+    {
+      "epoch": 1.2002556455048998,
+      "grad_norm": 33.0,
+      "learning_rate": 8.696465289607754e-07,
+      "loss": 0.5714,
+      "step": 14085
+    },
+    {
+      "epoch": 1.2003408606731998,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.694864647989975e-07,
+      "loss": 0.4568,
+      "step": 14086
+    },
+    {
+      "epoch": 1.2004260758414997,
+      "grad_norm": 23.125,
+      "learning_rate": 8.693264075134818e-07,
+      "loss": 0.3078,
+      "step": 14087
+    },
+    {
+      "epoch": 1.2005112910097997,
+      "grad_norm": 31.375,
+      "learning_rate": 8.691663571071211e-07,
+      "loss": 0.6994,
+      "step": 14088
+    },
+    {
+      "epoch": 1.2005965061780997,
+      "grad_norm": 14.75,
+      "learning_rate": 8.690063135828069e-07,
+      "loss": 0.7324,
+      "step": 14089
+    },
+    {
+      "epoch": 1.2006817213463996,
+      "grad_norm": 16.25,
+      "learning_rate": 8.688462769434316e-07,
+      "loss": 0.7098,
+      "step": 14090
+    },
+    {
+      "epoch": 1.2007669365146996,
+      "grad_norm": 17.125,
+      "learning_rate": 8.686862471918877e-07,
+      "loss": 0.4589,
+      "step": 14091
+    },
+    {
+      "epoch": 1.2008521516829995,
+      "grad_norm": 38.0,
+      "learning_rate": 8.685262243310665e-07,
+      "loss": 0.7908,
+      "step": 14092
+    },
+    {
+      "epoch": 1.2009373668512995,
+      "grad_norm": 23.75,
+      "learning_rate": 8.683662083638601e-07,
+      "loss": 1.1645,
+      "step": 14093
+    },
+    {
+      "epoch": 1.2010225820195994,
+      "grad_norm": 18.75,
+      "learning_rate": 8.6820619929316e-07,
+      "loss": 0.7679,
+      "step": 14094
+    },
+    {
+      "epoch": 1.2011077971878994,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.680461971218579e-07,
+      "loss": 0.4885,
+      "step": 14095
+    },
+    {
+      "epoch": 1.2011930123561994,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.678862018528455e-07,
+      "loss": 0.4537,
+      "step": 14096
+    },
+    {
+      "epoch": 1.2012782275244993,
+      "grad_norm": 21.625,
+      "learning_rate": 8.677262134890133e-07,
+      "loss": 0.8193,
+      "step": 14097
+    },
+    {
+      "epoch": 1.2013634426927993,
+      "grad_norm": 20.125,
+      "learning_rate": 8.675662320332526e-07,
+      "loss": 0.514,
+      "step": 14098
+    },
+    {
+      "epoch": 1.2014486578610992,
+      "grad_norm": 18.5,
+      "learning_rate": 8.674062574884551e-07,
+      "loss": 0.8436,
+      "step": 14099
+    },
+    {
+      "epoch": 1.2015338730293992,
+      "grad_norm": 16.125,
+      "learning_rate": 8.672462898575107e-07,
+      "loss": 0.9233,
+      "step": 14100
+    },
+    {
+      "epoch": 1.2016190881976991,
+      "grad_norm": 18.625,
+      "learning_rate": 8.670863291433113e-07,
+      "loss": 1.1457,
+      "step": 14101
+    },
+    {
+      "epoch": 1.201704303365999,
+      "grad_norm": 16.875,
+      "learning_rate": 8.669263753487464e-07,
+      "loss": 0.7729,
+      "step": 14102
+    },
+    {
+      "epoch": 1.201789518534299,
+      "grad_norm": 17.5,
+      "learning_rate": 8.667664284767075e-07,
+      "loss": 0.9133,
+      "step": 14103
+    },
+    {
+      "epoch": 1.201874733702599,
+      "grad_norm": 32.5,
+      "learning_rate": 8.666064885300848e-07,
+      "loss": 0.6118,
+      "step": 14104
+    },
+    {
+      "epoch": 1.201959948870899,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.664465555117681e-07,
+      "loss": 0.3658,
+      "step": 14105
+    },
+    {
+      "epoch": 1.202045164039199,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.662866294246479e-07,
+      "loss": 0.6481,
+      "step": 14106
+    },
+    {
+      "epoch": 1.2021303792074989,
+      "grad_norm": 17.625,
+      "learning_rate": 8.661267102716139e-07,
+      "loss": 0.4409,
+      "step": 14107
+    },
+    {
+      "epoch": 1.2022155943757988,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.659667980555567e-07,
+      "loss": 0.314,
+      "step": 14108
+    },
+    {
+      "epoch": 1.2023008095440988,
+      "grad_norm": 19.625,
+      "learning_rate": 8.658068927793651e-07,
+      "loss": 0.6957,
+      "step": 14109
+    },
+    {
+      "epoch": 1.2023860247123987,
+      "grad_norm": 14.625,
+      "learning_rate": 8.656469944459297e-07,
+      "loss": 0.4927,
+      "step": 14110
+    },
+    {
+      "epoch": 1.2024712398806987,
+      "grad_norm": 13.0,
+      "learning_rate": 8.654871030581394e-07,
+      "loss": 0.3295,
+      "step": 14111
+    },
+    {
+      "epoch": 1.2025564550489987,
+      "grad_norm": 13.0,
+      "learning_rate": 8.653272186188844e-07,
+      "loss": 0.5615,
+      "step": 14112
+    },
+    {
+      "epoch": 1.2026416702172986,
+      "grad_norm": 17.375,
+      "learning_rate": 8.651673411310526e-07,
+      "loss": 0.7884,
+      "step": 14113
+    },
+    {
+      "epoch": 1.2027268853855986,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.650074705975345e-07,
+      "loss": 0.5495,
+      "step": 14114
+    },
+    {
+      "epoch": 1.2028121005538985,
+      "grad_norm": 18.25,
+      "learning_rate": 8.64847607021218e-07,
+      "loss": 0.7949,
+      "step": 14115
+    },
+    {
+      "epoch": 1.2028973157221985,
+      "grad_norm": 13.875,
+      "learning_rate": 8.646877504049929e-07,
+      "loss": 0.6935,
+      "step": 14116
+    },
+    {
+      "epoch": 1.2029825308904984,
+      "grad_norm": 20.125,
+      "learning_rate": 8.645279007517474e-07,
+      "loss": 0.632,
+      "step": 14117
+    },
+    {
+      "epoch": 1.2030677460587984,
+      "grad_norm": 28.25,
+      "learning_rate": 8.643680580643704e-07,
+      "loss": 1.6578,
+      "step": 14118
+    },
+    {
+      "epoch": 1.2031529612270984,
+      "grad_norm": 15.75,
+      "learning_rate": 8.642082223457504e-07,
+      "loss": 0.7916,
+      "step": 14119
+    },
+    {
+      "epoch": 1.2032381763953983,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.640483935987762e-07,
+      "loss": 0.5402,
+      "step": 14120
+    },
+    {
+      "epoch": 1.2033233915636983,
+      "grad_norm": 17.25,
+      "learning_rate": 8.638885718263352e-07,
+      "loss": 0.8929,
+      "step": 14121
+    },
+    {
+      "epoch": 1.2034086067319982,
+      "grad_norm": 13.375,
+      "learning_rate": 8.637287570313159e-07,
+      "loss": 0.2864,
+      "step": 14122
+    },
+    {
+      "epoch": 1.2034938219002982,
+      "grad_norm": 17.625,
+      "learning_rate": 8.635689492166066e-07,
+      "loss": 0.5802,
+      "step": 14123
+    },
+    {
+      "epoch": 1.2035790370685981,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.634091483850946e-07,
+      "loss": 0.5603,
+      "step": 14124
+    },
+    {
+      "epoch": 1.203664252236898,
+      "grad_norm": 12.125,
+      "learning_rate": 8.632493545396683e-07,
+      "loss": 0.3803,
+      "step": 14125
+    },
+    {
+      "epoch": 1.203749467405198,
+      "grad_norm": 14.75,
+      "learning_rate": 8.630895676832151e-07,
+      "loss": 0.4944,
+      "step": 14126
+    },
+    {
+      "epoch": 1.203834682573498,
+      "grad_norm": 15.3125,
+      "learning_rate": 8.629297878186224e-07,
+      "loss": 0.6526,
+      "step": 14127
+    },
+    {
+      "epoch": 1.203919897741798,
+      "grad_norm": 23.5,
+      "learning_rate": 8.627700149487782e-07,
+      "loss": 0.8024,
+      "step": 14128
+    },
+    {
+      "epoch": 1.204005112910098,
+      "grad_norm": 19.875,
+      "learning_rate": 8.626102490765688e-07,
+      "loss": 0.563,
+      "step": 14129
+    },
+    {
+      "epoch": 1.2040903280783979,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.624504902048815e-07,
+      "loss": 0.7062,
+      "step": 14130
+    },
+    {
+      "epoch": 1.2041755432466978,
+      "grad_norm": 19.125,
+      "learning_rate": 8.622907383366039e-07,
+      "loss": 0.8156,
+      "step": 14131
+    },
+    {
+      "epoch": 1.2042607584149978,
+      "grad_norm": 9.8125,
+      "learning_rate": 8.621309934746222e-07,
+      "loss": 0.2526,
+      "step": 14132
+    },
+    {
+      "epoch": 1.2043459735832978,
+      "grad_norm": 17.125,
+      "learning_rate": 8.619712556218236e-07,
+      "loss": 0.5359,
+      "step": 14133
+    },
+    {
+      "epoch": 1.2044311887515977,
+      "grad_norm": 17.0,
+      "learning_rate": 8.618115247810945e-07,
+      "loss": 1.0013,
+      "step": 14134
+    },
+    {
+      "epoch": 1.2045164039198977,
+      "grad_norm": 16.375,
+      "learning_rate": 8.616518009553218e-07,
+      "loss": 0.8414,
+      "step": 14135
+    },
+    {
+      "epoch": 1.2046016190881976,
+      "grad_norm": 22.25,
+      "learning_rate": 8.614920841473917e-07,
+      "loss": 0.729,
+      "step": 14136
+    },
+    {
+      "epoch": 1.2046868342564976,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.613323743601899e-07,
+      "loss": 0.5106,
+      "step": 14137
+    },
+    {
+      "epoch": 1.2047720494247975,
+      "grad_norm": 24.0,
+      "learning_rate": 8.611726715966032e-07,
+      "loss": 0.9207,
+      "step": 14138
+    },
+    {
+      "epoch": 1.2048572645930975,
+      "grad_norm": 23.5,
+      "learning_rate": 8.61012975859517e-07,
+      "loss": 0.7182,
+      "step": 14139
+    },
+    {
+      "epoch": 1.2049424797613975,
+      "grad_norm": 17.5,
+      "learning_rate": 8.608532871518177e-07,
+      "loss": 0.6125,
+      "step": 14140
+    },
+    {
+      "epoch": 1.2050276949296974,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.606936054763907e-07,
+      "loss": 0.7645,
+      "step": 14141
+    },
+    {
+      "epoch": 1.2051129100979974,
+      "grad_norm": 21.75,
+      "learning_rate": 8.605339308361219e-07,
+      "loss": 0.6365,
+      "step": 14142
+    },
+    {
+      "epoch": 1.2051981252662973,
+      "grad_norm": 12.25,
+      "learning_rate": 8.603742632338963e-07,
+      "loss": 0.3891,
+      "step": 14143
+    },
+    {
+      "epoch": 1.2052833404345973,
+      "grad_norm": 17.25,
+      "learning_rate": 8.602146026726005e-07,
+      "loss": 0.8285,
+      "step": 14144
+    },
+    {
+      "epoch": 1.2053685556028972,
+      "grad_norm": 12.5,
+      "learning_rate": 8.600549491551179e-07,
+      "loss": 0.3931,
+      "step": 14145
+    },
+    {
+      "epoch": 1.2054537707711972,
+      "grad_norm": 16.125,
+      "learning_rate": 8.598953026843347e-07,
+      "loss": 0.7789,
+      "step": 14146
+    },
+    {
+      "epoch": 1.2055389859394972,
+      "grad_norm": 17.25,
+      "learning_rate": 8.597356632631354e-07,
+      "loss": 0.6328,
+      "step": 14147
+    },
+    {
+      "epoch": 1.205624201107797,
+      "grad_norm": 12.375,
+      "learning_rate": 8.595760308944056e-07,
+      "loss": 0.1813,
+      "step": 14148
+    },
+    {
+      "epoch": 1.205709416276097,
+      "grad_norm": 21.375,
+      "learning_rate": 8.594164055810289e-07,
+      "loss": 0.8747,
+      "step": 14149
+    },
+    {
+      "epoch": 1.205794631444397,
+      "grad_norm": 20.375,
+      "learning_rate": 8.592567873258911e-07,
+      "loss": 0.6306,
+      "step": 14150
+    },
+    {
+      "epoch": 1.205879846612697,
+      "grad_norm": 20.75,
+      "learning_rate": 8.590971761318756e-07,
+      "loss": 0.7399,
+      "step": 14151
+    },
+    {
+      "epoch": 1.205965061780997,
+      "grad_norm": 17.5,
+      "learning_rate": 8.589375720018675e-07,
+      "loss": 0.6904,
+      "step": 14152
+    },
+    {
+      "epoch": 1.206050276949297,
+      "grad_norm": 17.0,
+      "learning_rate": 8.58777974938751e-07,
+      "loss": 0.5246,
+      "step": 14153
+    },
+    {
+      "epoch": 1.2061354921175969,
+      "grad_norm": 18.75,
+      "learning_rate": 8.586183849454094e-07,
+      "loss": 0.6141,
+      "step": 14154
+    },
+    {
+      "epoch": 1.2062207072858968,
+      "grad_norm": 15.0,
+      "learning_rate": 8.584588020247275e-07,
+      "loss": 0.5386,
+      "step": 14155
+    },
+    {
+      "epoch": 1.2063059224541968,
+      "grad_norm": 32.25,
+      "learning_rate": 8.582992261795883e-07,
+      "loss": 0.7592,
+      "step": 14156
+    },
+    {
+      "epoch": 1.2063911376224967,
+      "grad_norm": 16.375,
+      "learning_rate": 8.581396574128764e-07,
+      "loss": 0.7365,
+      "step": 14157
+    },
+    {
+      "epoch": 1.2064763527907967,
+      "grad_norm": 16.25,
+      "learning_rate": 8.579800957274748e-07,
+      "loss": 0.5283,
+      "step": 14158
+    },
+    {
+      "epoch": 1.2065615679590966,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.57820541126267e-07,
+      "loss": 0.4268,
+      "step": 14159
+    },
+    {
+      "epoch": 1.2066467831273966,
+      "grad_norm": 19.0,
+      "learning_rate": 8.576609936121363e-07,
+      "loss": 0.9223,
+      "step": 14160
+    },
+    {
+      "epoch": 1.2067319982956966,
+      "grad_norm": 14.0625,
+      "learning_rate": 8.575014531879669e-07,
+      "loss": 0.7523,
+      "step": 14161
+    },
+    {
+      "epoch": 1.2068172134639965,
+      "grad_norm": 30.5,
+      "learning_rate": 8.573419198566399e-07,
+      "loss": 0.7288,
+      "step": 14162
+    },
+    {
+      "epoch": 1.2069024286322965,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.571823936210397e-07,
+      "loss": 0.7984,
+      "step": 14163
+    },
+    {
+      "epoch": 1.2069876438005964,
+      "grad_norm": 11.75,
+      "learning_rate": 8.570228744840482e-07,
+      "loss": 0.3422,
+      "step": 14164
+    },
+    {
+      "epoch": 1.2070728589688964,
+      "grad_norm": 21.25,
+      "learning_rate": 8.568633624485489e-07,
+      "loss": 0.9414,
+      "step": 14165
+    },
+    {
+      "epoch": 1.2071580741371963,
+      "grad_norm": 17.125,
+      "learning_rate": 8.567038575174237e-07,
+      "loss": 0.488,
+      "step": 14166
+    },
+    {
+      "epoch": 1.2072432893054963,
+      "grad_norm": 15.6875,
+      "learning_rate": 8.565443596935555e-07,
+      "loss": 0.8764,
+      "step": 14167
+    },
+    {
+      "epoch": 1.2073285044737962,
+      "grad_norm": 19.75,
+      "learning_rate": 8.563848689798262e-07,
+      "loss": 0.7281,
+      "step": 14168
+    },
+    {
+      "epoch": 1.2074137196420962,
+      "grad_norm": 18.875,
+      "learning_rate": 8.562253853791186e-07,
+      "loss": 0.8583,
+      "step": 14169
+    },
+    {
+      "epoch": 1.2074989348103962,
+      "grad_norm": 16.375,
+      "learning_rate": 8.560659088943141e-07,
+      "loss": 0.5999,
+      "step": 14170
+    },
+    {
+      "epoch": 1.2075841499786961,
+      "grad_norm": 24.875,
+      "learning_rate": 8.559064395282942e-07,
+      "loss": 0.6782,
+      "step": 14171
+    },
+    {
+      "epoch": 1.207669365146996,
+      "grad_norm": 18.5,
+      "learning_rate": 8.557469772839419e-07,
+      "loss": 0.6869,
+      "step": 14172
+    },
+    {
+      "epoch": 1.207754580315296,
+      "grad_norm": 17.0,
+      "learning_rate": 8.555875221641377e-07,
+      "loss": 0.394,
+      "step": 14173
+    },
+    {
+      "epoch": 1.207839795483596,
+      "grad_norm": 13.125,
+      "learning_rate": 8.554280741717639e-07,
+      "loss": 0.4108,
+      "step": 14174
+    },
+    {
+      "epoch": 1.207925010651896,
+      "grad_norm": 12.75,
+      "learning_rate": 8.552686333097011e-07,
+      "loss": 0.4517,
+      "step": 14175
+    },
+    {
+      "epoch": 1.208010225820196,
+      "grad_norm": 11.8125,
+      "learning_rate": 8.551091995808316e-07,
+      "loss": 0.4063,
+      "step": 14176
+    },
+    {
+      "epoch": 1.2080954409884959,
+      "grad_norm": 13.5,
+      "learning_rate": 8.54949772988036e-07,
+      "loss": 0.3967,
+      "step": 14177
+    },
+    {
+      "epoch": 1.2081806561567958,
+      "grad_norm": 21.375,
+      "learning_rate": 8.54790353534195e-07,
+      "loss": 0.8404,
+      "step": 14178
+    },
+    {
+      "epoch": 1.2082658713250958,
+      "grad_norm": 23.0,
+      "learning_rate": 8.546309412221895e-07,
+      "loss": 0.9861,
+      "step": 14179
+    },
+    {
+      "epoch": 1.2083510864933957,
+      "grad_norm": 21.625,
+      "learning_rate": 8.544715360549008e-07,
+      "loss": 0.3546,
+      "step": 14180
+    },
+    {
+      "epoch": 1.2084363016616957,
+      "grad_norm": 18.75,
+      "learning_rate": 8.543121380352088e-07,
+      "loss": 0.5103,
+      "step": 14181
+    },
+    {
+      "epoch": 1.2085215168299956,
+      "grad_norm": 16.125,
+      "learning_rate": 8.541527471659948e-07,
+      "loss": 0.7188,
+      "step": 14182
+    },
+    {
+      "epoch": 1.2086067319982956,
+      "grad_norm": 15.25,
+      "learning_rate": 8.539933634501384e-07,
+      "loss": 0.5575,
+      "step": 14183
+    },
+    {
+      "epoch": 1.2086919471665956,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.538339868905203e-07,
+      "loss": 0.4004,
+      "step": 14184
+    },
+    {
+      "epoch": 1.2087771623348955,
+      "grad_norm": 11.625,
+      "learning_rate": 8.536746174900206e-07,
+      "loss": 0.2434,
+      "step": 14185
+    },
+    {
+      "epoch": 1.2088623775031955,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.535152552515188e-07,
+      "loss": 0.48,
+      "step": 14186
+    },
+    {
+      "epoch": 1.2089475926714954,
+      "grad_norm": 13.25,
+      "learning_rate": 8.533559001778952e-07,
+      "loss": 0.5442,
+      "step": 14187
+    },
+    {
+      "epoch": 1.2090328078397954,
+      "grad_norm": 14.0625,
+      "learning_rate": 8.531965522720292e-07,
+      "loss": 0.2867,
+      "step": 14188
+    },
+    {
+      "epoch": 1.2091180230080953,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.530372115368006e-07,
+      "loss": 0.4273,
+      "step": 14189
+    },
+    {
+      "epoch": 1.2092032381763955,
+      "grad_norm": 14.0625,
+      "learning_rate": 8.528778779750886e-07,
+      "loss": 0.5339,
+      "step": 14190
+    },
+    {
+      "epoch": 1.2092884533446955,
+      "grad_norm": 15.25,
+      "learning_rate": 8.527185515897729e-07,
+      "loss": 0.683,
+      "step": 14191
+    },
+    {
+      "epoch": 1.2093736685129954,
+      "grad_norm": 17.125,
+      "learning_rate": 8.525592323837325e-07,
+      "loss": 0.4142,
+      "step": 14192
+    },
+    {
+      "epoch": 1.2094588836812954,
+      "grad_norm": 11.5,
+      "learning_rate": 8.523999203598467e-07,
+      "loss": 0.3446,
+      "step": 14193
+    },
+    {
+      "epoch": 1.2095440988495954,
+      "grad_norm": 10.9375,
+      "learning_rate": 8.522406155209937e-07,
+      "loss": 0.4636,
+      "step": 14194
+    },
+    {
+      "epoch": 1.2096293140178953,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.520813178700528e-07,
+      "loss": 0.2821,
+      "step": 14195
+    },
+    {
+      "epoch": 1.2097145291861953,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.519220274099024e-07,
+      "loss": 0.5829,
+      "step": 14196
+    },
+    {
+      "epoch": 1.2097997443544952,
+      "grad_norm": 19.0,
+      "learning_rate": 8.517627441434217e-07,
+      "loss": 0.5798,
+      "step": 14197
+    },
+    {
+      "epoch": 1.2098849595227952,
+      "grad_norm": 18.75,
+      "learning_rate": 8.516034680734883e-07,
+      "loss": 0.8952,
+      "step": 14198
+    },
+    {
+      "epoch": 1.2099701746910951,
+      "grad_norm": 12.375,
+      "learning_rate": 8.514441992029812e-07,
+      "loss": 0.5965,
+      "step": 14199
+    },
+    {
+      "epoch": 1.210055389859395,
+      "grad_norm": 13.0,
+      "learning_rate": 8.512849375347779e-07,
+      "loss": 0.4492,
+      "step": 14200
+    },
+    {
+      "epoch": 1.210140605027695,
+      "grad_norm": 14.75,
+      "learning_rate": 8.511256830717571e-07,
+      "loss": 0.3748,
+      "step": 14201
+    },
+    {
+      "epoch": 1.210225820195995,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.509664358167963e-07,
+      "loss": 0.3002,
+      "step": 14202
+    },
+    {
+      "epoch": 1.210311035364295,
+      "grad_norm": 15.875,
+      "learning_rate": 8.508071957727728e-07,
+      "loss": 0.856,
+      "step": 14203
+    },
+    {
+      "epoch": 1.210396250532595,
+      "grad_norm": 14.25,
+      "learning_rate": 8.50647962942565e-07,
+      "loss": 0.4694,
+      "step": 14204
+    },
+    {
+      "epoch": 1.2104814657008949,
+      "grad_norm": 16.75,
+      "learning_rate": 8.504887373290499e-07,
+      "loss": 0.6661,
+      "step": 14205
+    },
+    {
+      "epoch": 1.2105666808691948,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.503295189351052e-07,
+      "loss": 0.614,
+      "step": 14206
+    },
+    {
+      "epoch": 1.2106518960374948,
+      "grad_norm": 17.0,
+      "learning_rate": 8.501703077636081e-07,
+      "loss": 0.4606,
+      "step": 14207
+    },
+    {
+      "epoch": 1.2107371112057947,
+      "grad_norm": 15.5,
+      "learning_rate": 8.500111038174352e-07,
+      "loss": 0.6667,
+      "step": 14208
+    },
+    {
+      "epoch": 1.2108223263740947,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.498519070994646e-07,
+      "loss": 0.4457,
+      "step": 14209
+    },
+    {
+      "epoch": 1.2109075415423947,
+      "grad_norm": 14.375,
+      "learning_rate": 8.49692717612572e-07,
+      "loss": 0.5797,
+      "step": 14210
+    },
+    {
+      "epoch": 1.2109927567106946,
+      "grad_norm": 15.125,
+      "learning_rate": 8.495335353596343e-07,
+      "loss": 0.7663,
+      "step": 14211
+    },
+    {
+      "epoch": 1.2110779718789946,
+      "grad_norm": 24.0,
+      "learning_rate": 8.493743603435286e-07,
+      "loss": 1.0169,
+      "step": 14212
+    },
+    {
+      "epoch": 1.2111631870472945,
+      "grad_norm": 19.875,
+      "learning_rate": 8.492151925671307e-07,
+      "loss": 0.8912,
+      "step": 14213
+    },
+    {
+      "epoch": 1.2112484022155945,
+      "grad_norm": 14.125,
+      "learning_rate": 8.490560320333177e-07,
+      "loss": 0.4532,
+      "step": 14214
+    },
+    {
+      "epoch": 1.2113336173838944,
+      "grad_norm": 12.75,
+      "learning_rate": 8.488968787449652e-07,
+      "loss": 0.4928,
+      "step": 14215
+    },
+    {
+      "epoch": 1.2114188325521944,
+      "grad_norm": 16.75,
+      "learning_rate": 8.487377327049495e-07,
+      "loss": 0.6004,
+      "step": 14216
+    },
+    {
+      "epoch": 1.2115040477204944,
+      "grad_norm": 19.5,
+      "learning_rate": 8.485785939161463e-07,
+      "loss": 0.929,
+      "step": 14217
+    },
+    {
+      "epoch": 1.2115892628887943,
+      "grad_norm": 14.9375,
+      "learning_rate": 8.48419462381432e-07,
+      "loss": 0.7083,
+      "step": 14218
+    },
+    {
+      "epoch": 1.2116744780570943,
+      "grad_norm": 18.25,
+      "learning_rate": 8.482603381036816e-07,
+      "loss": 0.7162,
+      "step": 14219
+    },
+    {
+      "epoch": 1.2117596932253942,
+      "grad_norm": 12.4375,
+      "learning_rate": 8.481012210857708e-07,
+      "loss": 0.5165,
+      "step": 14220
+    },
+    {
+      "epoch": 1.2118449083936942,
+      "grad_norm": 18.25,
+      "learning_rate": 8.479421113305753e-07,
+      "loss": 0.6443,
+      "step": 14221
+    },
+    {
+      "epoch": 1.2119301235619941,
+      "grad_norm": 21.375,
+      "learning_rate": 8.477830088409699e-07,
+      "loss": 0.6465,
+      "step": 14222
+    },
+    {
+      "epoch": 1.212015338730294,
+      "grad_norm": 17.75,
+      "learning_rate": 8.4762391361983e-07,
+      "loss": 0.678,
+      "step": 14223
+    },
+    {
+      "epoch": 1.212100553898594,
+      "grad_norm": 12.125,
+      "learning_rate": 8.474648256700307e-07,
+      "loss": 0.2822,
+      "step": 14224
+    },
+    {
+      "epoch": 1.212185769066894,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.473057449944464e-07,
+      "loss": 0.5982,
+      "step": 14225
+    },
+    {
+      "epoch": 1.212270984235194,
+      "grad_norm": 16.5,
+      "learning_rate": 8.471466715959531e-07,
+      "loss": 0.5527,
+      "step": 14226
+    },
+    {
+      "epoch": 1.212356199403494,
+      "grad_norm": 16.75,
+      "learning_rate": 8.46987605477424e-07,
+      "loss": 0.6016,
+      "step": 14227
+    },
+    {
+      "epoch": 1.2124414145717939,
+      "grad_norm": 14.875,
+      "learning_rate": 8.46828546641734e-07,
+      "loss": 0.3789,
+      "step": 14228
+    },
+    {
+      "epoch": 1.2125266297400938,
+      "grad_norm": 19.625,
+      "learning_rate": 8.466694950917579e-07,
+      "loss": 0.669,
+      "step": 14229
+    },
+    {
+      "epoch": 1.2126118449083938,
+      "grad_norm": 12.625,
+      "learning_rate": 8.465104508303691e-07,
+      "loss": 0.4945,
+      "step": 14230
+    },
+    {
+      "epoch": 1.2126970600766938,
+      "grad_norm": 81.0,
+      "learning_rate": 8.463514138604425e-07,
+      "loss": 0.3029,
+      "step": 14231
+    },
+    {
+      "epoch": 1.2127822752449937,
+      "grad_norm": 11.875,
+      "learning_rate": 8.461923841848517e-07,
+      "loss": 0.4501,
+      "step": 14232
+    },
+    {
+      "epoch": 1.2128674904132937,
+      "grad_norm": 12.375,
+      "learning_rate": 8.460333618064708e-07,
+      "loss": 0.4638,
+      "step": 14233
+    },
+    {
+      "epoch": 1.2129527055815936,
+      "grad_norm": 19.0,
+      "learning_rate": 8.458743467281734e-07,
+      "loss": 0.7756,
+      "step": 14234
+    },
+    {
+      "epoch": 1.2130379207498936,
+      "grad_norm": 20.875,
+      "learning_rate": 8.457153389528325e-07,
+      "loss": 0.7872,
+      "step": 14235
+    },
+    {
+      "epoch": 1.2131231359181935,
+      "grad_norm": 14.0,
+      "learning_rate": 8.455563384833221e-07,
+      "loss": 0.845,
+      "step": 14236
+    },
+    {
+      "epoch": 1.2132083510864935,
+      "grad_norm": 23.0,
+      "learning_rate": 8.453973453225156e-07,
+      "loss": 0.755,
+      "step": 14237
+    },
+    {
+      "epoch": 1.2132935662547935,
+      "grad_norm": 15.0,
+      "learning_rate": 8.452383594732855e-07,
+      "loss": 0.6139,
+      "step": 14238
+    },
+    {
+      "epoch": 1.2133787814230934,
+      "grad_norm": 15.5,
+      "learning_rate": 8.450793809385055e-07,
+      "loss": 0.5804,
+      "step": 14239
+    },
+    {
+      "epoch": 1.2134639965913934,
+      "grad_norm": 18.0,
+      "learning_rate": 8.449204097210481e-07,
+      "loss": 0.4057,
+      "step": 14240
+    },
+    {
+      "epoch": 1.2135492117596933,
+      "grad_norm": 18.75,
+      "learning_rate": 8.447614458237865e-07,
+      "loss": 0.7369,
+      "step": 14241
+    },
+    {
+      "epoch": 1.2136344269279933,
+      "grad_norm": 16.875,
+      "learning_rate": 8.446024892495932e-07,
+      "loss": 0.6297,
+      "step": 14242
+    },
+    {
+      "epoch": 1.2137196420962932,
+      "grad_norm": 14.25,
+      "learning_rate": 8.444435400013401e-07,
+      "loss": 0.5253,
+      "step": 14243
+    },
+    {
+      "epoch": 1.2138048572645932,
+      "grad_norm": 22.375,
+      "learning_rate": 8.442845980819005e-07,
+      "loss": 0.5096,
+      "step": 14244
+    },
+    {
+      "epoch": 1.2138900724328932,
+      "grad_norm": 16.25,
+      "learning_rate": 8.441256634941458e-07,
+      "loss": 0.7037,
+      "step": 14245
+    },
+    {
+      "epoch": 1.213975287601193,
+      "grad_norm": 15.25,
+      "learning_rate": 8.439667362409487e-07,
+      "loss": 0.3959,
+      "step": 14246
+    },
+    {
+      "epoch": 1.214060502769493,
+      "grad_norm": 18.875,
+      "learning_rate": 8.438078163251808e-07,
+      "loss": 0.8084,
+      "step": 14247
+    },
+    {
+      "epoch": 1.214145717937793,
+      "grad_norm": 16.125,
+      "learning_rate": 8.436489037497142e-07,
+      "loss": 0.4822,
+      "step": 14248
+    },
+    {
+      "epoch": 1.214230933106093,
+      "grad_norm": 13.625,
+      "learning_rate": 8.434899985174205e-07,
+      "loss": 0.7781,
+      "step": 14249
+    },
+    {
+      "epoch": 1.214316148274393,
+      "grad_norm": 13.5,
+      "learning_rate": 8.433311006311718e-07,
+      "loss": 0.6064,
+      "step": 14250
+    },
+    {
+      "epoch": 1.214401363442693,
+      "grad_norm": 13.125,
+      "learning_rate": 8.431722100938386e-07,
+      "loss": 0.5313,
+      "step": 14251
+    },
+    {
+      "epoch": 1.2144865786109929,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.430133269082928e-07,
+      "loss": 0.3832,
+      "step": 14252
+    },
+    {
+      "epoch": 1.2145717937792928,
+      "grad_norm": 16.375,
+      "learning_rate": 8.428544510774052e-07,
+      "loss": 1.0254,
+      "step": 14253
+    },
+    {
+      "epoch": 1.2146570089475928,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.426955826040473e-07,
+      "loss": 0.4287,
+      "step": 14254
+    },
+    {
+      "epoch": 1.2147422241158927,
+      "grad_norm": 16.25,
+      "learning_rate": 8.425367214910896e-07,
+      "loss": 0.7628,
+      "step": 14255
+    },
+    {
+      "epoch": 1.2148274392841927,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.423778677414032e-07,
+      "loss": 0.5263,
+      "step": 14256
+    },
+    {
+      "epoch": 1.2149126544524926,
+      "grad_norm": 16.875,
+      "learning_rate": 8.422190213578585e-07,
+      "loss": 0.8243,
+      "step": 14257
+    },
+    {
+      "epoch": 1.2149978696207926,
+      "grad_norm": 15.125,
+      "learning_rate": 8.420601823433266e-07,
+      "loss": 0.4406,
+      "step": 14258
+    },
+    {
+      "epoch": 1.2150830847890925,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.41901350700677e-07,
+      "loss": 0.6284,
+      "step": 14259
+    },
+    {
+      "epoch": 1.2151682999573925,
+      "grad_norm": 13.0,
+      "learning_rate": 8.417425264327804e-07,
+      "loss": 0.6977,
+      "step": 14260
+    },
+    {
+      "epoch": 1.2152535151256925,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.415837095425069e-07,
+      "loss": 0.7631,
+      "step": 14261
+    },
+    {
+      "epoch": 1.2153387302939924,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.414249000327265e-07,
+      "loss": 0.5302,
+      "step": 14262
+    },
+    {
+      "epoch": 1.2154239454622924,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.412660979063091e-07,
+      "loss": 0.6646,
+      "step": 14263
+    },
+    {
+      "epoch": 1.2155091606305923,
+      "grad_norm": 16.875,
+      "learning_rate": 8.41107303166124e-07,
+      "loss": 0.8917,
+      "step": 14264
+    },
+    {
+      "epoch": 1.2155943757988923,
+      "grad_norm": 13.375,
+      "learning_rate": 8.409485158150414e-07,
+      "loss": 0.4421,
+      "step": 14265
+    },
+    {
+      "epoch": 1.2156795909671922,
+      "grad_norm": 19.5,
+      "learning_rate": 8.407897358559308e-07,
+      "loss": 0.8879,
+      "step": 14266
+    },
+    {
+      "epoch": 1.2157648061354922,
+      "grad_norm": 13.125,
+      "learning_rate": 8.406309632916608e-07,
+      "loss": 0.5283,
+      "step": 14267
+    },
+    {
+      "epoch": 1.2158500213037922,
+      "grad_norm": 18.25,
+      "learning_rate": 8.404721981251007e-07,
+      "loss": 0.8967,
+      "step": 14268
+    },
+    {
+      "epoch": 1.2159352364720921,
+      "grad_norm": 17.375,
+      "learning_rate": 8.403134403591201e-07,
+      "loss": 0.7255,
+      "step": 14269
+    },
+    {
+      "epoch": 1.216020451640392,
+      "grad_norm": 17.625,
+      "learning_rate": 8.401546899965873e-07,
+      "loss": 0.6983,
+      "step": 14270
+    },
+    {
+      "epoch": 1.216105666808692,
+      "grad_norm": 20.75,
+      "learning_rate": 8.399959470403718e-07,
+      "loss": 0.7689,
+      "step": 14271
+    },
+    {
+      "epoch": 1.216190881976992,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.398372114933413e-07,
+      "loss": 0.5063,
+      "step": 14272
+    },
+    {
+      "epoch": 1.216276097145292,
+      "grad_norm": 17.625,
+      "learning_rate": 8.396784833583653e-07,
+      "loss": 0.8025,
+      "step": 14273
+    },
+    {
+      "epoch": 1.216361312313592,
+      "grad_norm": 19.375,
+      "learning_rate": 8.395197626383119e-07,
+      "loss": 0.9521,
+      "step": 14274
+    },
+    {
+      "epoch": 1.2164465274818919,
+      "grad_norm": 14.5,
+      "learning_rate": 8.393610493360487e-07,
+      "loss": 0.4882,
+      "step": 14275
+    },
+    {
+      "epoch": 1.2165317426501918,
+      "grad_norm": 15.3125,
+      "learning_rate": 8.392023434544444e-07,
+      "loss": 0.5282,
+      "step": 14276
+    },
+    {
+      "epoch": 1.2166169578184918,
+      "grad_norm": 18.375,
+      "learning_rate": 8.390436449963667e-07,
+      "loss": 0.8183,
+      "step": 14277
+    },
+    {
+      "epoch": 1.2167021729867917,
+      "grad_norm": 18.625,
+      "learning_rate": 8.388849539646837e-07,
+      "loss": 0.8769,
+      "step": 14278
+    },
+    {
+      "epoch": 1.2167873881550917,
+      "grad_norm": 17.375,
+      "learning_rate": 8.387262703622629e-07,
+      "loss": 0.9322,
+      "step": 14279
+    },
+    {
+      "epoch": 1.2168726033233916,
+      "grad_norm": 19.125,
+      "learning_rate": 8.385675941919722e-07,
+      "loss": 0.7885,
+      "step": 14280
+    },
+    {
+      "epoch": 1.2169578184916916,
+      "grad_norm": 17.125,
+      "learning_rate": 8.384089254566785e-07,
+      "loss": 0.6721,
+      "step": 14281
+    },
+    {
+      "epoch": 1.2170430336599916,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.382502641592502e-07,
+      "loss": 0.6553,
+      "step": 14282
+    },
+    {
+      "epoch": 1.2171282488282915,
+      "grad_norm": 20.25,
+      "learning_rate": 8.380916103025527e-07,
+      "loss": 1.0162,
+      "step": 14283
+    },
+    {
+      "epoch": 1.2172134639965915,
+      "grad_norm": 13.875,
+      "learning_rate": 8.379329638894546e-07,
+      "loss": 0.3203,
+      "step": 14284
+    },
+    {
+      "epoch": 1.2172986791648914,
+      "grad_norm": 17.125,
+      "learning_rate": 8.377743249228218e-07,
+      "loss": 0.3779,
+      "step": 14285
+    },
+    {
+      "epoch": 1.2173838943331914,
+      "grad_norm": 18.625,
+      "learning_rate": 8.376156934055216e-07,
+      "loss": 0.6119,
+      "step": 14286
+    },
+    {
+      "epoch": 1.2174691095014913,
+      "grad_norm": 18.25,
+      "learning_rate": 8.374570693404205e-07,
+      "loss": 0.8944,
+      "step": 14287
+    },
+    {
+      "epoch": 1.2175543246697913,
+      "grad_norm": 16.375,
+      "learning_rate": 8.37298452730385e-07,
+      "loss": 0.4667,
+      "step": 14288
+    },
+    {
+      "epoch": 1.2176395398380913,
+      "grad_norm": 18.375,
+      "learning_rate": 8.371398435782813e-07,
+      "loss": 0.4449,
+      "step": 14289
+    },
+    {
+      "epoch": 1.2177247550063912,
+      "grad_norm": 20.625,
+      "learning_rate": 8.369812418869761e-07,
+      "loss": 0.998,
+      "step": 14290
+    },
+    {
+      "epoch": 1.2178099701746912,
+      "grad_norm": 14.5,
+      "learning_rate": 8.368226476593355e-07,
+      "loss": 0.731,
+      "step": 14291
+    },
+    {
+      "epoch": 1.2178951853429911,
+      "grad_norm": 18.75,
+      "learning_rate": 8.366640608982245e-07,
+      "loss": 0.4268,
+      "step": 14292
+    },
+    {
+      "epoch": 1.217980400511291,
+      "grad_norm": 16.375,
+      "learning_rate": 8.365054816065099e-07,
+      "loss": 0.6161,
+      "step": 14293
+    },
+    {
+      "epoch": 1.218065615679591,
+      "grad_norm": 19.625,
+      "learning_rate": 8.363469097870566e-07,
+      "loss": 0.5126,
+      "step": 14294
+    },
+    {
+      "epoch": 1.218150830847891,
+      "grad_norm": 16.5,
+      "learning_rate": 8.361883454427311e-07,
+      "loss": 0.397,
+      "step": 14295
+    },
+    {
+      "epoch": 1.218236046016191,
+      "grad_norm": 15.5,
+      "learning_rate": 8.36029788576398e-07,
+      "loss": 0.6022,
+      "step": 14296
+    },
+    {
+      "epoch": 1.218321261184491,
+      "grad_norm": 14.0625,
+      "learning_rate": 8.358712391909231e-07,
+      "loss": 0.4326,
+      "step": 14297
+    },
+    {
+      "epoch": 1.2184064763527909,
+      "grad_norm": 19.625,
+      "learning_rate": 8.35712697289171e-07,
+      "loss": 0.6293,
+      "step": 14298
+    },
+    {
+      "epoch": 1.2184916915210908,
+      "grad_norm": 15.25,
+      "learning_rate": 8.355541628740079e-07,
+      "loss": 0.5858,
+      "step": 14299
+    },
+    {
+      "epoch": 1.2185769066893908,
+      "grad_norm": 18.5,
+      "learning_rate": 8.353956359482973e-07,
+      "loss": 0.642,
+      "step": 14300
+    },
+    {
+      "epoch": 1.2186621218576907,
+      "grad_norm": 14.875,
+      "learning_rate": 8.352371165149045e-07,
+      "loss": 0.4365,
+      "step": 14301
+    },
+    {
+      "epoch": 1.2187473370259907,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.350786045766939e-07,
+      "loss": 0.5058,
+      "step": 14302
+    },
+    {
+      "epoch": 1.2188325521942907,
+      "grad_norm": 16.25,
+      "learning_rate": 8.349201001365304e-07,
+      "loss": 0.7715,
+      "step": 14303
+    },
+    {
+      "epoch": 1.2189177673625906,
+      "grad_norm": 41.75,
+      "learning_rate": 8.347616031972778e-07,
+      "loss": 0.7642,
+      "step": 14304
+    },
+    {
+      "epoch": 1.2190029825308906,
+      "grad_norm": 11.625,
+      "learning_rate": 8.346031137618008e-07,
+      "loss": 0.372,
+      "step": 14305
+    },
+    {
+      "epoch": 1.2190881976991905,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.34444631832963e-07,
+      "loss": 0.4241,
+      "step": 14306
+    },
+    {
+      "epoch": 1.2191734128674905,
+      "grad_norm": 10.875,
+      "learning_rate": 8.34286157413629e-07,
+      "loss": 0.3683,
+      "step": 14307
+    },
+    {
+      "epoch": 1.2192586280357904,
+      "grad_norm": 18.375,
+      "learning_rate": 8.341276905066621e-07,
+      "loss": 0.8122,
+      "step": 14308
+    },
+    {
+      "epoch": 1.2193438432040904,
+      "grad_norm": 10.875,
+      "learning_rate": 8.339692311149257e-07,
+      "loss": 0.2567,
+      "step": 14309
+    },
+    {
+      "epoch": 1.2194290583723904,
+      "grad_norm": 11.0625,
+      "learning_rate": 8.338107792412839e-07,
+      "loss": 0.3534,
+      "step": 14310
+    },
+    {
+      "epoch": 1.2195142735406903,
+      "grad_norm": 25.125,
+      "learning_rate": 8.336523348885997e-07,
+      "loss": 0.8438,
+      "step": 14311
+    },
+    {
+      "epoch": 1.2195994887089903,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.334938980597365e-07,
+      "loss": 0.3333,
+      "step": 14312
+    },
+    {
+      "epoch": 1.2196847038772902,
+      "grad_norm": 25.125,
+      "learning_rate": 8.333354687575573e-07,
+      "loss": 0.9861,
+      "step": 14313
+    },
+    {
+      "epoch": 1.2197699190455902,
+      "grad_norm": 16.25,
+      "learning_rate": 8.331770469849254e-07,
+      "loss": 0.5639,
+      "step": 14314
+    },
+    {
+      "epoch": 1.2198551342138901,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.330186327447036e-07,
+      "loss": 0.4092,
+      "step": 14315
+    },
+    {
+      "epoch": 1.21994034938219,
+      "grad_norm": 24.375,
+      "learning_rate": 8.328602260397541e-07,
+      "loss": 0.7765,
+      "step": 14316
+    },
+    {
+      "epoch": 1.22002556455049,
+      "grad_norm": 10.125,
+      "learning_rate": 8.327018268729398e-07,
+      "loss": 0.3123,
+      "step": 14317
+    },
+    {
+      "epoch": 1.22011077971879,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.325434352471231e-07,
+      "loss": 0.4499,
+      "step": 14318
+    },
+    {
+      "epoch": 1.22019599488709,
+      "grad_norm": 12.875,
+      "learning_rate": 8.323850511651662e-07,
+      "loss": 0.4656,
+      "step": 14319
+    },
+    {
+      "epoch": 1.22028121005539,
+      "grad_norm": 16.25,
+      "learning_rate": 8.322266746299316e-07,
+      "loss": 0.6067,
+      "step": 14320
+    },
+    {
+      "epoch": 1.2203664252236899,
+      "grad_norm": 20.875,
+      "learning_rate": 8.320683056442807e-07,
+      "loss": 0.5858,
+      "step": 14321
+    },
+    {
+      "epoch": 1.2204516403919898,
+      "grad_norm": 15.5,
+      "learning_rate": 8.319099442110762e-07,
+      "loss": 0.4778,
+      "step": 14322
+    },
+    {
+      "epoch": 1.2205368555602898,
+      "grad_norm": 13.625,
+      "learning_rate": 8.317515903331796e-07,
+      "loss": 0.5522,
+      "step": 14323
+    },
+    {
+      "epoch": 1.2206220707285897,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.315932440134518e-07,
+      "loss": 0.2924,
+      "step": 14324
+    },
+    {
+      "epoch": 1.2207072858968897,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.314349052547552e-07,
+      "loss": 0.6461,
+      "step": 14325
+    },
+    {
+      "epoch": 1.2207925010651897,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.312765740599505e-07,
+      "loss": 0.5876,
+      "step": 14326
+    },
+    {
+      "epoch": 1.2208777162334896,
+      "grad_norm": 16.375,
+      "learning_rate": 8.311182504318993e-07,
+      "loss": 0.6497,
+      "step": 14327
+    },
+    {
+      "epoch": 1.2209629314017896,
+      "grad_norm": 17.625,
+      "learning_rate": 8.309599343734623e-07,
+      "loss": 0.7302,
+      "step": 14328
+    },
+    {
+      "epoch": 1.2210481465700895,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.308016258875009e-07,
+      "loss": 0.5807,
+      "step": 14329
+    },
+    {
+      "epoch": 1.2211333617383895,
+      "grad_norm": 16.25,
+      "learning_rate": 8.306433249768759e-07,
+      "loss": 0.5513,
+      "step": 14330
+    },
+    {
+      "epoch": 1.2212185769066894,
+      "grad_norm": 19.25,
+      "learning_rate": 8.304850316444477e-07,
+      "loss": 0.8578,
+      "step": 14331
+    },
+    {
+      "epoch": 1.2213037920749894,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.303267458930763e-07,
+      "loss": 0.4727,
+      "step": 14332
+    },
+    {
+      "epoch": 1.2213890072432894,
+      "grad_norm": 14.875,
+      "learning_rate": 8.30168467725623e-07,
+      "loss": 0.3806,
+      "step": 14333
+    },
+    {
+      "epoch": 1.2214742224115893,
+      "grad_norm": 14.75,
+      "learning_rate": 8.300101971449474e-07,
+      "loss": 0.6695,
+      "step": 14334
+    },
+    {
+      "epoch": 1.2215594375798893,
+      "grad_norm": 13.875,
+      "learning_rate": 8.298519341539102e-07,
+      "loss": 0.6018,
+      "step": 14335
+    },
+    {
+      "epoch": 1.2216446527481892,
+      "grad_norm": 17.75,
+      "learning_rate": 8.296936787553706e-07,
+      "loss": 0.5728,
+      "step": 14336
+    },
+    {
+      "epoch": 1.2217298679164892,
+      "grad_norm": 12.4375,
+      "learning_rate": 8.295354309521893e-07,
+      "loss": 0.4016,
+      "step": 14337
+    },
+    {
+      "epoch": 1.2218150830847891,
+      "grad_norm": 11.0,
+      "learning_rate": 8.293771907472252e-07,
+      "loss": 0.3203,
+      "step": 14338
+    },
+    {
+      "epoch": 1.221900298253089,
+      "grad_norm": 20.875,
+      "learning_rate": 8.292189581433388e-07,
+      "loss": 1.08,
+      "step": 14339
+    },
+    {
+      "epoch": 1.221985513421389,
+      "grad_norm": 13.5,
+      "learning_rate": 8.290607331433886e-07,
+      "loss": 0.4541,
+      "step": 14340
+    },
+    {
+      "epoch": 1.222070728589689,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.289025157502343e-07,
+      "loss": 0.7774,
+      "step": 14341
+    },
+    {
+      "epoch": 1.222155943757989,
+      "grad_norm": 17.75,
+      "learning_rate": 8.287443059667349e-07,
+      "loss": 0.6292,
+      "step": 14342
+    },
+    {
+      "epoch": 1.222241158926289,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.285861037957493e-07,
+      "loss": 0.231,
+      "step": 14343
+    },
+    {
+      "epoch": 1.2223263740945889,
+      "grad_norm": 19.0,
+      "learning_rate": 8.284279092401368e-07,
+      "loss": 0.6922,
+      "step": 14344
+    },
+    {
+      "epoch": 1.2224115892628888,
+      "grad_norm": 22.875,
+      "learning_rate": 8.282697223027561e-07,
+      "loss": 1.0454,
+      "step": 14345
+    },
+    {
+      "epoch": 1.2224968044311888,
+      "grad_norm": 20.375,
+      "learning_rate": 8.281115429864652e-07,
+      "loss": 0.7021,
+      "step": 14346
+    },
+    {
+      "epoch": 1.2225820195994888,
+      "grad_norm": 17.25,
+      "learning_rate": 8.279533712941234e-07,
+      "loss": 0.6982,
+      "step": 14347
+    },
+    {
+      "epoch": 1.2226672347677887,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.277952072285886e-07,
+      "loss": 0.568,
+      "step": 14348
+    },
+    {
+      "epoch": 1.2227524499360887,
+      "grad_norm": 14.75,
+      "learning_rate": 8.276370507927184e-07,
+      "loss": 0.2369,
+      "step": 14349
+    },
+    {
+      "epoch": 1.2228376651043886,
+      "grad_norm": 12.8125,
+      "learning_rate": 8.274789019893718e-07,
+      "loss": 0.5487,
+      "step": 14350
+    },
+    {
+      "epoch": 1.2229228802726886,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.273207608214061e-07,
+      "loss": 0.6964,
+      "step": 14351
+    },
+    {
+      "epoch": 1.2230080954409885,
+      "grad_norm": 23.75,
+      "learning_rate": 8.271626272916793e-07,
+      "loss": 0.7589,
+      "step": 14352
+    },
+    {
+      "epoch": 1.2230933106092885,
+      "grad_norm": 15.5,
+      "learning_rate": 8.270045014030492e-07,
+      "loss": 0.4066,
+      "step": 14353
+    },
+    {
+      "epoch": 1.2231785257775885,
+      "grad_norm": 16.25,
+      "learning_rate": 8.268463831583731e-07,
+      "loss": 0.5047,
+      "step": 14354
+    },
+    {
+      "epoch": 1.2232637409458884,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.266882725605083e-07,
+      "loss": 0.3694,
+      "step": 14355
+    },
+    {
+      "epoch": 1.2233489561141884,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.265301696123124e-07,
+      "loss": 0.377,
+      "step": 14356
+    },
+    {
+      "epoch": 1.2234341712824883,
+      "grad_norm": 22.875,
+      "learning_rate": 8.263720743166418e-07,
+      "loss": 0.7289,
+      "step": 14357
+    },
+    {
+      "epoch": 1.2235193864507883,
+      "grad_norm": 13.0,
+      "learning_rate": 8.262139866763538e-07,
+      "loss": 0.5662,
+      "step": 14358
+    },
+    {
+      "epoch": 1.2236046016190882,
+      "grad_norm": 27.0,
+      "learning_rate": 8.260559066943053e-07,
+      "loss": 1.1001,
+      "step": 14359
+    },
+    {
+      "epoch": 1.2236898167873882,
+      "grad_norm": 14.375,
+      "learning_rate": 8.258978343733532e-07,
+      "loss": 0.5106,
+      "step": 14360
+    },
+    {
+      "epoch": 1.2237750319556882,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.25739769716353e-07,
+      "loss": 0.3155,
+      "step": 14361
+    },
+    {
+      "epoch": 1.223860247123988,
+      "grad_norm": 18.0,
+      "learning_rate": 8.255817127261623e-07,
+      "loss": 0.4861,
+      "step": 14362
+    },
+    {
+      "epoch": 1.223945462292288,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.254236634056365e-07,
+      "loss": 0.5578,
+      "step": 14363
+    },
+    {
+      "epoch": 1.224030677460588,
+      "grad_norm": 14.375,
+      "learning_rate": 8.252656217576327e-07,
+      "loss": 0.7464,
+      "step": 14364
+    },
+    {
+      "epoch": 1.224115892628888,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.251075877850058e-07,
+      "loss": 0.3164,
+      "step": 14365
+    },
+    {
+      "epoch": 1.224201107797188,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.249495614906118e-07,
+      "loss": 0.4481,
+      "step": 14366
+    },
+    {
+      "epoch": 1.224286322965488,
+      "grad_norm": 14.0625,
+      "learning_rate": 8.24791542877307e-07,
+      "loss": 0.6494,
+      "step": 14367
+    },
+    {
+      "epoch": 1.2243715381337879,
+      "grad_norm": 16.5,
+      "learning_rate": 8.246335319479462e-07,
+      "loss": 0.8217,
+      "step": 14368
+    },
+    {
+      "epoch": 1.2244567533020878,
+      "grad_norm": 16.25,
+      "learning_rate": 8.244755287053857e-07,
+      "loss": 0.732,
+      "step": 14369
+    },
+    {
+      "epoch": 1.2245419684703878,
+      "grad_norm": 17.125,
+      "learning_rate": 8.243175331524799e-07,
+      "loss": 0.5777,
+      "step": 14370
+    },
+    {
+      "epoch": 1.2246271836386877,
+      "grad_norm": 23.25,
+      "learning_rate": 8.241595452920845e-07,
+      "loss": 0.4402,
+      "step": 14371
+    },
+    {
+      "epoch": 1.2247123988069877,
+      "grad_norm": 28.25,
+      "learning_rate": 8.240015651270547e-07,
+      "loss": 1.1677,
+      "step": 14372
+    },
+    {
+      "epoch": 1.2247976139752876,
+      "grad_norm": 20.125,
+      "learning_rate": 8.238435926602445e-07,
+      "loss": 0.7971,
+      "step": 14373
+    },
+    {
+      "epoch": 1.2248828291435876,
+      "grad_norm": 20.5,
+      "learning_rate": 8.236856278945094e-07,
+      "loss": 0.8393,
+      "step": 14374
+    },
+    {
+      "epoch": 1.2249680443118875,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.235276708327035e-07,
+      "loss": 0.4389,
+      "step": 14375
+    },
+    {
+      "epoch": 1.2250532594801875,
+      "grad_norm": 15.125,
+      "learning_rate": 8.233697214776814e-07,
+      "loss": 0.5732,
+      "step": 14376
+    },
+    {
+      "epoch": 1.2251384746484875,
+      "grad_norm": 18.375,
+      "learning_rate": 8.232117798322975e-07,
+      "loss": 0.8083,
+      "step": 14377
+    },
+    {
+      "epoch": 1.2252236898167874,
+      "grad_norm": 13.875,
+      "learning_rate": 8.230538458994058e-07,
+      "loss": 0.4711,
+      "step": 14378
+    },
+    {
+      "epoch": 1.2253089049850874,
+      "grad_norm": 14.375,
+      "learning_rate": 8.228959196818606e-07,
+      "loss": 0.5149,
+      "step": 14379
+    },
+    {
+      "epoch": 1.2253941201533873,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.227380011825159e-07,
+      "loss": 0.6433,
+      "step": 14380
+    },
+    {
+      "epoch": 1.2254793353216873,
+      "grad_norm": 56.25,
+      "learning_rate": 8.225800904042247e-07,
+      "loss": 0.7149,
+      "step": 14381
+    },
+    {
+      "epoch": 1.2255645504899872,
+      "grad_norm": 21.125,
+      "learning_rate": 8.224221873498414e-07,
+      "loss": 0.8643,
+      "step": 14382
+    },
+    {
+      "epoch": 1.2256497656582872,
+      "grad_norm": 24.125,
+      "learning_rate": 8.222642920222187e-07,
+      "loss": 0.8243,
+      "step": 14383
+    },
+    {
+      "epoch": 1.2257349808265872,
+      "grad_norm": 18.25,
+      "learning_rate": 8.221064044242107e-07,
+      "loss": 0.6944,
+      "step": 14384
+    },
+    {
+      "epoch": 1.2258201959948871,
+      "grad_norm": 24.625,
+      "learning_rate": 8.219485245586701e-07,
+      "loss": 0.9646,
+      "step": 14385
+    },
+    {
+      "epoch": 1.225905411163187,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.217906524284502e-07,
+      "loss": 0.7437,
+      "step": 14386
+    },
+    {
+      "epoch": 1.225990626331487,
+      "grad_norm": 18.5,
+      "learning_rate": 8.216327880364036e-07,
+      "loss": 0.5139,
+      "step": 14387
+    },
+    {
+      "epoch": 1.226075841499787,
+      "grad_norm": 21.75,
+      "learning_rate": 8.214749313853837e-07,
+      "loss": 1.1702,
+      "step": 14388
+    },
+    {
+      "epoch": 1.226161056668087,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.213170824782426e-07,
+      "loss": 0.7668,
+      "step": 14389
+    },
+    {
+      "epoch": 1.226246271836387,
+      "grad_norm": 16.625,
+      "learning_rate": 8.211592413178329e-07,
+      "loss": 0.703,
+      "step": 14390
+    },
+    {
+      "epoch": 1.2263314870046869,
+      "grad_norm": 17.125,
+      "learning_rate": 8.210014079070066e-07,
+      "loss": 0.8535,
+      "step": 14391
+    },
+    {
+      "epoch": 1.2264167021729868,
+      "grad_norm": 24.375,
+      "learning_rate": 8.208435822486165e-07,
+      "loss": 0.6355,
+      "step": 14392
+    },
+    {
+      "epoch": 1.2265019173412868,
+      "grad_norm": 23.0,
+      "learning_rate": 8.206857643455142e-07,
+      "loss": 0.7368,
+      "step": 14393
+    },
+    {
+      "epoch": 1.2265871325095867,
+      "grad_norm": 23.25,
+      "learning_rate": 8.20527954200552e-07,
+      "loss": 0.8731,
+      "step": 14394
+    },
+    {
+      "epoch": 1.2266723476778867,
+      "grad_norm": 17.125,
+      "learning_rate": 8.203701518165814e-07,
+      "loss": 0.9526,
+      "step": 14395
+    },
+    {
+      "epoch": 1.2267575628461866,
+      "grad_norm": 16.0,
+      "learning_rate": 8.202123571964547e-07,
+      "loss": 0.7149,
+      "step": 14396
+    },
+    {
+      "epoch": 1.2268427780144866,
+      "grad_norm": 18.75,
+      "learning_rate": 8.200545703430227e-07,
+      "loss": 0.6252,
+      "step": 14397
+    },
+    {
+      "epoch": 1.2269279931827866,
+      "grad_norm": 17.0,
+      "learning_rate": 8.198967912591366e-07,
+      "loss": 0.7001,
+      "step": 14398
+    },
+    {
+      "epoch": 1.2270132083510865,
+      "grad_norm": 13.5,
+      "learning_rate": 8.197390199476482e-07,
+      "loss": 0.6932,
+      "step": 14399
+    },
+    {
+      "epoch": 1.2270984235193865,
+      "grad_norm": 13.0,
+      "learning_rate": 8.195812564114081e-07,
+      "loss": 0.4058,
+      "step": 14400
+    },
+    {
+      "epoch": 1.2271836386876864,
+      "grad_norm": 10.0,
+      "learning_rate": 8.194235006532678e-07,
+      "loss": 0.2526,
+      "step": 14401
+    },
+    {
+      "epoch": 1.2272688538559864,
+      "grad_norm": 14.875,
+      "learning_rate": 8.192657526760777e-07,
+      "loss": 0.6565,
+      "step": 14402
+    },
+    {
+      "epoch": 1.2273540690242863,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.191080124826886e-07,
+      "loss": 0.4241,
+      "step": 14403
+    },
+    {
+      "epoch": 1.2274392841925863,
+      "grad_norm": 16.125,
+      "learning_rate": 8.189502800759515e-07,
+      "loss": 0.5987,
+      "step": 14404
+    },
+    {
+      "epoch": 1.2275244993608863,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.187925554587159e-07,
+      "loss": 0.2314,
+      "step": 14405
+    },
+    {
+      "epoch": 1.2276097145291862,
+      "grad_norm": 13.75,
+      "learning_rate": 8.186348386338322e-07,
+      "loss": 0.3858,
+      "step": 14406
+    },
+    {
+      "epoch": 1.2276949296974862,
+      "grad_norm": 16.625,
+      "learning_rate": 8.18477129604151e-07,
+      "loss": 0.3744,
+      "step": 14407
+    },
+    {
+      "epoch": 1.2277801448657861,
+      "grad_norm": 21.125,
+      "learning_rate": 8.183194283725218e-07,
+      "loss": 0.5223,
+      "step": 14408
+    },
+    {
+      "epoch": 1.227865360034086,
+      "grad_norm": 18.625,
+      "learning_rate": 8.181617349417949e-07,
+      "loss": 0.7838,
+      "step": 14409
+    },
+    {
+      "epoch": 1.227950575202386,
+      "grad_norm": 18.25,
+      "learning_rate": 8.180040493148193e-07,
+      "loss": 0.6126,
+      "step": 14410
+    },
+    {
+      "epoch": 1.228035790370686,
+      "grad_norm": 18.25,
+      "learning_rate": 8.178463714944452e-07,
+      "loss": 0.7033,
+      "step": 14411
+    },
+    {
+      "epoch": 1.228121005538986,
+      "grad_norm": 17.125,
+      "learning_rate": 8.176887014835222e-07,
+      "loss": 0.4473,
+      "step": 14412
+    },
+    {
+      "epoch": 1.228206220707286,
+      "grad_norm": 21.125,
+      "learning_rate": 8.175310392848984e-07,
+      "loss": 0.7426,
+      "step": 14413
+    },
+    {
+      "epoch": 1.2282914358755859,
+      "grad_norm": 14.5,
+      "learning_rate": 8.173733849014238e-07,
+      "loss": 0.7649,
+      "step": 14414
+    },
+    {
+      "epoch": 1.2283766510438858,
+      "grad_norm": 15.875,
+      "learning_rate": 8.17215738335947e-07,
+      "loss": 0.8257,
+      "step": 14415
+    },
+    {
+      "epoch": 1.2284618662121858,
+      "grad_norm": 15.125,
+      "learning_rate": 8.170580995913173e-07,
+      "loss": 0.5833,
+      "step": 14416
+    },
+    {
+      "epoch": 1.2285470813804857,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.169004686703828e-07,
+      "loss": 0.7669,
+      "step": 14417
+    },
+    {
+      "epoch": 1.2286322965487857,
+      "grad_norm": 19.625,
+      "learning_rate": 8.167428455759926e-07,
+      "loss": 0.6515,
+      "step": 14418
+    },
+    {
+      "epoch": 1.2287175117170857,
+      "grad_norm": 13.875,
+      "learning_rate": 8.165852303109946e-07,
+      "loss": 0.4103,
+      "step": 14419
+    },
+    {
+      "epoch": 1.2288027268853856,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.164276228782375e-07,
+      "loss": 0.2408,
+      "step": 14420
+    },
+    {
+      "epoch": 1.2288879420536856,
+      "grad_norm": 15.5,
+      "learning_rate": 8.162700232805696e-07,
+      "loss": 0.73,
+      "step": 14421
+    },
+    {
+      "epoch": 1.2289731572219855,
+      "grad_norm": 31.25,
+      "learning_rate": 8.161124315208382e-07,
+      "loss": 0.8404,
+      "step": 14422
+    },
+    {
+      "epoch": 1.2290583723902855,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.159548476018913e-07,
+      "loss": 0.387,
+      "step": 14423
+    },
+    {
+      "epoch": 1.2291435875585854,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.157972715265771e-07,
+      "loss": 0.6301,
+      "step": 14424
+    },
+    {
+      "epoch": 1.2292288027268854,
+      "grad_norm": 11.375,
+      "learning_rate": 8.156397032977426e-07,
+      "loss": 0.2258,
+      "step": 14425
+    },
+    {
+      "epoch": 1.2293140178951854,
+      "grad_norm": 18.75,
+      "learning_rate": 8.154821429182355e-07,
+      "loss": 1.0408,
+      "step": 14426
+    },
+    {
+      "epoch": 1.2293992330634853,
+      "grad_norm": 13.875,
+      "learning_rate": 8.153245903909026e-07,
+      "loss": 0.607,
+      "step": 14427
+    },
+    {
+      "epoch": 1.2294844482317853,
+      "grad_norm": 10.625,
+      "learning_rate": 8.15167045718592e-07,
+      "loss": 0.2481,
+      "step": 14428
+    },
+    {
+      "epoch": 1.2295696634000852,
+      "grad_norm": 26.625,
+      "learning_rate": 8.150095089041504e-07,
+      "loss": 1.0483,
+      "step": 14429
+    },
+    {
+      "epoch": 1.2296548785683852,
+      "grad_norm": 17.125,
+      "learning_rate": 8.148519799504238e-07,
+      "loss": 0.5979,
+      "step": 14430
+    },
+    {
+      "epoch": 1.2297400937366851,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.146944588602598e-07,
+      "loss": 0.7043,
+      "step": 14431
+    },
+    {
+      "epoch": 1.229825308904985,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.145369456365044e-07,
+      "loss": 0.4015,
+      "step": 14432
+    },
+    {
+      "epoch": 1.229910524073285,
+      "grad_norm": 13.0,
+      "learning_rate": 8.143794402820044e-07,
+      "loss": 0.4216,
+      "step": 14433
+    },
+    {
+      "epoch": 1.229995739241585,
+      "grad_norm": 13.5625,
+      "learning_rate": 8.142219427996059e-07,
+      "loss": 0.4798,
+      "step": 14434
+    },
+    {
+      "epoch": 1.230080954409885,
+      "grad_norm": 15.875,
+      "learning_rate": 8.140644531921553e-07,
+      "loss": 0.6938,
+      "step": 14435
+    },
+    {
+      "epoch": 1.230166169578185,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.139069714624981e-07,
+      "loss": 0.5764,
+      "step": 14436
+    },
+    {
+      "epoch": 1.2302513847464849,
+      "grad_norm": 11.125,
+      "learning_rate": 8.137494976134813e-07,
+      "loss": 0.3643,
+      "step": 14437
+    },
+    {
+      "epoch": 1.2303365999147848,
+      "grad_norm": 14.9375,
+      "learning_rate": 8.13592031647949e-07,
+      "loss": 0.5891,
+      "step": 14438
+    },
+    {
+      "epoch": 1.2304218150830848,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.134345735687476e-07,
+      "loss": 0.6498,
+      "step": 14439
+    },
+    {
+      "epoch": 1.2305070302513847,
+      "grad_norm": 17.0,
+      "learning_rate": 8.132771233787223e-07,
+      "loss": 0.7822,
+      "step": 14440
+    },
+    {
+      "epoch": 1.2305922454196847,
+      "grad_norm": 18.125,
+      "learning_rate": 8.131196810807189e-07,
+      "loss": 0.494,
+      "step": 14441
+    },
+    {
+      "epoch": 1.2306774605879847,
+      "grad_norm": 17.25,
+      "learning_rate": 8.129622466775819e-07,
+      "loss": 0.7106,
+      "step": 14442
+    },
+    {
+      "epoch": 1.2307626757562846,
+      "grad_norm": 17.25,
+      "learning_rate": 8.128048201721567e-07,
+      "loss": 0.7046,
+      "step": 14443
+    },
+    {
+      "epoch": 1.2308478909245846,
+      "grad_norm": 15.5,
+      "learning_rate": 8.126474015672878e-07,
+      "loss": 0.7518,
+      "step": 14444
+    },
+    {
+      "epoch": 1.2309331060928845,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.124899908658207e-07,
+      "loss": 0.6698,
+      "step": 14445
+    },
+    {
+      "epoch": 1.2310183212611845,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.123325880705992e-07,
+      "loss": 0.7122,
+      "step": 14446
+    },
+    {
+      "epoch": 1.2311035364294844,
+      "grad_norm": 17.625,
+      "learning_rate": 8.121751931844676e-07,
+      "loss": 0.6943,
+      "step": 14447
+    },
+    {
+      "epoch": 1.2311887515977844,
+      "grad_norm": 20.5,
+      "learning_rate": 8.12017806210271e-07,
+      "loss": 0.8931,
+      "step": 14448
+    },
+    {
+      "epoch": 1.2312739667660844,
+      "grad_norm": 22.625,
+      "learning_rate": 8.118604271508526e-07,
+      "loss": 1.0771,
+      "step": 14449
+    },
+    {
+      "epoch": 1.2313591819343843,
+      "grad_norm": 15.5,
+      "learning_rate": 8.117030560090572e-07,
+      "loss": 0.7453,
+      "step": 14450
+    },
+    {
+      "epoch": 1.2314443971026843,
+      "grad_norm": 16.375,
+      "learning_rate": 8.11545692787728e-07,
+      "loss": 0.4901,
+      "step": 14451
+    },
+    {
+      "epoch": 1.2315296122709842,
+      "grad_norm": 19.625,
+      "learning_rate": 8.113883374897093e-07,
+      "loss": 0.8172,
+      "step": 14452
+    },
+    {
+      "epoch": 1.2316148274392842,
+      "grad_norm": 13.875,
+      "learning_rate": 8.112309901178448e-07,
+      "loss": 0.5826,
+      "step": 14453
+    },
+    {
+      "epoch": 1.2317000426075841,
+      "grad_norm": 27.0,
+      "learning_rate": 8.110736506749769e-07,
+      "loss": 0.7465,
+      "step": 14454
+    },
+    {
+      "epoch": 1.231785257775884,
+      "grad_norm": 20.25,
+      "learning_rate": 8.109163191639496e-07,
+      "loss": 0.5639,
+      "step": 14455
+    },
+    {
+      "epoch": 1.231870472944184,
+      "grad_norm": 14.8125,
+      "learning_rate": 8.107589955876061e-07,
+      "loss": 0.5941,
+      "step": 14456
+    },
+    {
+      "epoch": 1.231955688112484,
+      "grad_norm": 18.625,
+      "learning_rate": 8.10601679948789e-07,
+      "loss": 0.8677,
+      "step": 14457
+    },
+    {
+      "epoch": 1.232040903280784,
+      "grad_norm": 10.375,
+      "learning_rate": 8.104443722503416e-07,
+      "loss": 0.3252,
+      "step": 14458
+    },
+    {
+      "epoch": 1.232126118449084,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.102870724951062e-07,
+      "loss": 0.4186,
+      "step": 14459
+    },
+    {
+      "epoch": 1.2322113336173839,
+      "grad_norm": 21.625,
+      "learning_rate": 8.101297806859257e-07,
+      "loss": 0.6037,
+      "step": 14460
+    },
+    {
+      "epoch": 1.2322965487856838,
+      "grad_norm": 14.875,
+      "learning_rate": 8.099724968256429e-07,
+      "loss": 0.4335,
+      "step": 14461
+    },
+    {
+      "epoch": 1.2323817639539838,
+      "grad_norm": 19.0,
+      "learning_rate": 8.098152209170988e-07,
+      "loss": 1.0173,
+      "step": 14462
+    },
+    {
+      "epoch": 1.2324669791222838,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.096579529631367e-07,
+      "loss": 0.3258,
+      "step": 14463
+    },
+    {
+      "epoch": 1.2325521942905837,
+      "grad_norm": 15.875,
+      "learning_rate": 8.095006929665979e-07,
+      "loss": 0.7742,
+      "step": 14464
+    },
+    {
+      "epoch": 1.2326374094588837,
+      "grad_norm": 12.75,
+      "learning_rate": 8.093434409303247e-07,
+      "loss": 0.437,
+      "step": 14465
+    },
+    {
+      "epoch": 1.2327226246271836,
+      "grad_norm": 13.75,
+      "learning_rate": 8.091861968571585e-07,
+      "loss": 0.5079,
+      "step": 14466
+    },
+    {
+      "epoch": 1.2328078397954836,
+      "grad_norm": 13.0,
+      "learning_rate": 8.090289607499412e-07,
+      "loss": 0.4581,
+      "step": 14467
+    },
+    {
+      "epoch": 1.2328930549637835,
+      "grad_norm": 16.625,
+      "learning_rate": 8.088717326115137e-07,
+      "loss": 0.7145,
+      "step": 14468
+    },
+    {
+      "epoch": 1.2329782701320835,
+      "grad_norm": 19.625,
+      "learning_rate": 8.087145124447185e-07,
+      "loss": 0.713,
+      "step": 14469
+    },
+    {
+      "epoch": 1.2330634853003835,
+      "grad_norm": 11.4375,
+      "learning_rate": 8.085573002523948e-07,
+      "loss": 0.8115,
+      "step": 14470
+    },
+    {
+      "epoch": 1.2331487004686834,
+      "grad_norm": 15.8125,
+      "learning_rate": 8.084000960373851e-07,
+      "loss": 0.6105,
+      "step": 14471
+    },
+    {
+      "epoch": 1.2332339156369834,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.082428998025293e-07,
+      "loss": 0.3658,
+      "step": 14472
+    },
+    {
+      "epoch": 1.2333191308052833,
+      "grad_norm": 23.5,
+      "learning_rate": 8.080857115506688e-07,
+      "loss": 0.8776,
+      "step": 14473
+    },
+    {
+      "epoch": 1.2334043459735833,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.079285312846437e-07,
+      "loss": 0.3254,
+      "step": 14474
+    },
+    {
+      "epoch": 1.2334895611418832,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.077713590072947e-07,
+      "loss": 0.5601,
+      "step": 14475
+    },
+    {
+      "epoch": 1.2335747763101832,
+      "grad_norm": 10.9375,
+      "learning_rate": 8.076141947214619e-07,
+      "loss": 0.5177,
+      "step": 14476
+    },
+    {
+      "epoch": 1.2336599914784832,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.074570384299857e-07,
+      "loss": 0.3548,
+      "step": 14477
+    },
+    {
+      "epoch": 1.233745206646783,
+      "grad_norm": 20.125,
+      "learning_rate": 8.072998901357057e-07,
+      "loss": 0.7749,
+      "step": 14478
+    },
+    {
+      "epoch": 1.233830421815083,
+      "grad_norm": 16.375,
+      "learning_rate": 8.071427498414616e-07,
+      "loss": 0.5837,
+      "step": 14479
+    },
+    {
+      "epoch": 1.233915636983383,
+      "grad_norm": 19.5,
+      "learning_rate": 8.069856175500937e-07,
+      "loss": 0.9381,
+      "step": 14480
+    },
+    {
+      "epoch": 1.234000852151683,
+      "grad_norm": 11.8125,
+      "learning_rate": 8.068284932644407e-07,
+      "loss": 0.529,
+      "step": 14481
+    },
+    {
+      "epoch": 1.234086067319983,
+      "grad_norm": 16.625,
+      "learning_rate": 8.06671376987343e-07,
+      "loss": 0.7009,
+      "step": 14482
+    },
+    {
+      "epoch": 1.234171282488283,
+      "grad_norm": 17.375,
+      "learning_rate": 8.065142687216387e-07,
+      "loss": 0.7,
+      "step": 14483
+    },
+    {
+      "epoch": 1.2342564976565829,
+      "grad_norm": 16.75,
+      "learning_rate": 8.063571684701682e-07,
+      "loss": 0.6999,
+      "step": 14484
+    },
+    {
+      "epoch": 1.2343417128248828,
+      "grad_norm": 15.4375,
+      "learning_rate": 8.062000762357697e-07,
+      "loss": 0.5865,
+      "step": 14485
+    },
+    {
+      "epoch": 1.2344269279931828,
+      "grad_norm": 11.25,
+      "learning_rate": 8.06042992021282e-07,
+      "loss": 0.4193,
+      "step": 14486
+    },
+    {
+      "epoch": 1.2345121431614827,
+      "grad_norm": 13.625,
+      "learning_rate": 8.058859158295435e-07,
+      "loss": 0.4003,
+      "step": 14487
+    },
+    {
+      "epoch": 1.2345973583297827,
+      "grad_norm": 16.875,
+      "learning_rate": 8.057288476633935e-07,
+      "loss": 0.6689,
+      "step": 14488
+    },
+    {
+      "epoch": 1.2346825734980826,
+      "grad_norm": 22.125,
+      "learning_rate": 8.055717875256697e-07,
+      "loss": 0.7927,
+      "step": 14489
+    },
+    {
+      "epoch": 1.2347677886663826,
+      "grad_norm": 15.25,
+      "learning_rate": 8.054147354192107e-07,
+      "loss": 0.4423,
+      "step": 14490
+    },
+    {
+      "epoch": 1.2348530038346825,
+      "grad_norm": 15.5,
+      "learning_rate": 8.052576913468544e-07,
+      "loss": 0.5075,
+      "step": 14491
+    },
+    {
+      "epoch": 1.2349382190029825,
+      "grad_norm": 16.625,
+      "learning_rate": 8.051006553114391e-07,
+      "loss": 0.5374,
+      "step": 14492
+    },
+    {
+      "epoch": 1.2350234341712825,
+      "grad_norm": 15.625,
+      "learning_rate": 8.04943627315802e-07,
+      "loss": 0.7394,
+      "step": 14493
+    },
+    {
+      "epoch": 1.2351086493395824,
+      "grad_norm": 15.25,
+      "learning_rate": 8.047866073627816e-07,
+      "loss": 0.4514,
+      "step": 14494
+    },
+    {
+      "epoch": 1.2351938645078824,
+      "grad_norm": 17.25,
+      "learning_rate": 8.046295954552147e-07,
+      "loss": 0.7461,
+      "step": 14495
+    },
+    {
+      "epoch": 1.2352790796761823,
+      "grad_norm": 17.25,
+      "learning_rate": 8.044725915959387e-07,
+      "loss": 0.8528,
+      "step": 14496
+    },
+    {
+      "epoch": 1.2353642948444823,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.04315595787791e-07,
+      "loss": 0.2919,
+      "step": 14497
+    },
+    {
+      "epoch": 1.2354495100127822,
+      "grad_norm": 15.6875,
+      "learning_rate": 8.041586080336085e-07,
+      "loss": 0.5813,
+      "step": 14498
+    },
+    {
+      "epoch": 1.2355347251810822,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.040016283362286e-07,
+      "loss": 0.6932,
+      "step": 14499
+    },
+    {
+      "epoch": 1.2356199403493822,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.038446566984876e-07,
+      "loss": 0.3093,
+      "step": 14500
+    },
+    {
+      "epoch": 1.2357051555176821,
+      "grad_norm": 22.375,
+      "learning_rate": 8.036876931232222e-07,
+      "loss": 0.9861,
+      "step": 14501
+    },
+    {
+      "epoch": 1.235790370685982,
+      "grad_norm": 14.4375,
+      "learning_rate": 8.035307376132695e-07,
+      "loss": 0.5617,
+      "step": 14502
+    },
+    {
+      "epoch": 1.235875585854282,
+      "grad_norm": 15.5,
+      "learning_rate": 8.03373790171465e-07,
+      "loss": 0.8113,
+      "step": 14503
+    },
+    {
+      "epoch": 1.235960801022582,
+      "grad_norm": 16.625,
+      "learning_rate": 8.032168508006451e-07,
+      "loss": 0.743,
+      "step": 14504
+    },
+    {
+      "epoch": 1.236046016190882,
+      "grad_norm": 19.875,
+      "learning_rate": 8.030599195036462e-07,
+      "loss": 0.6202,
+      "step": 14505
+    },
+    {
+      "epoch": 1.236131231359182,
+      "grad_norm": 21.125,
+      "learning_rate": 8.029029962833037e-07,
+      "loss": 0.9888,
+      "step": 14506
+    },
+    {
+      "epoch": 1.2362164465274819,
+      "grad_norm": 11.5625,
+      "learning_rate": 8.027460811424537e-07,
+      "loss": 0.3226,
+      "step": 14507
+    },
+    {
+      "epoch": 1.2363016616957818,
+      "grad_norm": 15.75,
+      "learning_rate": 8.025891740839317e-07,
+      "loss": 0.7289,
+      "step": 14508
+    },
+    {
+      "epoch": 1.2363868768640818,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.024322751105736e-07,
+      "loss": 0.9043,
+      "step": 14509
+    },
+    {
+      "epoch": 1.2364720920323817,
+      "grad_norm": 20.625,
+      "learning_rate": 8.022753842252144e-07,
+      "loss": 0.8845,
+      "step": 14510
+    },
+    {
+      "epoch": 1.2365573072006817,
+      "grad_norm": 16.875,
+      "learning_rate": 8.021185014306888e-07,
+      "loss": 0.5755,
+      "step": 14511
+    },
+    {
+      "epoch": 1.2366425223689816,
+      "grad_norm": 16.5,
+      "learning_rate": 8.019616267298324e-07,
+      "loss": 0.4918,
+      "step": 14512
+    },
+    {
+      "epoch": 1.2367277375372816,
+      "grad_norm": 25.375,
+      "learning_rate": 8.0180476012548e-07,
+      "loss": 0.615,
+      "step": 14513
+    },
+    {
+      "epoch": 1.2368129527055816,
+      "grad_norm": 14.5,
+      "learning_rate": 8.016479016204662e-07,
+      "loss": 0.7477,
+      "step": 14514
+    },
+    {
+      "epoch": 1.2368981678738815,
+      "grad_norm": 16.875,
+      "learning_rate": 8.014910512176258e-07,
+      "loss": 0.7153,
+      "step": 14515
+    },
+    {
+      "epoch": 1.2369833830421815,
+      "grad_norm": 15.5,
+      "learning_rate": 8.013342089197928e-07,
+      "loss": 0.5963,
+      "step": 14516
+    },
+    {
+      "epoch": 1.2370685982104814,
+      "grad_norm": 16.75,
+      "learning_rate": 8.011773747298022e-07,
+      "loss": 0.4796,
+      "step": 14517
+    },
+    {
+      "epoch": 1.2371538133787814,
+      "grad_norm": 25.875,
+      "learning_rate": 8.01020548650488e-07,
+      "loss": 0.8417,
+      "step": 14518
+    },
+    {
+      "epoch": 1.2372390285470813,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.008637306846834e-07,
+      "loss": 0.7155,
+      "step": 14519
+    },
+    {
+      "epoch": 1.2373242437153813,
+      "grad_norm": 13.25,
+      "learning_rate": 8.00706920835223e-07,
+      "loss": 0.7163,
+      "step": 14520
+    },
+    {
+      "epoch": 1.2374094588836813,
+      "grad_norm": 18.25,
+      "learning_rate": 8.005501191049403e-07,
+      "loss": 0.5048,
+      "step": 14521
+    },
+    {
+      "epoch": 1.2374946740519812,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.00393325496669e-07,
+      "loss": 0.5039,
+      "step": 14522
+    },
+    {
+      "epoch": 1.2375798892202812,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.002365400132423e-07,
+      "loss": 0.5394,
+      "step": 14523
+    },
+    {
+      "epoch": 1.2376651043885811,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.000797626574938e-07,
+      "loss": 0.2966,
+      "step": 14524
+    },
+    {
+      "epoch": 1.237750319556881,
+      "grad_norm": 17.75,
+      "learning_rate": 7.999229934322561e-07,
+      "loss": 0.6193,
+      "step": 14525
+    },
+    {
+      "epoch": 1.237835534725181,
+      "grad_norm": 17.375,
+      "learning_rate": 7.997662323403629e-07,
+      "loss": 0.813,
+      "step": 14526
+    },
+    {
+      "epoch": 1.237920749893481,
+      "grad_norm": 25.25,
+      "learning_rate": 7.996094793846465e-07,
+      "loss": 0.6272,
+      "step": 14527
+    },
+    {
+      "epoch": 1.238005965061781,
+      "grad_norm": 16.5,
+      "learning_rate": 7.994527345679395e-07,
+      "loss": 0.6968,
+      "step": 14528
+    },
+    {
+      "epoch": 1.238091180230081,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.992959978930749e-07,
+      "loss": 0.4154,
+      "step": 14529
+    },
+    {
+      "epoch": 1.2381763953983809,
+      "grad_norm": 13.375,
+      "learning_rate": 7.991392693628848e-07,
+      "loss": 0.34,
+      "step": 14530
+    },
+    {
+      "epoch": 1.2382616105666808,
+      "grad_norm": 12.625,
+      "learning_rate": 7.989825489802011e-07,
+      "loss": 0.3067,
+      "step": 14531
+    },
+    {
+      "epoch": 1.2383468257349808,
+      "grad_norm": 13.125,
+      "learning_rate": 7.988258367478568e-07,
+      "loss": 0.5455,
+      "step": 14532
+    },
+    {
+      "epoch": 1.2384320409032807,
+      "grad_norm": 17.0,
+      "learning_rate": 7.98669132668683e-07,
+      "loss": 0.6555,
+      "step": 14533
+    },
+    {
+      "epoch": 1.2385172560715807,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.985124367455122e-07,
+      "loss": 0.7298,
+      "step": 14534
+    },
+    {
+      "epoch": 1.2386024712398807,
+      "grad_norm": 23.125,
+      "learning_rate": 7.983557489811754e-07,
+      "loss": 1.0131,
+      "step": 14535
+    },
+    {
+      "epoch": 1.2386876864081806,
+      "grad_norm": 24.75,
+      "learning_rate": 7.981990693785041e-07,
+      "loss": 0.7111,
+      "step": 14536
+    },
+    {
+      "epoch": 1.2387729015764806,
+      "grad_norm": 10.5,
+      "learning_rate": 7.980423979403304e-07,
+      "loss": 0.2446,
+      "step": 14537
+    },
+    {
+      "epoch": 1.2388581167447805,
+      "grad_norm": 24.125,
+      "learning_rate": 7.978857346694846e-07,
+      "loss": 1.0147,
+      "step": 14538
+    },
+    {
+      "epoch": 1.2389433319130805,
+      "grad_norm": 14.625,
+      "learning_rate": 7.977290795687985e-07,
+      "loss": 0.739,
+      "step": 14539
+    },
+    {
+      "epoch": 1.2390285470813804,
+      "grad_norm": 15.0,
+      "learning_rate": 7.975724326411025e-07,
+      "loss": 0.3423,
+      "step": 14540
+    },
+    {
+      "epoch": 1.2391137622496804,
+      "grad_norm": 8.9375,
+      "learning_rate": 7.97415793889228e-07,
+      "loss": 0.2164,
+      "step": 14541
+    },
+    {
+      "epoch": 1.2391989774179804,
+      "grad_norm": 20.125,
+      "learning_rate": 7.972591633160051e-07,
+      "loss": 0.4979,
+      "step": 14542
+    },
+    {
+      "epoch": 1.2392841925862803,
+      "grad_norm": 9.9375,
+      "learning_rate": 7.971025409242641e-07,
+      "loss": 0.9076,
+      "step": 14543
+    },
+    {
+      "epoch": 1.2393694077545803,
+      "grad_norm": 17.25,
+      "learning_rate": 7.96945926716836e-07,
+      "loss": 0.4437,
+      "step": 14544
+    },
+    {
+      "epoch": 1.2394546229228802,
+      "grad_norm": 13.0625,
+      "learning_rate": 7.967893206965505e-07,
+      "loss": 0.2974,
+      "step": 14545
+    },
+    {
+      "epoch": 1.2395398380911802,
+      "grad_norm": 37.25,
+      "learning_rate": 7.966327228662373e-07,
+      "loss": 1.3901,
+      "step": 14546
+    },
+    {
+      "epoch": 1.2396250532594801,
+      "grad_norm": 14.375,
+      "learning_rate": 7.964761332287271e-07,
+      "loss": 0.6615,
+      "step": 14547
+    },
+    {
+      "epoch": 1.23971026842778,
+      "grad_norm": 18.875,
+      "learning_rate": 7.963195517868488e-07,
+      "loss": 0.7583,
+      "step": 14548
+    },
+    {
+      "epoch": 1.23979548359608,
+      "grad_norm": 12.5,
+      "learning_rate": 7.961629785434329e-07,
+      "loss": 0.3806,
+      "step": 14549
+    },
+    {
+      "epoch": 1.23988069876438,
+      "grad_norm": 12.375,
+      "learning_rate": 7.960064135013085e-07,
+      "loss": 0.3351,
+      "step": 14550
+    },
+    {
+      "epoch": 1.23996591393268,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.958498566633041e-07,
+      "loss": 0.4748,
+      "step": 14551
+    },
+    {
+      "epoch": 1.24005112910098,
+      "grad_norm": 13.1875,
+      "learning_rate": 7.956933080322499e-07,
+      "loss": 0.4207,
+      "step": 14552
+    },
+    {
+      "epoch": 1.2401363442692799,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.955367676109742e-07,
+      "loss": 0.4808,
+      "step": 14553
+    },
+    {
+      "epoch": 1.2402215594375798,
+      "grad_norm": 14.125,
+      "learning_rate": 7.953802354023063e-07,
+      "loss": 0.3672,
+      "step": 14554
+    },
+    {
+      "epoch": 1.2403067746058798,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.952237114090744e-07,
+      "loss": 0.5251,
+      "step": 14555
+    },
+    {
+      "epoch": 1.2403919897741797,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.950671956341077e-07,
+      "loss": 0.5081,
+      "step": 14556
+    },
+    {
+      "epoch": 1.2404772049424797,
+      "grad_norm": 24.25,
+      "learning_rate": 7.949106880802341e-07,
+      "loss": 0.9129,
+      "step": 14557
+    },
+    {
+      "epoch": 1.2405624201107797,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.94754188750282e-07,
+      "loss": 0.5264,
+      "step": 14558
+    },
+    {
+      "epoch": 1.2406476352790796,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.945976976470798e-07,
+      "loss": 0.519,
+      "step": 14559
+    },
+    {
+      "epoch": 1.2407328504473796,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.94441214773455e-07,
+      "loss": 0.4413,
+      "step": 14560
+    },
+    {
+      "epoch": 1.2408180656156795,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.942847401322353e-07,
+      "loss": 0.4321,
+      "step": 14561
+    },
+    {
+      "epoch": 1.2409032807839795,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.941282737262487e-07,
+      "loss": 0.6936,
+      "step": 14562
+    },
+    {
+      "epoch": 1.2409884959522794,
+      "grad_norm": 17.5,
+      "learning_rate": 7.939718155583225e-07,
+      "loss": 0.6167,
+      "step": 14563
+    },
+    {
+      "epoch": 1.2410737111205794,
+      "grad_norm": 17.0,
+      "learning_rate": 7.938153656312844e-07,
+      "loss": 0.7481,
+      "step": 14564
+    },
+    {
+      "epoch": 1.2411589262888794,
+      "grad_norm": 20.25,
+      "learning_rate": 7.936589239479611e-07,
+      "loss": 1.0365,
+      "step": 14565
+    },
+    {
+      "epoch": 1.2412441414571793,
+      "grad_norm": 18.875,
+      "learning_rate": 7.935024905111802e-07,
+      "loss": 0.7845,
+      "step": 14566
+    },
+    {
+      "epoch": 1.2413293566254793,
+      "grad_norm": 29.75,
+      "learning_rate": 7.933460653237684e-07,
+      "loss": 0.9666,
+      "step": 14567
+    },
+    {
+      "epoch": 1.2414145717937792,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.93189648388552e-07,
+      "loss": 0.7358,
+      "step": 14568
+    },
+    {
+      "epoch": 1.2414997869620792,
+      "grad_norm": 19.0,
+      "learning_rate": 7.930332397083585e-07,
+      "loss": 0.4884,
+      "step": 14569
+    },
+    {
+      "epoch": 1.2415850021303791,
+      "grad_norm": 14.5,
+      "learning_rate": 7.928768392860134e-07,
+      "loss": 0.3494,
+      "step": 14570
+    },
+    {
+      "epoch": 1.241670217298679,
+      "grad_norm": 18.625,
+      "learning_rate": 7.927204471243438e-07,
+      "loss": 0.6889,
+      "step": 14571
+    },
+    {
+      "epoch": 1.241755432466979,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.925640632261753e-07,
+      "loss": 0.3671,
+      "step": 14572
+    },
+    {
+      "epoch": 1.241840647635279,
+      "grad_norm": 20.875,
+      "learning_rate": 7.924076875943343e-07,
+      "loss": 0.6434,
+      "step": 14573
+    },
+    {
+      "epoch": 1.241925862803579,
+      "grad_norm": 20.625,
+      "learning_rate": 7.922513202316465e-07,
+      "loss": 1.2731,
+      "step": 14574
+    },
+    {
+      "epoch": 1.242011077971879,
+      "grad_norm": 31.25,
+      "learning_rate": 7.920949611409382e-07,
+      "loss": 0.8002,
+      "step": 14575
+    },
+    {
+      "epoch": 1.2420962931401789,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.919386103250337e-07,
+      "loss": 0.8563,
+      "step": 14576
+    },
+    {
+      "epoch": 1.2421815083084788,
+      "grad_norm": 9.25,
+      "learning_rate": 7.917822677867595e-07,
+      "loss": 0.2108,
+      "step": 14577
+    },
+    {
+      "epoch": 1.2422667234767788,
+      "grad_norm": 16.125,
+      "learning_rate": 7.916259335289401e-07,
+      "loss": 0.7071,
+      "step": 14578
+    },
+    {
+      "epoch": 1.2423519386450788,
+      "grad_norm": 30.625,
+      "learning_rate": 7.914696075544013e-07,
+      "loss": 0.6715,
+      "step": 14579
+    },
+    {
+      "epoch": 1.2424371538133787,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.913132898659676e-07,
+      "loss": 0.6789,
+      "step": 14580
+    },
+    {
+      "epoch": 1.2425223689816787,
+      "grad_norm": 13.375,
+      "learning_rate": 7.911569804664643e-07,
+      "loss": 0.2534,
+      "step": 14581
+    },
+    {
+      "epoch": 1.2426075841499786,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.910006793587152e-07,
+      "loss": 0.359,
+      "step": 14582
+    },
+    {
+      "epoch": 1.2426927993182786,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.908443865455462e-07,
+      "loss": 0.7314,
+      "step": 14583
+    },
+    {
+      "epoch": 1.2427780144865785,
+      "grad_norm": 13.9375,
+      "learning_rate": 7.906881020297805e-07,
+      "loss": 0.6248,
+      "step": 14584
+    },
+    {
+      "epoch": 1.2428632296548785,
+      "grad_norm": 17.125,
+      "learning_rate": 7.905318258142425e-07,
+      "loss": 0.6161,
+      "step": 14585
+    },
+    {
+      "epoch": 1.2429484448231785,
+      "grad_norm": 16.875,
+      "learning_rate": 7.903755579017565e-07,
+      "loss": 0.5821,
+      "step": 14586
+    },
+    {
+      "epoch": 1.2430336599914784,
+      "grad_norm": 12.75,
+      "learning_rate": 7.902192982951462e-07,
+      "loss": 0.3288,
+      "step": 14587
+    },
+    {
+      "epoch": 1.2431188751597784,
+      "grad_norm": 19.25,
+      "learning_rate": 7.900630469972359e-07,
+      "loss": 1.1231,
+      "step": 14588
+    },
+    {
+      "epoch": 1.2432040903280783,
+      "grad_norm": 14.75,
+      "learning_rate": 7.899068040108484e-07,
+      "loss": 0.6788,
+      "step": 14589
+    },
+    {
+      "epoch": 1.2432893054963783,
+      "grad_norm": 13.125,
+      "learning_rate": 7.89750569338808e-07,
+      "loss": 0.4566,
+      "step": 14590
+    },
+    {
+      "epoch": 1.2433745206646782,
+      "grad_norm": 11.6875,
+      "learning_rate": 7.895943429839379e-07,
+      "loss": 0.3042,
+      "step": 14591
+    },
+    {
+      "epoch": 1.2434597358329782,
+      "grad_norm": 16.875,
+      "learning_rate": 7.894381249490605e-07,
+      "loss": 0.5243,
+      "step": 14592
+    },
+    {
+      "epoch": 1.2435449510012782,
+      "grad_norm": 14.75,
+      "learning_rate": 7.892819152369993e-07,
+      "loss": 0.4836,
+      "step": 14593
+    },
+    {
+      "epoch": 1.243630166169578,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.891257138505773e-07,
+      "loss": 0.5922,
+      "step": 14594
+    },
+    {
+      "epoch": 1.243715381337878,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.88969520792617e-07,
+      "loss": 0.4425,
+      "step": 14595
+    },
+    {
+      "epoch": 1.243800596506178,
+      "grad_norm": 13.625,
+      "learning_rate": 7.888133360659412e-07,
+      "loss": 0.4358,
+      "step": 14596
+    },
+    {
+      "epoch": 1.243885811674478,
+      "grad_norm": 18.0,
+      "learning_rate": 7.886571596733719e-07,
+      "loss": 0.5741,
+      "step": 14597
+    },
+    {
+      "epoch": 1.243971026842778,
+      "grad_norm": 10.6875,
+      "learning_rate": 7.88500991617732e-07,
+      "loss": 0.3579,
+      "step": 14598
+    },
+    {
+      "epoch": 1.244056242011078,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.883448319018433e-07,
+      "loss": 0.6104,
+      "step": 14599
+    },
+    {
+      "epoch": 1.2441414571793779,
+      "grad_norm": 16.5,
+      "learning_rate": 7.881886805285274e-07,
+      "loss": 0.5722,
+      "step": 14600
+    },
+    {
+      "epoch": 1.2442266723476778,
+      "grad_norm": 18.0,
+      "learning_rate": 7.880325375006066e-07,
+      "loss": 0.8445,
+      "step": 14601
+    },
+    {
+      "epoch": 1.2443118875159778,
+      "grad_norm": 20.375,
+      "learning_rate": 7.878764028209023e-07,
+      "loss": 0.8089,
+      "step": 14602
+    },
+    {
+      "epoch": 1.2443971026842777,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.877202764922362e-07,
+      "loss": 0.5218,
+      "step": 14603
+    },
+    {
+      "epoch": 1.2444823178525777,
+      "grad_norm": 26.0,
+      "learning_rate": 7.875641585174293e-07,
+      "loss": 1.1188,
+      "step": 14604
+    },
+    {
+      "epoch": 1.2445675330208776,
+      "grad_norm": 24.625,
+      "learning_rate": 7.874080488993033e-07,
+      "loss": 0.9768,
+      "step": 14605
+    },
+    {
+      "epoch": 1.2446527481891776,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.872519476406789e-07,
+      "loss": 1.0397,
+      "step": 14606
+    },
+    {
+      "epoch": 1.2447379633574776,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.870958547443779e-07,
+      "loss": 0.6703,
+      "step": 14607
+    },
+    {
+      "epoch": 1.2448231785257775,
+      "grad_norm": 22.5,
+      "learning_rate": 7.869397702132195e-07,
+      "loss": 0.6299,
+      "step": 14608
+    },
+    {
+      "epoch": 1.2449083936940775,
+      "grad_norm": 15.875,
+      "learning_rate": 7.867836940500252e-07,
+      "loss": 0.7937,
+      "step": 14609
+    },
+    {
+      "epoch": 1.2449936088623774,
+      "grad_norm": 12.5,
+      "learning_rate": 7.866276262576153e-07,
+      "loss": 0.2633,
+      "step": 14610
+    },
+    {
+      "epoch": 1.2450788240306774,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.864715668388103e-07,
+      "loss": 0.4869,
+      "step": 14611
+    },
+    {
+      "epoch": 1.2451640391989773,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.863155157964299e-07,
+      "loss": 0.6115,
+      "step": 14612
+    },
+    {
+      "epoch": 1.2452492543672773,
+      "grad_norm": 18.375,
+      "learning_rate": 7.861594731332948e-07,
+      "loss": 0.513,
+      "step": 14613
+    },
+    {
+      "epoch": 1.2453344695355772,
+      "grad_norm": 21.875,
+      "learning_rate": 7.860034388522239e-07,
+      "loss": 0.5601,
+      "step": 14614
+    },
+    {
+      "epoch": 1.2454196847038772,
+      "grad_norm": 19.25,
+      "learning_rate": 7.858474129560382e-07,
+      "loss": 0.7062,
+      "step": 14615
+    },
+    {
+      "epoch": 1.2455048998721772,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.856913954475562e-07,
+      "loss": 0.4848,
+      "step": 14616
+    },
+    {
+      "epoch": 1.2455901150404771,
+      "grad_norm": 35.25,
+      "learning_rate": 7.855353863295974e-07,
+      "loss": 0.843,
+      "step": 14617
+    },
+    {
+      "epoch": 1.245675330208777,
+      "grad_norm": 21.375,
+      "learning_rate": 7.853793856049814e-07,
+      "loss": 0.7631,
+      "step": 14618
+    },
+    {
+      "epoch": 1.245760545377077,
+      "grad_norm": 15.5,
+      "learning_rate": 7.852233932765269e-07,
+      "loss": 0.7349,
+      "step": 14619
+    },
+    {
+      "epoch": 1.245845760545377,
+      "grad_norm": 13.25,
+      "learning_rate": 7.850674093470533e-07,
+      "loss": 0.5253,
+      "step": 14620
+    },
+    {
+      "epoch": 1.245930975713677,
+      "grad_norm": 15.75,
+      "learning_rate": 7.849114338193788e-07,
+      "loss": 0.5871,
+      "step": 14621
+    },
+    {
+      "epoch": 1.246016190881977,
+      "grad_norm": 14.0,
+      "learning_rate": 7.847554666963226e-07,
+      "loss": 0.8007,
+      "step": 14622
+    },
+    {
+      "epoch": 1.2461014060502769,
+      "grad_norm": 18.0,
+      "learning_rate": 7.845995079807031e-07,
+      "loss": 0.8773,
+      "step": 14623
+    },
+    {
+      "epoch": 1.2461866212185768,
+      "grad_norm": 12.625,
+      "learning_rate": 7.844435576753387e-07,
+      "loss": 0.4596,
+      "step": 14624
+    },
+    {
+      "epoch": 1.2462718363868768,
+      "grad_norm": 30.625,
+      "learning_rate": 7.842876157830467e-07,
+      "loss": 0.5978,
+      "step": 14625
+    },
+    {
+      "epoch": 1.2463570515551767,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.841316823066463e-07,
+      "loss": 0.5941,
+      "step": 14626
+    },
+    {
+      "epoch": 1.2464422667234767,
+      "grad_norm": 16.75,
+      "learning_rate": 7.839757572489545e-07,
+      "loss": 0.6234,
+      "step": 14627
+    },
+    {
+      "epoch": 1.2465274818917766,
+      "grad_norm": 15.0,
+      "learning_rate": 7.838198406127896e-07,
+      "loss": 0.6071,
+      "step": 14628
+    },
+    {
+      "epoch": 1.2466126970600766,
+      "grad_norm": 12.0,
+      "learning_rate": 7.836639324009687e-07,
+      "loss": 0.2458,
+      "step": 14629
+    },
+    {
+      "epoch": 1.2466979122283766,
+      "grad_norm": 16.125,
+      "learning_rate": 7.8350803261631e-07,
+      "loss": 0.5096,
+      "step": 14630
+    },
+    {
+      "epoch": 1.2467831273966765,
+      "grad_norm": 12.375,
+      "learning_rate": 7.833521412616297e-07,
+      "loss": 0.3598,
+      "step": 14631
+    },
+    {
+      "epoch": 1.2468683425649765,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.831962583397459e-07,
+      "loss": 0.6971,
+      "step": 14632
+    },
+    {
+      "epoch": 1.2469535577332764,
+      "grad_norm": 16.0,
+      "learning_rate": 7.830403838534751e-07,
+      "loss": 0.6217,
+      "step": 14633
+    },
+    {
+      "epoch": 1.2470387729015764,
+      "grad_norm": 14.625,
+      "learning_rate": 7.828845178056338e-07,
+      "loss": 0.502,
+      "step": 14634
+    },
+    {
+      "epoch": 1.2471239880698763,
+      "grad_norm": 13.0625,
+      "learning_rate": 7.827286601990391e-07,
+      "loss": 0.4965,
+      "step": 14635
+    },
+    {
+      "epoch": 1.2472092032381763,
+      "grad_norm": 15.125,
+      "learning_rate": 7.825728110365073e-07,
+      "loss": 0.7139,
+      "step": 14636
+    },
+    {
+      "epoch": 1.2472944184064763,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.824169703208551e-07,
+      "loss": 0.7267,
+      "step": 14637
+    },
+    {
+      "epoch": 1.2473796335747762,
+      "grad_norm": 13.5,
+      "learning_rate": 7.822611380548983e-07,
+      "loss": 0.4705,
+      "step": 14638
+    },
+    {
+      "epoch": 1.2474648487430762,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.82105314241453e-07,
+      "loss": 0.439,
+      "step": 14639
+    },
+    {
+      "epoch": 1.2475500639113761,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.819494988833357e-07,
+      "loss": 0.3843,
+      "step": 14640
+    },
+    {
+      "epoch": 1.247635279079676,
+      "grad_norm": 16.125,
+      "learning_rate": 7.817936919833611e-07,
+      "loss": 0.7717,
+      "step": 14641
+    },
+    {
+      "epoch": 1.247720494247976,
+      "grad_norm": 16.125,
+      "learning_rate": 7.816378935443453e-07,
+      "loss": 0.7799,
+      "step": 14642
+    },
+    {
+      "epoch": 1.247805709416276,
+      "grad_norm": 18.75,
+      "learning_rate": 7.814821035691039e-07,
+      "loss": 0.5608,
+      "step": 14643
+    },
+    {
+      "epoch": 1.247890924584576,
+      "grad_norm": 16.875,
+      "learning_rate": 7.813263220604517e-07,
+      "loss": 0.5795,
+      "step": 14644
+    },
+    {
+      "epoch": 1.247976139752876,
+      "grad_norm": 9.75,
+      "learning_rate": 7.811705490212044e-07,
+      "loss": 0.2612,
+      "step": 14645
+    },
+    {
+      "epoch": 1.2480613549211759,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.810147844541766e-07,
+      "loss": 0.6701,
+      "step": 14646
+    },
+    {
+      "epoch": 1.2481465700894758,
+      "grad_norm": 22.5,
+      "learning_rate": 7.808590283621834e-07,
+      "loss": 1.0441,
+      "step": 14647
+    },
+    {
+      "epoch": 1.2482317852577758,
+      "grad_norm": 18.75,
+      "learning_rate": 7.807032807480397e-07,
+      "loss": 0.8453,
+      "step": 14648
+    },
+    {
+      "epoch": 1.2483170004260757,
+      "grad_norm": 13.75,
+      "learning_rate": 7.805475416145589e-07,
+      "loss": 0.5625,
+      "step": 14649
+    },
+    {
+      "epoch": 1.2484022155943757,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.803918109645566e-07,
+      "loss": 0.5269,
+      "step": 14650
+    },
+    {
+      "epoch": 1.2484874307626757,
+      "grad_norm": 17.375,
+      "learning_rate": 7.80236088800846e-07,
+      "loss": 0.5127,
+      "step": 14651
+    },
+    {
+      "epoch": 1.2485726459309756,
+      "grad_norm": 14.0,
+      "learning_rate": 7.80080375126242e-07,
+      "loss": 0.4743,
+      "step": 14652
+    },
+    {
+      "epoch": 1.2486578610992756,
+      "grad_norm": 17.25,
+      "learning_rate": 7.799246699435583e-07,
+      "loss": 0.5051,
+      "step": 14653
+    },
+    {
+      "epoch": 1.2487430762675755,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.797689732556082e-07,
+      "loss": 0.2845,
+      "step": 14654
+    },
+    {
+      "epoch": 1.2488282914358755,
+      "grad_norm": 14.5,
+      "learning_rate": 7.796132850652058e-07,
+      "loss": 0.4436,
+      "step": 14655
+    },
+    {
+      "epoch": 1.2489135066041754,
+      "grad_norm": 17.125,
+      "learning_rate": 7.794576053751646e-07,
+      "loss": 0.6844,
+      "step": 14656
+    },
+    {
+      "epoch": 1.2489987217724754,
+      "grad_norm": 16.875,
+      "learning_rate": 7.79301934188297e-07,
+      "loss": 0.6172,
+      "step": 14657
+    },
+    {
+      "epoch": 1.2490839369407754,
+      "grad_norm": 16.5,
+      "learning_rate": 7.791462715074173e-07,
+      "loss": 0.919,
+      "step": 14658
+    },
+    {
+      "epoch": 1.2491691521090753,
+      "grad_norm": 17.875,
+      "learning_rate": 7.789906173353376e-07,
+      "loss": 0.6099,
+      "step": 14659
+    },
+    {
+      "epoch": 1.2492543672773753,
+      "grad_norm": 24.375,
+      "learning_rate": 7.788349716748714e-07,
+      "loss": 0.6801,
+      "step": 14660
+    },
+    {
+      "epoch": 1.2493395824456752,
+      "grad_norm": 14.875,
+      "learning_rate": 7.786793345288307e-07,
+      "loss": 0.628,
+      "step": 14661
+    },
+    {
+      "epoch": 1.2494247976139752,
+      "grad_norm": 12.25,
+      "learning_rate": 7.785237059000286e-07,
+      "loss": 0.7181,
+      "step": 14662
+    },
+    {
+      "epoch": 1.2495100127822751,
+      "grad_norm": 23.25,
+      "learning_rate": 7.78368085791277e-07,
+      "loss": 0.7345,
+      "step": 14663
+    },
+    {
+      "epoch": 1.249595227950575,
+      "grad_norm": 18.125,
+      "learning_rate": 7.78212474205389e-07,
+      "loss": 0.8377,
+      "step": 14664
+    },
+    {
+      "epoch": 1.249680443118875,
+      "grad_norm": 12.75,
+      "learning_rate": 7.780568711451756e-07,
+      "loss": 0.4615,
+      "step": 14665
+    },
+    {
+      "epoch": 1.249765658287175,
+      "grad_norm": 19.375,
+      "learning_rate": 7.779012766134489e-07,
+      "loss": 0.4634,
+      "step": 14666
+    },
+    {
+      "epoch": 1.249850873455475,
+      "grad_norm": 21.875,
+      "learning_rate": 7.777456906130211e-07,
+      "loss": 0.8786,
+      "step": 14667
+    },
+    {
+      "epoch": 1.249936088623775,
+      "grad_norm": 22.75,
+      "learning_rate": 7.775901131467036e-07,
+      "loss": 0.5075,
+      "step": 14668
+    },
+    {
+      "epoch": 1.2500213037920749,
+      "grad_norm": 26.25,
+      "learning_rate": 7.774345442173076e-07,
+      "loss": 0.8304,
+      "step": 14669
+    },
+    {
+      "epoch": 1.2501065189603748,
+      "grad_norm": 16.875,
+      "learning_rate": 7.772789838276445e-07,
+      "loss": 0.7385,
+      "step": 14670
+    },
+    {
+      "epoch": 1.2501917341286748,
+      "grad_norm": 16.5,
+      "learning_rate": 7.771234319805257e-07,
+      "loss": 0.7161,
+      "step": 14671
+    },
+    {
+      "epoch": 1.2502769492969747,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.769678886787624e-07,
+      "loss": 0.7023,
+      "step": 14672
+    },
+    {
+      "epoch": 1.2503621644652747,
+      "grad_norm": 25.5,
+      "learning_rate": 7.768123539251646e-07,
+      "loss": 0.7773,
+      "step": 14673
+    },
+    {
+      "epoch": 1.2504473796335747,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.766568277225433e-07,
+      "loss": 0.5063,
+      "step": 14674
+    },
+    {
+      "epoch": 1.2505325948018746,
+      "grad_norm": 16.5,
+      "learning_rate": 7.765013100737092e-07,
+      "loss": 0.6495,
+      "step": 14675
+    },
+    {
+      "epoch": 1.2506178099701746,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.763458009814725e-07,
+      "loss": 0.481,
+      "step": 14676
+    },
+    {
+      "epoch": 1.2507030251384745,
+      "grad_norm": 20.125,
+      "learning_rate": 7.761903004486438e-07,
+      "loss": 0.7506,
+      "step": 14677
+    },
+    {
+      "epoch": 1.2507882403067745,
+      "grad_norm": 20.875,
+      "learning_rate": 7.760348084780323e-07,
+      "loss": 0.6135,
+      "step": 14678
+    },
+    {
+      "epoch": 1.2508734554750744,
+      "grad_norm": 32.25,
+      "learning_rate": 7.758793250724488e-07,
+      "loss": 0.7601,
+      "step": 14679
+    },
+    {
+      "epoch": 1.2509586706433744,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.757238502347028e-07,
+      "loss": 0.4277,
+      "step": 14680
+    },
+    {
+      "epoch": 1.2510438858116744,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.755683839676034e-07,
+      "loss": 0.4344,
+      "step": 14681
+    },
+    {
+      "epoch": 1.2511291009799743,
+      "grad_norm": 15.125,
+      "learning_rate": 7.754129262739607e-07,
+      "loss": 0.4443,
+      "step": 14682
+    },
+    {
+      "epoch": 1.2512143161482743,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.752574771565835e-07,
+      "loss": 0.838,
+      "step": 14683
+    },
+    {
+      "epoch": 1.2512995313165742,
+      "grad_norm": 17.0,
+      "learning_rate": 7.751020366182807e-07,
+      "loss": 0.712,
+      "step": 14684
+    },
+    {
+      "epoch": 1.2513847464848742,
+      "grad_norm": 16.25,
+      "learning_rate": 7.749466046618622e-07,
+      "loss": 0.6198,
+      "step": 14685
+    },
+    {
+      "epoch": 1.2514699616531741,
+      "grad_norm": 16.625,
+      "learning_rate": 7.747911812901359e-07,
+      "loss": 0.6036,
+      "step": 14686
+    },
+    {
+      "epoch": 1.251555176821474,
+      "grad_norm": 27.125,
+      "learning_rate": 7.74635766505911e-07,
+      "loss": 0.6246,
+      "step": 14687
+    },
+    {
+      "epoch": 1.251640391989774,
+      "grad_norm": 17.375,
+      "learning_rate": 7.74480360311996e-07,
+      "loss": 0.8158,
+      "step": 14688
+    },
+    {
+      "epoch": 1.251725607158074,
+      "grad_norm": 21.75,
+      "learning_rate": 7.743249627111987e-07,
+      "loss": 0.7725,
+      "step": 14689
+    },
+    {
+      "epoch": 1.251810822326374,
+      "grad_norm": 20.875,
+      "learning_rate": 7.741695737063278e-07,
+      "loss": 0.8485,
+      "step": 14690
+    },
+    {
+      "epoch": 1.251896037494674,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.74014193300191e-07,
+      "loss": 0.4862,
+      "step": 14691
+    },
+    {
+      "epoch": 1.2519812526629739,
+      "grad_norm": 19.125,
+      "learning_rate": 7.738588214955967e-07,
+      "loss": 0.6126,
+      "step": 14692
+    },
+    {
+      "epoch": 1.2520664678312738,
+      "grad_norm": 16.25,
+      "learning_rate": 7.737034582953519e-07,
+      "loss": 0.3129,
+      "step": 14693
+    },
+    {
+      "epoch": 1.2521516829995738,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.73548103702265e-07,
+      "loss": 0.6435,
+      "step": 14694
+    },
+    {
+      "epoch": 1.2522368981678738,
+      "grad_norm": 17.5,
+      "learning_rate": 7.733927577191425e-07,
+      "loss": 0.8859,
+      "step": 14695
+    },
+    {
+      "epoch": 1.2523221133361737,
+      "grad_norm": 12.625,
+      "learning_rate": 7.732374203487925e-07,
+      "loss": 0.455,
+      "step": 14696
+    },
+    {
+      "epoch": 1.2524073285044737,
+      "grad_norm": 22.125,
+      "learning_rate": 7.73082091594022e-07,
+      "loss": 0.8317,
+      "step": 14697
+    },
+    {
+      "epoch": 1.2524925436727736,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.729267714576374e-07,
+      "loss": 0.7693,
+      "step": 14698
+    },
+    {
+      "epoch": 1.2525777588410736,
+      "grad_norm": 17.75,
+      "learning_rate": 7.727714599424454e-07,
+      "loss": 0.7912,
+      "step": 14699
+    },
+    {
+      "epoch": 1.2526629740093735,
+      "grad_norm": 12.375,
+      "learning_rate": 7.726161570512534e-07,
+      "loss": 0.6755,
+      "step": 14700
+    },
+    {
+      "epoch": 1.2527481891776735,
+      "grad_norm": 17.75,
+      "learning_rate": 7.724608627868673e-07,
+      "loss": 0.6475,
+      "step": 14701
+    },
+    {
+      "epoch": 1.2528334043459735,
+      "grad_norm": 13.375,
+      "learning_rate": 7.723055771520938e-07,
+      "loss": 0.6995,
+      "step": 14702
+    },
+    {
+      "epoch": 1.2529186195142734,
+      "grad_norm": 18.25,
+      "learning_rate": 7.721503001497385e-07,
+      "loss": 0.6928,
+      "step": 14703
+    },
+    {
+      "epoch": 1.2530038346825734,
+      "grad_norm": 29.25,
+      "learning_rate": 7.719950317826082e-07,
+      "loss": 0.4541,
+      "step": 14704
+    },
+    {
+      "epoch": 1.2530890498508733,
+      "grad_norm": 14.875,
+      "learning_rate": 7.718397720535086e-07,
+      "loss": 0.6209,
+      "step": 14705
+    },
+    {
+      "epoch": 1.2531742650191733,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.716845209652446e-07,
+      "loss": 0.7165,
+      "step": 14706
+    },
+    {
+      "epoch": 1.2532594801874732,
+      "grad_norm": 15.5,
+      "learning_rate": 7.715292785206224e-07,
+      "loss": 0.5183,
+      "step": 14707
+    },
+    {
+      "epoch": 1.2533446953557732,
+      "grad_norm": 22.0,
+      "learning_rate": 7.713740447224471e-07,
+      "loss": 0.7164,
+      "step": 14708
+    },
+    {
+      "epoch": 1.2534299105240732,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.712188195735245e-07,
+      "loss": 0.4587,
+      "step": 14709
+    },
+    {
+      "epoch": 1.2535151256923731,
+      "grad_norm": 21.625,
+      "learning_rate": 7.710636030766589e-07,
+      "loss": 1.2628,
+      "step": 14710
+    },
+    {
+      "epoch": 1.253600340860673,
+      "grad_norm": 11.8125,
+      "learning_rate": 7.70908395234656e-07,
+      "loss": 0.3048,
+      "step": 14711
+    },
+    {
+      "epoch": 1.253685556028973,
+      "grad_norm": 18.5,
+      "learning_rate": 7.707531960503198e-07,
+      "loss": 0.4974,
+      "step": 14712
+    },
+    {
+      "epoch": 1.253770771197273,
+      "grad_norm": 18.125,
+      "learning_rate": 7.705980055264559e-07,
+      "loss": 0.7327,
+      "step": 14713
+    },
+    {
+      "epoch": 1.253855986365573,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.704428236658676e-07,
+      "loss": 0.5049,
+      "step": 14714
+    },
+    {
+      "epoch": 1.253941201533873,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.702876504713599e-07,
+      "loss": 0.6619,
+      "step": 14715
+    },
+    {
+      "epoch": 1.2540264167021729,
+      "grad_norm": 12.3125,
+      "learning_rate": 7.701324859457364e-07,
+      "loss": 0.5473,
+      "step": 14716
+    },
+    {
+      "epoch": 1.2541116318704728,
+      "grad_norm": 24.25,
+      "learning_rate": 7.699773300918017e-07,
+      "loss": 0.8938,
+      "step": 14717
+    },
+    {
+      "epoch": 1.2541968470387728,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.698221829123592e-07,
+      "loss": 0.6725,
+      "step": 14718
+    },
+    {
+      "epoch": 1.2542820622070727,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.696670444102128e-07,
+      "loss": 0.6757,
+      "step": 14719
+    },
+    {
+      "epoch": 1.2543672773753727,
+      "grad_norm": 19.625,
+      "learning_rate": 7.695119145881657e-07,
+      "loss": 0.4822,
+      "step": 14720
+    },
+    {
+      "epoch": 1.2544524925436726,
+      "grad_norm": 13.75,
+      "learning_rate": 7.693567934490221e-07,
+      "loss": 0.4454,
+      "step": 14721
+    },
+    {
+      "epoch": 1.2545377077119726,
+      "grad_norm": 18.75,
+      "learning_rate": 7.692016809955842e-07,
+      "loss": 0.682,
+      "step": 14722
+    },
+    {
+      "epoch": 1.2546229228802726,
+      "grad_norm": 17.125,
+      "learning_rate": 7.690465772306552e-07,
+      "loss": 1.0014,
+      "step": 14723
+    },
+    {
+      "epoch": 1.2547081380485725,
+      "grad_norm": 14.375,
+      "learning_rate": 7.688914821570384e-07,
+      "loss": 0.6769,
+      "step": 14724
+    },
+    {
+      "epoch": 1.2547933532168725,
+      "grad_norm": 19.0,
+      "learning_rate": 7.68736395777536e-07,
+      "loss": 0.594,
+      "step": 14725
+    },
+    {
+      "epoch": 1.2548785683851726,
+      "grad_norm": 20.125,
+      "learning_rate": 7.685813180949514e-07,
+      "loss": 0.8626,
+      "step": 14726
+    },
+    {
+      "epoch": 1.2549637835534726,
+      "grad_norm": 21.875,
+      "learning_rate": 7.684262491120858e-07,
+      "loss": 0.669,
+      "step": 14727
+    },
+    {
+      "epoch": 1.2550489987217726,
+      "grad_norm": 15.5,
+      "learning_rate": 7.682711888317427e-07,
+      "loss": 0.6762,
+      "step": 14728
+    },
+    {
+      "epoch": 1.2551342138900725,
+      "grad_norm": 20.125,
+      "learning_rate": 7.681161372567239e-07,
+      "loss": 0.4479,
+      "step": 14729
+    },
+    {
+      "epoch": 1.2552194290583725,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.679610943898308e-07,
+      "loss": 0.5165,
+      "step": 14730
+    },
+    {
+      "epoch": 1.2553046442266724,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.678060602338651e-07,
+      "loss": 0.7642,
+      "step": 14731
+    },
+    {
+      "epoch": 1.2553898593949724,
+      "grad_norm": 15.125,
+      "learning_rate": 7.676510347916291e-07,
+      "loss": 0.3226,
+      "step": 14732
+    },
+    {
+      "epoch": 1.2554750745632723,
+      "grad_norm": 16.25,
+      "learning_rate": 7.674960180659235e-07,
+      "loss": 0.3955,
+      "step": 14733
+    },
+    {
+      "epoch": 1.2555602897315723,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.673410100595505e-07,
+      "loss": 0.7236,
+      "step": 14734
+    },
+    {
+      "epoch": 1.2556455048998723,
+      "grad_norm": 15.75,
+      "learning_rate": 7.671860107753105e-07,
+      "loss": 0.6195,
+      "step": 14735
+    },
+    {
+      "epoch": 1.2557307200681722,
+      "grad_norm": 18.0,
+      "learning_rate": 7.670310202160052e-07,
+      "loss": 0.6392,
+      "step": 14736
+    },
+    {
+      "epoch": 1.2558159352364722,
+      "grad_norm": 12.0,
+      "learning_rate": 7.66876038384435e-07,
+      "loss": 0.2498,
+      "step": 14737
+    },
+    {
+      "epoch": 1.2559011504047721,
+      "grad_norm": 22.25,
+      "learning_rate": 7.667210652834004e-07,
+      "loss": 1.0402,
+      "step": 14738
+    },
+    {
+      "epoch": 1.255986365573072,
+      "grad_norm": 13.5,
+      "learning_rate": 7.665661009157022e-07,
+      "loss": 0.5617,
+      "step": 14739
+    },
+    {
+      "epoch": 1.256071580741372,
+      "grad_norm": 17.75,
+      "learning_rate": 7.664111452841402e-07,
+      "loss": 0.5832,
+      "step": 14740
+    },
+    {
+      "epoch": 1.256156795909672,
+      "grad_norm": 18.625,
+      "learning_rate": 7.662561983915157e-07,
+      "loss": 0.9422,
+      "step": 14741
+    },
+    {
+      "epoch": 1.256242011077972,
+      "grad_norm": 14.5625,
+      "learning_rate": 7.661012602406278e-07,
+      "loss": 0.6766,
+      "step": 14742
+    },
+    {
+      "epoch": 1.256327226246272,
+      "grad_norm": 11.8125,
+      "learning_rate": 7.659463308342768e-07,
+      "loss": 0.2537,
+      "step": 14743
+    },
+    {
+      "epoch": 1.2564124414145719,
+      "grad_norm": 15.75,
+      "learning_rate": 7.657914101752622e-07,
+      "loss": 0.5953,
+      "step": 14744
+    },
+    {
+      "epoch": 1.2564976565828718,
+      "grad_norm": 16.75,
+      "learning_rate": 7.656364982663847e-07,
+      "loss": 0.5801,
+      "step": 14745
+    },
+    {
+      "epoch": 1.2565828717511718,
+      "grad_norm": 15.6875,
+      "learning_rate": 7.654815951104419e-07,
+      "loss": 0.4965,
+      "step": 14746
+    },
+    {
+      "epoch": 1.2566680869194717,
+      "grad_norm": 15.125,
+      "learning_rate": 7.65326700710234e-07,
+      "loss": 0.3834,
+      "step": 14747
+    },
+    {
+      "epoch": 1.2567533020877717,
+      "grad_norm": 12.75,
+      "learning_rate": 7.651718150685597e-07,
+      "loss": 0.3145,
+      "step": 14748
+    },
+    {
+      "epoch": 1.2568385172560717,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.650169381882186e-07,
+      "loss": 0.4912,
+      "step": 14749
+    },
+    {
+      "epoch": 1.2569237324243716,
+      "grad_norm": 16.625,
+      "learning_rate": 7.648620700720088e-07,
+      "loss": 0.534,
+      "step": 14750
+    },
+    {
+      "epoch": 1.2570089475926716,
+      "grad_norm": 16.625,
+      "learning_rate": 7.647072107227297e-07,
+      "loss": 0.3808,
+      "step": 14751
+    },
+    {
+      "epoch": 1.2570941627609715,
+      "grad_norm": 10.5625,
+      "learning_rate": 7.64552360143179e-07,
+      "loss": 0.358,
+      "step": 14752
+    },
+    {
+      "epoch": 1.2571793779292715,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.643975183361555e-07,
+      "loss": 0.5785,
+      "step": 14753
+    },
+    {
+      "epoch": 1.2572645930975714,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.642426853044573e-07,
+      "loss": 0.3668,
+      "step": 14754
+    },
+    {
+      "epoch": 1.2573498082658714,
+      "grad_norm": 11.75,
+      "learning_rate": 7.640878610508817e-07,
+      "loss": 0.3408,
+      "step": 14755
+    },
+    {
+      "epoch": 1.2574350234341714,
+      "grad_norm": 14.6875,
+      "learning_rate": 7.639330455782275e-07,
+      "loss": 0.4472,
+      "step": 14756
+    },
+    {
+      "epoch": 1.2575202386024713,
+      "grad_norm": 14.375,
+      "learning_rate": 7.637782388892916e-07,
+      "loss": 0.7079,
+      "step": 14757
+    },
+    {
+      "epoch": 1.2576054537707713,
+      "grad_norm": 17.625,
+      "learning_rate": 7.63623440986872e-07,
+      "loss": 0.6583,
+      "step": 14758
+    },
+    {
+      "epoch": 1.2576906689390712,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.634686518737659e-07,
+      "loss": 0.5979,
+      "step": 14759
+    },
+    {
+      "epoch": 1.2577758841073712,
+      "grad_norm": 17.25,
+      "learning_rate": 7.633138715527706e-07,
+      "loss": 0.8534,
+      "step": 14760
+    },
+    {
+      "epoch": 1.2578610992756711,
+      "grad_norm": 15.5,
+      "learning_rate": 7.631591000266829e-07,
+      "loss": 0.5489,
+      "step": 14761
+    },
+    {
+      "epoch": 1.257946314443971,
+      "grad_norm": 13.875,
+      "learning_rate": 7.630043372983005e-07,
+      "loss": 0.491,
+      "step": 14762
+    },
+    {
+      "epoch": 1.258031529612271,
+      "grad_norm": 17.375,
+      "learning_rate": 7.628495833704186e-07,
+      "loss": 0.8262,
+      "step": 14763
+    },
+    {
+      "epoch": 1.258116744780571,
+      "grad_norm": 14.375,
+      "learning_rate": 7.62694838245835e-07,
+      "loss": 0.5647,
+      "step": 14764
+    },
+    {
+      "epoch": 1.258201959948871,
+      "grad_norm": 20.75,
+      "learning_rate": 7.625401019273454e-07,
+      "loss": 0.5822,
+      "step": 14765
+    },
+    {
+      "epoch": 1.258287175117171,
+      "grad_norm": 17.5,
+      "learning_rate": 7.623853744177466e-07,
+      "loss": 0.6623,
+      "step": 14766
+    },
+    {
+      "epoch": 1.2583723902854709,
+      "grad_norm": 13.1875,
+      "learning_rate": 7.622306557198341e-07,
+      "loss": 0.4083,
+      "step": 14767
+    },
+    {
+      "epoch": 1.2584576054537708,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.620759458364043e-07,
+      "loss": 0.4822,
+      "step": 14768
+    },
+    {
+      "epoch": 1.2585428206220708,
+      "grad_norm": 20.875,
+      "learning_rate": 7.619212447702528e-07,
+      "loss": 0.6139,
+      "step": 14769
+    },
+    {
+      "epoch": 1.2586280357903707,
+      "grad_norm": 18.75,
+      "learning_rate": 7.617665525241757e-07,
+      "loss": 0.7264,
+      "step": 14770
+    },
+    {
+      "epoch": 1.2587132509586707,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.616118691009675e-07,
+      "loss": 0.6282,
+      "step": 14771
+    },
+    {
+      "epoch": 1.2587984661269707,
+      "grad_norm": 12.375,
+      "learning_rate": 7.614571945034238e-07,
+      "loss": 0.4528,
+      "step": 14772
+    },
+    {
+      "epoch": 1.2588836812952706,
+      "grad_norm": 16.875,
+      "learning_rate": 7.613025287343401e-07,
+      "loss": 0.4314,
+      "step": 14773
+    },
+    {
+      "epoch": 1.2589688964635706,
+      "grad_norm": 16.75,
+      "learning_rate": 7.611478717965109e-07,
+      "loss": 0.596,
+      "step": 14774
+    },
+    {
+      "epoch": 1.2590541116318705,
+      "grad_norm": 13.0,
+      "learning_rate": 7.609932236927315e-07,
+      "loss": 0.3991,
+      "step": 14775
+    },
+    {
+      "epoch": 1.2591393268001705,
+      "grad_norm": 16.875,
+      "learning_rate": 7.608385844257959e-07,
+      "loss": 0.4711,
+      "step": 14776
+    },
+    {
+      "epoch": 1.2592245419684704,
+      "grad_norm": 13.625,
+      "learning_rate": 7.606839539984995e-07,
+      "loss": 0.3357,
+      "step": 14777
+    },
+    {
+      "epoch": 1.2593097571367704,
+      "grad_norm": 16.5,
+      "learning_rate": 7.60529332413636e-07,
+      "loss": 0.7104,
+      "step": 14778
+    },
+    {
+      "epoch": 1.2593949723050704,
+      "grad_norm": 21.375,
+      "learning_rate": 7.603747196739997e-07,
+      "loss": 0.998,
+      "step": 14779
+    },
+    {
+      "epoch": 1.2594801874733703,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.602201157823841e-07,
+      "loss": 0.729,
+      "step": 14780
+    },
+    {
+      "epoch": 1.2595654026416703,
+      "grad_norm": 16.375,
+      "learning_rate": 7.600655207415838e-07,
+      "loss": 0.6283,
+      "step": 14781
+    },
+    {
+      "epoch": 1.2596506178099702,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.59910934554392e-07,
+      "loss": 0.2892,
+      "step": 14782
+    },
+    {
+      "epoch": 1.2597358329782702,
+      "grad_norm": 16.75,
+      "learning_rate": 7.597563572236027e-07,
+      "loss": 0.8029,
+      "step": 14783
+    },
+    {
+      "epoch": 1.2598210481465701,
+      "grad_norm": 18.5,
+      "learning_rate": 7.596017887520088e-07,
+      "loss": 0.737,
+      "step": 14784
+    },
+    {
+      "epoch": 1.25990626331487,
+      "grad_norm": 16.125,
+      "learning_rate": 7.59447229142404e-07,
+      "loss": 0.6369,
+      "step": 14785
+    },
+    {
+      "epoch": 1.25999147848317,
+      "grad_norm": 18.25,
+      "learning_rate": 7.592926783975813e-07,
+      "loss": 0.3987,
+      "step": 14786
+    },
+    {
+      "epoch": 1.26007669365147,
+      "grad_norm": 38.25,
+      "learning_rate": 7.591381365203327e-07,
+      "loss": 0.6789,
+      "step": 14787
+    },
+    {
+      "epoch": 1.26016190881977,
+      "grad_norm": 17.75,
+      "learning_rate": 7.58983603513452e-07,
+      "loss": 0.5868,
+      "step": 14788
+    },
+    {
+      "epoch": 1.26024712398807,
+      "grad_norm": 19.0,
+      "learning_rate": 7.58829079379731e-07,
+      "loss": 0.9919,
+      "step": 14789
+    },
+    {
+      "epoch": 1.2603323391563699,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.586745641219626e-07,
+      "loss": 0.5793,
+      "step": 14790
+    },
+    {
+      "epoch": 1.2604175543246698,
+      "grad_norm": 19.125,
+      "learning_rate": 7.585200577429387e-07,
+      "loss": 0.6396,
+      "step": 14791
+    },
+    {
+      "epoch": 1.2605027694929698,
+      "grad_norm": 11.5,
+      "learning_rate": 7.58365560245452e-07,
+      "loss": 0.3634,
+      "step": 14792
+    },
+    {
+      "epoch": 1.2605879846612698,
+      "grad_norm": 17.125,
+      "learning_rate": 7.582110716322938e-07,
+      "loss": 0.5278,
+      "step": 14793
+    },
+    {
+      "epoch": 1.2606731998295697,
+      "grad_norm": 15.75,
+      "learning_rate": 7.580565919062563e-07,
+      "loss": 0.4552,
+      "step": 14794
+    },
+    {
+      "epoch": 1.2607584149978697,
+      "grad_norm": 20.75,
+      "learning_rate": 7.579021210701304e-07,
+      "loss": 0.7806,
+      "step": 14795
+    },
+    {
+      "epoch": 1.2608436301661696,
+      "grad_norm": 18.5,
+      "learning_rate": 7.577476591267084e-07,
+      "loss": 0.7404,
+      "step": 14796
+    },
+    {
+      "epoch": 1.2609288453344696,
+      "grad_norm": 27.125,
+      "learning_rate": 7.575932060787808e-07,
+      "loss": 1.1459,
+      "step": 14797
+    },
+    {
+      "epoch": 1.2610140605027695,
+      "grad_norm": 14.625,
+      "learning_rate": 7.574387619291396e-07,
+      "loss": 0.6001,
+      "step": 14798
+    },
+    {
+      "epoch": 1.2610992756710695,
+      "grad_norm": 12.375,
+      "learning_rate": 7.572843266805747e-07,
+      "loss": 0.5163,
+      "step": 14799
+    },
+    {
+      "epoch": 1.2611844908393695,
+      "grad_norm": 15.6875,
+      "learning_rate": 7.571299003358779e-07,
+      "loss": 0.8591,
+      "step": 14800
+    },
+    {
+      "epoch": 1.2612697060076694,
+      "grad_norm": 18.625,
+      "learning_rate": 7.569754828978393e-07,
+      "loss": 0.7369,
+      "step": 14801
+    },
+    {
+      "epoch": 1.2613549211759694,
+      "grad_norm": 17.375,
+      "learning_rate": 7.568210743692497e-07,
+      "loss": 0.5494,
+      "step": 14802
+    },
+    {
+      "epoch": 1.2614401363442693,
+      "grad_norm": 16.0,
+      "learning_rate": 7.566666747528992e-07,
+      "loss": 0.6382,
+      "step": 14803
+    },
+    {
+      "epoch": 1.2615253515125693,
+      "grad_norm": 12.875,
+      "learning_rate": 7.565122840515776e-07,
+      "loss": 0.4514,
+      "step": 14804
+    },
+    {
+      "epoch": 1.2616105666808692,
+      "grad_norm": 18.25,
+      "learning_rate": 7.563579022680756e-07,
+      "loss": 0.5869,
+      "step": 14805
+    },
+    {
+      "epoch": 1.2616957818491692,
+      "grad_norm": 14.375,
+      "learning_rate": 7.562035294051824e-07,
+      "loss": 0.5325,
+      "step": 14806
+    },
+    {
+      "epoch": 1.2617809970174692,
+      "grad_norm": 17.0,
+      "learning_rate": 7.560491654656882e-07,
+      "loss": 0.4941,
+      "step": 14807
+    },
+    {
+      "epoch": 1.261866212185769,
+      "grad_norm": 11.875,
+      "learning_rate": 7.558948104523825e-07,
+      "loss": 0.415,
+      "step": 14808
+    },
+    {
+      "epoch": 1.261951427354069,
+      "grad_norm": 17.125,
+      "learning_rate": 7.557404643680538e-07,
+      "loss": 0.6768,
+      "step": 14809
+    },
+    {
+      "epoch": 1.262036642522369,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.555861272154927e-07,
+      "loss": 0.3011,
+      "step": 14810
+    },
+    {
+      "epoch": 1.262121857690669,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.554317989974871e-07,
+      "loss": 0.5505,
+      "step": 14811
+    },
+    {
+      "epoch": 1.262207072858969,
+      "grad_norm": 19.0,
+      "learning_rate": 7.552774797168261e-07,
+      "loss": 0.4301,
+      "step": 14812
+    },
+    {
+      "epoch": 1.262292288027269,
+      "grad_norm": 17.875,
+      "learning_rate": 7.551231693762987e-07,
+      "loss": 0.5846,
+      "step": 14813
+    },
+    {
+      "epoch": 1.2623775031955689,
+      "grad_norm": 10.9375,
+      "learning_rate": 7.549688679786929e-07,
+      "loss": 0.3565,
+      "step": 14814
+    },
+    {
+      "epoch": 1.2624627183638688,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.548145755267979e-07,
+      "loss": 0.3375,
+      "step": 14815
+    },
+    {
+      "epoch": 1.2625479335321688,
+      "grad_norm": 18.375,
+      "learning_rate": 7.546602920234011e-07,
+      "loss": 0.5924,
+      "step": 14816
+    },
+    {
+      "epoch": 1.2626331487004687,
+      "grad_norm": 24.5,
+      "learning_rate": 7.545060174712912e-07,
+      "loss": 1.164,
+      "step": 14817
+    },
+    {
+      "epoch": 1.2627183638687687,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.54351751873256e-07,
+      "loss": 0.5373,
+      "step": 14818
+    },
+    {
+      "epoch": 1.2628035790370686,
+      "grad_norm": 14.5625,
+      "learning_rate": 7.541974952320827e-07,
+      "loss": 0.3873,
+      "step": 14819
+    },
+    {
+      "epoch": 1.2628887942053686,
+      "grad_norm": 16.125,
+      "learning_rate": 7.540432475505593e-07,
+      "loss": 0.6975,
+      "step": 14820
+    },
+    {
+      "epoch": 1.2629740093736685,
+      "grad_norm": 22.375,
+      "learning_rate": 7.538890088314731e-07,
+      "loss": 0.7877,
+      "step": 14821
+    },
+    {
+      "epoch": 1.2630592245419685,
+      "grad_norm": 17.25,
+      "learning_rate": 7.537347790776114e-07,
+      "loss": 0.7288,
+      "step": 14822
+    },
+    {
+      "epoch": 1.2631444397102685,
+      "grad_norm": 12.375,
+      "learning_rate": 7.535805582917616e-07,
+      "loss": 0.2239,
+      "step": 14823
+    },
+    {
+      "epoch": 1.2632296548785684,
+      "grad_norm": 17.375,
+      "learning_rate": 7.534263464767097e-07,
+      "loss": 0.6621,
+      "step": 14824
+    },
+    {
+      "epoch": 1.2633148700468684,
+      "grad_norm": 16.625,
+      "learning_rate": 7.532721436352436e-07,
+      "loss": 0.5718,
+      "step": 14825
+    },
+    {
+      "epoch": 1.2634000852151683,
+      "grad_norm": 13.875,
+      "learning_rate": 7.53117949770149e-07,
+      "loss": 0.48,
+      "step": 14826
+    },
+    {
+      "epoch": 1.2634853003834683,
+      "grad_norm": 15.875,
+      "learning_rate": 7.529637648842131e-07,
+      "loss": 0.6214,
+      "step": 14827
+    },
+    {
+      "epoch": 1.2635705155517682,
+      "grad_norm": 22.0,
+      "learning_rate": 7.528095889802217e-07,
+      "loss": 0.7493,
+      "step": 14828
+    },
+    {
+      "epoch": 1.2636557307200682,
+      "grad_norm": 10.8125,
+      "learning_rate": 7.526554220609605e-07,
+      "loss": 0.2894,
+      "step": 14829
+    },
+    {
+      "epoch": 1.2637409458883682,
+      "grad_norm": 23.0,
+      "learning_rate": 7.525012641292165e-07,
+      "loss": 0.6788,
+      "step": 14830
+    },
+    {
+      "epoch": 1.2638261610566681,
+      "grad_norm": 12.75,
+      "learning_rate": 7.523471151877744e-07,
+      "loss": 0.5408,
+      "step": 14831
+    },
+    {
+      "epoch": 1.263911376224968,
+      "grad_norm": 12.3125,
+      "learning_rate": 7.521929752394209e-07,
+      "loss": 0.5056,
+      "step": 14832
+    },
+    {
+      "epoch": 1.263996591393268,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.520388442869403e-07,
+      "loss": 0.629,
+      "step": 14833
+    },
+    {
+      "epoch": 1.264081806561568,
+      "grad_norm": 16.5,
+      "learning_rate": 7.518847223331191e-07,
+      "loss": 0.6272,
+      "step": 14834
+    },
+    {
+      "epoch": 1.264167021729868,
+      "grad_norm": 18.75,
+      "learning_rate": 7.51730609380742e-07,
+      "loss": 0.6811,
+      "step": 14835
+    },
+    {
+      "epoch": 1.264252236898168,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.515765054325932e-07,
+      "loss": 0.5443,
+      "step": 14836
+    },
+    {
+      "epoch": 1.2643374520664679,
+      "grad_norm": 16.75,
+      "learning_rate": 7.514224104914586e-07,
+      "loss": 0.5802,
+      "step": 14837
+    },
+    {
+      "epoch": 1.2644226672347678,
+      "grad_norm": 17.875,
+      "learning_rate": 7.512683245601224e-07,
+      "loss": 0.9035,
+      "step": 14838
+    },
+    {
+      "epoch": 1.2645078824030678,
+      "grad_norm": 13.125,
+      "learning_rate": 7.511142476413687e-07,
+      "loss": 0.6229,
+      "step": 14839
+    },
+    {
+      "epoch": 1.2645930975713677,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.509601797379824e-07,
+      "loss": 0.484,
+      "step": 14840
+    },
+    {
+      "epoch": 1.2646783127396677,
+      "grad_norm": 23.125,
+      "learning_rate": 7.508061208527474e-07,
+      "loss": 0.7213,
+      "step": 14841
+    },
+    {
+      "epoch": 1.2647635279079676,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.506520709884479e-07,
+      "loss": 0.6746,
+      "step": 14842
+    },
+    {
+      "epoch": 1.2648487430762676,
+      "grad_norm": 18.5,
+      "learning_rate": 7.50498030147868e-07,
+      "loss": 0.9253,
+      "step": 14843
+    },
+    {
+      "epoch": 1.2649339582445676,
+      "grad_norm": 17.375,
+      "learning_rate": 7.503439983337904e-07,
+      "loss": 0.5983,
+      "step": 14844
+    },
+    {
+      "epoch": 1.2650191734128675,
+      "grad_norm": 13.5,
+      "learning_rate": 7.501899755489994e-07,
+      "loss": 0.446,
+      "step": 14845
+    },
+    {
+      "epoch": 1.2651043885811675,
+      "grad_norm": 15.0,
+      "learning_rate": 7.50035961796278e-07,
+      "loss": 0.5689,
+      "step": 14846
+    },
+    {
+      "epoch": 1.2651896037494674,
+      "grad_norm": 14.5,
+      "learning_rate": 7.498819570784098e-07,
+      "loss": 0.448,
+      "step": 14847
+    },
+    {
+      "epoch": 1.2652748189177674,
+      "grad_norm": 12.25,
+      "learning_rate": 7.497279613981773e-07,
+      "loss": 0.3957,
+      "step": 14848
+    },
+    {
+      "epoch": 1.2653600340860673,
+      "grad_norm": 20.5,
+      "learning_rate": 7.495739747583636e-07,
+      "loss": 0.6283,
+      "step": 14849
+    },
+    {
+      "epoch": 1.2654452492543673,
+      "grad_norm": 17.0,
+      "learning_rate": 7.494199971617513e-07,
+      "loss": 0.9095,
+      "step": 14850
+    },
+    {
+      "epoch": 1.2655304644226673,
+      "grad_norm": 11.875,
+      "learning_rate": 7.492660286111237e-07,
+      "loss": 0.503,
+      "step": 14851
+    },
+    {
+      "epoch": 1.2656156795909672,
+      "grad_norm": 21.0,
+      "learning_rate": 7.49112069109262e-07,
+      "loss": 0.8372,
+      "step": 14852
+    },
+    {
+      "epoch": 1.2657008947592672,
+      "grad_norm": 10.9375,
+      "learning_rate": 7.489581186589489e-07,
+      "loss": 0.3218,
+      "step": 14853
+    },
+    {
+      "epoch": 1.2657861099275671,
+      "grad_norm": 36.25,
+      "learning_rate": 7.488041772629663e-07,
+      "loss": 0.7869,
+      "step": 14854
+    },
+    {
+      "epoch": 1.265871325095867,
+      "grad_norm": 16.0,
+      "learning_rate": 7.486502449240965e-07,
+      "loss": 0.723,
+      "step": 14855
+    },
+    {
+      "epoch": 1.265956540264167,
+      "grad_norm": 16.5,
+      "learning_rate": 7.484963216451205e-07,
+      "loss": 0.5091,
+      "step": 14856
+    },
+    {
+      "epoch": 1.266041755432467,
+      "grad_norm": 15.6875,
+      "learning_rate": 7.483424074288207e-07,
+      "loss": 0.5811,
+      "step": 14857
+    },
+    {
+      "epoch": 1.266126970600767,
+      "grad_norm": 11.625,
+      "learning_rate": 7.481885022779775e-07,
+      "loss": 0.2233,
+      "step": 14858
+    },
+    {
+      "epoch": 1.266212185769067,
+      "grad_norm": 15.375,
+      "learning_rate": 7.480346061953735e-07,
+      "loss": 0.7656,
+      "step": 14859
+    },
+    {
+      "epoch": 1.2662974009373669,
+      "grad_norm": 14.25,
+      "learning_rate": 7.478807191837885e-07,
+      "loss": 0.3737,
+      "step": 14860
+    },
+    {
+      "epoch": 1.2663826161056668,
+      "grad_norm": 18.875,
+      "learning_rate": 7.477268412460037e-07,
+      "loss": 0.8529,
+      "step": 14861
+    },
+    {
+      "epoch": 1.2664678312739668,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.475729723848e-07,
+      "loss": 0.6417,
+      "step": 14862
+    },
+    {
+      "epoch": 1.2665530464422667,
+      "grad_norm": 11.8125,
+      "learning_rate": 7.474191126029579e-07,
+      "loss": 0.3944,
+      "step": 14863
+    },
+    {
+      "epoch": 1.2666382616105667,
+      "grad_norm": 11.625,
+      "learning_rate": 7.47265261903258e-07,
+      "loss": 0.5039,
+      "step": 14864
+    },
+    {
+      "epoch": 1.2667234767788667,
+      "grad_norm": 19.375,
+      "learning_rate": 7.4711142028848e-07,
+      "loss": 0.695,
+      "step": 14865
+    },
+    {
+      "epoch": 1.2668086919471666,
+      "grad_norm": 13.875,
+      "learning_rate": 7.469575877614047e-07,
+      "loss": 0.6105,
+      "step": 14866
+    },
+    {
+      "epoch": 1.2668939071154666,
+      "grad_norm": 14.5625,
+      "learning_rate": 7.468037643248119e-07,
+      "loss": 0.4175,
+      "step": 14867
+    },
+    {
+      "epoch": 1.2669791222837665,
+      "grad_norm": 14.6875,
+      "learning_rate": 7.466499499814807e-07,
+      "loss": 0.7527,
+      "step": 14868
+    },
+    {
+      "epoch": 1.2670643374520665,
+      "grad_norm": 22.125,
+      "learning_rate": 7.464961447341909e-07,
+      "loss": 0.5648,
+      "step": 14869
+    },
+    {
+      "epoch": 1.2671495526203664,
+      "grad_norm": 18.5,
+      "learning_rate": 7.463423485857224e-07,
+      "loss": 0.7705,
+      "step": 14870
+    },
+    {
+      "epoch": 1.2672347677886664,
+      "grad_norm": 13.375,
+      "learning_rate": 7.461885615388536e-07,
+      "loss": 0.5381,
+      "step": 14871
+    },
+    {
+      "epoch": 1.2673199829569664,
+      "grad_norm": 13.0,
+      "learning_rate": 7.460347835963647e-07,
+      "loss": 0.4936,
+      "step": 14872
+    },
+    {
+      "epoch": 1.2674051981252663,
+      "grad_norm": 20.875,
+      "learning_rate": 7.458810147610336e-07,
+      "loss": 0.8413,
+      "step": 14873
+    },
+    {
+      "epoch": 1.2674904132935663,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.457272550356399e-07,
+      "loss": 0.7646,
+      "step": 14874
+    },
+    {
+      "epoch": 1.2675756284618662,
+      "grad_norm": 16.375,
+      "learning_rate": 7.455735044229617e-07,
+      "loss": 0.7063,
+      "step": 14875
+    },
+    {
+      "epoch": 1.2676608436301662,
+      "grad_norm": 15.5,
+      "learning_rate": 7.454197629257773e-07,
+      "loss": 0.6136,
+      "step": 14876
+    },
+    {
+      "epoch": 1.2677460587984661,
+      "grad_norm": 13.5,
+      "learning_rate": 7.452660305468653e-07,
+      "loss": 0.3719,
+      "step": 14877
+    },
+    {
+      "epoch": 1.267831273966766,
+      "grad_norm": 17.875,
+      "learning_rate": 7.451123072890037e-07,
+      "loss": 0.8116,
+      "step": 14878
+    },
+    {
+      "epoch": 1.267916489135066,
+      "grad_norm": 16.125,
+      "learning_rate": 7.449585931549704e-07,
+      "loss": 0.5316,
+      "step": 14879
+    },
+    {
+      "epoch": 1.268001704303366,
+      "grad_norm": 16.125,
+      "learning_rate": 7.44804888147543e-07,
+      "loss": 0.4795,
+      "step": 14880
+    },
+    {
+      "epoch": 1.268086919471666,
+      "grad_norm": 20.5,
+      "learning_rate": 7.446511922694994e-07,
+      "loss": 0.5703,
+      "step": 14881
+    },
+    {
+      "epoch": 1.268172134639966,
+      "grad_norm": 11.9375,
+      "learning_rate": 7.444975055236169e-07,
+      "loss": 0.35,
+      "step": 14882
+    },
+    {
+      "epoch": 1.2682573498082659,
+      "grad_norm": 14.0,
+      "learning_rate": 7.443438279126734e-07,
+      "loss": 0.5845,
+      "step": 14883
+    },
+    {
+      "epoch": 1.2683425649765658,
+      "grad_norm": 21.875,
+      "learning_rate": 7.441901594394446e-07,
+      "loss": 0.7929,
+      "step": 14884
+    },
+    {
+      "epoch": 1.2684277801448658,
+      "grad_norm": 12.25,
+      "learning_rate": 7.440365001067087e-07,
+      "loss": 0.3127,
+      "step": 14885
+    },
+    {
+      "epoch": 1.2685129953131657,
+      "grad_norm": 14.25,
+      "learning_rate": 7.438828499172416e-07,
+      "loss": 0.3749,
+      "step": 14886
+    },
+    {
+      "epoch": 1.2685982104814657,
+      "grad_norm": 11.6875,
+      "learning_rate": 7.437292088738207e-07,
+      "loss": 0.5223,
+      "step": 14887
+    },
+    {
+      "epoch": 1.2686834256497657,
+      "grad_norm": 15.5,
+      "learning_rate": 7.435755769792218e-07,
+      "loss": 0.5193,
+      "step": 14888
+    },
+    {
+      "epoch": 1.2687686408180656,
+      "grad_norm": 11.25,
+      "learning_rate": 7.434219542362217e-07,
+      "loss": 0.3711,
+      "step": 14889
+    },
+    {
+      "epoch": 1.2688538559863656,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.432683406475962e-07,
+      "loss": 0.3422,
+      "step": 14890
+    },
+    {
+      "epoch": 1.2689390711546655,
+      "grad_norm": 18.5,
+      "learning_rate": 7.431147362161217e-07,
+      "loss": 0.9403,
+      "step": 14891
+    },
+    {
+      "epoch": 1.2690242863229655,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.429611409445733e-07,
+      "loss": 0.3545,
+      "step": 14892
+    },
+    {
+      "epoch": 1.2691095014912654,
+      "grad_norm": 16.25,
+      "learning_rate": 7.42807554835727e-07,
+      "loss": 0.5702,
+      "step": 14893
+    },
+    {
+      "epoch": 1.2691947166595654,
+      "grad_norm": 22.875,
+      "learning_rate": 7.42653977892358e-07,
+      "loss": 0.5754,
+      "step": 14894
+    },
+    {
+      "epoch": 1.2692799318278654,
+      "grad_norm": 17.75,
+      "learning_rate": 7.42500410117242e-07,
+      "loss": 0.7299,
+      "step": 14895
+    },
+    {
+      "epoch": 1.2693651469961653,
+      "grad_norm": 15.125,
+      "learning_rate": 7.423468515131538e-07,
+      "loss": 0.6058,
+      "step": 14896
+    },
+    {
+      "epoch": 1.2694503621644653,
+      "grad_norm": 17.875,
+      "learning_rate": 7.421933020828684e-07,
+      "loss": 0.8696,
+      "step": 14897
+    },
+    {
+      "epoch": 1.2695355773327652,
+      "grad_norm": 15.75,
+      "learning_rate": 7.42039761829161e-07,
+      "loss": 0.6012,
+      "step": 14898
+    },
+    {
+      "epoch": 1.2696207925010652,
+      "grad_norm": 19.125,
+      "learning_rate": 7.418862307548055e-07,
+      "loss": 0.8615,
+      "step": 14899
+    },
+    {
+      "epoch": 1.2697060076693651,
+      "grad_norm": 14.625,
+      "learning_rate": 7.417327088625776e-07,
+      "loss": 0.624,
+      "step": 14900
+    },
+    {
+      "epoch": 1.269791222837665,
+      "grad_norm": 11.75,
+      "learning_rate": 7.4157919615525e-07,
+      "loss": 0.2909,
+      "step": 14901
+    },
+    {
+      "epoch": 1.269876438005965,
+      "grad_norm": 16.625,
+      "learning_rate": 7.414256926355978e-07,
+      "loss": 0.8275,
+      "step": 14902
+    },
+    {
+      "epoch": 1.269961653174265,
+      "grad_norm": 20.0,
+      "learning_rate": 7.412721983063944e-07,
+      "loss": 0.3769,
+      "step": 14903
+    },
+    {
+      "epoch": 1.270046868342565,
+      "grad_norm": 19.25,
+      "learning_rate": 7.411187131704143e-07,
+      "loss": 0.7938,
+      "step": 14904
+    },
+    {
+      "epoch": 1.270132083510865,
+      "grad_norm": 23.75,
+      "learning_rate": 7.409652372304306e-07,
+      "loss": 0.9351,
+      "step": 14905
+    },
+    {
+      "epoch": 1.2702172986791649,
+      "grad_norm": 16.875,
+      "learning_rate": 7.40811770489217e-07,
+      "loss": 0.5219,
+      "step": 14906
+    },
+    {
+      "epoch": 1.2703025138474648,
+      "grad_norm": 14.5625,
+      "learning_rate": 7.406583129495466e-07,
+      "loss": 0.6657,
+      "step": 14907
+    },
+    {
+      "epoch": 1.2703877290157648,
+      "grad_norm": 15.25,
+      "learning_rate": 7.405048646141932e-07,
+      "loss": 0.4659,
+      "step": 14908
+    },
+    {
+      "epoch": 1.2704729441840648,
+      "grad_norm": 16.125,
+      "learning_rate": 7.40351425485929e-07,
+      "loss": 0.5535,
+      "step": 14909
+    },
+    {
+      "epoch": 1.2705581593523647,
+      "grad_norm": 21.625,
+      "learning_rate": 7.401979955675267e-07,
+      "loss": 0.8114,
+      "step": 14910
+    },
+    {
+      "epoch": 1.2706433745206647,
+      "grad_norm": 13.25,
+      "learning_rate": 7.400445748617595e-07,
+      "loss": 0.6602,
+      "step": 14911
+    },
+    {
+      "epoch": 1.2707285896889646,
+      "grad_norm": 16.125,
+      "learning_rate": 7.398911633713994e-07,
+      "loss": 0.6098,
+      "step": 14912
+    },
+    {
+      "epoch": 1.2708138048572646,
+      "grad_norm": 20.0,
+      "learning_rate": 7.397377610992193e-07,
+      "loss": 1.0105,
+      "step": 14913
+    },
+    {
+      "epoch": 1.2708990200255645,
+      "grad_norm": 16.5,
+      "learning_rate": 7.395843680479906e-07,
+      "loss": 0.5999,
+      "step": 14914
+    },
+    {
+      "epoch": 1.2709842351938645,
+      "grad_norm": 24.0,
+      "learning_rate": 7.394309842204858e-07,
+      "loss": 0.8848,
+      "step": 14915
+    },
+    {
+      "epoch": 1.2710694503621645,
+      "grad_norm": 13.125,
+      "learning_rate": 7.392776096194769e-07,
+      "loss": 0.4414,
+      "step": 14916
+    },
+    {
+      "epoch": 1.2711546655304644,
+      "grad_norm": 25.0,
+      "learning_rate": 7.391242442477349e-07,
+      "loss": 0.6606,
+      "step": 14917
+    },
+    {
+      "epoch": 1.2712398806987644,
+      "grad_norm": 18.125,
+      "learning_rate": 7.389708881080312e-07,
+      "loss": 0.5772,
+      "step": 14918
+    },
+    {
+      "epoch": 1.2713250958670643,
+      "grad_norm": 14.375,
+      "learning_rate": 7.388175412031379e-07,
+      "loss": 0.4497,
+      "step": 14919
+    },
+    {
+      "epoch": 1.2714103110353643,
+      "grad_norm": 19.125,
+      "learning_rate": 7.386642035358252e-07,
+      "loss": 0.4895,
+      "step": 14920
+    },
+    {
+      "epoch": 1.2714955262036642,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.385108751088649e-07,
+      "loss": 0.5445,
+      "step": 14921
+    },
+    {
+      "epoch": 1.2715807413719642,
+      "grad_norm": 14.0,
+      "learning_rate": 7.383575559250271e-07,
+      "loss": 0.4578,
+      "step": 14922
+    },
+    {
+      "epoch": 1.2716659565402642,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.382042459870831e-07,
+      "loss": 0.3584,
+      "step": 14923
+    },
+    {
+      "epoch": 1.271751171708564,
+      "grad_norm": 13.25,
+      "learning_rate": 7.380509452978032e-07,
+      "loss": 0.5988,
+      "step": 14924
+    },
+    {
+      "epoch": 1.271836386876864,
+      "grad_norm": 17.25,
+      "learning_rate": 7.378976538599569e-07,
+      "loss": 0.6437,
+      "step": 14925
+    },
+    {
+      "epoch": 1.271921602045164,
+      "grad_norm": 14.25,
+      "learning_rate": 7.377443716763155e-07,
+      "loss": 0.4798,
+      "step": 14926
+    },
+    {
+      "epoch": 1.272006817213464,
+      "grad_norm": 6.5,
+      "learning_rate": 7.375910987496479e-07,
+      "loss": 0.107,
+      "step": 14927
+    },
+    {
+      "epoch": 1.272092032381764,
+      "grad_norm": 19.875,
+      "learning_rate": 7.374378350827248e-07,
+      "loss": 0.9183,
+      "step": 14928
+    },
+    {
+      "epoch": 1.272177247550064,
+      "grad_norm": 19.75,
+      "learning_rate": 7.372845806783152e-07,
+      "loss": 0.8643,
+      "step": 14929
+    },
+    {
+      "epoch": 1.2722624627183639,
+      "grad_norm": 13.625,
+      "learning_rate": 7.37131335539189e-07,
+      "loss": 0.5338,
+      "step": 14930
+    },
+    {
+      "epoch": 1.2723476778866638,
+      "grad_norm": 14.6875,
+      "learning_rate": 7.369780996681153e-07,
+      "loss": 0.3288,
+      "step": 14931
+    },
+    {
+      "epoch": 1.2724328930549638,
+      "grad_norm": 10.75,
+      "learning_rate": 7.368248730678634e-07,
+      "loss": 0.3136,
+      "step": 14932
+    },
+    {
+      "epoch": 1.2725181082232637,
+      "grad_norm": 19.75,
+      "learning_rate": 7.366716557412019e-07,
+      "loss": 0.7263,
+      "step": 14933
+    },
+    {
+      "epoch": 1.2726033233915637,
+      "grad_norm": 17.5,
+      "learning_rate": 7.365184476908995e-07,
+      "loss": 0.9171,
+      "step": 14934
+    },
+    {
+      "epoch": 1.2726885385598636,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.363652489197252e-07,
+      "loss": 0.8063,
+      "step": 14935
+    },
+    {
+      "epoch": 1.2727737537281636,
+      "grad_norm": 21.25,
+      "learning_rate": 7.362120594304476e-07,
+      "loss": 0.5757,
+      "step": 14936
+    },
+    {
+      "epoch": 1.2728589688964635,
+      "grad_norm": 19.875,
+      "learning_rate": 7.360588792258344e-07,
+      "loss": 0.9526,
+      "step": 14937
+    },
+    {
+      "epoch": 1.2729441840647635,
+      "grad_norm": 27.875,
+      "learning_rate": 7.359057083086543e-07,
+      "loss": 1.1087,
+      "step": 14938
+    },
+    {
+      "epoch": 1.2730293992330635,
+      "grad_norm": 9.875,
+      "learning_rate": 7.357525466816748e-07,
+      "loss": 0.2731,
+      "step": 14939
+    },
+    {
+      "epoch": 1.2731146144013634,
+      "grad_norm": 38.75,
+      "learning_rate": 7.355993943476644e-07,
+      "loss": 0.7923,
+      "step": 14940
+    },
+    {
+      "epoch": 1.2731998295696634,
+      "grad_norm": 26.125,
+      "learning_rate": 7.3544625130939e-07,
+      "loss": 0.7759,
+      "step": 14941
+    },
+    {
+      "epoch": 1.2732850447379633,
+      "grad_norm": 17.0,
+      "learning_rate": 7.352931175696191e-07,
+      "loss": 0.5691,
+      "step": 14942
+    },
+    {
+      "epoch": 1.2733702599062633,
+      "grad_norm": 17.125,
+      "learning_rate": 7.351399931311191e-07,
+      "loss": 0.7022,
+      "step": 14943
+    },
+    {
+      "epoch": 1.2734554750745632,
+      "grad_norm": 15.125,
+      "learning_rate": 7.349868779966571e-07,
+      "loss": 0.6164,
+      "step": 14944
+    },
+    {
+      "epoch": 1.2735406902428632,
+      "grad_norm": 10.875,
+      "learning_rate": 7.348337721690004e-07,
+      "loss": 0.2313,
+      "step": 14945
+    },
+    {
+      "epoch": 1.2736259054111632,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.346806756509153e-07,
+      "loss": 0.5515,
+      "step": 14946
+    },
+    {
+      "epoch": 1.2737111205794631,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.345275884451686e-07,
+      "loss": 0.466,
+      "step": 14947
+    },
+    {
+      "epoch": 1.273796335747763,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.343745105545269e-07,
+      "loss": 0.7773,
+      "step": 14948
+    },
+    {
+      "epoch": 1.273881550916063,
+      "grad_norm": 16.375,
+      "learning_rate": 7.342214419817561e-07,
+      "loss": 0.7301,
+      "step": 14949
+    },
+    {
+      "epoch": 1.273966766084363,
+      "grad_norm": 10.6875,
+      "learning_rate": 7.340683827296224e-07,
+      "loss": 0.2507,
+      "step": 14950
+    },
+    {
+      "epoch": 1.274051981252663,
+      "grad_norm": 18.375,
+      "learning_rate": 7.339153328008922e-07,
+      "loss": 0.4547,
+      "step": 14951
+    },
+    {
+      "epoch": 1.274137196420963,
+      "grad_norm": 15.125,
+      "learning_rate": 7.337622921983303e-07,
+      "loss": 0.5418,
+      "step": 14952
+    },
+    {
+      "epoch": 1.2742224115892629,
+      "grad_norm": 16.375,
+      "learning_rate": 7.336092609247034e-07,
+      "loss": 0.3821,
+      "step": 14953
+    },
+    {
+      "epoch": 1.2743076267575628,
+      "grad_norm": 15.125,
+      "learning_rate": 7.334562389827762e-07,
+      "loss": 0.5002,
+      "step": 14954
+    },
+    {
+      "epoch": 1.2743928419258628,
+      "grad_norm": 18.125,
+      "learning_rate": 7.333032263753142e-07,
+      "loss": 0.7862,
+      "step": 14955
+    },
+    {
+      "epoch": 1.2744780570941627,
+      "grad_norm": 15.125,
+      "learning_rate": 7.33150223105083e-07,
+      "loss": 0.7382,
+      "step": 14956
+    },
+    {
+      "epoch": 1.2745632722624627,
+      "grad_norm": 16.25,
+      "learning_rate": 7.329972291748464e-07,
+      "loss": 0.3691,
+      "step": 14957
+    },
+    {
+      "epoch": 1.2746484874307626,
+      "grad_norm": 18.125,
+      "learning_rate": 7.328442445873699e-07,
+      "loss": 0.555,
+      "step": 14958
+    },
+    {
+      "epoch": 1.2747337025990626,
+      "grad_norm": 22.875,
+      "learning_rate": 7.326912693454177e-07,
+      "loss": 0.9039,
+      "step": 14959
+    },
+    {
+      "epoch": 1.2748189177673626,
+      "grad_norm": 16.375,
+      "learning_rate": 7.325383034517549e-07,
+      "loss": 0.4336,
+      "step": 14960
+    },
+    {
+      "epoch": 1.2749041329356625,
+      "grad_norm": 15.0,
+      "learning_rate": 7.323853469091452e-07,
+      "loss": 0.6374,
+      "step": 14961
+    },
+    {
+      "epoch": 1.2749893481039625,
+      "grad_norm": 18.0,
+      "learning_rate": 7.322323997203525e-07,
+      "loss": 0.8151,
+      "step": 14962
+    },
+    {
+      "epoch": 1.2750745632722624,
+      "grad_norm": 18.75,
+      "learning_rate": 7.320794618881412e-07,
+      "loss": 0.4047,
+      "step": 14963
+    },
+    {
+      "epoch": 1.2751597784405624,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.319265334152748e-07,
+      "loss": 0.4835,
+      "step": 14964
+    },
+    {
+      "epoch": 1.2752449936088623,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.317736143045171e-07,
+      "loss": 0.5499,
+      "step": 14965
+    },
+    {
+      "epoch": 1.2753302087771623,
+      "grad_norm": 14.5,
+      "learning_rate": 7.316207045586313e-07,
+      "loss": 0.7171,
+      "step": 14966
+    },
+    {
+      "epoch": 1.2754154239454623,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.314678041803802e-07,
+      "loss": 0.4786,
+      "step": 14967
+    },
+    {
+      "epoch": 1.2755006391137622,
+      "grad_norm": 15.875,
+      "learning_rate": 7.313149131725277e-07,
+      "loss": 0.5722,
+      "step": 14968
+    },
+    {
+      "epoch": 1.2755858542820622,
+      "grad_norm": 16.125,
+      "learning_rate": 7.31162031537836e-07,
+      "loss": 0.6202,
+      "step": 14969
+    },
+    {
+      "epoch": 1.2756710694503621,
+      "grad_norm": 16.625,
+      "learning_rate": 7.310091592790683e-07,
+      "loss": 0.5454,
+      "step": 14970
+    },
+    {
+      "epoch": 1.275756284618662,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.308562963989869e-07,
+      "loss": 0.4022,
+      "step": 14971
+    },
+    {
+      "epoch": 1.275841499786962,
+      "grad_norm": 16.375,
+      "learning_rate": 7.307034429003542e-07,
+      "loss": 0.4146,
+      "step": 14972
+    },
+    {
+      "epoch": 1.275926714955262,
+      "grad_norm": 15.625,
+      "learning_rate": 7.30550598785933e-07,
+      "loss": 0.4525,
+      "step": 14973
+    },
+    {
+      "epoch": 1.276011930123562,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.303977640584842e-07,
+      "loss": 0.5165,
+      "step": 14974
+    },
+    {
+      "epoch": 1.276097145291862,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.302449387207707e-07,
+      "loss": 0.8541,
+      "step": 14975
+    },
+    {
+      "epoch": 1.2761823604601619,
+      "grad_norm": 19.75,
+      "learning_rate": 7.300921227755537e-07,
+      "loss": 0.2242,
+      "step": 14976
+    },
+    {
+      "epoch": 1.2762675756284618,
+      "grad_norm": 13.125,
+      "learning_rate": 7.299393162255946e-07,
+      "loss": 0.5531,
+      "step": 14977
+    },
+    {
+      "epoch": 1.2763527907967618,
+      "grad_norm": 15.75,
+      "learning_rate": 7.297865190736552e-07,
+      "loss": 0.4541,
+      "step": 14978
+    },
+    {
+      "epoch": 1.2764380059650617,
+      "grad_norm": 21.625,
+      "learning_rate": 7.296337313224965e-07,
+      "loss": 0.7248,
+      "step": 14979
+    },
+    {
+      "epoch": 1.2765232211333617,
+      "grad_norm": 16.125,
+      "learning_rate": 7.294809529748794e-07,
+      "loss": 0.5078,
+      "step": 14980
+    },
+    {
+      "epoch": 1.2766084363016617,
+      "grad_norm": 22.125,
+      "learning_rate": 7.293281840335653e-07,
+      "loss": 0.5162,
+      "step": 14981
+    },
+    {
+      "epoch": 1.2766936514699616,
+      "grad_norm": 12.0,
+      "learning_rate": 7.291754245013147e-07,
+      "loss": 0.3598,
+      "step": 14982
+    },
+    {
+      "epoch": 1.2767788666382616,
+      "grad_norm": 22.125,
+      "learning_rate": 7.290226743808873e-07,
+      "loss": 1.011,
+      "step": 14983
+    },
+    {
+      "epoch": 1.2768640818065615,
+      "grad_norm": 15.875,
+      "learning_rate": 7.288699336750441e-07,
+      "loss": 0.7183,
+      "step": 14984
+    },
+    {
+      "epoch": 1.2769492969748615,
+      "grad_norm": 33.25,
+      "learning_rate": 7.287172023865452e-07,
+      "loss": 0.7252,
+      "step": 14985
+    },
+    {
+      "epoch": 1.2770345121431614,
+      "grad_norm": 14.875,
+      "learning_rate": 7.28564480518151e-07,
+      "loss": 0.7955,
+      "step": 14986
+    },
+    {
+      "epoch": 1.2771197273114614,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.284117680726208e-07,
+      "loss": 0.7062,
+      "step": 14987
+    },
+    {
+      "epoch": 1.2772049424797614,
+      "grad_norm": 10.375,
+      "learning_rate": 7.282590650527141e-07,
+      "loss": 0.1982,
+      "step": 14988
+    },
+    {
+      "epoch": 1.2772901576480613,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.281063714611914e-07,
+      "loss": 0.5037,
+      "step": 14989
+    },
+    {
+      "epoch": 1.2773753728163613,
+      "grad_norm": 21.375,
+      "learning_rate": 7.27953687300811e-07,
+      "loss": 0.5884,
+      "step": 14990
+    },
+    {
+      "epoch": 1.2774605879846612,
+      "grad_norm": 14.25,
+      "learning_rate": 7.278010125743327e-07,
+      "loss": 0.55,
+      "step": 14991
+    },
+    {
+      "epoch": 1.2775458031529612,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.276483472845148e-07,
+      "loss": 0.6246,
+      "step": 14992
+    },
+    {
+      "epoch": 1.2776310183212611,
+      "grad_norm": 15.25,
+      "learning_rate": 7.274956914341166e-07,
+      "loss": 0.8062,
+      "step": 14993
+    },
+    {
+      "epoch": 1.277716233489561,
+      "grad_norm": 14.875,
+      "learning_rate": 7.273430450258973e-07,
+      "loss": 0.6742,
+      "step": 14994
+    },
+    {
+      "epoch": 1.277801448657861,
+      "grad_norm": 19.25,
+      "learning_rate": 7.27190408062614e-07,
+      "loss": 0.7489,
+      "step": 14995
+    },
+    {
+      "epoch": 1.277886663826161,
+      "grad_norm": 18.125,
+      "learning_rate": 7.270377805470258e-07,
+      "loss": 0.7734,
+      "step": 14996
+    },
+    {
+      "epoch": 1.277971878994461,
+      "grad_norm": 15.625,
+      "learning_rate": 7.268851624818913e-07,
+      "loss": 0.6351,
+      "step": 14997
+    },
+    {
+      "epoch": 1.278057094162761,
+      "grad_norm": 11.6875,
+      "learning_rate": 7.267325538699674e-07,
+      "loss": 0.3188,
+      "step": 14998
+    },
+    {
+      "epoch": 1.2781423093310609,
+      "grad_norm": 13.25,
+      "learning_rate": 7.265799547140132e-07,
+      "loss": 0.5073,
+      "step": 14999
+    },
+    {
+      "epoch": 1.2782275244993608,
+      "grad_norm": 20.25,
+      "learning_rate": 7.264273650167847e-07,
+      "loss": 0.6926,
+      "step": 15000
+    },
+    {
+      "epoch": 1.2783127396676608,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.262747847810404e-07,
+      "loss": 0.5763,
+      "step": 15001
+    },
+    {
+      "epoch": 1.2783979548359607,
+      "grad_norm": 18.375,
+      "learning_rate": 7.261222140095374e-07,
+      "loss": 0.6478,
+      "step": 15002
+    },
+    {
+      "epoch": 1.2784831700042607,
+      "grad_norm": 19.375,
+      "learning_rate": 7.259696527050332e-07,
+      "loss": 0.7442,
+      "step": 15003
+    },
+    {
+      "epoch": 1.2785683851725607,
+      "grad_norm": 15.875,
+      "learning_rate": 7.258171008702839e-07,
+      "loss": 0.9312,
+      "step": 15004
+    },
+    {
+      "epoch": 1.2786536003408606,
+      "grad_norm": 19.75,
+      "learning_rate": 7.256645585080472e-07,
+      "loss": 0.966,
+      "step": 15005
+    },
+    {
+      "epoch": 1.2787388155091606,
+      "grad_norm": 16.125,
+      "learning_rate": 7.255120256210786e-07,
+      "loss": 0.5236,
+      "step": 15006
+    },
+    {
+      "epoch": 1.2788240306774605,
+      "grad_norm": 14.5625,
+      "learning_rate": 7.253595022121357e-07,
+      "loss": 0.5449,
+      "step": 15007
+    },
+    {
+      "epoch": 1.2789092458457605,
+      "grad_norm": 10.5625,
+      "learning_rate": 7.252069882839735e-07,
+      "loss": 0.3431,
+      "step": 15008
+    },
+    {
+      "epoch": 1.2789944610140604,
+      "grad_norm": 15.625,
+      "learning_rate": 7.250544838393489e-07,
+      "loss": 0.5911,
+      "step": 15009
+    },
+    {
+      "epoch": 1.2790796761823604,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.249019888810177e-07,
+      "loss": 0.357,
+      "step": 15010
+    },
+    {
+      "epoch": 1.2791648913506604,
+      "grad_norm": 16.875,
+      "learning_rate": 7.247495034117361e-07,
+      "loss": 0.7202,
+      "step": 15011
+    },
+    {
+      "epoch": 1.2792501065189603,
+      "grad_norm": 18.875,
+      "learning_rate": 7.245970274342584e-07,
+      "loss": 0.4696,
+      "step": 15012
+    },
+    {
+      "epoch": 1.2793353216872603,
+      "grad_norm": 10.75,
+      "learning_rate": 7.244445609513414e-07,
+      "loss": 0.2505,
+      "step": 15013
+    },
+    {
+      "epoch": 1.2794205368555602,
+      "grad_norm": 15.6875,
+      "learning_rate": 7.242921039657393e-07,
+      "loss": 0.4527,
+      "step": 15014
+    },
+    {
+      "epoch": 1.2795057520238602,
+      "grad_norm": 19.625,
+      "learning_rate": 7.241396564802074e-07,
+      "loss": 0.6797,
+      "step": 15015
+    },
+    {
+      "epoch": 1.2795909671921601,
+      "grad_norm": 38.5,
+      "learning_rate": 7.239872184975011e-07,
+      "loss": 0.7694,
+      "step": 15016
+    },
+    {
+      "epoch": 1.27967618236046,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.238347900203744e-07,
+      "loss": 0.625,
+      "step": 15017
+    },
+    {
+      "epoch": 1.27976139752876,
+      "grad_norm": 18.0,
+      "learning_rate": 7.236823710515819e-07,
+      "loss": 0.5112,
+      "step": 15018
+    },
+    {
+      "epoch": 1.27984661269706,
+      "grad_norm": 20.25,
+      "learning_rate": 7.235299615938785e-07,
+      "loss": 0.6654,
+      "step": 15019
+    },
+    {
+      "epoch": 1.27993182786536,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.233775616500185e-07,
+      "loss": 0.4417,
+      "step": 15020
+    },
+    {
+      "epoch": 1.28001704303366,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.232251712227554e-07,
+      "loss": 0.5317,
+      "step": 15021
+    },
+    {
+      "epoch": 1.2801022582019599,
+      "grad_norm": 17.75,
+      "learning_rate": 7.230727903148428e-07,
+      "loss": 0.8129,
+      "step": 15022
+    },
+    {
+      "epoch": 1.2801874733702598,
+      "grad_norm": 19.375,
+      "learning_rate": 7.229204189290347e-07,
+      "loss": 0.5367,
+      "step": 15023
+    },
+    {
+      "epoch": 1.2802726885385598,
+      "grad_norm": 12.375,
+      "learning_rate": 7.227680570680851e-07,
+      "loss": 0.3759,
+      "step": 15024
+    },
+    {
+      "epoch": 1.2803579037068598,
+      "grad_norm": 20.5,
+      "learning_rate": 7.226157047347465e-07,
+      "loss": 0.6855,
+      "step": 15025
+    },
+    {
+      "epoch": 1.2804431188751597,
+      "grad_norm": 13.125,
+      "learning_rate": 7.224633619317723e-07,
+      "loss": 0.7443,
+      "step": 15026
+    },
+    {
+      "epoch": 1.2805283340434597,
+      "grad_norm": 15.0,
+      "learning_rate": 7.223110286619157e-07,
+      "loss": 0.4937,
+      "step": 15027
+    },
+    {
+      "epoch": 1.2806135492117596,
+      "grad_norm": 13.375,
+      "learning_rate": 7.2215870492793e-07,
+      "loss": 0.5423,
+      "step": 15028
+    },
+    {
+      "epoch": 1.2806987643800596,
+      "grad_norm": 18.5,
+      "learning_rate": 7.220063907325666e-07,
+      "loss": 0.8018,
+      "step": 15029
+    },
+    {
+      "epoch": 1.2807839795483595,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.218540860785795e-07,
+      "loss": 0.6262,
+      "step": 15030
+    },
+    {
+      "epoch": 1.2808691947166595,
+      "grad_norm": 24.375,
+      "learning_rate": 7.217017909687193e-07,
+      "loss": 0.5389,
+      "step": 15031
+    },
+    {
+      "epoch": 1.2809544098849595,
+      "grad_norm": 18.375,
+      "learning_rate": 7.215495054057392e-07,
+      "loss": 0.6472,
+      "step": 15032
+    },
+    {
+      "epoch": 1.2810396250532594,
+      "grad_norm": 22.625,
+      "learning_rate": 7.213972293923913e-07,
+      "loss": 0.8111,
+      "step": 15033
+    },
+    {
+      "epoch": 1.2811248402215594,
+      "grad_norm": 21.5,
+      "learning_rate": 7.212449629314267e-07,
+      "loss": 0.8046,
+      "step": 15034
+    },
+    {
+      "epoch": 1.2812100553898593,
+      "grad_norm": 14.0,
+      "learning_rate": 7.210927060255971e-07,
+      "loss": 0.3659,
+      "step": 15035
+    },
+    {
+      "epoch": 1.2812952705581593,
+      "grad_norm": 9.375,
+      "learning_rate": 7.209404586776543e-07,
+      "loss": 0.2725,
+      "step": 15036
+    },
+    {
+      "epoch": 1.2813804857264592,
+      "grad_norm": 24.5,
+      "learning_rate": 7.2078822089035e-07,
+      "loss": 0.6344,
+      "step": 15037
+    },
+    {
+      "epoch": 1.2814657008947592,
+      "grad_norm": 23.5,
+      "learning_rate": 7.206359926664345e-07,
+      "loss": 1.3018,
+      "step": 15038
+    },
+    {
+      "epoch": 1.2815509160630592,
+      "grad_norm": 14.875,
+      "learning_rate": 7.204837740086584e-07,
+      "loss": 0.6955,
+      "step": 15039
+    },
+    {
+      "epoch": 1.281636131231359,
+      "grad_norm": 16.375,
+      "learning_rate": 7.203315649197731e-07,
+      "loss": 0.919,
+      "step": 15040
+    },
+    {
+      "epoch": 1.281721346399659,
+      "grad_norm": 21.75,
+      "learning_rate": 7.201793654025297e-07,
+      "loss": 0.8063,
+      "step": 15041
+    },
+    {
+      "epoch": 1.281806561567959,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.200271754596771e-07,
+      "loss": 0.5939,
+      "step": 15042
+    },
+    {
+      "epoch": 1.281891776736259,
+      "grad_norm": 14.125,
+      "learning_rate": 7.198749950939667e-07,
+      "loss": 0.6959,
+      "step": 15043
+    },
+    {
+      "epoch": 1.281976991904559,
+      "grad_norm": 16.625,
+      "learning_rate": 7.19722824308148e-07,
+      "loss": 0.4886,
+      "step": 15044
+    },
+    {
+      "epoch": 1.282062207072859,
+      "grad_norm": 16.125,
+      "learning_rate": 7.195706631049718e-07,
+      "loss": 0.6447,
+      "step": 15045
+    },
+    {
+      "epoch": 1.2821474222411589,
+      "grad_norm": 29.25,
+      "learning_rate": 7.194185114871868e-07,
+      "loss": 0.6003,
+      "step": 15046
+    },
+    {
+      "epoch": 1.2822326374094588,
+      "grad_norm": 14.625,
+      "learning_rate": 7.192663694575428e-07,
+      "loss": 0.7729,
+      "step": 15047
+    },
+    {
+      "epoch": 1.2823178525777588,
+      "grad_norm": 19.0,
+      "learning_rate": 7.191142370187887e-07,
+      "loss": 0.5621,
+      "step": 15048
+    },
+    {
+      "epoch": 1.2824030677460587,
+      "grad_norm": 20.375,
+      "learning_rate": 7.189621141736744e-07,
+      "loss": 0.6954,
+      "step": 15049
+    },
+    {
+      "epoch": 1.2824882829143587,
+      "grad_norm": 20.0,
+      "learning_rate": 7.188100009249494e-07,
+      "loss": 0.5112,
+      "step": 15050
+    },
+    {
+      "epoch": 1.2825734980826586,
+      "grad_norm": 10.75,
+      "learning_rate": 7.18657897275361e-07,
+      "loss": 0.2816,
+      "step": 15051
+    },
+    {
+      "epoch": 1.2826587132509586,
+      "grad_norm": 17.375,
+      "learning_rate": 7.185058032276589e-07,
+      "loss": 0.7607,
+      "step": 15052
+    },
+    {
+      "epoch": 1.2827439284192586,
+      "grad_norm": 11.3125,
+      "learning_rate": 7.183537187845913e-07,
+      "loss": 0.3999,
+      "step": 15053
+    },
+    {
+      "epoch": 1.2828291435875585,
+      "grad_norm": 15.75,
+      "learning_rate": 7.182016439489076e-07,
+      "loss": 0.4299,
+      "step": 15054
+    },
+    {
+      "epoch": 1.2829143587558585,
+      "grad_norm": 12.25,
+      "learning_rate": 7.180495787233541e-07,
+      "loss": 0.3483,
+      "step": 15055
+    },
+    {
+      "epoch": 1.2829995739241584,
+      "grad_norm": 30.75,
+      "learning_rate": 7.178975231106794e-07,
+      "loss": 0.7989,
+      "step": 15056
+    },
+    {
+      "epoch": 1.2830847890924584,
+      "grad_norm": 19.875,
+      "learning_rate": 7.177454771136316e-07,
+      "loss": 0.6729,
+      "step": 15057
+    },
+    {
+      "epoch": 1.2831700042607583,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.175934407349589e-07,
+      "loss": 0.5917,
+      "step": 15058
+    },
+    {
+      "epoch": 1.2832552194290583,
+      "grad_norm": 14.625,
+      "learning_rate": 7.174414139774075e-07,
+      "loss": 0.615,
+      "step": 15059
+    },
+    {
+      "epoch": 1.2833404345973582,
+      "grad_norm": 21.0,
+      "learning_rate": 7.172893968437252e-07,
+      "loss": 0.8324,
+      "step": 15060
+    },
+    {
+      "epoch": 1.2834256497656582,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.171373893366594e-07,
+      "loss": 0.7575,
+      "step": 15061
+    },
+    {
+      "epoch": 1.2835108649339582,
+      "grad_norm": 23.5,
+      "learning_rate": 7.169853914589572e-07,
+      "loss": 0.9775,
+      "step": 15062
+    },
+    {
+      "epoch": 1.2835960801022581,
+      "grad_norm": 16.375,
+      "learning_rate": 7.168334032133651e-07,
+      "loss": 0.7289,
+      "step": 15063
+    },
+    {
+      "epoch": 1.2836812952705583,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.166814246026288e-07,
+      "loss": 0.6016,
+      "step": 15064
+    },
+    {
+      "epoch": 1.2837665104388583,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.165294556294955e-07,
+      "loss": 0.6615,
+      "step": 15065
+    },
+    {
+      "epoch": 1.2838517256071582,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.163774962967114e-07,
+      "loss": 0.4791,
+      "step": 15066
+    },
+    {
+      "epoch": 1.2839369407754582,
+      "grad_norm": 20.0,
+      "learning_rate": 7.162255466070231e-07,
+      "loss": 1.0902,
+      "step": 15067
+    },
+    {
+      "epoch": 1.2840221559437581,
+      "grad_norm": 18.25,
+      "learning_rate": 7.160736065631755e-07,
+      "loss": 0.8088,
+      "step": 15068
+    },
+    {
+      "epoch": 1.284107371112058,
+      "grad_norm": 10.375,
+      "learning_rate": 7.159216761679144e-07,
+      "loss": 0.4919,
+      "step": 15069
+    },
+    {
+      "epoch": 1.284192586280358,
+      "grad_norm": 13.875,
+      "learning_rate": 7.157697554239864e-07,
+      "loss": 0.5626,
+      "step": 15070
+    },
+    {
+      "epoch": 1.284277801448658,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.156178443341362e-07,
+      "loss": 0.5965,
+      "step": 15071
+    },
+    {
+      "epoch": 1.284363016616958,
+      "grad_norm": 13.375,
+      "learning_rate": 7.154659429011084e-07,
+      "loss": 0.4038,
+      "step": 15072
+    },
+    {
+      "epoch": 1.284448231785258,
+      "grad_norm": 14.5,
+      "learning_rate": 7.153140511276485e-07,
+      "loss": 0.6984,
+      "step": 15073
+    },
+    {
+      "epoch": 1.2845334469535579,
+      "grad_norm": 16.375,
+      "learning_rate": 7.151621690165013e-07,
+      "loss": 0.5937,
+      "step": 15074
+    },
+    {
+      "epoch": 1.2846186621218578,
+      "grad_norm": 13.1875,
+      "learning_rate": 7.15010296570412e-07,
+      "loss": 0.6077,
+      "step": 15075
+    },
+    {
+      "epoch": 1.2847038772901578,
+      "grad_norm": 18.5,
+      "learning_rate": 7.148584337921242e-07,
+      "loss": 0.6863,
+      "step": 15076
+    },
+    {
+      "epoch": 1.2847890924584577,
+      "grad_norm": 17.0,
+      "learning_rate": 7.147065806843826e-07,
+      "loss": 0.7406,
+      "step": 15077
+    },
+    {
+      "epoch": 1.2848743076267577,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.145547372499319e-07,
+      "loss": 0.5194,
+      "step": 15078
+    },
+    {
+      "epoch": 1.2849595227950577,
+      "grad_norm": 18.75,
+      "learning_rate": 7.144029034915149e-07,
+      "loss": 0.5087,
+      "step": 15079
+    },
+    {
+      "epoch": 1.2850447379633576,
+      "grad_norm": 12.1875,
+      "learning_rate": 7.142510794118767e-07,
+      "loss": 0.507,
+      "step": 15080
+    },
+    {
+      "epoch": 1.2851299531316576,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.140992650137596e-07,
+      "loss": 0.6032,
+      "step": 15081
+    },
+    {
+      "epoch": 1.2852151682999575,
+      "grad_norm": 13.75,
+      "learning_rate": 7.139474602999077e-07,
+      "loss": 0.4407,
+      "step": 15082
+    },
+    {
+      "epoch": 1.2853003834682575,
+      "grad_norm": 17.375,
+      "learning_rate": 7.137956652730644e-07,
+      "loss": 0.5077,
+      "step": 15083
+    },
+    {
+      "epoch": 1.2853855986365574,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.13643879935973e-07,
+      "loss": 0.5777,
+      "step": 15084
+    },
+    {
+      "epoch": 1.2854708138048574,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.134921042913753e-07,
+      "loss": 0.4061,
+      "step": 15085
+    },
+    {
+      "epoch": 1.2855560289731573,
+      "grad_norm": 14.625,
+      "learning_rate": 7.133403383420156e-07,
+      "loss": 0.5027,
+      "step": 15086
+    },
+    {
+      "epoch": 1.2856412441414573,
+      "grad_norm": 13.1875,
+      "learning_rate": 7.131885820906351e-07,
+      "loss": 0.3994,
+      "step": 15087
+    },
+    {
+      "epoch": 1.2857264593097573,
+      "grad_norm": 15.25,
+      "learning_rate": 7.13036835539977e-07,
+      "loss": 0.6309,
+      "step": 15088
+    },
+    {
+      "epoch": 1.2858116744780572,
+      "grad_norm": 14.75,
+      "learning_rate": 7.12885098692783e-07,
+      "loss": 0.3851,
+      "step": 15089
+    },
+    {
+      "epoch": 1.2858968896463572,
+      "grad_norm": 11.9375,
+      "learning_rate": 7.127333715517954e-07,
+      "loss": 0.3319,
+      "step": 15090
+    },
+    {
+      "epoch": 1.2859821048146571,
+      "grad_norm": 22.875,
+      "learning_rate": 7.125816541197559e-07,
+      "loss": 0.6869,
+      "step": 15091
+    },
+    {
+      "epoch": 1.286067319982957,
+      "grad_norm": 16.625,
+      "learning_rate": 7.124299463994067e-07,
+      "loss": 0.4867,
+      "step": 15092
+    },
+    {
+      "epoch": 1.286152535151257,
+      "grad_norm": 17.0,
+      "learning_rate": 7.122782483934885e-07,
+      "loss": 0.9977,
+      "step": 15093
+    },
+    {
+      "epoch": 1.286237750319557,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.121265601047436e-07,
+      "loss": 0.4308,
+      "step": 15094
+    },
+    {
+      "epoch": 1.286322965487857,
+      "grad_norm": 17.0,
+      "learning_rate": 7.119748815359123e-07,
+      "loss": 0.3849,
+      "step": 15095
+    },
+    {
+      "epoch": 1.286408180656157,
+      "grad_norm": 18.25,
+      "learning_rate": 7.118232126897358e-07,
+      "loss": 0.5573,
+      "step": 15096
+    },
+    {
+      "epoch": 1.2864933958244569,
+      "grad_norm": 10.9375,
+      "learning_rate": 7.116715535689556e-07,
+      "loss": 0.3496,
+      "step": 15097
+    },
+    {
+      "epoch": 1.2865786109927568,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.11519904176311e-07,
+      "loss": 0.5815,
+      "step": 15098
+    },
+    {
+      "epoch": 1.2866638261610568,
+      "grad_norm": 17.375,
+      "learning_rate": 7.113682645145436e-07,
+      "loss": 0.7165,
+      "step": 15099
+    },
+    {
+      "epoch": 1.2867490413293567,
+      "grad_norm": 13.875,
+      "learning_rate": 7.112166345863936e-07,
+      "loss": 0.5607,
+      "step": 15100
+    },
+    {
+      "epoch": 1.2868342564976567,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.110650143946001e-07,
+      "loss": 0.7267,
+      "step": 15101
+    },
+    {
+      "epoch": 1.2869194716659567,
+      "grad_norm": 16.625,
+      "learning_rate": 7.109134039419041e-07,
+      "loss": 0.7336,
+      "step": 15102
+    },
+    {
+      "epoch": 1.2870046868342566,
+      "grad_norm": 16.25,
+      "learning_rate": 7.107618032310453e-07,
+      "loss": 0.8601,
+      "step": 15103
+    },
+    {
+      "epoch": 1.2870899020025566,
+      "grad_norm": 16.375,
+      "learning_rate": 7.106102122647624e-07,
+      "loss": 0.5396,
+      "step": 15104
+    },
+    {
+      "epoch": 1.2871751171708565,
+      "grad_norm": 14.0625,
+      "learning_rate": 7.104586310457959e-07,
+      "loss": 0.4023,
+      "step": 15105
+    },
+    {
+      "epoch": 1.2872603323391565,
+      "grad_norm": 18.125,
+      "learning_rate": 7.103070595768841e-07,
+      "loss": 0.6298,
+      "step": 15106
+    },
+    {
+      "epoch": 1.2873455475074564,
+      "grad_norm": 10.1875,
+      "learning_rate": 7.101554978607663e-07,
+      "loss": 0.2957,
+      "step": 15107
+    },
+    {
+      "epoch": 1.2874307626757564,
+      "grad_norm": 15.625,
+      "learning_rate": 7.100039459001816e-07,
+      "loss": 0.6626,
+      "step": 15108
+    },
+    {
+      "epoch": 1.2875159778440564,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.098524036978691e-07,
+      "loss": 0.5311,
+      "step": 15109
+    },
+    {
+      "epoch": 1.2876011930123563,
+      "grad_norm": 13.9375,
+      "learning_rate": 7.097008712565662e-07,
+      "loss": 0.5253,
+      "step": 15110
+    },
+    {
+      "epoch": 1.2876864081806563,
+      "grad_norm": 13.25,
+      "learning_rate": 7.095493485790126e-07,
+      "loss": 0.338,
+      "step": 15111
+    },
+    {
+      "epoch": 1.2877716233489562,
+      "grad_norm": 18.0,
+      "learning_rate": 7.093978356679451e-07,
+      "loss": 0.7971,
+      "step": 15112
+    },
+    {
+      "epoch": 1.2878568385172562,
+      "grad_norm": 14.1875,
+      "learning_rate": 7.092463325261023e-07,
+      "loss": 0.5011,
+      "step": 15113
+    },
+    {
+      "epoch": 1.2879420536855561,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.090948391562225e-07,
+      "loss": 0.6159,
+      "step": 15114
+    },
+    {
+      "epoch": 1.288027268853856,
+      "grad_norm": 23.375,
+      "learning_rate": 7.089433555610426e-07,
+      "loss": 0.6046,
+      "step": 15115
+    },
+    {
+      "epoch": 1.288112484022156,
+      "grad_norm": 12.3125,
+      "learning_rate": 7.087918817433001e-07,
+      "loss": 0.5112,
+      "step": 15116
+    },
+    {
+      "epoch": 1.288197699190456,
+      "grad_norm": 18.125,
+      "learning_rate": 7.086404177057331e-07,
+      "loss": 0.287,
+      "step": 15117
+    },
+    {
+      "epoch": 1.288282914358756,
+      "grad_norm": 14.375,
+      "learning_rate": 7.084889634510776e-07,
+      "loss": 0.5479,
+      "step": 15118
+    },
+    {
+      "epoch": 1.288368129527056,
+      "grad_norm": 12.375,
+      "learning_rate": 7.083375189820717e-07,
+      "loss": 0.3994,
+      "step": 15119
+    },
+    {
+      "epoch": 1.2884533446953559,
+      "grad_norm": 17.625,
+      "learning_rate": 7.08186084301451e-07,
+      "loss": 0.5072,
+      "step": 15120
+    },
+    {
+      "epoch": 1.2885385598636558,
+      "grad_norm": 16.5,
+      "learning_rate": 7.080346594119524e-07,
+      "loss": 0.676,
+      "step": 15121
+    },
+    {
+      "epoch": 1.2886237750319558,
+      "grad_norm": 12.0,
+      "learning_rate": 7.07883244316313e-07,
+      "loss": 0.3134,
+      "step": 15122
+    },
+    {
+      "epoch": 1.2887089902002558,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.077318390172682e-07,
+      "loss": 0.5933,
+      "step": 15123
+    },
+    {
+      "epoch": 1.2887942053685557,
+      "grad_norm": 31.25,
+      "learning_rate": 7.075804435175543e-07,
+      "loss": 0.7032,
+      "step": 15124
+    },
+    {
+      "epoch": 1.2888794205368557,
+      "grad_norm": 20.0,
+      "learning_rate": 7.074290578199072e-07,
+      "loss": 0.6434,
+      "step": 15125
+    },
+    {
+      "epoch": 1.2889646357051556,
+      "grad_norm": 12.5625,
+      "learning_rate": 7.072776819270633e-07,
+      "loss": 0.4171,
+      "step": 15126
+    },
+    {
+      "epoch": 1.2890498508734556,
+      "grad_norm": 14.4375,
+      "learning_rate": 7.071263158417571e-07,
+      "loss": 0.3842,
+      "step": 15127
+    },
+    {
+      "epoch": 1.2891350660417555,
+      "grad_norm": 23.125,
+      "learning_rate": 7.06974959566724e-07,
+      "loss": 0.769,
+      "step": 15128
+    },
+    {
+      "epoch": 1.2892202812100555,
+      "grad_norm": 21.125,
+      "learning_rate": 7.068236131046994e-07,
+      "loss": 0.3633,
+      "step": 15129
+    },
+    {
+      "epoch": 1.2893054963783555,
+      "grad_norm": 11.5,
+      "learning_rate": 7.066722764584183e-07,
+      "loss": 0.2246,
+      "step": 15130
+    },
+    {
+      "epoch": 1.2893907115466554,
+      "grad_norm": 14.4375,
+      "learning_rate": 7.06520949630616e-07,
+      "loss": 0.4088,
+      "step": 15131
+    },
+    {
+      "epoch": 1.2894759267149554,
+      "grad_norm": 18.125,
+      "learning_rate": 7.063696326240262e-07,
+      "loss": 0.5391,
+      "step": 15132
+    },
+    {
+      "epoch": 1.2895611418832553,
+      "grad_norm": 13.5625,
+      "learning_rate": 7.062183254413838e-07,
+      "loss": 0.477,
+      "step": 15133
+    },
+    {
+      "epoch": 1.2896463570515553,
+      "grad_norm": 23.125,
+      "learning_rate": 7.060670280854234e-07,
+      "loss": 0.681,
+      "step": 15134
+    },
+    {
+      "epoch": 1.2897315722198552,
+      "grad_norm": 12.5,
+      "learning_rate": 7.05915740558879e-07,
+      "loss": 0.3855,
+      "step": 15135
+    },
+    {
+      "epoch": 1.2898167873881552,
+      "grad_norm": 18.125,
+      "learning_rate": 7.057644628644838e-07,
+      "loss": 0.8117,
+      "step": 15136
+    },
+    {
+      "epoch": 1.2899020025564552,
+      "grad_norm": 13.0,
+      "learning_rate": 7.056131950049718e-07,
+      "loss": 0.5801,
+      "step": 15137
+    },
+    {
+      "epoch": 1.289987217724755,
+      "grad_norm": 17.875,
+      "learning_rate": 7.05461936983077e-07,
+      "loss": 0.6438,
+      "step": 15138
+    },
+    {
+      "epoch": 1.290072432893055,
+      "grad_norm": 16.875,
+      "learning_rate": 7.053106888015329e-07,
+      "loss": 1.2043,
+      "step": 15139
+    },
+    {
+      "epoch": 1.290157648061355,
+      "grad_norm": 16.875,
+      "learning_rate": 7.051594504630721e-07,
+      "loss": 0.8353,
+      "step": 15140
+    },
+    {
+      "epoch": 1.290242863229655,
+      "grad_norm": 20.625,
+      "learning_rate": 7.050082219704276e-07,
+      "loss": 0.5865,
+      "step": 15141
+    },
+    {
+      "epoch": 1.290328078397955,
+      "grad_norm": 21.625,
+      "learning_rate": 7.04857003326333e-07,
+      "loss": 0.7834,
+      "step": 15142
+    },
+    {
+      "epoch": 1.290413293566255,
+      "grad_norm": 19.875,
+      "learning_rate": 7.047057945335206e-07,
+      "loss": 0.6894,
+      "step": 15143
+    },
+    {
+      "epoch": 1.2904985087345549,
+      "grad_norm": 13.5,
+      "learning_rate": 7.04554595594723e-07,
+      "loss": 0.2951,
+      "step": 15144
+    },
+    {
+      "epoch": 1.2905837239028548,
+      "grad_norm": 31.25,
+      "learning_rate": 7.04403406512672e-07,
+      "loss": 0.3598,
+      "step": 15145
+    },
+    {
+      "epoch": 1.2906689390711548,
+      "grad_norm": 21.625,
+      "learning_rate": 7.042522272900999e-07,
+      "loss": 0.6158,
+      "step": 15146
+    },
+    {
+      "epoch": 1.2907541542394547,
+      "grad_norm": 14.5,
+      "learning_rate": 7.041010579297394e-07,
+      "loss": 0.5333,
+      "step": 15147
+    },
+    {
+      "epoch": 1.2908393694077547,
+      "grad_norm": 20.875,
+      "learning_rate": 7.039498984343212e-07,
+      "loss": 0.694,
+      "step": 15148
+    },
+    {
+      "epoch": 1.2909245845760546,
+      "grad_norm": 18.75,
+      "learning_rate": 7.037987488065775e-07,
+      "loss": 0.7569,
+      "step": 15149
+    },
+    {
+      "epoch": 1.2910097997443546,
+      "grad_norm": 23.125,
+      "learning_rate": 7.036476090492395e-07,
+      "loss": 1.1767,
+      "step": 15150
+    },
+    {
+      "epoch": 1.2910950149126545,
+      "grad_norm": 17.0,
+      "learning_rate": 7.034964791650392e-07,
+      "loss": 0.5989,
+      "step": 15151
+    },
+    {
+      "epoch": 1.2911802300809545,
+      "grad_norm": 15.625,
+      "learning_rate": 7.033453591567071e-07,
+      "loss": 0.7496,
+      "step": 15152
+    },
+    {
+      "epoch": 1.2912654452492545,
+      "grad_norm": 13.0,
+      "learning_rate": 7.031942490269735e-07,
+      "loss": 0.57,
+      "step": 15153
+    },
+    {
+      "epoch": 1.2913506604175544,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.030431487785698e-07,
+      "loss": 0.719,
+      "step": 15154
+    },
+    {
+      "epoch": 1.2914358755858544,
+      "grad_norm": 14.875,
+      "learning_rate": 7.028920584142263e-07,
+      "loss": 0.502,
+      "step": 15155
+    },
+    {
+      "epoch": 1.2915210907541543,
+      "grad_norm": 20.75,
+      "learning_rate": 7.02740977936674e-07,
+      "loss": 0.6913,
+      "step": 15156
+    },
+    {
+      "epoch": 1.2916063059224543,
+      "grad_norm": 22.875,
+      "learning_rate": 7.025899073486419e-07,
+      "loss": 0.5518,
+      "step": 15157
+    },
+    {
+      "epoch": 1.2916915210907542,
+      "grad_norm": 14.375,
+      "learning_rate": 7.024388466528608e-07,
+      "loss": 0.6213,
+      "step": 15158
+    },
+    {
+      "epoch": 1.2917767362590542,
+      "grad_norm": 16.375,
+      "learning_rate": 7.022877958520611e-07,
+      "loss": 0.6652,
+      "step": 15159
+    },
+    {
+      "epoch": 1.2918619514273542,
+      "grad_norm": 10.5,
+      "learning_rate": 7.021367549489708e-07,
+      "loss": 0.2571,
+      "step": 15160
+    },
+    {
+      "epoch": 1.2919471665956541,
+      "grad_norm": 18.5,
+      "learning_rate": 7.019857239463209e-07,
+      "loss": 0.9149,
+      "step": 15161
+    },
+    {
+      "epoch": 1.292032381763954,
+      "grad_norm": 19.625,
+      "learning_rate": 7.018347028468395e-07,
+      "loss": 0.5043,
+      "step": 15162
+    },
+    {
+      "epoch": 1.292117596932254,
+      "grad_norm": 12.75,
+      "learning_rate": 7.016836916532563e-07,
+      "loss": 0.5463,
+      "step": 15163
+    },
+    {
+      "epoch": 1.292202812100554,
+      "grad_norm": 14.6875,
+      "learning_rate": 7.015326903683006e-07,
+      "loss": 0.5039,
+      "step": 15164
+    },
+    {
+      "epoch": 1.292288027268854,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.013816989947002e-07,
+      "loss": 0.5184,
+      "step": 15165
+    },
+    {
+      "epoch": 1.292373242437154,
+      "grad_norm": 18.75,
+      "learning_rate": 7.012307175351843e-07,
+      "loss": 0.6132,
+      "step": 15166
+    },
+    {
+      "epoch": 1.2924584576054539,
+      "grad_norm": 13.625,
+      "learning_rate": 7.01079745992481e-07,
+      "loss": 0.3651,
+      "step": 15167
+    },
+    {
+      "epoch": 1.2925436727737538,
+      "grad_norm": 16.75,
+      "learning_rate": 7.009287843693192e-07,
+      "loss": 0.3371,
+      "step": 15168
+    },
+    {
+      "epoch": 1.2926288879420538,
+      "grad_norm": 17.375,
+      "learning_rate": 7.007778326684264e-07,
+      "loss": 0.7953,
+      "step": 15169
+    },
+    {
+      "epoch": 1.2927141031103537,
+      "grad_norm": 16.0,
+      "learning_rate": 7.006268908925301e-07,
+      "loss": 0.8093,
+      "step": 15170
+    },
+    {
+      "epoch": 1.2927993182786537,
+      "grad_norm": 14.125,
+      "learning_rate": 7.004759590443583e-07,
+      "loss": 0.4036,
+      "step": 15171
+    },
+    {
+      "epoch": 1.2928845334469536,
+      "grad_norm": 16.75,
+      "learning_rate": 7.003250371266383e-07,
+      "loss": 0.5849,
+      "step": 15172
+    },
+    {
+      "epoch": 1.2929697486152536,
+      "grad_norm": 15.25,
+      "learning_rate": 7.001741251420982e-07,
+      "loss": 0.6896,
+      "step": 15173
+    },
+    {
+      "epoch": 1.2930549637835536,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.000232230934642e-07,
+      "loss": 0.4887,
+      "step": 15174
+    },
+    {
+      "epoch": 1.2931401789518535,
+      "grad_norm": 20.5,
+      "learning_rate": 6.998723309834634e-07,
+      "loss": 0.7003,
+      "step": 15175
+    },
+    {
+      "epoch": 1.2932253941201535,
+      "grad_norm": 13.8125,
+      "learning_rate": 6.997214488148235e-07,
+      "loss": 0.4471,
+      "step": 15176
+    },
+    {
+      "epoch": 1.2933106092884534,
+      "grad_norm": 19.0,
+      "learning_rate": 6.995705765902702e-07,
+      "loss": 0.6771,
+      "step": 15177
+    },
+    {
+      "epoch": 1.2933958244567534,
+      "grad_norm": 15.625,
+      "learning_rate": 6.994197143125297e-07,
+      "loss": 0.5664,
+      "step": 15178
+    },
+    {
+      "epoch": 1.2934810396250533,
+      "grad_norm": 16.75,
+      "learning_rate": 6.992688619843286e-07,
+      "loss": 0.5801,
+      "step": 15179
+    },
+    {
+      "epoch": 1.2935662547933533,
+      "grad_norm": 13.0,
+      "learning_rate": 6.991180196083929e-07,
+      "loss": 0.5631,
+      "step": 15180
+    },
+    {
+      "epoch": 1.2936514699616533,
+      "grad_norm": 15.0,
+      "learning_rate": 6.98967187187449e-07,
+      "loss": 0.7203,
+      "step": 15181
+    },
+    {
+      "epoch": 1.2937366851299532,
+      "grad_norm": 21.625,
+      "learning_rate": 6.988163647242216e-07,
+      "loss": 0.5454,
+      "step": 15182
+    },
+    {
+      "epoch": 1.2938219002982532,
+      "grad_norm": 16.625,
+      "learning_rate": 6.986655522214369e-07,
+      "loss": 0.8698,
+      "step": 15183
+    },
+    {
+      "epoch": 1.2939071154665531,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.985147496818205e-07,
+      "loss": 0.7245,
+      "step": 15184
+    },
+    {
+      "epoch": 1.293992330634853,
+      "grad_norm": 19.5,
+      "learning_rate": 6.983639571080966e-07,
+      "loss": 0.6196,
+      "step": 15185
+    },
+    {
+      "epoch": 1.294077545803153,
+      "grad_norm": 15.875,
+      "learning_rate": 6.982131745029913e-07,
+      "loss": 0.5569,
+      "step": 15186
+    },
+    {
+      "epoch": 1.294162760971453,
+      "grad_norm": 12.0,
+      "learning_rate": 6.980624018692282e-07,
+      "loss": 0.4614,
+      "step": 15187
+    },
+    {
+      "epoch": 1.294247976139753,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.979116392095325e-07,
+      "loss": 0.3476,
+      "step": 15188
+    },
+    {
+      "epoch": 1.294333191308053,
+      "grad_norm": 16.0,
+      "learning_rate": 6.977608865266289e-07,
+      "loss": 0.4849,
+      "step": 15189
+    },
+    {
+      "epoch": 1.2944184064763529,
+      "grad_norm": 17.75,
+      "learning_rate": 6.976101438232416e-07,
+      "loss": 0.8338,
+      "step": 15190
+    },
+    {
+      "epoch": 1.2945036216446528,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.974594111020941e-07,
+      "loss": 0.5814,
+      "step": 15191
+    },
+    {
+      "epoch": 1.2945888368129528,
+      "grad_norm": 13.5625,
+      "learning_rate": 6.973086883659111e-07,
+      "loss": 0.489,
+      "step": 15192
+    },
+    {
+      "epoch": 1.2946740519812527,
+      "grad_norm": 23.125,
+      "learning_rate": 6.971579756174154e-07,
+      "loss": 0.696,
+      "step": 15193
+    },
+    {
+      "epoch": 1.2947592671495527,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.970072728593316e-07,
+      "loss": 0.6349,
+      "step": 15194
+    },
+    {
+      "epoch": 1.2948444823178527,
+      "grad_norm": 15.375,
+      "learning_rate": 6.968565800943819e-07,
+      "loss": 0.5272,
+      "step": 15195
+    },
+    {
+      "epoch": 1.2949296974861526,
+      "grad_norm": 20.625,
+      "learning_rate": 6.9670589732529e-07,
+      "loss": 0.6985,
+      "step": 15196
+    },
+    {
+      "epoch": 1.2950149126544526,
+      "grad_norm": 20.5,
+      "learning_rate": 6.965552245547791e-07,
+      "loss": 0.5732,
+      "step": 15197
+    },
+    {
+      "epoch": 1.2951001278227525,
+      "grad_norm": 25.0,
+      "learning_rate": 6.964045617855721e-07,
+      "loss": 0.8428,
+      "step": 15198
+    },
+    {
+      "epoch": 1.2951853429910525,
+      "grad_norm": 17.25,
+      "learning_rate": 6.962539090203908e-07,
+      "loss": 0.7555,
+      "step": 15199
+    },
+    {
+      "epoch": 1.2952705581593524,
+      "grad_norm": 11.5625,
+      "learning_rate": 6.961032662619588e-07,
+      "loss": 0.3381,
+      "step": 15200
+    },
+    {
+      "epoch": 1.2953557733276524,
+      "grad_norm": 11.3125,
+      "learning_rate": 6.959526335129974e-07,
+      "loss": 0.4085,
+      "step": 15201
+    },
+    {
+      "epoch": 1.2954409884959524,
+      "grad_norm": 14.75,
+      "learning_rate": 6.958020107762289e-07,
+      "loss": 0.6015,
+      "step": 15202
+    },
+    {
+      "epoch": 1.2955262036642523,
+      "grad_norm": 12.625,
+      "learning_rate": 6.956513980543759e-07,
+      "loss": 0.5905,
+      "step": 15203
+    },
+    {
+      "epoch": 1.2956114188325523,
+      "grad_norm": 17.375,
+      "learning_rate": 6.955007953501589e-07,
+      "loss": 0.6573,
+      "step": 15204
+    },
+    {
+      "epoch": 1.2956966340008522,
+      "grad_norm": 18.625,
+      "learning_rate": 6.953502026663003e-07,
+      "loss": 0.6264,
+      "step": 15205
+    },
+    {
+      "epoch": 1.2957818491691522,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.951996200055214e-07,
+      "loss": 0.478,
+      "step": 15206
+    },
+    {
+      "epoch": 1.2958670643374521,
+      "grad_norm": 13.75,
+      "learning_rate": 6.950490473705435e-07,
+      "loss": 0.5811,
+      "step": 15207
+    },
+    {
+      "epoch": 1.295952279505752,
+      "grad_norm": 12.5,
+      "learning_rate": 6.948984847640875e-07,
+      "loss": 0.499,
+      "step": 15208
+    },
+    {
+      "epoch": 1.296037494674052,
+      "grad_norm": 16.25,
+      "learning_rate": 6.947479321888736e-07,
+      "loss": 0.8175,
+      "step": 15209
+    },
+    {
+      "epoch": 1.296122709842352,
+      "grad_norm": 19.125,
+      "learning_rate": 6.945973896476228e-07,
+      "loss": 0.7917,
+      "step": 15210
+    },
+    {
+      "epoch": 1.296207925010652,
+      "grad_norm": 25.875,
+      "learning_rate": 6.944468571430563e-07,
+      "loss": 0.9344,
+      "step": 15211
+    },
+    {
+      "epoch": 1.296293140178952,
+      "grad_norm": 17.25,
+      "learning_rate": 6.942963346778932e-07,
+      "loss": 0.5976,
+      "step": 15212
+    },
+    {
+      "epoch": 1.2963783553472519,
+      "grad_norm": 32.5,
+      "learning_rate": 6.941458222548542e-07,
+      "loss": 0.7255,
+      "step": 15213
+    },
+    {
+      "epoch": 1.2964635705155518,
+      "grad_norm": 14.0,
+      "learning_rate": 6.939953198766589e-07,
+      "loss": 0.5307,
+      "step": 15214
+    },
+    {
+      "epoch": 1.2965487856838518,
+      "grad_norm": 16.625,
+      "learning_rate": 6.93844827546028e-07,
+      "loss": 0.4613,
+      "step": 15215
+    },
+    {
+      "epoch": 1.2966340008521517,
+      "grad_norm": 16.375,
+      "learning_rate": 6.936943452656802e-07,
+      "loss": 0.5695,
+      "step": 15216
+    },
+    {
+      "epoch": 1.2967192160204517,
+      "grad_norm": 18.25,
+      "learning_rate": 6.935438730383345e-07,
+      "loss": 0.6086,
+      "step": 15217
+    },
+    {
+      "epoch": 1.2968044311887517,
+      "grad_norm": 13.375,
+      "learning_rate": 6.933934108667108e-07,
+      "loss": 0.4479,
+      "step": 15218
+    },
+    {
+      "epoch": 1.2968896463570516,
+      "grad_norm": 19.875,
+      "learning_rate": 6.932429587535277e-07,
+      "loss": 0.6651,
+      "step": 15219
+    },
+    {
+      "epoch": 1.2969748615253516,
+      "grad_norm": 17.125,
+      "learning_rate": 6.930925167015046e-07,
+      "loss": 0.7064,
+      "step": 15220
+    },
+    {
+      "epoch": 1.2970600766936515,
+      "grad_norm": 17.875,
+      "learning_rate": 6.929420847133594e-07,
+      "loss": 1.0303,
+      "step": 15221
+    },
+    {
+      "epoch": 1.2971452918619515,
+      "grad_norm": 12.0625,
+      "learning_rate": 6.927916627918107e-07,
+      "loss": 0.3923,
+      "step": 15222
+    },
+    {
+      "epoch": 1.2972305070302514,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.926412509395772e-07,
+      "loss": 0.484,
+      "step": 15223
+    },
+    {
+      "epoch": 1.2973157221985514,
+      "grad_norm": 14.25,
+      "learning_rate": 6.924908491593776e-07,
+      "loss": 0.5768,
+      "step": 15224
+    },
+    {
+      "epoch": 1.2974009373668514,
+      "grad_norm": 21.625,
+      "learning_rate": 6.923404574539282e-07,
+      "loss": 1.0559,
+      "step": 15225
+    },
+    {
+      "epoch": 1.2974861525351513,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.921900758259472e-07,
+      "loss": 0.524,
+      "step": 15226
+    },
+    {
+      "epoch": 1.2975713677034513,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.920397042781526e-07,
+      "loss": 0.7534,
+      "step": 15227
+    },
+    {
+      "epoch": 1.2976565828717512,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.918893428132623e-07,
+      "loss": 0.4391,
+      "step": 15228
+    },
+    {
+      "epoch": 1.2977417980400512,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.91738991433992e-07,
+      "loss": 0.4048,
+      "step": 15229
+    },
+    {
+      "epoch": 1.2978270132083511,
+      "grad_norm": 13.8125,
+      "learning_rate": 6.915886501430596e-07,
+      "loss": 0.4819,
+      "step": 15230
+    },
+    {
+      "epoch": 1.297912228376651,
+      "grad_norm": 22.25,
+      "learning_rate": 6.914383189431818e-07,
+      "loss": 0.7331,
+      "step": 15231
+    },
+    {
+      "epoch": 1.297997443544951,
+      "grad_norm": 22.25,
+      "learning_rate": 6.912879978370758e-07,
+      "loss": 0.9725,
+      "step": 15232
+    },
+    {
+      "epoch": 1.298082658713251,
+      "grad_norm": 14.6875,
+      "learning_rate": 6.911376868274574e-07,
+      "loss": 0.7844,
+      "step": 15233
+    },
+    {
+      "epoch": 1.298167873881551,
+      "grad_norm": 19.5,
+      "learning_rate": 6.909873859170427e-07,
+      "loss": 0.844,
+      "step": 15234
+    },
+    {
+      "epoch": 1.298253089049851,
+      "grad_norm": 14.4375,
+      "learning_rate": 6.90837095108548e-07,
+      "loss": 0.4212,
+      "step": 15235
+    },
+    {
+      "epoch": 1.2983383042181509,
+      "grad_norm": 19.125,
+      "learning_rate": 6.906868144046891e-07,
+      "loss": 0.607,
+      "step": 15236
+    },
+    {
+      "epoch": 1.2984235193864508,
+      "grad_norm": 28.875,
+      "learning_rate": 6.905365438081826e-07,
+      "loss": 0.6844,
+      "step": 15237
+    },
+    {
+      "epoch": 1.2985087345547508,
+      "grad_norm": 18.625,
+      "learning_rate": 6.903862833217427e-07,
+      "loss": 0.5836,
+      "step": 15238
+    },
+    {
+      "epoch": 1.2985939497230508,
+      "grad_norm": 16.125,
+      "learning_rate": 6.902360329480854e-07,
+      "loss": 0.5235,
+      "step": 15239
+    },
+    {
+      "epoch": 1.2986791648913507,
+      "grad_norm": 12.1875,
+      "learning_rate": 6.900857926899264e-07,
+      "loss": 0.3698,
+      "step": 15240
+    },
+    {
+      "epoch": 1.2987643800596507,
+      "grad_norm": 16.75,
+      "learning_rate": 6.899355625499801e-07,
+      "loss": 0.6695,
+      "step": 15241
+    },
+    {
+      "epoch": 1.2988495952279506,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.897853425309611e-07,
+      "loss": 0.7095,
+      "step": 15242
+    },
+    {
+      "epoch": 1.2989348103962506,
+      "grad_norm": 14.25,
+      "learning_rate": 6.896351326355841e-07,
+      "loss": 0.5926,
+      "step": 15243
+    },
+    {
+      "epoch": 1.2990200255645505,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.894849328665637e-07,
+      "loss": 0.5323,
+      "step": 15244
+    },
+    {
+      "epoch": 1.2991052407328505,
+      "grad_norm": 16.125,
+      "learning_rate": 6.893347432266148e-07,
+      "loss": 0.6236,
+      "step": 15245
+    },
+    {
+      "epoch": 1.2991904559011505,
+      "grad_norm": 20.25,
+      "learning_rate": 6.891845637184505e-07,
+      "loss": 0.4984,
+      "step": 15246
+    },
+    {
+      "epoch": 1.2992756710694504,
+      "grad_norm": 25.5,
+      "learning_rate": 6.89034394344785e-07,
+      "loss": 1.1318,
+      "step": 15247
+    },
+    {
+      "epoch": 1.2993608862377504,
+      "grad_norm": 17.25,
+      "learning_rate": 6.888842351083319e-07,
+      "loss": 0.8634,
+      "step": 15248
+    },
+    {
+      "epoch": 1.2994461014060503,
+      "grad_norm": 14.75,
+      "learning_rate": 6.887340860118056e-07,
+      "loss": 0.5517,
+      "step": 15249
+    },
+    {
+      "epoch": 1.2995313165743503,
+      "grad_norm": 25.375,
+      "learning_rate": 6.885839470579187e-07,
+      "loss": 0.7941,
+      "step": 15250
+    },
+    {
+      "epoch": 1.2996165317426502,
+      "grad_norm": 14.8125,
+      "learning_rate": 6.88433818249384e-07,
+      "loss": 0.5883,
+      "step": 15251
+    },
+    {
+      "epoch": 1.2997017469109502,
+      "grad_norm": 13.0,
+      "learning_rate": 6.882836995889147e-07,
+      "loss": 0.466,
+      "step": 15252
+    },
+    {
+      "epoch": 1.2997869620792502,
+      "grad_norm": 16.5,
+      "learning_rate": 6.881335910792242e-07,
+      "loss": 0.3988,
+      "step": 15253
+    },
+    {
+      "epoch": 1.29987217724755,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.879834927230247e-07,
+      "loss": 0.7602,
+      "step": 15254
+    },
+    {
+      "epoch": 1.29995739241585,
+      "grad_norm": 15.0,
+      "learning_rate": 6.878334045230285e-07,
+      "loss": 0.6885,
+      "step": 15255
+    },
+    {
+      "epoch": 1.30004260758415,
+      "grad_norm": 17.625,
+      "learning_rate": 6.87683326481948e-07,
+      "loss": 0.9881,
+      "step": 15256
+    },
+    {
+      "epoch": 1.30012782275245,
+      "grad_norm": 17.375,
+      "learning_rate": 6.875332586024955e-07,
+      "loss": 0.5717,
+      "step": 15257
+    },
+    {
+      "epoch": 1.30021303792075,
+      "grad_norm": 24.625,
+      "learning_rate": 6.873832008873828e-07,
+      "loss": 0.5091,
+      "step": 15258
+    },
+    {
+      "epoch": 1.30029825308905,
+      "grad_norm": 21.375,
+      "learning_rate": 6.872331533393209e-07,
+      "loss": 0.5149,
+      "step": 15259
+    },
+    {
+      "epoch": 1.3003834682573499,
+      "grad_norm": 21.375,
+      "learning_rate": 6.870831159610217e-07,
+      "loss": 0.9007,
+      "step": 15260
+    },
+    {
+      "epoch": 1.3004686834256498,
+      "grad_norm": 13.75,
+      "learning_rate": 6.869330887551967e-07,
+      "loss": 0.5163,
+      "step": 15261
+    },
+    {
+      "epoch": 1.3005538985939498,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.867830717245575e-07,
+      "loss": 0.5828,
+      "step": 15262
+    },
+    {
+      "epoch": 1.3006391137622497,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.866330648718142e-07,
+      "loss": 0.3296,
+      "step": 15263
+    },
+    {
+      "epoch": 1.3007243289305497,
+      "grad_norm": 13.5625,
+      "learning_rate": 6.864830681996777e-07,
+      "loss": 0.4653,
+      "step": 15264
+    },
+    {
+      "epoch": 1.3008095440988496,
+      "grad_norm": 12.625,
+      "learning_rate": 6.863330817108594e-07,
+      "loss": 0.3393,
+      "step": 15265
+    },
+    {
+      "epoch": 1.3008947592671496,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.861831054080686e-07,
+      "loss": 0.7411,
+      "step": 15266
+    },
+    {
+      "epoch": 1.3009799744354495,
+      "grad_norm": 16.75,
+      "learning_rate": 6.860331392940167e-07,
+      "loss": 0.4362,
+      "step": 15267
+    },
+    {
+      "epoch": 1.3010651896037495,
+      "grad_norm": 19.375,
+      "learning_rate": 6.858831833714125e-07,
+      "loss": 0.6598,
+      "step": 15268
+    },
+    {
+      "epoch": 1.3011504047720495,
+      "grad_norm": 18.0,
+      "learning_rate": 6.857332376429663e-07,
+      "loss": 0.4877,
+      "step": 15269
+    },
+    {
+      "epoch": 1.3012356199403494,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.855833021113886e-07,
+      "loss": 0.5588,
+      "step": 15270
+    },
+    {
+      "epoch": 1.3013208351086494,
+      "grad_norm": 19.0,
+      "learning_rate": 6.854333767793877e-07,
+      "loss": 0.5448,
+      "step": 15271
+    },
+    {
+      "epoch": 1.3014060502769493,
+      "grad_norm": 12.5,
+      "learning_rate": 6.852834616496733e-07,
+      "loss": 0.536,
+      "step": 15272
+    },
+    {
+      "epoch": 1.3014912654452493,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.85133556724955e-07,
+      "loss": 0.5267,
+      "step": 15273
+    },
+    {
+      "epoch": 1.3015764806135492,
+      "grad_norm": 17.125,
+      "learning_rate": 6.849836620079412e-07,
+      "loss": 0.6593,
+      "step": 15274
+    },
+    {
+      "epoch": 1.3016616957818492,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.848337775013409e-07,
+      "loss": 0.4954,
+      "step": 15275
+    },
+    {
+      "epoch": 1.3017469109501492,
+      "grad_norm": 17.875,
+      "learning_rate": 6.846839032078622e-07,
+      "loss": 1.0108,
+      "step": 15276
+    },
+    {
+      "epoch": 1.3018321261184491,
+      "grad_norm": 12.9375,
+      "learning_rate": 6.845340391302139e-07,
+      "loss": 0.2836,
+      "step": 15277
+    },
+    {
+      "epoch": 1.301917341286749,
+      "grad_norm": 13.75,
+      "learning_rate": 6.843841852711042e-07,
+      "loss": 0.6242,
+      "step": 15278
+    },
+    {
+      "epoch": 1.302002556455049,
+      "grad_norm": 19.5,
+      "learning_rate": 6.842343416332414e-07,
+      "loss": 0.9831,
+      "step": 15279
+    },
+    {
+      "epoch": 1.302087771623349,
+      "grad_norm": 13.0,
+      "learning_rate": 6.840845082193326e-07,
+      "loss": 0.3596,
+      "step": 15280
+    },
+    {
+      "epoch": 1.302172986791649,
+      "grad_norm": 13.5625,
+      "learning_rate": 6.839346850320862e-07,
+      "loss": 0.6221,
+      "step": 15281
+    },
+    {
+      "epoch": 1.302258201959949,
+      "grad_norm": 19.625,
+      "learning_rate": 6.837848720742088e-07,
+      "loss": 0.9982,
+      "step": 15282
+    },
+    {
+      "epoch": 1.3023434171282489,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.836350693484081e-07,
+      "loss": 0.5154,
+      "step": 15283
+    },
+    {
+      "epoch": 1.3024286322965488,
+      "grad_norm": 13.5625,
+      "learning_rate": 6.834852768573919e-07,
+      "loss": 0.674,
+      "step": 15284
+    },
+    {
+      "epoch": 1.3025138474648488,
+      "grad_norm": 13.25,
+      "learning_rate": 6.833354946038657e-07,
+      "loss": 0.3937,
+      "step": 15285
+    },
+    {
+      "epoch": 1.3025990626331487,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.831857225905369e-07,
+      "loss": 0.6341,
+      "step": 15286
+    },
+    {
+      "epoch": 1.3026842778014487,
+      "grad_norm": 19.625,
+      "learning_rate": 6.830359608201127e-07,
+      "loss": 0.4433,
+      "step": 15287
+    },
+    {
+      "epoch": 1.3027694929697486,
+      "grad_norm": 17.875,
+      "learning_rate": 6.828862092952985e-07,
+      "loss": 0.7904,
+      "step": 15288
+    },
+    {
+      "epoch": 1.3028547081380486,
+      "grad_norm": 16.25,
+      "learning_rate": 6.827364680188009e-07,
+      "loss": 0.574,
+      "step": 15289
+    },
+    {
+      "epoch": 1.3029399233063486,
+      "grad_norm": 12.625,
+      "learning_rate": 6.825867369933254e-07,
+      "loss": 0.3381,
+      "step": 15290
+    },
+    {
+      "epoch": 1.3030251384746485,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.824370162215783e-07,
+      "loss": 0.3572,
+      "step": 15291
+    },
+    {
+      "epoch": 1.3031103536429485,
+      "grad_norm": 16.75,
+      "learning_rate": 6.822873057062653e-07,
+      "loss": 0.6405,
+      "step": 15292
+    },
+    {
+      "epoch": 1.3031955688112484,
+      "grad_norm": 16.125,
+      "learning_rate": 6.821376054500911e-07,
+      "loss": 0.4735,
+      "step": 15293
+    },
+    {
+      "epoch": 1.3032807839795484,
+      "grad_norm": 13.5,
+      "learning_rate": 6.819879154557615e-07,
+      "loss": 0.6004,
+      "step": 15294
+    },
+    {
+      "epoch": 1.3033659991478483,
+      "grad_norm": 13.125,
+      "learning_rate": 6.818382357259814e-07,
+      "loss": 0.606,
+      "step": 15295
+    },
+    {
+      "epoch": 1.3034512143161483,
+      "grad_norm": 14.375,
+      "learning_rate": 6.816885662634561e-07,
+      "loss": 0.6028,
+      "step": 15296
+    },
+    {
+      "epoch": 1.3035364294844483,
+      "grad_norm": 11.875,
+      "learning_rate": 6.815389070708903e-07,
+      "loss": 0.3505,
+      "step": 15297
+    },
+    {
+      "epoch": 1.3036216446527482,
+      "grad_norm": 16.125,
+      "learning_rate": 6.813892581509872e-07,
+      "loss": 0.6732,
+      "step": 15298
+    },
+    {
+      "epoch": 1.3037068598210482,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.812396195064521e-07,
+      "loss": 0.5071,
+      "step": 15299
+    },
+    {
+      "epoch": 1.3037920749893481,
+      "grad_norm": 15.375,
+      "learning_rate": 6.810899911399895e-07,
+      "loss": 0.8301,
+      "step": 15300
+    },
+    {
+      "epoch": 1.303877290157648,
+      "grad_norm": 17.875,
+      "learning_rate": 6.809403730543023e-07,
+      "loss": 0.8184,
+      "step": 15301
+    },
+    {
+      "epoch": 1.303962505325948,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.807907652520948e-07,
+      "loss": 0.5431,
+      "step": 15302
+    },
+    {
+      "epoch": 1.304047720494248,
+      "grad_norm": 13.5,
+      "learning_rate": 6.806411677360705e-07,
+      "loss": 0.4563,
+      "step": 15303
+    },
+    {
+      "epoch": 1.304132935662548,
+      "grad_norm": 13.75,
+      "learning_rate": 6.804915805089335e-07,
+      "loss": 0.5589,
+      "step": 15304
+    },
+    {
+      "epoch": 1.304218150830848,
+      "grad_norm": 22.5,
+      "learning_rate": 6.803420035733857e-07,
+      "loss": 0.5234,
+      "step": 15305
+    },
+    {
+      "epoch": 1.3043033659991479,
+      "grad_norm": 14.5,
+      "learning_rate": 6.801924369321314e-07,
+      "loss": 0.8253,
+      "step": 15306
+    },
+    {
+      "epoch": 1.3043885811674478,
+      "grad_norm": 19.0,
+      "learning_rate": 6.80042880587872e-07,
+      "loss": 0.9325,
+      "step": 15307
+    },
+    {
+      "epoch": 1.3044737963357478,
+      "grad_norm": 17.5,
+      "learning_rate": 6.798933345433111e-07,
+      "loss": 0.7194,
+      "step": 15308
+    },
+    {
+      "epoch": 1.3045590115040477,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.797437988011513e-07,
+      "loss": 0.7592,
+      "step": 15309
+    },
+    {
+      "epoch": 1.3046442266723477,
+      "grad_norm": 22.875,
+      "learning_rate": 6.795942733640939e-07,
+      "loss": 0.7976,
+      "step": 15310
+    },
+    {
+      "epoch": 1.3047294418406477,
+      "grad_norm": 17.25,
+      "learning_rate": 6.794447582348418e-07,
+      "loss": 0.7394,
+      "step": 15311
+    },
+    {
+      "epoch": 1.3048146570089476,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.792952534160965e-07,
+      "loss": 0.71,
+      "step": 15312
+    },
+    {
+      "epoch": 1.3048998721772476,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.791457589105605e-07,
+      "loss": 0.4231,
+      "step": 15313
+    },
+    {
+      "epoch": 1.3049850873455475,
+      "grad_norm": 19.5,
+      "learning_rate": 6.789962747209345e-07,
+      "loss": 0.6185,
+      "step": 15314
+    },
+    {
+      "epoch": 1.3050703025138475,
+      "grad_norm": 11.6875,
+      "learning_rate": 6.788468008499196e-07,
+      "loss": 0.4752,
+      "step": 15315
+    },
+    {
+      "epoch": 1.3051555176821474,
+      "grad_norm": 15.875,
+      "learning_rate": 6.786973373002174e-07,
+      "loss": 0.7148,
+      "step": 15316
+    },
+    {
+      "epoch": 1.3052407328504474,
+      "grad_norm": 19.125,
+      "learning_rate": 6.785478840745294e-07,
+      "loss": 0.6581,
+      "step": 15317
+    },
+    {
+      "epoch": 1.3053259480187474,
+      "grad_norm": 18.875,
+      "learning_rate": 6.783984411755551e-07,
+      "loss": 0.6178,
+      "step": 15318
+    },
+    {
+      "epoch": 1.3054111631870473,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.782490086059957e-07,
+      "loss": 0.7363,
+      "step": 15319
+    },
+    {
+      "epoch": 1.3054963783553473,
+      "grad_norm": 19.75,
+      "learning_rate": 6.780995863685518e-07,
+      "loss": 0.633,
+      "step": 15320
+    },
+    {
+      "epoch": 1.3055815935236472,
+      "grad_norm": 13.375,
+      "learning_rate": 6.779501744659238e-07,
+      "loss": 0.3973,
+      "step": 15321
+    },
+    {
+      "epoch": 1.3056668086919472,
+      "grad_norm": 13.125,
+      "learning_rate": 6.778007729008115e-07,
+      "loss": 0.4486,
+      "step": 15322
+    },
+    {
+      "epoch": 1.3057520238602471,
+      "grad_norm": 14.5,
+      "learning_rate": 6.776513816759142e-07,
+      "loss": 0.5381,
+      "step": 15323
+    },
+    {
+      "epoch": 1.305837239028547,
+      "grad_norm": 18.875,
+      "learning_rate": 6.775020007939317e-07,
+      "loss": 0.9599,
+      "step": 15324
+    },
+    {
+      "epoch": 1.305922454196847,
+      "grad_norm": 15.625,
+      "learning_rate": 6.77352630257564e-07,
+      "loss": 0.6201,
+      "step": 15325
+    },
+    {
+      "epoch": 1.306007669365147,
+      "grad_norm": 16.125,
+      "learning_rate": 6.772032700695105e-07,
+      "loss": 0.4252,
+      "step": 15326
+    },
+    {
+      "epoch": 1.306092884533447,
+      "grad_norm": 17.375,
+      "learning_rate": 6.770539202324694e-07,
+      "loss": 0.6245,
+      "step": 15327
+    },
+    {
+      "epoch": 1.306178099701747,
+      "grad_norm": 15.25,
+      "learning_rate": 6.769045807491402e-07,
+      "loss": 0.4163,
+      "step": 15328
+    },
+    {
+      "epoch": 1.3062633148700469,
+      "grad_norm": 20.25,
+      "learning_rate": 6.767552516222214e-07,
+      "loss": 0.7515,
+      "step": 15329
+    },
+    {
+      "epoch": 1.3063485300383468,
+      "grad_norm": 24.875,
+      "learning_rate": 6.766059328544128e-07,
+      "loss": 0.557,
+      "step": 15330
+    },
+    {
+      "epoch": 1.3064337452066468,
+      "grad_norm": 14.875,
+      "learning_rate": 6.764566244484106e-07,
+      "loss": 0.5739,
+      "step": 15331
+    },
+    {
+      "epoch": 1.3065189603749467,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.763073264069137e-07,
+      "loss": 0.6625,
+      "step": 15332
+    },
+    {
+      "epoch": 1.3066041755432467,
+      "grad_norm": 17.625,
+      "learning_rate": 6.761580387326205e-07,
+      "loss": 0.9779,
+      "step": 15333
+    },
+    {
+      "epoch": 1.3066893907115467,
+      "grad_norm": 18.375,
+      "learning_rate": 6.76008761428229e-07,
+      "loss": 0.689,
+      "step": 15334
+    },
+    {
+      "epoch": 1.3067746058798466,
+      "grad_norm": 17.0,
+      "learning_rate": 6.758594944964358e-07,
+      "loss": 0.7297,
+      "step": 15335
+    },
+    {
+      "epoch": 1.3068598210481466,
+      "grad_norm": 15.5,
+      "learning_rate": 6.757102379399389e-07,
+      "loss": 0.3316,
+      "step": 15336
+    },
+    {
+      "epoch": 1.3069450362164465,
+      "grad_norm": 16.625,
+      "learning_rate": 6.755609917614356e-07,
+      "loss": 0.4785,
+      "step": 15337
+    },
+    {
+      "epoch": 1.3070302513847465,
+      "grad_norm": 10.8125,
+      "learning_rate": 6.754117559636233e-07,
+      "loss": 0.7587,
+      "step": 15338
+    },
+    {
+      "epoch": 1.3071154665530464,
+      "grad_norm": 11.3125,
+      "learning_rate": 6.752625305491983e-07,
+      "loss": 0.4599,
+      "step": 15339
+    },
+    {
+      "epoch": 1.3072006817213464,
+      "grad_norm": 13.875,
+      "learning_rate": 6.75113315520857e-07,
+      "loss": 0.6435,
+      "step": 15340
+    },
+    {
+      "epoch": 1.3072858968896464,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.74964110881296e-07,
+      "loss": 0.3003,
+      "step": 15341
+    },
+    {
+      "epoch": 1.3073711120579463,
+      "grad_norm": 12.875,
+      "learning_rate": 6.748149166332119e-07,
+      "loss": 0.4318,
+      "step": 15342
+    },
+    {
+      "epoch": 1.3074563272262463,
+      "grad_norm": 15.0,
+      "learning_rate": 6.74665732779301e-07,
+      "loss": 0.5902,
+      "step": 15343
+    },
+    {
+      "epoch": 1.3075415423945462,
+      "grad_norm": 18.375,
+      "learning_rate": 6.745165593222583e-07,
+      "loss": 0.894,
+      "step": 15344
+    },
+    {
+      "epoch": 1.3076267575628462,
+      "grad_norm": 15.625,
+      "learning_rate": 6.743673962647802e-07,
+      "loss": 0.5885,
+      "step": 15345
+    },
+    {
+      "epoch": 1.3077119727311461,
+      "grad_norm": 14.75,
+      "learning_rate": 6.742182436095626e-07,
+      "loss": 0.4738,
+      "step": 15346
+    },
+    {
+      "epoch": 1.307797187899446,
+      "grad_norm": 20.5,
+      "learning_rate": 6.740691013593002e-07,
+      "loss": 0.5926,
+      "step": 15347
+    },
+    {
+      "epoch": 1.307882403067746,
+      "grad_norm": 13.75,
+      "learning_rate": 6.739199695166879e-07,
+      "loss": 0.3915,
+      "step": 15348
+    },
+    {
+      "epoch": 1.307967618236046,
+      "grad_norm": 11.5,
+      "learning_rate": 6.73770848084421e-07,
+      "loss": 0.3932,
+      "step": 15349
+    },
+    {
+      "epoch": 1.308052833404346,
+      "grad_norm": 19.875,
+      "learning_rate": 6.736217370651944e-07,
+      "loss": 0.9585,
+      "step": 15350
+    },
+    {
+      "epoch": 1.308138048572646,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.734726364617031e-07,
+      "loss": 0.4649,
+      "step": 15351
+    },
+    {
+      "epoch": 1.3082232637409459,
+      "grad_norm": 11.75,
+      "learning_rate": 6.733235462766405e-07,
+      "loss": 0.3234,
+      "step": 15352
+    },
+    {
+      "epoch": 1.3083084789092458,
+      "grad_norm": 11.5625,
+      "learning_rate": 6.731744665127012e-07,
+      "loss": 0.3865,
+      "step": 15353
+    },
+    {
+      "epoch": 1.3083936940775458,
+      "grad_norm": 18.5,
+      "learning_rate": 6.730253971725798e-07,
+      "loss": 0.4972,
+      "step": 15354
+    },
+    {
+      "epoch": 1.3084789092458458,
+      "grad_norm": 17.125,
+      "learning_rate": 6.728763382589693e-07,
+      "loss": 1.1915,
+      "step": 15355
+    },
+    {
+      "epoch": 1.3085641244141457,
+      "grad_norm": 18.625,
+      "learning_rate": 6.727272897745641e-07,
+      "loss": 0.8562,
+      "step": 15356
+    },
+    {
+      "epoch": 1.3086493395824457,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.72578251722057e-07,
+      "loss": 0.4997,
+      "step": 15357
+    },
+    {
+      "epoch": 1.3087345547507456,
+      "grad_norm": 41.75,
+      "learning_rate": 6.724292241041413e-07,
+      "loss": 1.1714,
+      "step": 15358
+    },
+    {
+      "epoch": 1.3088197699190456,
+      "grad_norm": 18.25,
+      "learning_rate": 6.722802069235104e-07,
+      "loss": 0.3617,
+      "step": 15359
+    },
+    {
+      "epoch": 1.3089049850873455,
+      "grad_norm": 43.75,
+      "learning_rate": 6.721312001828576e-07,
+      "loss": 0.9277,
+      "step": 15360
+    },
+    {
+      "epoch": 1.3089902002556455,
+      "grad_norm": 9.5625,
+      "learning_rate": 6.719822038848745e-07,
+      "loss": 0.1717,
+      "step": 15361
+    },
+    {
+      "epoch": 1.3090754154239455,
+      "grad_norm": 17.375,
+      "learning_rate": 6.718332180322548e-07,
+      "loss": 0.6324,
+      "step": 15362
+    },
+    {
+      "epoch": 1.3091606305922454,
+      "grad_norm": 14.0,
+      "learning_rate": 6.716842426276898e-07,
+      "loss": 0.6823,
+      "step": 15363
+    },
+    {
+      "epoch": 1.3092458457605454,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.715352776738724e-07,
+      "loss": 0.7052,
+      "step": 15364
+    },
+    {
+      "epoch": 1.3093310609288453,
+      "grad_norm": 16.375,
+      "learning_rate": 6.713863231734939e-07,
+      "loss": 0.669,
+      "step": 15365
+    },
+    {
+      "epoch": 1.3094162760971453,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.712373791292463e-07,
+      "loss": 0.7044,
+      "step": 15366
+    },
+    {
+      "epoch": 1.3095014912654452,
+      "grad_norm": 14.5,
+      "learning_rate": 6.710884455438213e-07,
+      "loss": 0.5283,
+      "step": 15367
+    },
+    {
+      "epoch": 1.3095867064337452,
+      "grad_norm": 12.25,
+      "learning_rate": 6.709395224199107e-07,
+      "loss": 0.591,
+      "step": 15368
+    },
+    {
+      "epoch": 1.3096719216020452,
+      "grad_norm": 16.625,
+      "learning_rate": 6.707906097602048e-07,
+      "loss": 0.8888,
+      "step": 15369
+    },
+    {
+      "epoch": 1.309757136770345,
+      "grad_norm": 14.8125,
+      "learning_rate": 6.706417075673948e-07,
+      "loss": 0.5586,
+      "step": 15370
+    },
+    {
+      "epoch": 1.309842351938645,
+      "grad_norm": 14.25,
+      "learning_rate": 6.704928158441725e-07,
+      "loss": 0.5235,
+      "step": 15371
+    },
+    {
+      "epoch": 1.309927567106945,
+      "grad_norm": 14.375,
+      "learning_rate": 6.703439345932272e-07,
+      "loss": 0.5794,
+      "step": 15372
+    },
+    {
+      "epoch": 1.310012782275245,
+      "grad_norm": 16.5,
+      "learning_rate": 6.701950638172503e-07,
+      "loss": 0.412,
+      "step": 15373
+    },
+    {
+      "epoch": 1.310097997443545,
+      "grad_norm": 10.25,
+      "learning_rate": 6.700462035189312e-07,
+      "loss": 0.3629,
+      "step": 15374
+    },
+    {
+      "epoch": 1.310183212611845,
+      "grad_norm": 16.0,
+      "learning_rate": 6.698973537009604e-07,
+      "loss": 0.6084,
+      "step": 15375
+    },
+    {
+      "epoch": 1.3102684277801449,
+      "grad_norm": 20.0,
+      "learning_rate": 6.697485143660276e-07,
+      "loss": 0.6572,
+      "step": 15376
+    },
+    {
+      "epoch": 1.3103536429484448,
+      "grad_norm": 16.875,
+      "learning_rate": 6.695996855168232e-07,
+      "loss": 0.85,
+      "step": 15377
+    },
+    {
+      "epoch": 1.3104388581167448,
+      "grad_norm": 36.5,
+      "learning_rate": 6.694508671560357e-07,
+      "loss": 0.8257,
+      "step": 15378
+    },
+    {
+      "epoch": 1.3105240732850447,
+      "grad_norm": 13.1875,
+      "learning_rate": 6.693020592863553e-07,
+      "loss": 0.4859,
+      "step": 15379
+    },
+    {
+      "epoch": 1.3106092884533447,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.691532619104702e-07,
+      "loss": 0.5618,
+      "step": 15380
+    },
+    {
+      "epoch": 1.3106945036216446,
+      "grad_norm": 17.0,
+      "learning_rate": 6.690044750310701e-07,
+      "loss": 0.759,
+      "step": 15381
+    },
+    {
+      "epoch": 1.3107797187899446,
+      "grad_norm": 15.0,
+      "learning_rate": 6.688556986508429e-07,
+      "loss": 0.6163,
+      "step": 15382
+    },
+    {
+      "epoch": 1.3108649339582445,
+      "grad_norm": 23.375,
+      "learning_rate": 6.687069327724776e-07,
+      "loss": 1.1852,
+      "step": 15383
+    },
+    {
+      "epoch": 1.3109501491265445,
+      "grad_norm": 19.0,
+      "learning_rate": 6.685581773986627e-07,
+      "loss": 0.49,
+      "step": 15384
+    },
+    {
+      "epoch": 1.3110353642948445,
+      "grad_norm": 13.875,
+      "learning_rate": 6.684094325320866e-07,
+      "loss": 0.478,
+      "step": 15385
+    },
+    {
+      "epoch": 1.3111205794631444,
+      "grad_norm": 16.75,
+      "learning_rate": 6.682606981754364e-07,
+      "loss": 0.4598,
+      "step": 15386
+    },
+    {
+      "epoch": 1.3112057946314444,
+      "grad_norm": 19.375,
+      "learning_rate": 6.681119743314007e-07,
+      "loss": 0.9594,
+      "step": 15387
+    },
+    {
+      "epoch": 1.3112910097997443,
+      "grad_norm": 17.0,
+      "learning_rate": 6.679632610026665e-07,
+      "loss": 0.6137,
+      "step": 15388
+    },
+    {
+      "epoch": 1.3113762249680443,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.678145581919215e-07,
+      "loss": 0.4175,
+      "step": 15389
+    },
+    {
+      "epoch": 1.3114614401363442,
+      "grad_norm": 13.125,
+      "learning_rate": 6.676658659018532e-07,
+      "loss": 0.3571,
+      "step": 15390
+    },
+    {
+      "epoch": 1.3115466553046442,
+      "grad_norm": 12.875,
+      "learning_rate": 6.675171841351479e-07,
+      "loss": 0.299,
+      "step": 15391
+    },
+    {
+      "epoch": 1.3116318704729442,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.673685128944926e-07,
+      "loss": 0.5589,
+      "step": 15392
+    },
+    {
+      "epoch": 1.3117170856412441,
+      "grad_norm": 19.0,
+      "learning_rate": 6.672198521825748e-07,
+      "loss": 0.7522,
+      "step": 15393
+    },
+    {
+      "epoch": 1.311802300809544,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.670712020020801e-07,
+      "loss": 0.5869,
+      "step": 15394
+    },
+    {
+      "epoch": 1.311887515977844,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.669225623556949e-07,
+      "loss": 0.2488,
+      "step": 15395
+    },
+    {
+      "epoch": 1.311972731146144,
+      "grad_norm": 26.0,
+      "learning_rate": 6.667739332461051e-07,
+      "loss": 0.8275,
+      "step": 15396
+    },
+    {
+      "epoch": 1.312057946314444,
+      "grad_norm": 13.875,
+      "learning_rate": 6.666253146759968e-07,
+      "loss": 0.402,
+      "step": 15397
+    },
+    {
+      "epoch": 1.312143161482744,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.664767066480562e-07,
+      "loss": 0.6103,
+      "step": 15398
+    },
+    {
+      "epoch": 1.3122283766510439,
+      "grad_norm": 14.5,
+      "learning_rate": 6.663281091649679e-07,
+      "loss": 0.5477,
+      "step": 15399
+    },
+    {
+      "epoch": 1.3123135918193438,
+      "grad_norm": 15.75,
+      "learning_rate": 6.661795222294175e-07,
+      "loss": 0.7335,
+      "step": 15400
+    },
+    {
+      "epoch": 1.3123988069876438,
+      "grad_norm": 16.375,
+      "learning_rate": 6.660309458440901e-07,
+      "loss": 0.625,
+      "step": 15401
+    },
+    {
+      "epoch": 1.3124840221559437,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.658823800116714e-07,
+      "loss": 0.2921,
+      "step": 15402
+    },
+    {
+      "epoch": 1.3125692373242437,
+      "grad_norm": 18.75,
+      "learning_rate": 6.657338247348454e-07,
+      "loss": 0.6292,
+      "step": 15403
+    },
+    {
+      "epoch": 1.3126544524925436,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.655852800162963e-07,
+      "loss": 0.6587,
+      "step": 15404
+    },
+    {
+      "epoch": 1.3127396676608436,
+      "grad_norm": 13.5,
+      "learning_rate": 6.654367458587087e-07,
+      "loss": 0.3783,
+      "step": 15405
+    },
+    {
+      "epoch": 1.3128248828291436,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.652882222647669e-07,
+      "loss": 0.4794,
+      "step": 15406
+    },
+    {
+      "epoch": 1.3129100979974435,
+      "grad_norm": 16.5,
+      "learning_rate": 6.651397092371556e-07,
+      "loss": 0.5813,
+      "step": 15407
+    },
+    {
+      "epoch": 1.3129953131657435,
+      "grad_norm": 12.25,
+      "learning_rate": 6.649912067785572e-07,
+      "loss": 0.4377,
+      "step": 15408
+    },
+    {
+      "epoch": 1.3130805283340434,
+      "grad_norm": 14.875,
+      "learning_rate": 6.648427148916561e-07,
+      "loss": 0.555,
+      "step": 15409
+    },
+    {
+      "epoch": 1.3131657435023434,
+      "grad_norm": 14.625,
+      "learning_rate": 6.646942335791359e-07,
+      "loss": 0.5567,
+      "step": 15410
+    },
+    {
+      "epoch": 1.3132509586706433,
+      "grad_norm": 33.75,
+      "learning_rate": 6.645457628436797e-07,
+      "loss": 0.7929,
+      "step": 15411
+    },
+    {
+      "epoch": 1.3133361738389433,
+      "grad_norm": 12.25,
+      "learning_rate": 6.643973026879697e-07,
+      "loss": 0.9013,
+      "step": 15412
+    },
+    {
+      "epoch": 1.3134213890072433,
+      "grad_norm": 17.0,
+      "learning_rate": 6.642488531146891e-07,
+      "loss": 0.8858,
+      "step": 15413
+    },
+    {
+      "epoch": 1.3135066041755432,
+      "grad_norm": 14.0,
+      "learning_rate": 6.641004141265208e-07,
+      "loss": 0.7271,
+      "step": 15414
+    },
+    {
+      "epoch": 1.3135918193438432,
+      "grad_norm": 23.25,
+      "learning_rate": 6.63951985726148e-07,
+      "loss": 1.1887,
+      "step": 15415
+    },
+    {
+      "epoch": 1.3136770345121431,
+      "grad_norm": 17.75,
+      "learning_rate": 6.638035679162513e-07,
+      "loss": 0.6856,
+      "step": 15416
+    },
+    {
+      "epoch": 1.313762249680443,
+      "grad_norm": 15.75,
+      "learning_rate": 6.636551606995137e-07,
+      "loss": 0.8047,
+      "step": 15417
+    },
+    {
+      "epoch": 1.313847464848743,
+      "grad_norm": 17.375,
+      "learning_rate": 6.635067640786171e-07,
+      "loss": 0.7164,
+      "step": 15418
+    },
+    {
+      "epoch": 1.313932680017043,
+      "grad_norm": 24.0,
+      "learning_rate": 6.633583780562433e-07,
+      "loss": 0.7173,
+      "step": 15419
+    },
+    {
+      "epoch": 1.314017895185343,
+      "grad_norm": 18.0,
+      "learning_rate": 6.632100026350738e-07,
+      "loss": 0.811,
+      "step": 15420
+    },
+    {
+      "epoch": 1.314103110353643,
+      "grad_norm": 13.5,
+      "learning_rate": 6.630616378177889e-07,
+      "loss": 0.3032,
+      "step": 15421
+    },
+    {
+      "epoch": 1.3141883255219429,
+      "grad_norm": 19.0,
+      "learning_rate": 6.629132836070704e-07,
+      "loss": 0.9835,
+      "step": 15422
+    },
+    {
+      "epoch": 1.3142735406902428,
+      "grad_norm": 13.5,
+      "learning_rate": 6.627649400055994e-07,
+      "loss": 0.5907,
+      "step": 15423
+    },
+    {
+      "epoch": 1.3143587558585428,
+      "grad_norm": 16.75,
+      "learning_rate": 6.626166070160568e-07,
+      "loss": 0.5778,
+      "step": 15424
+    },
+    {
+      "epoch": 1.3144439710268427,
+      "grad_norm": 16.25,
+      "learning_rate": 6.624682846411224e-07,
+      "loss": 0.8411,
+      "step": 15425
+    },
+    {
+      "epoch": 1.3145291861951427,
+      "grad_norm": 10.0625,
+      "learning_rate": 6.623199728834767e-07,
+      "loss": 0.3694,
+      "step": 15426
+    },
+    {
+      "epoch": 1.3146144013634427,
+      "grad_norm": 14.5,
+      "learning_rate": 6.621716717458007e-07,
+      "loss": 0.5971,
+      "step": 15427
+    },
+    {
+      "epoch": 1.3146996165317426,
+      "grad_norm": 12.6875,
+      "learning_rate": 6.620233812307736e-07,
+      "loss": 0.3018,
+      "step": 15428
+    },
+    {
+      "epoch": 1.3147848317000426,
+      "grad_norm": 14.625,
+      "learning_rate": 6.618751013410748e-07,
+      "loss": 0.6595,
+      "step": 15429
+    },
+    {
+      "epoch": 1.3148700468683425,
+      "grad_norm": 22.25,
+      "learning_rate": 6.617268320793844e-07,
+      "loss": 0.5228,
+      "step": 15430
+    },
+    {
+      "epoch": 1.3149552620366425,
+      "grad_norm": 21.25,
+      "learning_rate": 6.615785734483818e-07,
+      "loss": 0.6278,
+      "step": 15431
+    },
+    {
+      "epoch": 1.3150404772049424,
+      "grad_norm": 20.5,
+      "learning_rate": 6.614303254507464e-07,
+      "loss": 0.7388,
+      "step": 15432
+    },
+    {
+      "epoch": 1.3151256923732424,
+      "grad_norm": 17.625,
+      "learning_rate": 6.612820880891563e-07,
+      "loss": 0.5935,
+      "step": 15433
+    },
+    {
+      "epoch": 1.3152109075415424,
+      "grad_norm": 16.0,
+      "learning_rate": 6.61133861366291e-07,
+      "loss": 0.505,
+      "step": 15434
+    },
+    {
+      "epoch": 1.3152961227098423,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.60985645284829e-07,
+      "loss": 0.5201,
+      "step": 15435
+    },
+    {
+      "epoch": 1.3153813378781423,
+      "grad_norm": 16.5,
+      "learning_rate": 6.60837439847449e-07,
+      "loss": 0.4904,
+      "step": 15436
+    },
+    {
+      "epoch": 1.3154665530464422,
+      "grad_norm": 18.625,
+      "learning_rate": 6.606892450568292e-07,
+      "loss": 0.6562,
+      "step": 15437
+    },
+    {
+      "epoch": 1.3155517682147422,
+      "grad_norm": 24.0,
+      "learning_rate": 6.605410609156465e-07,
+      "loss": 0.7016,
+      "step": 15438
+    },
+    {
+      "epoch": 1.3156369833830421,
+      "grad_norm": 16.5,
+      "learning_rate": 6.603928874265799e-07,
+      "loss": 0.4223,
+      "step": 15439
+    },
+    {
+      "epoch": 1.315722198551342,
+      "grad_norm": 25.5,
+      "learning_rate": 6.60244724592307e-07,
+      "loss": 0.5433,
+      "step": 15440
+    },
+    {
+      "epoch": 1.315807413719642,
+      "grad_norm": 12.8125,
+      "learning_rate": 6.600965724155045e-07,
+      "loss": 0.3768,
+      "step": 15441
+    },
+    {
+      "epoch": 1.315892628887942,
+      "grad_norm": 14.5,
+      "learning_rate": 6.599484308988502e-07,
+      "loss": 0.5223,
+      "step": 15442
+    },
+    {
+      "epoch": 1.315977844056242,
+      "grad_norm": 18.0,
+      "learning_rate": 6.598003000450212e-07,
+      "loss": 0.7828,
+      "step": 15443
+    },
+    {
+      "epoch": 1.316063059224542,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.596521798566946e-07,
+      "loss": 0.5904,
+      "step": 15444
+    },
+    {
+      "epoch": 1.3161482743928419,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.595040703365469e-07,
+      "loss": 0.6069,
+      "step": 15445
+    },
+    {
+      "epoch": 1.3162334895611418,
+      "grad_norm": 24.375,
+      "learning_rate": 6.593559714872538e-07,
+      "loss": 0.8387,
+      "step": 15446
+    },
+    {
+      "epoch": 1.3163187047294418,
+      "grad_norm": 21.25,
+      "learning_rate": 6.592078833114922e-07,
+      "loss": 0.5262,
+      "step": 15447
+    },
+    {
+      "epoch": 1.3164039198977417,
+      "grad_norm": 11.1875,
+      "learning_rate": 6.590598058119382e-07,
+      "loss": 0.3016,
+      "step": 15448
+    },
+    {
+      "epoch": 1.3164891350660417,
+      "grad_norm": 14.5,
+      "learning_rate": 6.589117389912684e-07,
+      "loss": 0.5915,
+      "step": 15449
+    },
+    {
+      "epoch": 1.3165743502343417,
+      "grad_norm": 19.125,
+      "learning_rate": 6.587636828521574e-07,
+      "loss": 0.5429,
+      "step": 15450
+    },
+    {
+      "epoch": 1.3166595654026416,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.58615637397281e-07,
+      "loss": 0.3568,
+      "step": 15451
+    },
+    {
+      "epoch": 1.3167447805709416,
+      "grad_norm": 22.25,
+      "learning_rate": 6.584676026293154e-07,
+      "loss": 0.505,
+      "step": 15452
+    },
+    {
+      "epoch": 1.3168299957392415,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.583195785509345e-07,
+      "loss": 0.6568,
+      "step": 15453
+    },
+    {
+      "epoch": 1.3169152109075415,
+      "grad_norm": 16.0,
+      "learning_rate": 6.581715651648142e-07,
+      "loss": 0.7261,
+      "step": 15454
+    },
+    {
+      "epoch": 1.3170004260758414,
+      "grad_norm": 11.0625,
+      "learning_rate": 6.580235624736283e-07,
+      "loss": 0.4172,
+      "step": 15455
+    },
+    {
+      "epoch": 1.3170856412441414,
+      "grad_norm": 16.125,
+      "learning_rate": 6.578755704800521e-07,
+      "loss": 0.6075,
+      "step": 15456
+    },
+    {
+      "epoch": 1.3171708564124414,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.577275891867602e-07,
+      "loss": 0.5733,
+      "step": 15457
+    },
+    {
+      "epoch": 1.3172560715807413,
+      "grad_norm": 15.875,
+      "learning_rate": 6.575796185964259e-07,
+      "loss": 0.6118,
+      "step": 15458
+    },
+    {
+      "epoch": 1.3173412867490413,
+      "grad_norm": 15.0,
+      "learning_rate": 6.574316587117236e-07,
+      "loss": 0.652,
+      "step": 15459
+    },
+    {
+      "epoch": 1.3174265019173412,
+      "grad_norm": 16.5,
+      "learning_rate": 6.572837095353274e-07,
+      "loss": 0.7427,
+      "step": 15460
+    },
+    {
+      "epoch": 1.3175117170856412,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.571357710699103e-07,
+      "loss": 0.6524,
+      "step": 15461
+    },
+    {
+      "epoch": 1.3175969322539411,
+      "grad_norm": 17.5,
+      "learning_rate": 6.569878433181465e-07,
+      "loss": 0.6554,
+      "step": 15462
+    },
+    {
+      "epoch": 1.317682147422241,
+      "grad_norm": 16.375,
+      "learning_rate": 6.568399262827079e-07,
+      "loss": 0.7124,
+      "step": 15463
+    },
+    {
+      "epoch": 1.317767362590541,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.566920199662686e-07,
+      "loss": 0.3508,
+      "step": 15464
+    },
+    {
+      "epoch": 1.317852577758841,
+      "grad_norm": 24.75,
+      "learning_rate": 6.565441243715011e-07,
+      "loss": 0.9498,
+      "step": 15465
+    },
+    {
+      "epoch": 1.317937792927141,
+      "grad_norm": 29.875,
+      "learning_rate": 6.563962395010784e-07,
+      "loss": 1.5344,
+      "step": 15466
+    },
+    {
+      "epoch": 1.318023008095441,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.56248365357672e-07,
+      "loss": 0.6284,
+      "step": 15467
+    },
+    {
+      "epoch": 1.3181082232637409,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.561005019439555e-07,
+      "loss": 0.5028,
+      "step": 15468
+    },
+    {
+      "epoch": 1.3181934384320408,
+      "grad_norm": 31.75,
+      "learning_rate": 6.559526492625995e-07,
+      "loss": 0.9941,
+      "step": 15469
+    },
+    {
+      "epoch": 1.3182786536003408,
+      "grad_norm": 12.5,
+      "learning_rate": 6.558048073162769e-07,
+      "loss": 0.4775,
+      "step": 15470
+    },
+    {
+      "epoch": 1.3183638687686408,
+      "grad_norm": 12.5625,
+      "learning_rate": 6.556569761076585e-07,
+      "loss": 0.3441,
+      "step": 15471
+    },
+    {
+      "epoch": 1.3184490839369407,
+      "grad_norm": 15.625,
+      "learning_rate": 6.555091556394163e-07,
+      "loss": 0.7119,
+      "step": 15472
+    },
+    {
+      "epoch": 1.3185342991052407,
+      "grad_norm": 16.25,
+      "learning_rate": 6.553613459142214e-07,
+      "loss": 0.6743,
+      "step": 15473
+    },
+    {
+      "epoch": 1.3186195142735406,
+      "grad_norm": 17.125,
+      "learning_rate": 6.552135469347453e-07,
+      "loss": 0.8856,
+      "step": 15474
+    },
+    {
+      "epoch": 1.3187047294418406,
+      "grad_norm": 20.375,
+      "learning_rate": 6.550657587036581e-07,
+      "loss": 0.7738,
+      "step": 15475
+    },
+    {
+      "epoch": 1.3187899446101405,
+      "grad_norm": 21.5,
+      "learning_rate": 6.549179812236314e-07,
+      "loss": 0.7166,
+      "step": 15476
+    },
+    {
+      "epoch": 1.3188751597784405,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.547702144973348e-07,
+      "loss": 0.6175,
+      "step": 15477
+    },
+    {
+      "epoch": 1.3189603749467405,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.546224585274388e-07,
+      "loss": 0.3014,
+      "step": 15478
+    },
+    {
+      "epoch": 1.3190455901150404,
+      "grad_norm": 13.125,
+      "learning_rate": 6.544747133166144e-07,
+      "loss": 0.5052,
+      "step": 15479
+    },
+    {
+      "epoch": 1.3191308052833404,
+      "grad_norm": 10.8125,
+      "learning_rate": 6.543269788675299e-07,
+      "loss": 0.5391,
+      "step": 15480
+    },
+    {
+      "epoch": 1.3192160204516403,
+      "grad_norm": 14.4375,
+      "learning_rate": 6.541792551828561e-07,
+      "loss": 0.516,
+      "step": 15481
+    },
+    {
+      "epoch": 1.3193012356199403,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.540315422652624e-07,
+      "loss": 0.6095,
+      "step": 15482
+    },
+    {
+      "epoch": 1.3193864507882402,
+      "grad_norm": 16.25,
+      "learning_rate": 6.538838401174183e-07,
+      "loss": 0.4185,
+      "step": 15483
+    },
+    {
+      "epoch": 1.3194716659565402,
+      "grad_norm": 14.125,
+      "learning_rate": 6.537361487419928e-07,
+      "loss": 0.5646,
+      "step": 15484
+    },
+    {
+      "epoch": 1.3195568811248402,
+      "grad_norm": 17.5,
+      "learning_rate": 6.535884681416542e-07,
+      "loss": 0.6418,
+      "step": 15485
+    },
+    {
+      "epoch": 1.31964209629314,
+      "grad_norm": 17.125,
+      "learning_rate": 6.534407983190717e-07,
+      "loss": 0.8113,
+      "step": 15486
+    },
+    {
+      "epoch": 1.31972731146144,
+      "grad_norm": 21.75,
+      "learning_rate": 6.532931392769143e-07,
+      "loss": 0.8686,
+      "step": 15487
+    },
+    {
+      "epoch": 1.31981252662974,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.531454910178494e-07,
+      "loss": 0.6186,
+      "step": 15488
+    },
+    {
+      "epoch": 1.31989774179804,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.529978535445458e-07,
+      "loss": 0.4389,
+      "step": 15489
+    },
+    {
+      "epoch": 1.31998295696634,
+      "grad_norm": 16.5,
+      "learning_rate": 6.528502268596712e-07,
+      "loss": 0.6286,
+      "step": 15490
+    },
+    {
+      "epoch": 1.32006817213464,
+      "grad_norm": 16.0,
+      "learning_rate": 6.527026109658941e-07,
+      "loss": 0.4586,
+      "step": 15491
+    },
+    {
+      "epoch": 1.3201533873029399,
+      "grad_norm": 10.75,
+      "learning_rate": 6.525550058658813e-07,
+      "loss": 0.35,
+      "step": 15492
+    },
+    {
+      "epoch": 1.3202386024712398,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.524074115622998e-07,
+      "loss": 0.5575,
+      "step": 15493
+    },
+    {
+      "epoch": 1.3203238176395398,
+      "grad_norm": 11.6875,
+      "learning_rate": 6.522598280578174e-07,
+      "loss": 0.3634,
+      "step": 15494
+    },
+    {
+      "epoch": 1.3204090328078397,
+      "grad_norm": 13.8125,
+      "learning_rate": 6.521122553551009e-07,
+      "loss": 0.3248,
+      "step": 15495
+    },
+    {
+      "epoch": 1.3204942479761397,
+      "grad_norm": 20.5,
+      "learning_rate": 6.519646934568177e-07,
+      "loss": 0.8944,
+      "step": 15496
+    },
+    {
+      "epoch": 1.3205794631444396,
+      "grad_norm": 17.0,
+      "learning_rate": 6.518171423656336e-07,
+      "loss": 0.6229,
+      "step": 15497
+    },
+    {
+      "epoch": 1.3206646783127396,
+      "grad_norm": 13.625,
+      "learning_rate": 6.516696020842151e-07,
+      "loss": 0.5192,
+      "step": 15498
+    },
+    {
+      "epoch": 1.3207498934810395,
+      "grad_norm": 20.5,
+      "learning_rate": 6.515220726152284e-07,
+      "loss": 1.0354,
+      "step": 15499
+    },
+    {
+      "epoch": 1.3208351086493395,
+      "grad_norm": 25.625,
+      "learning_rate": 6.513745539613403e-07,
+      "loss": 0.7525,
+      "step": 15500
+    },
+    {
+      "epoch": 1.3209203238176395,
+      "grad_norm": 14.8125,
+      "learning_rate": 6.512270461252158e-07,
+      "loss": 0.5107,
+      "step": 15501
+    },
+    {
+      "epoch": 1.3210055389859394,
+      "grad_norm": 17.625,
+      "learning_rate": 6.510795491095203e-07,
+      "loss": 0.8479,
+      "step": 15502
+    },
+    {
+      "epoch": 1.3210907541542394,
+      "grad_norm": 12.375,
+      "learning_rate": 6.509320629169197e-07,
+      "loss": 0.4749,
+      "step": 15503
+    },
+    {
+      "epoch": 1.3211759693225393,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.507845875500796e-07,
+      "loss": 0.3673,
+      "step": 15504
+    },
+    {
+      "epoch": 1.3212611844908393,
+      "grad_norm": 14.8125,
+      "learning_rate": 6.50637123011664e-07,
+      "loss": 0.4624,
+      "step": 15505
+    },
+    {
+      "epoch": 1.3213463996591392,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.504896693043384e-07,
+      "loss": 0.6702,
+      "step": 15506
+    },
+    {
+      "epoch": 1.3214316148274392,
+      "grad_norm": 19.25,
+      "learning_rate": 6.503422264307673e-07,
+      "loss": 0.8024,
+      "step": 15507
+    },
+    {
+      "epoch": 1.3215168299957392,
+      "grad_norm": 15.375,
+      "learning_rate": 6.501947943936156e-07,
+      "loss": 0.6587,
+      "step": 15508
+    },
+    {
+      "epoch": 1.3216020451640391,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.50047373195547e-07,
+      "loss": 0.4783,
+      "step": 15509
+    },
+    {
+      "epoch": 1.321687260332339,
+      "grad_norm": 14.5,
+      "learning_rate": 6.498999628392254e-07,
+      "loss": 0.7538,
+      "step": 15510
+    },
+    {
+      "epoch": 1.321772475500639,
+      "grad_norm": 12.5625,
+      "learning_rate": 6.497525633273148e-07,
+      "loss": 0.4592,
+      "step": 15511
+    },
+    {
+      "epoch": 1.321857690668939,
+      "grad_norm": 9.5625,
+      "learning_rate": 6.49605174662479e-07,
+      "loss": 0.17,
+      "step": 15512
+    },
+    {
+      "epoch": 1.321942905837239,
+      "grad_norm": 18.75,
+      "learning_rate": 6.49457796847382e-07,
+      "loss": 0.5805,
+      "step": 15513
+    },
+    {
+      "epoch": 1.322028121005539,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.493104298846859e-07,
+      "loss": 0.688,
+      "step": 15514
+    },
+    {
+      "epoch": 1.3221133361738389,
+      "grad_norm": 13.0625,
+      "learning_rate": 6.491630737770543e-07,
+      "loss": 0.4344,
+      "step": 15515
+    },
+    {
+      "epoch": 1.3221985513421388,
+      "grad_norm": 14.25,
+      "learning_rate": 6.490157285271504e-07,
+      "loss": 0.4381,
+      "step": 15516
+    },
+    {
+      "epoch": 1.3222837665104388,
+      "grad_norm": 19.125,
+      "learning_rate": 6.488683941376375e-07,
+      "loss": 0.8797,
+      "step": 15517
+    },
+    {
+      "epoch": 1.3223689816787387,
+      "grad_norm": 12.1875,
+      "learning_rate": 6.487210706111761e-07,
+      "loss": 0.377,
+      "step": 15518
+    },
+    {
+      "epoch": 1.3224541968470387,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.485737579504297e-07,
+      "loss": 0.6436,
+      "step": 15519
+    },
+    {
+      "epoch": 1.3225394120153386,
+      "grad_norm": 17.375,
+      "learning_rate": 6.484264561580603e-07,
+      "loss": 0.7755,
+      "step": 15520
+    },
+    {
+      "epoch": 1.3226246271836386,
+      "grad_norm": 14.875,
+      "learning_rate": 6.482791652367302e-07,
+      "loss": 0.7225,
+      "step": 15521
+    },
+    {
+      "epoch": 1.3227098423519386,
+      "grad_norm": 17.75,
+      "learning_rate": 6.481318851891002e-07,
+      "loss": 0.6543,
+      "step": 15522
+    },
+    {
+      "epoch": 1.3227950575202385,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.479846160178321e-07,
+      "loss": 0.6054,
+      "step": 15523
+    },
+    {
+      "epoch": 1.3228802726885385,
+      "grad_norm": 12.5,
+      "learning_rate": 6.478373577255875e-07,
+      "loss": 0.374,
+      "step": 15524
+    },
+    {
+      "epoch": 1.3229654878568384,
+      "grad_norm": 14.25,
+      "learning_rate": 6.476901103150279e-07,
+      "loss": 0.6184,
+      "step": 15525
+    },
+    {
+      "epoch": 1.3230507030251384,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.475428737888137e-07,
+      "loss": 0.7745,
+      "step": 15526
+    },
+    {
+      "epoch": 1.3231359181934383,
+      "grad_norm": 25.5,
+      "learning_rate": 6.473956481496049e-07,
+      "loss": 0.6662,
+      "step": 15527
+    },
+    {
+      "epoch": 1.3232211333617383,
+      "grad_norm": 12.875,
+      "learning_rate": 6.472484334000629e-07,
+      "loss": 0.3165,
+      "step": 15528
+    },
+    {
+      "epoch": 1.3233063485300383,
+      "grad_norm": 16.125,
+      "learning_rate": 6.471012295428478e-07,
+      "loss": 0.5944,
+      "step": 15529
+    },
+    {
+      "epoch": 1.3233915636983382,
+      "grad_norm": 14.375,
+      "learning_rate": 6.469540365806202e-07,
+      "loss": 0.3362,
+      "step": 15530
+    },
+    {
+      "epoch": 1.3234767788666382,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.468068545160391e-07,
+      "loss": 0.4027,
+      "step": 15531
+    },
+    {
+      "epoch": 1.3235619940349381,
+      "grad_norm": 14.4375,
+      "learning_rate": 6.466596833517647e-07,
+      "loss": 0.6648,
+      "step": 15532
+    },
+    {
+      "epoch": 1.323647209203238,
+      "grad_norm": 18.5,
+      "learning_rate": 6.465125230904568e-07,
+      "loss": 0.4131,
+      "step": 15533
+    },
+    {
+      "epoch": 1.323732424371538,
+      "grad_norm": 15.75,
+      "learning_rate": 6.463653737347748e-07,
+      "loss": 0.6232,
+      "step": 15534
+    },
+    {
+      "epoch": 1.323817639539838,
+      "grad_norm": 16.5,
+      "learning_rate": 6.462182352873768e-07,
+      "loss": 0.2563,
+      "step": 15535
+    },
+    {
+      "epoch": 1.323902854708138,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.460711077509224e-07,
+      "loss": 1.2542,
+      "step": 15536
+    },
+    {
+      "epoch": 1.323988069876438,
+      "grad_norm": 16.5,
+      "learning_rate": 6.459239911280706e-07,
+      "loss": 0.424,
+      "step": 15537
+    },
+    {
+      "epoch": 1.3240732850447379,
+      "grad_norm": 10.8125,
+      "learning_rate": 6.4577688542148e-07,
+      "loss": 0.3378,
+      "step": 15538
+    },
+    {
+      "epoch": 1.3241585002130378,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.456297906338083e-07,
+      "loss": 0.387,
+      "step": 15539
+    },
+    {
+      "epoch": 1.3242437153813378,
+      "grad_norm": 12.1875,
+      "learning_rate": 6.45482706767714e-07,
+      "loss": 0.4798,
+      "step": 15540
+    },
+    {
+      "epoch": 1.3243289305496377,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.453356338258555e-07,
+      "loss": 0.4521,
+      "step": 15541
+    },
+    {
+      "epoch": 1.3244141457179377,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.451885718108897e-07,
+      "loss": 0.3642,
+      "step": 15542
+    },
+    {
+      "epoch": 1.3244993608862377,
+      "grad_norm": 19.0,
+      "learning_rate": 6.450415207254751e-07,
+      "loss": 0.4517,
+      "step": 15543
+    },
+    {
+      "epoch": 1.3245845760545376,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.448944805722682e-07,
+      "loss": 0.3924,
+      "step": 15544
+    },
+    {
+      "epoch": 1.3246697912228376,
+      "grad_norm": 22.75,
+      "learning_rate": 6.447474513539262e-07,
+      "loss": 0.7529,
+      "step": 15545
+    },
+    {
+      "epoch": 1.3247550063911375,
+      "grad_norm": 23.0,
+      "learning_rate": 6.446004330731066e-07,
+      "loss": 0.9091,
+      "step": 15546
+    },
+    {
+      "epoch": 1.3248402215594375,
+      "grad_norm": 20.5,
+      "learning_rate": 6.444534257324664e-07,
+      "loss": 0.8651,
+      "step": 15547
+    },
+    {
+      "epoch": 1.3249254367277374,
+      "grad_norm": 11.75,
+      "learning_rate": 6.443064293346613e-07,
+      "loss": 0.4996,
+      "step": 15548
+    },
+    {
+      "epoch": 1.3250106518960374,
+      "grad_norm": 19.0,
+      "learning_rate": 6.441594438823485e-07,
+      "loss": 0.6837,
+      "step": 15549
+    },
+    {
+      "epoch": 1.3250958670643374,
+      "grad_norm": 22.25,
+      "learning_rate": 6.440124693781833e-07,
+      "loss": 0.6429,
+      "step": 15550
+    },
+    {
+      "epoch": 1.3251810822326373,
+      "grad_norm": 17.375,
+      "learning_rate": 6.438655058248225e-07,
+      "loss": 0.8109,
+      "step": 15551
+    },
+    {
+      "epoch": 1.3252662974009373,
+      "grad_norm": 15.125,
+      "learning_rate": 6.437185532249213e-07,
+      "loss": 0.6294,
+      "step": 15552
+    },
+    {
+      "epoch": 1.3253515125692372,
+      "grad_norm": 16.625,
+      "learning_rate": 6.435716115811352e-07,
+      "loss": 0.9326,
+      "step": 15553
+    },
+    {
+      "epoch": 1.3254367277375372,
+      "grad_norm": 17.125,
+      "learning_rate": 6.4342468089612e-07,
+      "loss": 0.6685,
+      "step": 15554
+    },
+    {
+      "epoch": 1.3255219429058371,
+      "grad_norm": 14.125,
+      "learning_rate": 6.432777611725315e-07,
+      "loss": 0.4013,
+      "step": 15555
+    },
+    {
+      "epoch": 1.325607158074137,
+      "grad_norm": 15.25,
+      "learning_rate": 6.431308524130233e-07,
+      "loss": 0.4833,
+      "step": 15556
+    },
+    {
+      "epoch": 1.325692373242437,
+      "grad_norm": 15.125,
+      "learning_rate": 6.429839546202513e-07,
+      "loss": 0.5507,
+      "step": 15557
+    },
+    {
+      "epoch": 1.325777588410737,
+      "grad_norm": 15.625,
+      "learning_rate": 6.428370677968691e-07,
+      "loss": 0.725,
+      "step": 15558
+    },
+    {
+      "epoch": 1.325862803579037,
+      "grad_norm": 15.5,
+      "learning_rate": 6.426901919455317e-07,
+      "loss": 0.3917,
+      "step": 15559
+    },
+    {
+      "epoch": 1.325948018747337,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.425433270688939e-07,
+      "loss": 0.5027,
+      "step": 15560
+    },
+    {
+      "epoch": 1.3260332339156369,
+      "grad_norm": 25.75,
+      "learning_rate": 6.423964731696084e-07,
+      "loss": 0.6568,
+      "step": 15561
+    },
+    {
+      "epoch": 1.3261184490839368,
+      "grad_norm": 14.5,
+      "learning_rate": 6.422496302503295e-07,
+      "loss": 0.3881,
+      "step": 15562
+    },
+    {
+      "epoch": 1.3262036642522368,
+      "grad_norm": 16.875,
+      "learning_rate": 6.421027983137115e-07,
+      "loss": 0.631,
+      "step": 15563
+    },
+    {
+      "epoch": 1.3262888794205367,
+      "grad_norm": 21.375,
+      "learning_rate": 6.419559773624068e-07,
+      "loss": 0.8545,
+      "step": 15564
+    },
+    {
+      "epoch": 1.3263740945888367,
+      "grad_norm": 35.5,
+      "learning_rate": 6.418091673990692e-07,
+      "loss": 1.1535,
+      "step": 15565
+    },
+    {
+      "epoch": 1.3264593097571367,
+      "grad_norm": 19.375,
+      "learning_rate": 6.416623684263515e-07,
+      "loss": 0.7635,
+      "step": 15566
+    },
+    {
+      "epoch": 1.3265445249254366,
+      "grad_norm": 18.75,
+      "learning_rate": 6.415155804469061e-07,
+      "loss": 0.6744,
+      "step": 15567
+    },
+    {
+      "epoch": 1.3266297400937366,
+      "grad_norm": 13.1875,
+      "learning_rate": 6.413688034633869e-07,
+      "loss": 0.5515,
+      "step": 15568
+    },
+    {
+      "epoch": 1.3267149552620365,
+      "grad_norm": 31.5,
+      "learning_rate": 6.412220374784448e-07,
+      "loss": 0.7286,
+      "step": 15569
+    },
+    {
+      "epoch": 1.3268001704303365,
+      "grad_norm": 73.5,
+      "learning_rate": 6.410752824947327e-07,
+      "loss": 0.3932,
+      "step": 15570
+    },
+    {
+      "epoch": 1.3268853855986364,
+      "grad_norm": 14.875,
+      "learning_rate": 6.409285385149025e-07,
+      "loss": 0.3961,
+      "step": 15571
+    },
+    {
+      "epoch": 1.3269706007669364,
+      "grad_norm": 18.625,
+      "learning_rate": 6.407818055416066e-07,
+      "loss": 0.6192,
+      "step": 15572
+    },
+    {
+      "epoch": 1.3270558159352364,
+      "grad_norm": 18.0,
+      "learning_rate": 6.406350835774956e-07,
+      "loss": 0.8179,
+      "step": 15573
+    },
+    {
+      "epoch": 1.3271410311035363,
+      "grad_norm": 15.125,
+      "learning_rate": 6.404883726252219e-07,
+      "loss": 0.5299,
+      "step": 15574
+    },
+    {
+      "epoch": 1.3272262462718363,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.403416726874356e-07,
+      "loss": 0.6936,
+      "step": 15575
+    },
+    {
+      "epoch": 1.3273114614401362,
+      "grad_norm": 16.25,
+      "learning_rate": 6.401949837667886e-07,
+      "loss": 0.8529,
+      "step": 15576
+    },
+    {
+      "epoch": 1.3273966766084362,
+      "grad_norm": 15.375,
+      "learning_rate": 6.400483058659319e-07,
+      "loss": 0.7732,
+      "step": 15577
+    },
+    {
+      "epoch": 1.3274818917767361,
+      "grad_norm": 15.5,
+      "learning_rate": 6.399016389875149e-07,
+      "loss": 0.5452,
+      "step": 15578
+    },
+    {
+      "epoch": 1.327567106945036,
+      "grad_norm": 13.5,
+      "learning_rate": 6.39754983134189e-07,
+      "loss": 0.3446,
+      "step": 15579
+    },
+    {
+      "epoch": 1.327652322113336,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.396083383086046e-07,
+      "loss": 0.31,
+      "step": 15580
+    },
+    {
+      "epoch": 1.327737537281636,
+      "grad_norm": 15.25,
+      "learning_rate": 6.394617045134109e-07,
+      "loss": 0.5258,
+      "step": 15581
+    },
+    {
+      "epoch": 1.327822752449936,
+      "grad_norm": 16.125,
+      "learning_rate": 6.393150817512586e-07,
+      "loss": 0.6396,
+      "step": 15582
+    },
+    {
+      "epoch": 1.327907967618236,
+      "grad_norm": 16.25,
+      "learning_rate": 6.391684700247963e-07,
+      "loss": 0.7739,
+      "step": 15583
+    },
+    {
+      "epoch": 1.3279931827865359,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.390218693366743e-07,
+      "loss": 0.6652,
+      "step": 15584
+    },
+    {
+      "epoch": 1.3280783979548358,
+      "grad_norm": 15.625,
+      "learning_rate": 6.388752796895417e-07,
+      "loss": 0.7387,
+      "step": 15585
+    },
+    {
+      "epoch": 1.3281636131231358,
+      "grad_norm": 16.125,
+      "learning_rate": 6.387287010860468e-07,
+      "loss": 0.7027,
+      "step": 15586
+    },
+    {
+      "epoch": 1.3282488282914358,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.385821335288392e-07,
+      "loss": 0.4567,
+      "step": 15587
+    },
+    {
+      "epoch": 1.3283340434597357,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.38435577020567e-07,
+      "loss": 0.6818,
+      "step": 15588
+    },
+    {
+      "epoch": 1.3284192586280357,
+      "grad_norm": 12.9375,
+      "learning_rate": 6.382890315638796e-07,
+      "loss": 0.4347,
+      "step": 15589
+    },
+    {
+      "epoch": 1.3285044737963356,
+      "grad_norm": 18.625,
+      "learning_rate": 6.381424971614244e-07,
+      "loss": 1.0646,
+      "step": 15590
+    },
+    {
+      "epoch": 1.3285896889646356,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.379959738158492e-07,
+      "loss": 0.728,
+      "step": 15591
+    },
+    {
+      "epoch": 1.3286749041329355,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.37849461529802e-07,
+      "loss": 0.3933,
+      "step": 15592
+    },
+    {
+      "epoch": 1.3287601193012355,
+      "grad_norm": 12.25,
+      "learning_rate": 6.377029603059312e-07,
+      "loss": 0.3987,
+      "step": 15593
+    },
+    {
+      "epoch": 1.3288453344695355,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.375564701468831e-07,
+      "loss": 0.4494,
+      "step": 15594
+    },
+    {
+      "epoch": 1.3289305496378354,
+      "grad_norm": 18.125,
+      "learning_rate": 6.374099910553054e-07,
+      "loss": 0.6786,
+      "step": 15595
+    },
+    {
+      "epoch": 1.3290157648061354,
+      "grad_norm": 18.5,
+      "learning_rate": 6.372635230338453e-07,
+      "loss": 0.6173,
+      "step": 15596
+    },
+    {
+      "epoch": 1.3291009799744353,
+      "grad_norm": 22.25,
+      "learning_rate": 6.371170660851498e-07,
+      "loss": 0.7306,
+      "step": 15597
+    },
+    {
+      "epoch": 1.3291861951427353,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.369706202118653e-07,
+      "loss": 0.4737,
+      "step": 15598
+    },
+    {
+      "epoch": 1.3292714103110352,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.368241854166377e-07,
+      "loss": 0.4495,
+      "step": 15599
+    },
+    {
+      "epoch": 1.3293566254793352,
+      "grad_norm": 19.75,
+      "learning_rate": 6.366777617021136e-07,
+      "loss": 0.7747,
+      "step": 15600
+    },
+    {
+      "epoch": 1.3294418406476352,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.365313490709391e-07,
+      "loss": 0.7337,
+      "step": 15601
+    },
+    {
+      "epoch": 1.329527055815935,
+      "grad_norm": 12.75,
+      "learning_rate": 6.363849475257605e-07,
+      "loss": 0.4389,
+      "step": 15602
+    },
+    {
+      "epoch": 1.329612270984235,
+      "grad_norm": 17.125,
+      "learning_rate": 6.362385570692225e-07,
+      "loss": 0.6661,
+      "step": 15603
+    },
+    {
+      "epoch": 1.329697486152535,
+      "grad_norm": 22.25,
+      "learning_rate": 6.360921777039709e-07,
+      "loss": 0.7444,
+      "step": 15604
+    },
+    {
+      "epoch": 1.329782701320835,
+      "grad_norm": 20.625,
+      "learning_rate": 6.35945809432651e-07,
+      "loss": 0.8543,
+      "step": 15605
+    },
+    {
+      "epoch": 1.329867916489135,
+      "grad_norm": 10.8125,
+      "learning_rate": 6.357994522579083e-07,
+      "loss": 0.3088,
+      "step": 15606
+    },
+    {
+      "epoch": 1.329953131657435,
+      "grad_norm": 13.8125,
+      "learning_rate": 6.35653106182387e-07,
+      "loss": 0.5507,
+      "step": 15607
+    },
+    {
+      "epoch": 1.3300383468257349,
+      "grad_norm": 14.0,
+      "learning_rate": 6.355067712087315e-07,
+      "loss": 0.404,
+      "step": 15608
+    },
+    {
+      "epoch": 1.3301235619940348,
+      "grad_norm": 12.9375,
+      "learning_rate": 6.353604473395866e-07,
+      "loss": 0.3057,
+      "step": 15609
+    },
+    {
+      "epoch": 1.3302087771623348,
+      "grad_norm": 13.1875,
+      "learning_rate": 6.35214134577597e-07,
+      "loss": 0.4884,
+      "step": 15610
+    },
+    {
+      "epoch": 1.3302939923306347,
+      "grad_norm": 13.75,
+      "learning_rate": 6.350678329254057e-07,
+      "loss": 0.4596,
+      "step": 15611
+    },
+    {
+      "epoch": 1.3303792074989347,
+      "grad_norm": 19.375,
+      "learning_rate": 6.349215423856568e-07,
+      "loss": 0.7621,
+      "step": 15612
+    },
+    {
+      "epoch": 1.3304644226672346,
+      "grad_norm": 33.0,
+      "learning_rate": 6.347752629609943e-07,
+      "loss": 1.1956,
+      "step": 15613
+    },
+    {
+      "epoch": 1.3305496378355346,
+      "grad_norm": 16.5,
+      "learning_rate": 6.34628994654062e-07,
+      "loss": 0.9324,
+      "step": 15614
+    },
+    {
+      "epoch": 1.3306348530038346,
+      "grad_norm": 13.25,
+      "learning_rate": 6.344827374675024e-07,
+      "loss": 0.3379,
+      "step": 15615
+    },
+    {
+      "epoch": 1.3307200681721345,
+      "grad_norm": 22.125,
+      "learning_rate": 6.343364914039583e-07,
+      "loss": 0.7926,
+      "step": 15616
+    },
+    {
+      "epoch": 1.3308052833404345,
+      "grad_norm": 20.5,
+      "learning_rate": 6.341902564660729e-07,
+      "loss": 0.6884,
+      "step": 15617
+    },
+    {
+      "epoch": 1.3308904985087344,
+      "grad_norm": 15.0,
+      "learning_rate": 6.340440326564888e-07,
+      "loss": 0.7279,
+      "step": 15618
+    },
+    {
+      "epoch": 1.3309757136770344,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.338978199778489e-07,
+      "loss": 0.5522,
+      "step": 15619
+    },
+    {
+      "epoch": 1.3310609288453343,
+      "grad_norm": 13.9375,
+      "learning_rate": 6.337516184327944e-07,
+      "loss": 0.4233,
+      "step": 15620
+    },
+    {
+      "epoch": 1.3311461440136343,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.33605428023968e-07,
+      "loss": 0.5019,
+      "step": 15621
+    },
+    {
+      "epoch": 1.3312313591819342,
+      "grad_norm": 18.75,
+      "learning_rate": 6.334592487540117e-07,
+      "loss": 0.8607,
+      "step": 15622
+    },
+    {
+      "epoch": 1.3313165743502342,
+      "grad_norm": 18.625,
+      "learning_rate": 6.333130806255668e-07,
+      "loss": 0.5629,
+      "step": 15623
+    },
+    {
+      "epoch": 1.3314017895185342,
+      "grad_norm": 17.0,
+      "learning_rate": 6.33166923641274e-07,
+      "loss": 0.6582,
+      "step": 15624
+    },
+    {
+      "epoch": 1.3314870046868341,
+      "grad_norm": 11.0,
+      "learning_rate": 6.330207778037754e-07,
+      "loss": 0.354,
+      "step": 15625
+    },
+    {
+      "epoch": 1.331572219855134,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.328746431157118e-07,
+      "loss": 0.6407,
+      "step": 15626
+    },
+    {
+      "epoch": 1.331657435023434,
+      "grad_norm": 18.625,
+      "learning_rate": 6.327285195797242e-07,
+      "loss": 0.6945,
+      "step": 15627
+    },
+    {
+      "epoch": 1.3317426501917342,
+      "grad_norm": 16.25,
+      "learning_rate": 6.325824071984526e-07,
+      "loss": 0.5635,
+      "step": 15628
+    },
+    {
+      "epoch": 1.3318278653600342,
+      "grad_norm": 12.875,
+      "learning_rate": 6.324363059745378e-07,
+      "loss": 0.4307,
+      "step": 15629
+    },
+    {
+      "epoch": 1.3319130805283341,
+      "grad_norm": 16.0,
+      "learning_rate": 6.322902159106205e-07,
+      "loss": 0.8672,
+      "step": 15630
+    },
+    {
+      "epoch": 1.331998295696634,
+      "grad_norm": 15.5,
+      "learning_rate": 6.321441370093396e-07,
+      "loss": 0.7528,
+      "step": 15631
+    },
+    {
+      "epoch": 1.332083510864934,
+      "grad_norm": 17.0,
+      "learning_rate": 6.31998069273336e-07,
+      "loss": 0.6148,
+      "step": 15632
+    },
+    {
+      "epoch": 1.332168726033234,
+      "grad_norm": 12.8125,
+      "learning_rate": 6.318520127052481e-07,
+      "loss": 0.4522,
+      "step": 15633
+    },
+    {
+      "epoch": 1.332253941201534,
+      "grad_norm": 19.0,
+      "learning_rate": 6.317059673077161e-07,
+      "loss": 0.8924,
+      "step": 15634
+    },
+    {
+      "epoch": 1.332339156369834,
+      "grad_norm": 16.5,
+      "learning_rate": 6.315599330833788e-07,
+      "loss": 0.615,
+      "step": 15635
+    },
+    {
+      "epoch": 1.3324243715381339,
+      "grad_norm": 14.125,
+      "learning_rate": 6.314139100348759e-07,
+      "loss": 0.4373,
+      "step": 15636
+    },
+    {
+      "epoch": 1.3325095867064338,
+      "grad_norm": 12.625,
+      "learning_rate": 6.312678981648454e-07,
+      "loss": 0.4595,
+      "step": 15637
+    },
+    {
+      "epoch": 1.3325948018747338,
+      "grad_norm": 12.5,
+      "learning_rate": 6.311218974759262e-07,
+      "loss": 0.3226,
+      "step": 15638
+    },
+    {
+      "epoch": 1.3326800170430337,
+      "grad_norm": 17.5,
+      "learning_rate": 6.30975907970757e-07,
+      "loss": 0.7782,
+      "step": 15639
+    },
+    {
+      "epoch": 1.3327652322113337,
+      "grad_norm": 16.375,
+      "learning_rate": 6.308299296519754e-07,
+      "loss": 0.5207,
+      "step": 15640
+    },
+    {
+      "epoch": 1.3328504473796337,
+      "grad_norm": 15.375,
+      "learning_rate": 6.306839625222194e-07,
+      "loss": 0.4249,
+      "step": 15641
+    },
+    {
+      "epoch": 1.3329356625479336,
+      "grad_norm": 13.75,
+      "learning_rate": 6.305380065841269e-07,
+      "loss": 0.6355,
+      "step": 15642
+    },
+    {
+      "epoch": 1.3330208777162336,
+      "grad_norm": 12.625,
+      "learning_rate": 6.303920618403356e-07,
+      "loss": 0.4853,
+      "step": 15643
+    },
+    {
+      "epoch": 1.3331060928845335,
+      "grad_norm": 26.0,
+      "learning_rate": 6.302461282934832e-07,
+      "loss": 0.9384,
+      "step": 15644
+    },
+    {
+      "epoch": 1.3331913080528335,
+      "grad_norm": 26.125,
+      "learning_rate": 6.301002059462059e-07,
+      "loss": 0.8621,
+      "step": 15645
+    },
+    {
+      "epoch": 1.3332765232211334,
+      "grad_norm": 13.625,
+      "learning_rate": 6.299542948011413e-07,
+      "loss": 0.5483,
+      "step": 15646
+    },
+    {
+      "epoch": 1.3333617383894334,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.298083948609268e-07,
+      "loss": 0.5552,
+      "step": 15647
+    },
+    {
+      "epoch": 1.3334469535577333,
+      "grad_norm": 16.5,
+      "learning_rate": 6.296625061281975e-07,
+      "loss": 0.8563,
+      "step": 15648
+    },
+    {
+      "epoch": 1.3335321687260333,
+      "grad_norm": 10.375,
+      "learning_rate": 6.295166286055912e-07,
+      "loss": 0.2181,
+      "step": 15649
+    },
+    {
+      "epoch": 1.3336173838943333,
+      "grad_norm": 18.5,
+      "learning_rate": 6.293707622957426e-07,
+      "loss": 0.4174,
+      "step": 15650
+    },
+    {
+      "epoch": 1.3337025990626332,
+      "grad_norm": 17.75,
+      "learning_rate": 6.292249072012888e-07,
+      "loss": 0.6716,
+      "step": 15651
+    },
+    {
+      "epoch": 1.3337878142309332,
+      "grad_norm": 13.75,
+      "learning_rate": 6.290790633248648e-07,
+      "loss": 0.4388,
+      "step": 15652
+    },
+    {
+      "epoch": 1.3338730293992331,
+      "grad_norm": 17.25,
+      "learning_rate": 6.289332306691072e-07,
+      "loss": 0.4727,
+      "step": 15653
+    },
+    {
+      "epoch": 1.333958244567533,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.2878740923665e-07,
+      "loss": 0.4555,
+      "step": 15654
+    },
+    {
+      "epoch": 1.334043459735833,
+      "grad_norm": 13.875,
+      "learning_rate": 6.286415990301297e-07,
+      "loss": 0.5666,
+      "step": 15655
+    },
+    {
+      "epoch": 1.334128674904133,
+      "grad_norm": 23.75,
+      "learning_rate": 6.284958000521799e-07,
+      "loss": 1.01,
+      "step": 15656
+    },
+    {
+      "epoch": 1.334213890072433,
+      "grad_norm": 16.875,
+      "learning_rate": 6.283500123054365e-07,
+      "loss": 0.7886,
+      "step": 15657
+    },
+    {
+      "epoch": 1.334299105240733,
+      "grad_norm": 10.9375,
+      "learning_rate": 6.28204235792533e-07,
+      "loss": 0.4274,
+      "step": 15658
+    },
+    {
+      "epoch": 1.3343843204090329,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.280584705161041e-07,
+      "loss": 0.5967,
+      "step": 15659
+    },
+    {
+      "epoch": 1.3344695355773328,
+      "grad_norm": 15.125,
+      "learning_rate": 6.279127164787843e-07,
+      "loss": 0.4639,
+      "step": 15660
+    },
+    {
+      "epoch": 1.3345547507456328,
+      "grad_norm": 33.25,
+      "learning_rate": 6.277669736832076e-07,
+      "loss": 0.9024,
+      "step": 15661
+    },
+    {
+      "epoch": 1.3346399659139327,
+      "grad_norm": 17.75,
+      "learning_rate": 6.276212421320069e-07,
+      "loss": 0.6282,
+      "step": 15662
+    },
+    {
+      "epoch": 1.3347251810822327,
+      "grad_norm": 12.3125,
+      "learning_rate": 6.274755218278167e-07,
+      "loss": 0.3909,
+      "step": 15663
+    },
+    {
+      "epoch": 1.3348103962505327,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.273298127732693e-07,
+      "loss": 0.3907,
+      "step": 15664
+    },
+    {
+      "epoch": 1.3348956114188326,
+      "grad_norm": 18.75,
+      "learning_rate": 6.271841149709984e-07,
+      "loss": 0.4667,
+      "step": 15665
+    },
+    {
+      "epoch": 1.3349808265871326,
+      "grad_norm": 21.75,
+      "learning_rate": 6.270384284236373e-07,
+      "loss": 0.4605,
+      "step": 15666
+    },
+    {
+      "epoch": 1.3350660417554325,
+      "grad_norm": 15.0,
+      "learning_rate": 6.268927531338178e-07,
+      "loss": 0.5187,
+      "step": 15667
+    },
+    {
+      "epoch": 1.3351512569237325,
+      "grad_norm": 16.875,
+      "learning_rate": 6.267470891041727e-07,
+      "loss": 0.9368,
+      "step": 15668
+    },
+    {
+      "epoch": 1.3352364720920324,
+      "grad_norm": 15.25,
+      "learning_rate": 6.266014363373347e-07,
+      "loss": 0.4576,
+      "step": 15669
+    },
+    {
+      "epoch": 1.3353216872603324,
+      "grad_norm": 14.6875,
+      "learning_rate": 6.264557948359358e-07,
+      "loss": 0.4103,
+      "step": 15670
+    },
+    {
+      "epoch": 1.3354069024286324,
+      "grad_norm": 18.0,
+      "learning_rate": 6.263101646026079e-07,
+      "loss": 0.6336,
+      "step": 15671
+    },
+    {
+      "epoch": 1.3354921175969323,
+      "grad_norm": 16.875,
+      "learning_rate": 6.261645456399818e-07,
+      "loss": 0.8661,
+      "step": 15672
+    },
+    {
+      "epoch": 1.3355773327652323,
+      "grad_norm": 19.875,
+      "learning_rate": 6.260189379506898e-07,
+      "loss": 0.576,
+      "step": 15673
+    },
+    {
+      "epoch": 1.3356625479335322,
+      "grad_norm": 19.875,
+      "learning_rate": 6.258733415373636e-07,
+      "loss": 0.969,
+      "step": 15674
+    },
+    {
+      "epoch": 1.3357477631018322,
+      "grad_norm": 10.8125,
+      "learning_rate": 6.257277564026332e-07,
+      "loss": 0.3217,
+      "step": 15675
+    },
+    {
+      "epoch": 1.3358329782701321,
+      "grad_norm": 25.875,
+      "learning_rate": 6.2558218254913e-07,
+      "loss": 1.048,
+      "step": 15676
+    },
+    {
+      "epoch": 1.335918193438432,
+      "grad_norm": 11.0625,
+      "learning_rate": 6.254366199794847e-07,
+      "loss": 0.3052,
+      "step": 15677
+    },
+    {
+      "epoch": 1.336003408606732,
+      "grad_norm": 18.375,
+      "learning_rate": 6.252910686963282e-07,
+      "loss": 0.6116,
+      "step": 15678
+    },
+    {
+      "epoch": 1.336088623775032,
+      "grad_norm": 16.25,
+      "learning_rate": 6.251455287022901e-07,
+      "loss": 0.5934,
+      "step": 15679
+    },
+    {
+      "epoch": 1.336173838943332,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.250000000000004e-07,
+      "loss": 0.5959,
+      "step": 15680
+    },
+    {
+      "epoch": 1.336259054111632,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.248544825920889e-07,
+      "loss": 0.493,
+      "step": 15681
+    },
+    {
+      "epoch": 1.3363442692799319,
+      "grad_norm": 13.375,
+      "learning_rate": 6.247089764811857e-07,
+      "loss": 0.5888,
+      "step": 15682
+    },
+    {
+      "epoch": 1.3364294844482318,
+      "grad_norm": 16.625,
+      "learning_rate": 6.245634816699206e-07,
+      "loss": 0.6976,
+      "step": 15683
+    },
+    {
+      "epoch": 1.3365146996165318,
+      "grad_norm": 15.25,
+      "learning_rate": 6.244179981609217e-07,
+      "loss": 0.9093,
+      "step": 15684
+    },
+    {
+      "epoch": 1.3365999147848318,
+      "grad_norm": 12.625,
+      "learning_rate": 6.242725259568189e-07,
+      "loss": 0.3541,
+      "step": 15685
+    },
+    {
+      "epoch": 1.3366851299531317,
+      "grad_norm": 13.375,
+      "learning_rate": 6.241270650602406e-07,
+      "loss": 0.3877,
+      "step": 15686
+    },
+    {
+      "epoch": 1.3367703451214317,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.239816154738166e-07,
+      "loss": 0.6845,
+      "step": 15687
+    },
+    {
+      "epoch": 1.3368555602897316,
+      "grad_norm": 13.75,
+      "learning_rate": 6.238361772001733e-07,
+      "loss": 0.5618,
+      "step": 15688
+    },
+    {
+      "epoch": 1.3369407754580316,
+      "grad_norm": 15.25,
+      "learning_rate": 6.2369075024194e-07,
+      "loss": 0.6768,
+      "step": 15689
+    },
+    {
+      "epoch": 1.3370259906263315,
+      "grad_norm": 16.5,
+      "learning_rate": 6.235453346017444e-07,
+      "loss": 0.7249,
+      "step": 15690
+    },
+    {
+      "epoch": 1.3371112057946315,
+      "grad_norm": 18.0,
+      "learning_rate": 6.233999302822154e-07,
+      "loss": 0.5653,
+      "step": 15691
+    },
+    {
+      "epoch": 1.3371964209629315,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.232545372859791e-07,
+      "loss": 0.3707,
+      "step": 15692
+    },
+    {
+      "epoch": 1.3372816361312314,
+      "grad_norm": 13.9375,
+      "learning_rate": 6.231091556156634e-07,
+      "loss": 0.5516,
+      "step": 15693
+    },
+    {
+      "epoch": 1.3373668512995314,
+      "grad_norm": 17.375,
+      "learning_rate": 6.22963785273896e-07,
+      "loss": 0.7924,
+      "step": 15694
+    },
+    {
+      "epoch": 1.3374520664678313,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.228184262633037e-07,
+      "loss": 0.5161,
+      "step": 15695
+    },
+    {
+      "epoch": 1.3375372816361313,
+      "grad_norm": 19.125,
+      "learning_rate": 6.226730785865133e-07,
+      "loss": 0.7232,
+      "step": 15696
+    },
+    {
+      "epoch": 1.3376224968044312,
+      "grad_norm": 17.0,
+      "learning_rate": 6.225277422461507e-07,
+      "loss": 0.8308,
+      "step": 15697
+    },
+    {
+      "epoch": 1.3377077119727312,
+      "grad_norm": 11.75,
+      "learning_rate": 6.223824172448427e-07,
+      "loss": 0.4218,
+      "step": 15698
+    },
+    {
+      "epoch": 1.3377929271410312,
+      "grad_norm": 21.125,
+      "learning_rate": 6.222371035852156e-07,
+      "loss": 0.7345,
+      "step": 15699
+    },
+    {
+      "epoch": 1.337878142309331,
+      "grad_norm": 13.125,
+      "learning_rate": 6.220918012698957e-07,
+      "loss": 0.25,
+      "step": 15700
+    },
+    {
+      "epoch": 1.337963357477631,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.219465103015079e-07,
+      "loss": 0.6474,
+      "step": 15701
+    },
+    {
+      "epoch": 1.338048572645931,
+      "grad_norm": 21.125,
+      "learning_rate": 6.218012306826783e-07,
+      "loss": 0.7271,
+      "step": 15702
+    },
+    {
+      "epoch": 1.338133787814231,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.216559624160326e-07,
+      "loss": 0.4292,
+      "step": 15703
+    },
+    {
+      "epoch": 1.338219002982531,
+      "grad_norm": 14.6875,
+      "learning_rate": 6.215107055041955e-07,
+      "loss": 0.4744,
+      "step": 15704
+    },
+    {
+      "epoch": 1.338304218150831,
+      "grad_norm": 21.5,
+      "learning_rate": 6.213654599497914e-07,
+      "loss": 0.7093,
+      "step": 15705
+    },
+    {
+      "epoch": 1.3383894333191309,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.212202257554458e-07,
+      "loss": 0.5499,
+      "step": 15706
+    },
+    {
+      "epoch": 1.3384746484874308,
+      "grad_norm": 19.875,
+      "learning_rate": 6.210750029237827e-07,
+      "loss": 0.9656,
+      "step": 15707
+    },
+    {
+      "epoch": 1.3385598636557308,
+      "grad_norm": 18.5,
+      "learning_rate": 6.209297914574274e-07,
+      "loss": 0.9574,
+      "step": 15708
+    },
+    {
+      "epoch": 1.3386450788240307,
+      "grad_norm": 14.625,
+      "learning_rate": 6.207845913590028e-07,
+      "loss": 0.71,
+      "step": 15709
+    },
+    {
+      "epoch": 1.3387302939923307,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.206394026311334e-07,
+      "loss": 0.5585,
+      "step": 15710
+    },
+    {
+      "epoch": 1.3388155091606306,
+      "grad_norm": 16.25,
+      "learning_rate": 6.204942252764429e-07,
+      "loss": 0.5981,
+      "step": 15711
+    },
+    {
+      "epoch": 1.3389007243289306,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.203490592975552e-07,
+      "loss": 0.7555,
+      "step": 15712
+    },
+    {
+      "epoch": 1.3389859394972305,
+      "grad_norm": 20.625,
+      "learning_rate": 6.202039046970932e-07,
+      "loss": 0.6557,
+      "step": 15713
+    },
+    {
+      "epoch": 1.3390711546655305,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.200587614776795e-07,
+      "loss": 0.5673,
+      "step": 15714
+    },
+    {
+      "epoch": 1.3391563698338305,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.199136296419372e-07,
+      "loss": 0.3766,
+      "step": 15715
+    },
+    {
+      "epoch": 1.3392415850021304,
+      "grad_norm": 16.75,
+      "learning_rate": 6.197685091924895e-07,
+      "loss": 0.6367,
+      "step": 15716
+    },
+    {
+      "epoch": 1.3393268001704304,
+      "grad_norm": 17.125,
+      "learning_rate": 6.196234001319589e-07,
+      "loss": 0.7573,
+      "step": 15717
+    },
+    {
+      "epoch": 1.3394120153387303,
+      "grad_norm": 18.75,
+      "learning_rate": 6.19478302462967e-07,
+      "loss": 0.8693,
+      "step": 15718
+    },
+    {
+      "epoch": 1.3394972305070303,
+      "grad_norm": 22.75,
+      "learning_rate": 6.19333216188136e-07,
+      "loss": 1.0281,
+      "step": 15719
+    },
+    {
+      "epoch": 1.3395824456753302,
+      "grad_norm": 16.0,
+      "learning_rate": 6.191881413100886e-07,
+      "loss": 0.5265,
+      "step": 15720
+    },
+    {
+      "epoch": 1.3396676608436302,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.190430778314457e-07,
+      "loss": 0.5743,
+      "step": 15721
+    },
+    {
+      "epoch": 1.3397528760119302,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.188980257548286e-07,
+      "loss": 0.3437,
+      "step": 15722
+    },
+    {
+      "epoch": 1.3398380911802301,
+      "grad_norm": 16.625,
+      "learning_rate": 6.187529850828587e-07,
+      "loss": 0.5188,
+      "step": 15723
+    },
+    {
+      "epoch": 1.33992330634853,
+      "grad_norm": 12.3125,
+      "learning_rate": 6.186079558181571e-07,
+      "loss": 0.516,
+      "step": 15724
+    },
+    {
+      "epoch": 1.34000852151683,
+      "grad_norm": 16.5,
+      "learning_rate": 6.184629379633452e-07,
+      "loss": 0.6354,
+      "step": 15725
+    },
+    {
+      "epoch": 1.34009373668513,
+      "grad_norm": 12.75,
+      "learning_rate": 6.183179315210425e-07,
+      "loss": 0.3984,
+      "step": 15726
+    },
+    {
+      "epoch": 1.34017895185343,
+      "grad_norm": 25.375,
+      "learning_rate": 6.181729364938699e-07,
+      "loss": 1.2726,
+      "step": 15727
+    },
+    {
+      "epoch": 1.34026416702173,
+      "grad_norm": 18.0,
+      "learning_rate": 6.180279528844482e-07,
+      "loss": 0.767,
+      "step": 15728
+    },
+    {
+      "epoch": 1.3403493821900299,
+      "grad_norm": 16.375,
+      "learning_rate": 6.178829806953966e-07,
+      "loss": 0.7979,
+      "step": 15729
+    },
+    {
+      "epoch": 1.3404345973583298,
+      "grad_norm": 20.5,
+      "learning_rate": 6.177380199293355e-07,
+      "loss": 0.707,
+      "step": 15730
+    },
+    {
+      "epoch": 1.3405198125266298,
+      "grad_norm": 20.375,
+      "learning_rate": 6.175930705888838e-07,
+      "loss": 0.6921,
+      "step": 15731
+    },
+    {
+      "epoch": 1.3406050276949297,
+      "grad_norm": 17.5,
+      "learning_rate": 6.174481326766611e-07,
+      "loss": 0.813,
+      "step": 15732
+    },
+    {
+      "epoch": 1.3406902428632297,
+      "grad_norm": 16.75,
+      "learning_rate": 6.173032061952873e-07,
+      "loss": 0.5818,
+      "step": 15733
+    },
+    {
+      "epoch": 1.3407754580315296,
+      "grad_norm": 17.125,
+      "learning_rate": 6.171582911473803e-07,
+      "loss": 0.7163,
+      "step": 15734
+    },
+    {
+      "epoch": 1.3408606731998296,
+      "grad_norm": 17.875,
+      "learning_rate": 6.170133875355594e-07,
+      "loss": 0.7903,
+      "step": 15735
+    },
+    {
+      "epoch": 1.3409458883681296,
+      "grad_norm": 13.1875,
+      "learning_rate": 6.168684953624434e-07,
+      "loss": 0.3583,
+      "step": 15736
+    },
+    {
+      "epoch": 1.3410311035364295,
+      "grad_norm": 16.25,
+      "learning_rate": 6.167236146306501e-07,
+      "loss": 0.493,
+      "step": 15737
+    },
+    {
+      "epoch": 1.3411163187047295,
+      "grad_norm": 17.0,
+      "learning_rate": 6.165787453427982e-07,
+      "loss": 0.9997,
+      "step": 15738
+    },
+    {
+      "epoch": 1.3412015338730294,
+      "grad_norm": 16.0,
+      "learning_rate": 6.16433887501505e-07,
+      "loss": 0.439,
+      "step": 15739
+    },
+    {
+      "epoch": 1.3412867490413294,
+      "grad_norm": 15.125,
+      "learning_rate": 6.162890411093885e-07,
+      "loss": 0.4338,
+      "step": 15740
+    },
+    {
+      "epoch": 1.3413719642096293,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.161442061690662e-07,
+      "loss": 0.5005,
+      "step": 15741
+    },
+    {
+      "epoch": 1.3414571793779293,
+      "grad_norm": 16.625,
+      "learning_rate": 6.15999382683156e-07,
+      "loss": 0.4024,
+      "step": 15742
+    },
+    {
+      "epoch": 1.3415423945462293,
+      "grad_norm": 16.75,
+      "learning_rate": 6.15854570654274e-07,
+      "loss": 0.3623,
+      "step": 15743
+    },
+    {
+      "epoch": 1.3416276097145292,
+      "grad_norm": 12.5,
+      "learning_rate": 6.157097700850381e-07,
+      "loss": 0.4578,
+      "step": 15744
+    },
+    {
+      "epoch": 1.3417128248828292,
+      "grad_norm": 12.1875,
+      "learning_rate": 6.155649809780641e-07,
+      "loss": 0.4782,
+      "step": 15745
+    },
+    {
+      "epoch": 1.3417980400511291,
+      "grad_norm": 18.5,
+      "learning_rate": 6.154202033359686e-07,
+      "loss": 0.5755,
+      "step": 15746
+    },
+    {
+      "epoch": 1.341883255219429,
+      "grad_norm": 18.875,
+      "learning_rate": 6.15275437161369e-07,
+      "loss": 0.4891,
+      "step": 15747
+    },
+    {
+      "epoch": 1.341968470387729,
+      "grad_norm": 18.25,
+      "learning_rate": 6.151306824568797e-07,
+      "loss": 0.4314,
+      "step": 15748
+    },
+    {
+      "epoch": 1.342053685556029,
+      "grad_norm": 24.125,
+      "learning_rate": 6.149859392251176e-07,
+      "loss": 0.8643,
+      "step": 15749
+    },
+    {
+      "epoch": 1.342138900724329,
+      "grad_norm": 36.25,
+      "learning_rate": 6.148412074686984e-07,
+      "loss": 0.674,
+      "step": 15750
+    },
+    {
+      "epoch": 1.342224115892629,
+      "grad_norm": 23.0,
+      "learning_rate": 6.146964871902369e-07,
+      "loss": 0.6592,
+      "step": 15751
+    },
+    {
+      "epoch": 1.3423093310609289,
+      "grad_norm": 16.125,
+      "learning_rate": 6.145517783923493e-07,
+      "loss": 0.6739,
+      "step": 15752
+    },
+    {
+      "epoch": 1.3423945462292288,
+      "grad_norm": 18.0,
+      "learning_rate": 6.144070810776495e-07,
+      "loss": 0.5365,
+      "step": 15753
+    },
+    {
+      "epoch": 1.3424797613975288,
+      "grad_norm": 20.75,
+      "learning_rate": 6.142623952487528e-07,
+      "loss": 0.5746,
+      "step": 15754
+    },
+    {
+      "epoch": 1.3425649765658287,
+      "grad_norm": 9.0625,
+      "learning_rate": 6.141177209082745e-07,
+      "loss": 0.2225,
+      "step": 15755
+    },
+    {
+      "epoch": 1.3426501917341287,
+      "grad_norm": 17.0,
+      "learning_rate": 6.13973058058828e-07,
+      "loss": 0.8738,
+      "step": 15756
+    },
+    {
+      "epoch": 1.3427354069024287,
+      "grad_norm": 17.875,
+      "learning_rate": 6.138284067030277e-07,
+      "loss": 0.5033,
+      "step": 15757
+    },
+    {
+      "epoch": 1.3428206220707286,
+      "grad_norm": 14.375,
+      "learning_rate": 6.136837668434879e-07,
+      "loss": 0.5787,
+      "step": 15758
+    },
+    {
+      "epoch": 1.3429058372390286,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.135391384828229e-07,
+      "loss": 0.2792,
+      "step": 15759
+    },
+    {
+      "epoch": 1.3429910524073285,
+      "grad_norm": 13.5,
+      "learning_rate": 6.133945216236456e-07,
+      "loss": 0.4844,
+      "step": 15760
+    },
+    {
+      "epoch": 1.3430762675756285,
+      "grad_norm": 25.875,
+      "learning_rate": 6.132499162685689e-07,
+      "loss": 0.4957,
+      "step": 15761
+    },
+    {
+      "epoch": 1.3431614827439284,
+      "grad_norm": 16.75,
+      "learning_rate": 6.131053224202066e-07,
+      "loss": 0.6553,
+      "step": 15762
+    },
+    {
+      "epoch": 1.3432466979122284,
+      "grad_norm": 14.75,
+      "learning_rate": 6.12960740081172e-07,
+      "loss": 0.7764,
+      "step": 15763
+    },
+    {
+      "epoch": 1.3433319130805284,
+      "grad_norm": 14.75,
+      "learning_rate": 6.128161692540769e-07,
+      "loss": 0.6774,
+      "step": 15764
+    },
+    {
+      "epoch": 1.3434171282488283,
+      "grad_norm": 17.75,
+      "learning_rate": 6.126716099415345e-07,
+      "loss": 0.6884,
+      "step": 15765
+    },
+    {
+      "epoch": 1.3435023434171283,
+      "grad_norm": 20.125,
+      "learning_rate": 6.125270621461568e-07,
+      "loss": 0.7481,
+      "step": 15766
+    },
+    {
+      "epoch": 1.3435875585854282,
+      "grad_norm": 13.875,
+      "learning_rate": 6.123825258705568e-07,
+      "loss": 0.4403,
+      "step": 15767
+    },
+    {
+      "epoch": 1.3436727737537282,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.122380011173457e-07,
+      "loss": 0.3084,
+      "step": 15768
+    },
+    {
+      "epoch": 1.3437579889220281,
+      "grad_norm": 23.5,
+      "learning_rate": 6.120934878891346e-07,
+      "loss": 0.5549,
+      "step": 15769
+    },
+    {
+      "epoch": 1.343843204090328,
+      "grad_norm": 18.25,
+      "learning_rate": 6.119489861885357e-07,
+      "loss": 0.7841,
+      "step": 15770
+    },
+    {
+      "epoch": 1.343928419258628,
+      "grad_norm": 18.0,
+      "learning_rate": 6.118044960181603e-07,
+      "loss": 0.5578,
+      "step": 15771
+    },
+    {
+      "epoch": 1.344013634426928,
+      "grad_norm": 13.25,
+      "learning_rate": 6.116600173806199e-07,
+      "loss": 0.5436,
+      "step": 15772
+    },
+    {
+      "epoch": 1.344098849595228,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.115155502785243e-07,
+      "loss": 0.3165,
+      "step": 15773
+    },
+    {
+      "epoch": 1.344184064763528,
+      "grad_norm": 16.875,
+      "learning_rate": 6.113710947144847e-07,
+      "loss": 0.5749,
+      "step": 15774
+    },
+    {
+      "epoch": 1.3442692799318279,
+      "grad_norm": 16.875,
+      "learning_rate": 6.112266506911118e-07,
+      "loss": 0.6834,
+      "step": 15775
+    },
+    {
+      "epoch": 1.3443544951001278,
+      "grad_norm": 17.0,
+      "learning_rate": 6.110822182110162e-07,
+      "loss": 0.6688,
+      "step": 15776
+    },
+    {
+      "epoch": 1.3444397102684278,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.109377972768071e-07,
+      "loss": 0.5449,
+      "step": 15777
+    },
+    {
+      "epoch": 1.3445249254367277,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.107933878910943e-07,
+      "loss": 0.4237,
+      "step": 15778
+    },
+    {
+      "epoch": 1.3446101406050277,
+      "grad_norm": 14.375,
+      "learning_rate": 6.106489900564879e-07,
+      "loss": 0.7421,
+      "step": 15779
+    },
+    {
+      "epoch": 1.3446953557733277,
+      "grad_norm": 22.125,
+      "learning_rate": 6.105046037755974e-07,
+      "loss": 0.9309,
+      "step": 15780
+    },
+    {
+      "epoch": 1.3447805709416276,
+      "grad_norm": 16.75,
+      "learning_rate": 6.103602290510315e-07,
+      "loss": 0.5672,
+      "step": 15781
+    },
+    {
+      "epoch": 1.3448657861099276,
+      "grad_norm": 13.375,
+      "learning_rate": 6.102158658853994e-07,
+      "loss": 0.4657,
+      "step": 15782
+    },
+    {
+      "epoch": 1.3449510012782275,
+      "grad_norm": 20.375,
+      "learning_rate": 6.100715142813098e-07,
+      "loss": 0.5265,
+      "step": 15783
+    },
+    {
+      "epoch": 1.3450362164465275,
+      "grad_norm": 33.75,
+      "learning_rate": 6.099271742413721e-07,
+      "loss": 0.7699,
+      "step": 15784
+    },
+    {
+      "epoch": 1.3451214316148274,
+      "grad_norm": 15.125,
+      "learning_rate": 6.09782845768194e-07,
+      "loss": 0.4127,
+      "step": 15785
+    },
+    {
+      "epoch": 1.3452066467831274,
+      "grad_norm": 22.375,
+      "learning_rate": 6.09638528864383e-07,
+      "loss": 0.9485,
+      "step": 15786
+    },
+    {
+      "epoch": 1.3452918619514274,
+      "grad_norm": 16.25,
+      "learning_rate": 6.09494223532548e-07,
+      "loss": 0.7417,
+      "step": 15787
+    },
+    {
+      "epoch": 1.3453770771197273,
+      "grad_norm": 9.4375,
+      "learning_rate": 6.093499297752962e-07,
+      "loss": 0.2085,
+      "step": 15788
+    },
+    {
+      "epoch": 1.3454622922880273,
+      "grad_norm": 12.375,
+      "learning_rate": 6.09205647595236e-07,
+      "loss": 0.4626,
+      "step": 15789
+    },
+    {
+      "epoch": 1.3455475074563272,
+      "grad_norm": 19.75,
+      "learning_rate": 6.090613769949735e-07,
+      "loss": 0.6636,
+      "step": 15790
+    },
+    {
+      "epoch": 1.3456327226246272,
+      "grad_norm": 18.75,
+      "learning_rate": 6.089171179771167e-07,
+      "loss": 0.4606,
+      "step": 15791
+    },
+    {
+      "epoch": 1.3457179377929271,
+      "grad_norm": 12.8125,
+      "learning_rate": 6.087728705442721e-07,
+      "loss": 0.4842,
+      "step": 15792
+    },
+    {
+      "epoch": 1.345803152961227,
+      "grad_norm": 16.625,
+      "learning_rate": 6.086286346990474e-07,
+      "loss": 0.4564,
+      "step": 15793
+    },
+    {
+      "epoch": 1.345888368129527,
+      "grad_norm": 18.5,
+      "learning_rate": 6.084844104440472e-07,
+      "loss": 0.9401,
+      "step": 15794
+    },
+    {
+      "epoch": 1.345973583297827,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.083401977818792e-07,
+      "loss": 0.5254,
+      "step": 15795
+    },
+    {
+      "epoch": 1.346058798466127,
+      "grad_norm": 17.875,
+      "learning_rate": 6.081959967151487e-07,
+      "loss": 0.5924,
+      "step": 15796
+    },
+    {
+      "epoch": 1.346144013634427,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.080518072464624e-07,
+      "loss": 0.5881,
+      "step": 15797
+    },
+    {
+      "epoch": 1.3462292288027269,
+      "grad_norm": 17.875,
+      "learning_rate": 6.079076293784252e-07,
+      "loss": 0.6528,
+      "step": 15798
+    },
+    {
+      "epoch": 1.3463144439710268,
+      "grad_norm": 12.9375,
+      "learning_rate": 6.077634631136427e-07,
+      "loss": 0.5699,
+      "step": 15799
+    },
+    {
+      "epoch": 1.3463996591393268,
+      "grad_norm": 15.375,
+      "learning_rate": 6.076193084547202e-07,
+      "loss": 0.4941,
+      "step": 15800
+    },
+    {
+      "epoch": 1.3464848743076268,
+      "grad_norm": 11.625,
+      "learning_rate": 6.074751654042632e-07,
+      "loss": 0.3549,
+      "step": 15801
+    },
+    {
+      "epoch": 1.3465700894759267,
+      "grad_norm": 25.25,
+      "learning_rate": 6.073310339648762e-07,
+      "loss": 0.7699,
+      "step": 15802
+    },
+    {
+      "epoch": 1.3466553046442267,
+      "grad_norm": 16.125,
+      "learning_rate": 6.071869141391631e-07,
+      "loss": 0.736,
+      "step": 15803
+    },
+    {
+      "epoch": 1.3467405198125266,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.07042805929729e-07,
+      "loss": 0.5478,
+      "step": 15804
+    },
+    {
+      "epoch": 1.3468257349808266,
+      "grad_norm": 20.625,
+      "learning_rate": 6.068987093391778e-07,
+      "loss": 1.0594,
+      "step": 15805
+    },
+    {
+      "epoch": 1.3469109501491265,
+      "grad_norm": 28.5,
+      "learning_rate": 6.067546243701142e-07,
+      "loss": 0.7303,
+      "step": 15806
+    },
+    {
+      "epoch": 1.3469961653174265,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.066105510251408e-07,
+      "loss": 0.6478,
+      "step": 15807
+    },
+    {
+      "epoch": 1.3470813804857265,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.064664893068617e-07,
+      "loss": 0.5998,
+      "step": 15808
+    },
+    {
+      "epoch": 1.3471665956540264,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.063224392178808e-07,
+      "loss": 0.8845,
+      "step": 15809
+    },
+    {
+      "epoch": 1.3472518108223264,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.061784007608007e-07,
+      "loss": 0.4496,
+      "step": 15810
+    },
+    {
+      "epoch": 1.3473370259906263,
+      "grad_norm": 10.3125,
+      "learning_rate": 6.060343739382237e-07,
+      "loss": 0.3702,
+      "step": 15811
+    },
+    {
+      "epoch": 1.3474222411589263,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.058903587527533e-07,
+      "loss": 0.4573,
+      "step": 15812
+    },
+    {
+      "epoch": 1.3475074563272262,
+      "grad_norm": 15.375,
+      "learning_rate": 6.057463552069917e-07,
+      "loss": 0.6626,
+      "step": 15813
+    },
+    {
+      "epoch": 1.3475926714955262,
+      "grad_norm": 12.9375,
+      "learning_rate": 6.056023633035418e-07,
+      "loss": 0.3749,
+      "step": 15814
+    },
+    {
+      "epoch": 1.3476778866638262,
+      "grad_norm": 28.0,
+      "learning_rate": 6.054583830450049e-07,
+      "loss": 0.9343,
+      "step": 15815
+    },
+    {
+      "epoch": 1.347763101832126,
+      "grad_norm": 11.625,
+      "learning_rate": 6.05314414433983e-07,
+      "loss": 0.3671,
+      "step": 15816
+    },
+    {
+      "epoch": 1.347848317000426,
+      "grad_norm": 17.625,
+      "learning_rate": 6.051704574730784e-07,
+      "loss": 0.4965,
+      "step": 15817
+    },
+    {
+      "epoch": 1.347933532168726,
+      "grad_norm": 18.0,
+      "learning_rate": 6.050265121648915e-07,
+      "loss": 0.6225,
+      "step": 15818
+    },
+    {
+      "epoch": 1.348018747337026,
+      "grad_norm": 14.25,
+      "learning_rate": 6.048825785120247e-07,
+      "loss": 0.3876,
+      "step": 15819
+    },
+    {
+      "epoch": 1.348103962505326,
+      "grad_norm": 16.75,
+      "learning_rate": 6.047386565170778e-07,
+      "loss": 0.7574,
+      "step": 15820
+    },
+    {
+      "epoch": 1.348189177673626,
+      "grad_norm": 17.0,
+      "learning_rate": 6.045947461826524e-07,
+      "loss": 0.6694,
+      "step": 15821
+    },
+    {
+      "epoch": 1.3482743928419259,
+      "grad_norm": 13.75,
+      "learning_rate": 6.044508475113488e-07,
+      "loss": 0.4245,
+      "step": 15822
+    },
+    {
+      "epoch": 1.3483596080102258,
+      "grad_norm": 17.125,
+      "learning_rate": 6.043069605057682e-07,
+      "loss": 0.7726,
+      "step": 15823
+    },
+    {
+      "epoch": 1.3484448231785258,
+      "grad_norm": 16.625,
+      "learning_rate": 6.041630851685095e-07,
+      "loss": 0.7617,
+      "step": 15824
+    },
+    {
+      "epoch": 1.3485300383468257,
+      "grad_norm": 16.125,
+      "learning_rate": 6.040192215021737e-07,
+      "loss": 0.4532,
+      "step": 15825
+    },
+    {
+      "epoch": 1.3486152535151257,
+      "grad_norm": 12.5,
+      "learning_rate": 6.038753695093598e-07,
+      "loss": 0.5056,
+      "step": 15826
+    },
+    {
+      "epoch": 1.3487004686834256,
+      "grad_norm": 13.625,
+      "learning_rate": 6.037315291926682e-07,
+      "loss": 0.4505,
+      "step": 15827
+    },
+    {
+      "epoch": 1.3487856838517256,
+      "grad_norm": 16.25,
+      "learning_rate": 6.035877005546971e-07,
+      "loss": 1.1094,
+      "step": 15828
+    },
+    {
+      "epoch": 1.3488708990200255,
+      "grad_norm": 14.125,
+      "learning_rate": 6.034438835980465e-07,
+      "loss": 0.6765,
+      "step": 15829
+    },
+    {
+      "epoch": 1.3489561141883255,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.033000783253148e-07,
+      "loss": 0.3139,
+      "step": 15830
+    },
+    {
+      "epoch": 1.3490413293566255,
+      "grad_norm": 26.625,
+      "learning_rate": 6.031562847391017e-07,
+      "loss": 0.8411,
+      "step": 15831
+    },
+    {
+      "epoch": 1.3491265445249254,
+      "grad_norm": 27.875,
+      "learning_rate": 6.030125028420045e-07,
+      "loss": 1.1612,
+      "step": 15832
+    },
+    {
+      "epoch": 1.3492117596932254,
+      "grad_norm": 13.875,
+      "learning_rate": 6.028687326366224e-07,
+      "loss": 0.6284,
+      "step": 15833
+    },
+    {
+      "epoch": 1.3492969748615253,
+      "grad_norm": 21.625,
+      "learning_rate": 6.027249741255525e-07,
+      "loss": 0.8207,
+      "step": 15834
+    },
+    {
+      "epoch": 1.3493821900298253,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.025812273113933e-07,
+      "loss": 0.8499,
+      "step": 15835
+    },
+    {
+      "epoch": 1.3494674051981252,
+      "grad_norm": 21.25,
+      "learning_rate": 6.024374921967427e-07,
+      "loss": 0.7364,
+      "step": 15836
+    },
+    {
+      "epoch": 1.3495526203664252,
+      "grad_norm": 16.375,
+      "learning_rate": 6.022937687841975e-07,
+      "loss": 0.6,
+      "step": 15837
+    },
+    {
+      "epoch": 1.3496378355347252,
+      "grad_norm": 17.375,
+      "learning_rate": 6.02150057076355e-07,
+      "loss": 0.6208,
+      "step": 15838
+    },
+    {
+      "epoch": 1.3497230507030251,
+      "grad_norm": 18.5,
+      "learning_rate": 6.020063570758126e-07,
+      "loss": 0.3062,
+      "step": 15839
+    },
+    {
+      "epoch": 1.349808265871325,
+      "grad_norm": 19.875,
+      "learning_rate": 6.018626687851674e-07,
+      "loss": 0.8002,
+      "step": 15840
+    },
+    {
+      "epoch": 1.349893481039625,
+      "grad_norm": 16.25,
+      "learning_rate": 6.017189922070149e-07,
+      "loss": 0.6893,
+      "step": 15841
+    },
+    {
+      "epoch": 1.349978696207925,
+      "grad_norm": 19.375,
+      "learning_rate": 6.015753273439527e-07,
+      "loss": 0.7421,
+      "step": 15842
+    },
+    {
+      "epoch": 1.350063911376225,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.01431674198576e-07,
+      "loss": 0.5787,
+      "step": 15843
+    },
+    {
+      "epoch": 1.350149126544525,
+      "grad_norm": 18.25,
+      "learning_rate": 6.012880327734814e-07,
+      "loss": 0.7974,
+      "step": 15844
+    },
+    {
+      "epoch": 1.3502343417128249,
+      "grad_norm": 17.375,
+      "learning_rate": 6.01144403071264e-07,
+      "loss": 1.1023,
+      "step": 15845
+    },
+    {
+      "epoch": 1.3503195568811248,
+      "grad_norm": 18.0,
+      "learning_rate": 6.010007850945197e-07,
+      "loss": 0.6663,
+      "step": 15846
+    },
+    {
+      "epoch": 1.3504047720494248,
+      "grad_norm": 14.125,
+      "learning_rate": 6.00857178845844e-07,
+      "loss": 0.5281,
+      "step": 15847
+    },
+    {
+      "epoch": 1.3504899872177247,
+      "grad_norm": 17.625,
+      "learning_rate": 6.007135843278321e-07,
+      "loss": 0.785,
+      "step": 15848
+    },
+    {
+      "epoch": 1.3505752023860247,
+      "grad_norm": 19.875,
+      "learning_rate": 6.005700015430782e-07,
+      "loss": 0.9119,
+      "step": 15849
+    },
+    {
+      "epoch": 1.3506604175543246,
+      "grad_norm": 16.25,
+      "learning_rate": 6.00426430494178e-07,
+      "loss": 0.712,
+      "step": 15850
+    },
+    {
+      "epoch": 1.3507456327226246,
+      "grad_norm": 17.125,
+      "learning_rate": 6.002828711837248e-07,
+      "loss": 0.8806,
+      "step": 15851
+    },
+    {
+      "epoch": 1.3508308478909246,
+      "grad_norm": 12.8125,
+      "learning_rate": 6.001393236143135e-07,
+      "loss": 0.3643,
+      "step": 15852
+    },
+    {
+      "epoch": 1.3509160630592245,
+      "grad_norm": 12.9375,
+      "learning_rate": 5.999957877885386e-07,
+      "loss": 0.2465,
+      "step": 15853
+    },
+    {
+      "epoch": 1.3510012782275245,
+      "grad_norm": 16.625,
+      "learning_rate": 5.998522637089931e-07,
+      "loss": 0.5329,
+      "step": 15854
+    },
+    {
+      "epoch": 1.3510864933958244,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.997087513782707e-07,
+      "loss": 0.6052,
+      "step": 15855
+    },
+    {
+      "epoch": 1.3511717085641244,
+      "grad_norm": 14.25,
+      "learning_rate": 5.995652507989657e-07,
+      "loss": 0.4683,
+      "step": 15856
+    },
+    {
+      "epoch": 1.3512569237324243,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.994217619736702e-07,
+      "loss": 0.2995,
+      "step": 15857
+    },
+    {
+      "epoch": 1.3513421389007243,
+      "grad_norm": 18.625,
+      "learning_rate": 5.992782849049782e-07,
+      "loss": 0.9606,
+      "step": 15858
+    },
+    {
+      "epoch": 1.3514273540690243,
+      "grad_norm": 19.25,
+      "learning_rate": 5.991348195954813e-07,
+      "loss": 0.7654,
+      "step": 15859
+    },
+    {
+      "epoch": 1.3515125692373242,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.989913660477728e-07,
+      "loss": 0.7139,
+      "step": 15860
+    },
+    {
+      "epoch": 1.3515977844056242,
+      "grad_norm": 15.4375,
+      "learning_rate": 5.988479242644455e-07,
+      "loss": 0.6454,
+      "step": 15861
+    },
+    {
+      "epoch": 1.3516829995739241,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.987044942480904e-07,
+      "loss": 0.4046,
+      "step": 15862
+    },
+    {
+      "epoch": 1.351768214742224,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.985610760013001e-07,
+      "loss": 0.7737,
+      "step": 15863
+    },
+    {
+      "epoch": 1.351853429910524,
+      "grad_norm": 12.0625,
+      "learning_rate": 5.984176695266662e-07,
+      "loss": 0.3874,
+      "step": 15864
+    },
+    {
+      "epoch": 1.351938645078824,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.982742748267807e-07,
+      "loss": 0.3902,
+      "step": 15865
+    },
+    {
+      "epoch": 1.352023860247124,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.981308919042344e-07,
+      "loss": 0.5255,
+      "step": 15866
+    },
+    {
+      "epoch": 1.352109075415424,
+      "grad_norm": 18.75,
+      "learning_rate": 5.97987520761618e-07,
+      "loss": 0.8196,
+      "step": 15867
+    },
+    {
+      "epoch": 1.3521942905837239,
+      "grad_norm": 11.875,
+      "learning_rate": 5.978441614015228e-07,
+      "loss": 0.3243,
+      "step": 15868
+    },
+    {
+      "epoch": 1.3522795057520238,
+      "grad_norm": 15.125,
+      "learning_rate": 5.977008138265392e-07,
+      "loss": 0.537,
+      "step": 15869
+    },
+    {
+      "epoch": 1.3523647209203238,
+      "grad_norm": 17.75,
+      "learning_rate": 5.975574780392583e-07,
+      "loss": 0.832,
+      "step": 15870
+    },
+    {
+      "epoch": 1.3524499360886237,
+      "grad_norm": 17.75,
+      "learning_rate": 5.974141540422696e-07,
+      "loss": 0.8173,
+      "step": 15871
+    },
+    {
+      "epoch": 1.3525351512569237,
+      "grad_norm": 11.8125,
+      "learning_rate": 5.972708418381631e-07,
+      "loss": 0.3605,
+      "step": 15872
+    },
+    {
+      "epoch": 1.3526203664252237,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.971275414295295e-07,
+      "loss": 0.5674,
+      "step": 15873
+    },
+    {
+      "epoch": 1.3527055815935236,
+      "grad_norm": 26.75,
+      "learning_rate": 5.969842528189577e-07,
+      "loss": 0.6826,
+      "step": 15874
+    },
+    {
+      "epoch": 1.3527907967618236,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.968409760090367e-07,
+      "loss": 0.583,
+      "step": 15875
+    },
+    {
+      "epoch": 1.3528760119301235,
+      "grad_norm": 17.125,
+      "learning_rate": 5.96697711002356e-07,
+      "loss": 0.7757,
+      "step": 15876
+    },
+    {
+      "epoch": 1.3529612270984235,
+      "grad_norm": 13.25,
+      "learning_rate": 5.965544578015046e-07,
+      "loss": 0.5738,
+      "step": 15877
+    },
+    {
+      "epoch": 1.3530464422667234,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.964112164090718e-07,
+      "loss": 0.6923,
+      "step": 15878
+    },
+    {
+      "epoch": 1.3531316574350234,
+      "grad_norm": 15.5,
+      "learning_rate": 5.962679868276449e-07,
+      "loss": 0.7008,
+      "step": 15879
+    },
+    {
+      "epoch": 1.3532168726033234,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.96124769059813e-07,
+      "loss": 0.334,
+      "step": 15880
+    },
+    {
+      "epoch": 1.3533020877716233,
+      "grad_norm": 13.1875,
+      "learning_rate": 5.95981563108164e-07,
+      "loss": 0.4529,
+      "step": 15881
+    },
+    {
+      "epoch": 1.3533873029399233,
+      "grad_norm": 15.875,
+      "learning_rate": 5.958383689752863e-07,
+      "loss": 0.7905,
+      "step": 15882
+    },
+    {
+      "epoch": 1.3534725181082232,
+      "grad_norm": 22.75,
+      "learning_rate": 5.956951866637671e-07,
+      "loss": 0.7251,
+      "step": 15883
+    },
+    {
+      "epoch": 1.3535577332765232,
+      "grad_norm": 16.25,
+      "learning_rate": 5.955520161761934e-07,
+      "loss": 0.4628,
+      "step": 15884
+    },
+    {
+      "epoch": 1.3536429484448231,
+      "grad_norm": 17.125,
+      "learning_rate": 5.954088575151527e-07,
+      "loss": 0.9171,
+      "step": 15885
+    },
+    {
+      "epoch": 1.353728163613123,
+      "grad_norm": 20.125,
+      "learning_rate": 5.952657106832328e-07,
+      "loss": 0.8455,
+      "step": 15886
+    },
+    {
+      "epoch": 1.353813378781423,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.951225756830194e-07,
+      "loss": 0.7517,
+      "step": 15887
+    },
+    {
+      "epoch": 1.353898593949723,
+      "grad_norm": 12.625,
+      "learning_rate": 5.949794525170996e-07,
+      "loss": 0.5419,
+      "step": 15888
+    },
+    {
+      "epoch": 1.353983809118023,
+      "grad_norm": 10.8125,
+      "learning_rate": 5.948363411880598e-07,
+      "loss": 0.2975,
+      "step": 15889
+    },
+    {
+      "epoch": 1.354069024286323,
+      "grad_norm": 9.875,
+      "learning_rate": 5.946932416984864e-07,
+      "loss": 0.2283,
+      "step": 15890
+    },
+    {
+      "epoch": 1.3541542394546229,
+      "grad_norm": 13.5,
+      "learning_rate": 5.945501540509654e-07,
+      "loss": 0.3838,
+      "step": 15891
+    },
+    {
+      "epoch": 1.3542394546229228,
+      "grad_norm": 18.75,
+      "learning_rate": 5.944070782480815e-07,
+      "loss": 0.8425,
+      "step": 15892
+    },
+    {
+      "epoch": 1.3543246697912228,
+      "grad_norm": 17.125,
+      "learning_rate": 5.94264014292421e-07,
+      "loss": 0.8381,
+      "step": 15893
+    },
+    {
+      "epoch": 1.3544098849595227,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.941209621865691e-07,
+      "loss": 0.4299,
+      "step": 15894
+    },
+    {
+      "epoch": 1.3544951001278227,
+      "grad_norm": 17.625,
+      "learning_rate": 5.939779219331115e-07,
+      "loss": 0.731,
+      "step": 15895
+    },
+    {
+      "epoch": 1.3545803152961227,
+      "grad_norm": 19.375,
+      "learning_rate": 5.938348935346322e-07,
+      "loss": 0.8002,
+      "step": 15896
+    },
+    {
+      "epoch": 1.3546655304644226,
+      "grad_norm": 17.875,
+      "learning_rate": 5.93691876993716e-07,
+      "loss": 0.7089,
+      "step": 15897
+    },
+    {
+      "epoch": 1.3547507456327226,
+      "grad_norm": 15.5625,
+      "learning_rate": 5.935488723129481e-07,
+      "loss": 0.4617,
+      "step": 15898
+    },
+    {
+      "epoch": 1.3548359608010225,
+      "grad_norm": 16.0,
+      "learning_rate": 5.934058794949117e-07,
+      "loss": 0.5794,
+      "step": 15899
+    },
+    {
+      "epoch": 1.3549211759693225,
+      "grad_norm": 17.625,
+      "learning_rate": 5.932628985421918e-07,
+      "loss": 0.6229,
+      "step": 15900
+    },
+    {
+      "epoch": 1.3550063911376224,
+      "grad_norm": 17.75,
+      "learning_rate": 5.931199294573713e-07,
+      "loss": 0.7742,
+      "step": 15901
+    },
+    {
+      "epoch": 1.3550916063059224,
+      "grad_norm": 23.0,
+      "learning_rate": 5.929769722430341e-07,
+      "loss": 0.9518,
+      "step": 15902
+    },
+    {
+      "epoch": 1.3551768214742224,
+      "grad_norm": 17.625,
+      "learning_rate": 5.928340269017642e-07,
+      "loss": 0.5512,
+      "step": 15903
+    },
+    {
+      "epoch": 1.3552620366425223,
+      "grad_norm": 16.25,
+      "learning_rate": 5.926910934361439e-07,
+      "loss": 0.6373,
+      "step": 15904
+    },
+    {
+      "epoch": 1.3553472518108223,
+      "grad_norm": 15.6875,
+      "learning_rate": 5.925481718487565e-07,
+      "loss": 0.9097,
+      "step": 15905
+    },
+    {
+      "epoch": 1.3554324669791222,
+      "grad_norm": 12.3125,
+      "learning_rate": 5.924052621421847e-07,
+      "loss": 0.2453,
+      "step": 15906
+    },
+    {
+      "epoch": 1.3555176821474222,
+      "grad_norm": 15.75,
+      "learning_rate": 5.922623643190114e-07,
+      "loss": 0.6213,
+      "step": 15907
+    },
+    {
+      "epoch": 1.3556028973157221,
+      "grad_norm": 15.125,
+      "learning_rate": 5.921194783818188e-07,
+      "loss": 0.5509,
+      "step": 15908
+    },
+    {
+      "epoch": 1.355688112484022,
+      "grad_norm": 13.5,
+      "learning_rate": 5.919766043331883e-07,
+      "loss": 0.28,
+      "step": 15909
+    },
+    {
+      "epoch": 1.355773327652322,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.918337421757022e-07,
+      "loss": 0.4226,
+      "step": 15910
+    },
+    {
+      "epoch": 1.355858542820622,
+      "grad_norm": 19.0,
+      "learning_rate": 5.916908919119421e-07,
+      "loss": 0.5925,
+      "step": 15911
+    },
+    {
+      "epoch": 1.355943757988922,
+      "grad_norm": 22.375,
+      "learning_rate": 5.915480535444903e-07,
+      "loss": 1.2797,
+      "step": 15912
+    },
+    {
+      "epoch": 1.356028973157222,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.914052270759265e-07,
+      "loss": 0.6694,
+      "step": 15913
+    },
+    {
+      "epoch": 1.3561141883255219,
+      "grad_norm": 14.875,
+      "learning_rate": 5.912624125088328e-07,
+      "loss": 0.8276,
+      "step": 15914
+    },
+    {
+      "epoch": 1.3561994034938218,
+      "grad_norm": 28.125,
+      "learning_rate": 5.9111960984579e-07,
+      "loss": 0.8987,
+      "step": 15915
+    },
+    {
+      "epoch": 1.3562846186621218,
+      "grad_norm": 15.375,
+      "learning_rate": 5.909768190893784e-07,
+      "loss": 0.5611,
+      "step": 15916
+    },
+    {
+      "epoch": 1.3563698338304218,
+      "grad_norm": 20.375,
+      "learning_rate": 5.908340402421781e-07,
+      "loss": 0.2776,
+      "step": 15917
+    },
+    {
+      "epoch": 1.3564550489987217,
+      "grad_norm": 12.25,
+      "learning_rate": 5.906912733067693e-07,
+      "loss": 0.5888,
+      "step": 15918
+    },
+    {
+      "epoch": 1.3565402641670217,
+      "grad_norm": 15.0,
+      "learning_rate": 5.905485182857323e-07,
+      "loss": 0.6105,
+      "step": 15919
+    },
+    {
+      "epoch": 1.3566254793353216,
+      "grad_norm": 13.5,
+      "learning_rate": 5.904057751816473e-07,
+      "loss": 0.6149,
+      "step": 15920
+    },
+    {
+      "epoch": 1.3567106945036216,
+      "grad_norm": 15.0,
+      "learning_rate": 5.902630439970926e-07,
+      "loss": 0.3949,
+      "step": 15921
+    },
+    {
+      "epoch": 1.3567959096719215,
+      "grad_norm": 24.875,
+      "learning_rate": 5.901203247346481e-07,
+      "loss": 1.1055,
+      "step": 15922
+    },
+    {
+      "epoch": 1.3568811248402215,
+      "grad_norm": 16.125,
+      "learning_rate": 5.899776173968934e-07,
+      "loss": 0.8457,
+      "step": 15923
+    },
+    {
+      "epoch": 1.3569663400085215,
+      "grad_norm": 14.5,
+      "learning_rate": 5.898349219864064e-07,
+      "loss": 0.6871,
+      "step": 15924
+    },
+    {
+      "epoch": 1.3570515551768214,
+      "grad_norm": 14.75,
+      "learning_rate": 5.896922385057666e-07,
+      "loss": 0.6489,
+      "step": 15925
+    },
+    {
+      "epoch": 1.3571367703451214,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.895495669575515e-07,
+      "loss": 0.2933,
+      "step": 15926
+    },
+    {
+      "epoch": 1.3572219855134213,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.894069073443401e-07,
+      "loss": 0.5319,
+      "step": 15927
+    },
+    {
+      "epoch": 1.3573072006817213,
+      "grad_norm": 11.5,
+      "learning_rate": 5.892642596687101e-07,
+      "loss": 0.3115,
+      "step": 15928
+    },
+    {
+      "epoch": 1.3573924158500212,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.891216239332397e-07,
+      "loss": 0.6167,
+      "step": 15929
+    },
+    {
+      "epoch": 1.3574776310183212,
+      "grad_norm": 14.875,
+      "learning_rate": 5.889790001405058e-07,
+      "loss": 0.481,
+      "step": 15930
+    },
+    {
+      "epoch": 1.3575628461866212,
+      "grad_norm": 15.875,
+      "learning_rate": 5.888363882930865e-07,
+      "loss": 0.8069,
+      "step": 15931
+    },
+    {
+      "epoch": 1.357648061354921,
+      "grad_norm": 16.375,
+      "learning_rate": 5.886937883935579e-07,
+      "loss": 0.8347,
+      "step": 15932
+    },
+    {
+      "epoch": 1.357733276523221,
+      "grad_norm": 15.0,
+      "learning_rate": 5.885512004444981e-07,
+      "loss": 0.6962,
+      "step": 15933
+    },
+    {
+      "epoch": 1.357818491691521,
+      "grad_norm": 15.75,
+      "learning_rate": 5.884086244484828e-07,
+      "loss": 0.5843,
+      "step": 15934
+    },
+    {
+      "epoch": 1.357903706859821,
+      "grad_norm": 12.25,
+      "learning_rate": 5.882660604080889e-07,
+      "loss": 0.4408,
+      "step": 15935
+    },
+    {
+      "epoch": 1.357988922028121,
+      "grad_norm": 18.625,
+      "learning_rate": 5.881235083258928e-07,
+      "loss": 0.4395,
+      "step": 15936
+    },
+    {
+      "epoch": 1.358074137196421,
+      "grad_norm": 12.75,
+      "learning_rate": 5.879809682044708e-07,
+      "loss": 0.7589,
+      "step": 15937
+    },
+    {
+      "epoch": 1.3581593523647209,
+      "grad_norm": 18.125,
+      "learning_rate": 5.878384400463981e-07,
+      "loss": 0.5083,
+      "step": 15938
+    },
+    {
+      "epoch": 1.3582445675330208,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.87695923854251e-07,
+      "loss": 0.5216,
+      "step": 15939
+    },
+    {
+      "epoch": 1.3583297827013208,
+      "grad_norm": 15.25,
+      "learning_rate": 5.875534196306042e-07,
+      "loss": 0.4046,
+      "step": 15940
+    },
+    {
+      "epoch": 1.3584149978696207,
+      "grad_norm": 16.125,
+      "learning_rate": 5.87410927378033e-07,
+      "loss": 0.6059,
+      "step": 15941
+    },
+    {
+      "epoch": 1.3585002130379207,
+      "grad_norm": 15.375,
+      "learning_rate": 5.872684470991134e-07,
+      "loss": 0.4912,
+      "step": 15942
+    },
+    {
+      "epoch": 1.3585854282062206,
+      "grad_norm": 13.0625,
+      "learning_rate": 5.871259787964189e-07,
+      "loss": 0.5477,
+      "step": 15943
+    },
+    {
+      "epoch": 1.3586706433745206,
+      "grad_norm": 12.6875,
+      "learning_rate": 5.869835224725243e-07,
+      "loss": 0.2879,
+      "step": 15944
+    },
+    {
+      "epoch": 1.3587558585428205,
+      "grad_norm": 11.6875,
+      "learning_rate": 5.868410781300044e-07,
+      "loss": 0.2783,
+      "step": 15945
+    },
+    {
+      "epoch": 1.3588410737111205,
+      "grad_norm": 17.75,
+      "learning_rate": 5.866986457714332e-07,
+      "loss": 0.5897,
+      "step": 15946
+    },
+    {
+      "epoch": 1.3589262888794205,
+      "grad_norm": 15.75,
+      "learning_rate": 5.865562253993848e-07,
+      "loss": 0.618,
+      "step": 15947
+    },
+    {
+      "epoch": 1.3590115040477204,
+      "grad_norm": 23.0,
+      "learning_rate": 5.86413817016432e-07,
+      "loss": 0.6742,
+      "step": 15948
+    },
+    {
+      "epoch": 1.3590967192160204,
+      "grad_norm": 14.125,
+      "learning_rate": 5.862714206251488e-07,
+      "loss": 0.5644,
+      "step": 15949
+    },
+    {
+      "epoch": 1.3591819343843203,
+      "grad_norm": 30.125,
+      "learning_rate": 5.86129036228109e-07,
+      "loss": 0.6692,
+      "step": 15950
+    },
+    {
+      "epoch": 1.3592671495526203,
+      "grad_norm": 26.0,
+      "learning_rate": 5.859866638278845e-07,
+      "loss": 0.3863,
+      "step": 15951
+    },
+    {
+      "epoch": 1.3593523647209202,
+      "grad_norm": 11.0,
+      "learning_rate": 5.858443034270488e-07,
+      "loss": 0.3295,
+      "step": 15952
+    },
+    {
+      "epoch": 1.3594375798892202,
+      "grad_norm": 15.125,
+      "learning_rate": 5.857019550281744e-07,
+      "loss": 0.4746,
+      "step": 15953
+    },
+    {
+      "epoch": 1.3595227950575202,
+      "grad_norm": 17.125,
+      "learning_rate": 5.85559618633834e-07,
+      "loss": 0.4506,
+      "step": 15954
+    },
+    {
+      "epoch": 1.3596080102258201,
+      "grad_norm": 14.0625,
+      "learning_rate": 5.854172942465995e-07,
+      "loss": 0.3714,
+      "step": 15955
+    },
+    {
+      "epoch": 1.35969322539412,
+      "grad_norm": 15.75,
+      "learning_rate": 5.852749818690425e-07,
+      "loss": 0.6052,
+      "step": 15956
+    },
+    {
+      "epoch": 1.35977844056242,
+      "grad_norm": 14.5,
+      "learning_rate": 5.851326815037347e-07,
+      "loss": 0.69,
+      "step": 15957
+    },
+    {
+      "epoch": 1.35986365573072,
+      "grad_norm": 18.625,
+      "learning_rate": 5.849903931532482e-07,
+      "loss": 0.5228,
+      "step": 15958
+    },
+    {
+      "epoch": 1.35994887089902,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.848481168201543e-07,
+      "loss": 0.424,
+      "step": 15959
+    },
+    {
+      "epoch": 1.36003408606732,
+      "grad_norm": 13.75,
+      "learning_rate": 5.847058525070235e-07,
+      "loss": 0.6474,
+      "step": 15960
+    },
+    {
+      "epoch": 1.3601193012356199,
+      "grad_norm": 16.75,
+      "learning_rate": 5.845636002164267e-07,
+      "loss": 0.6883,
+      "step": 15961
+    },
+    {
+      "epoch": 1.3602045164039198,
+      "grad_norm": 15.5625,
+      "learning_rate": 5.844213599509349e-07,
+      "loss": 0.5883,
+      "step": 15962
+    },
+    {
+      "epoch": 1.3602897315722198,
+      "grad_norm": 18.125,
+      "learning_rate": 5.842791317131194e-07,
+      "loss": 0.5319,
+      "step": 15963
+    },
+    {
+      "epoch": 1.3603749467405197,
+      "grad_norm": 18.375,
+      "learning_rate": 5.841369155055485e-07,
+      "loss": 0.7101,
+      "step": 15964
+    },
+    {
+      "epoch": 1.3604601619088197,
+      "grad_norm": 17.125,
+      "learning_rate": 5.839947113307931e-07,
+      "loss": 0.6981,
+      "step": 15965
+    },
+    {
+      "epoch": 1.3605453770771196,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.838525191914229e-07,
+      "loss": 0.4287,
+      "step": 15966
+    },
+    {
+      "epoch": 1.3606305922454198,
+      "grad_norm": 21.75,
+      "learning_rate": 5.837103390900081e-07,
+      "loss": 0.6906,
+      "step": 15967
+    },
+    {
+      "epoch": 1.3607158074137198,
+      "grad_norm": 17.0,
+      "learning_rate": 5.83568171029117e-07,
+      "loss": 0.7192,
+      "step": 15968
+    },
+    {
+      "epoch": 1.3608010225820197,
+      "grad_norm": 15.25,
+      "learning_rate": 5.834260150113192e-07,
+      "loss": 0.4449,
+      "step": 15969
+    },
+    {
+      "epoch": 1.3608862377503197,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.832838710391836e-07,
+      "loss": 0.4807,
+      "step": 15970
+    },
+    {
+      "epoch": 1.3609714529186197,
+      "grad_norm": 17.25,
+      "learning_rate": 5.831417391152794e-07,
+      "loss": 0.7733,
+      "step": 15971
+    },
+    {
+      "epoch": 1.3610566680869196,
+      "grad_norm": 12.375,
+      "learning_rate": 5.829996192421744e-07,
+      "loss": 0.4813,
+      "step": 15972
+    },
+    {
+      "epoch": 1.3611418832552196,
+      "grad_norm": 15.125,
+      "learning_rate": 5.828575114224366e-07,
+      "loss": 0.8302,
+      "step": 15973
+    },
+    {
+      "epoch": 1.3612270984235195,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.827154156586346e-07,
+      "loss": 0.5226,
+      "step": 15974
+    },
+    {
+      "epoch": 1.3613123135918195,
+      "grad_norm": 14.875,
+      "learning_rate": 5.825733319533359e-07,
+      "loss": 0.6715,
+      "step": 15975
+    },
+    {
+      "epoch": 1.3613975287601194,
+      "grad_norm": 19.25,
+      "learning_rate": 5.824312603091087e-07,
+      "loss": 0.779,
+      "step": 15976
+    },
+    {
+      "epoch": 1.3614827439284194,
+      "grad_norm": 23.625,
+      "learning_rate": 5.822892007285194e-07,
+      "loss": 0.6616,
+      "step": 15977
+    },
+    {
+      "epoch": 1.3615679590967193,
+      "grad_norm": 11.5625,
+      "learning_rate": 5.821471532141357e-07,
+      "loss": 0.2809,
+      "step": 15978
+    },
+    {
+      "epoch": 1.3616531742650193,
+      "grad_norm": 10.125,
+      "learning_rate": 5.820051177685246e-07,
+      "loss": 0.3028,
+      "step": 15979
+    },
+    {
+      "epoch": 1.3617383894333193,
+      "grad_norm": 19.375,
+      "learning_rate": 5.818630943942537e-07,
+      "loss": 0.678,
+      "step": 15980
+    },
+    {
+      "epoch": 1.3618236046016192,
+      "grad_norm": 17.875,
+      "learning_rate": 5.817210830938876e-07,
+      "loss": 0.5908,
+      "step": 15981
+    },
+    {
+      "epoch": 1.3619088197699192,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.815790838699934e-07,
+      "loss": 0.5574,
+      "step": 15982
+    },
+    {
+      "epoch": 1.3619940349382191,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.814370967251373e-07,
+      "loss": 0.6033,
+      "step": 15983
+    },
+    {
+      "epoch": 1.362079250106519,
+      "grad_norm": 17.875,
+      "learning_rate": 5.812951216618858e-07,
+      "loss": 0.7202,
+      "step": 15984
+    },
+    {
+      "epoch": 1.362164465274819,
+      "grad_norm": 20.375,
+      "learning_rate": 5.811531586828033e-07,
+      "loss": 0.9735,
+      "step": 15985
+    },
+    {
+      "epoch": 1.362249680443119,
+      "grad_norm": 26.375,
+      "learning_rate": 5.810112077904559e-07,
+      "loss": 0.5277,
+      "step": 15986
+    },
+    {
+      "epoch": 1.362334895611419,
+      "grad_norm": 19.875,
+      "learning_rate": 5.808692689874086e-07,
+      "loss": 1.0465,
+      "step": 15987
+    },
+    {
+      "epoch": 1.362420110779719,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.80727342276227e-07,
+      "loss": 0.6369,
+      "step": 15988
+    },
+    {
+      "epoch": 1.3625053259480189,
+      "grad_norm": 19.375,
+      "learning_rate": 5.805854276594753e-07,
+      "loss": 0.8363,
+      "step": 15989
+    },
+    {
+      "epoch": 1.3625905411163188,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.804435251397176e-07,
+      "loss": 0.6467,
+      "step": 15990
+    },
+    {
+      "epoch": 1.3626757562846188,
+      "grad_norm": 15.25,
+      "learning_rate": 5.803016347195187e-07,
+      "loss": 0.8417,
+      "step": 15991
+    },
+    {
+      "epoch": 1.3627609714529187,
+      "grad_norm": 11.1875,
+      "learning_rate": 5.801597564014426e-07,
+      "loss": 0.315,
+      "step": 15992
+    },
+    {
+      "epoch": 1.3628461866212187,
+      "grad_norm": 13.875,
+      "learning_rate": 5.800178901880539e-07,
+      "loss": 0.3547,
+      "step": 15993
+    },
+    {
+      "epoch": 1.3629314017895187,
+      "grad_norm": 11.1875,
+      "learning_rate": 5.79876036081915e-07,
+      "loss": 0.9094,
+      "step": 15994
+    },
+    {
+      "epoch": 1.3630166169578186,
+      "grad_norm": 14.625,
+      "learning_rate": 5.797341940855901e-07,
+      "loss": 0.4358,
+      "step": 15995
+    },
+    {
+      "epoch": 1.3631018321261186,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.795923642016428e-07,
+      "loss": 0.9059,
+      "step": 15996
+    },
+    {
+      "epoch": 1.3631870472944185,
+      "grad_norm": 16.5,
+      "learning_rate": 5.794505464326357e-07,
+      "loss": 0.7799,
+      "step": 15997
+    },
+    {
+      "epoch": 1.3632722624627185,
+      "grad_norm": 13.375,
+      "learning_rate": 5.793087407811309e-07,
+      "loss": 0.4759,
+      "step": 15998
+    },
+    {
+      "epoch": 1.3633574776310184,
+      "grad_norm": 13.25,
+      "learning_rate": 5.791669472496916e-07,
+      "loss": 0.5153,
+      "step": 15999
+    },
+    {
+      "epoch": 1.3634426927993184,
+      "grad_norm": 19.5,
+      "learning_rate": 5.790251658408802e-07,
+      "loss": 0.7079,
+      "step": 16000
+    },
+    {
+      "epoch": 1.3635279079676184,
+      "grad_norm": 16.5,
+      "learning_rate": 5.788833965572592e-07,
+      "loss": 0.4141,
+      "step": 16001
+    },
+    {
+      "epoch": 1.3636131231359183,
+      "grad_norm": 16.25,
+      "learning_rate": 5.787416394013897e-07,
+      "loss": 0.5057,
+      "step": 16002
+    },
+    {
+      "epoch": 1.3636983383042183,
+      "grad_norm": 11.625,
+      "learning_rate": 5.785998943758336e-07,
+      "loss": 0.344,
+      "step": 16003
+    },
+    {
+      "epoch": 1.3637835534725182,
+      "grad_norm": 16.625,
+      "learning_rate": 5.784581614831532e-07,
+      "loss": 0.5798,
+      "step": 16004
+    },
+    {
+      "epoch": 1.3638687686408182,
+      "grad_norm": 18.625,
+      "learning_rate": 5.783164407259086e-07,
+      "loss": 0.5144,
+      "step": 16005
+    },
+    {
+      "epoch": 1.3639539838091181,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.781747321066617e-07,
+      "loss": 0.6022,
+      "step": 16006
+    },
+    {
+      "epoch": 1.364039198977418,
+      "grad_norm": 13.875,
+      "learning_rate": 5.780330356279728e-07,
+      "loss": 0.5603,
+      "step": 16007
+    },
+    {
+      "epoch": 1.364124414145718,
+      "grad_norm": 16.125,
+      "learning_rate": 5.778913512924024e-07,
+      "loss": 0.5789,
+      "step": 16008
+    },
+    {
+      "epoch": 1.364209629314018,
+      "grad_norm": 16.0,
+      "learning_rate": 5.777496791025111e-07,
+      "loss": 0.6167,
+      "step": 16009
+    },
+    {
+      "epoch": 1.364294844482318,
+      "grad_norm": 13.125,
+      "learning_rate": 5.776080190608596e-07,
+      "loss": 0.2654,
+      "step": 16010
+    },
+    {
+      "epoch": 1.364380059650618,
+      "grad_norm": 18.375,
+      "learning_rate": 5.77466371170007e-07,
+      "loss": 0.582,
+      "step": 16011
+    },
+    {
+      "epoch": 1.3644652748189179,
+      "grad_norm": 19.125,
+      "learning_rate": 5.773247354325137e-07,
+      "loss": 0.7123,
+      "step": 16012
+    },
+    {
+      "epoch": 1.3645504899872178,
+      "grad_norm": 17.25,
+      "learning_rate": 5.771831118509384e-07,
+      "loss": 0.4806,
+      "step": 16013
+    },
+    {
+      "epoch": 1.3646357051555178,
+      "grad_norm": 13.375,
+      "learning_rate": 5.770415004278413e-07,
+      "loss": 0.6413,
+      "step": 16014
+    },
+    {
+      "epoch": 1.3647209203238178,
+      "grad_norm": 15.875,
+      "learning_rate": 5.768999011657806e-07,
+      "loss": 0.4478,
+      "step": 16015
+    },
+    {
+      "epoch": 1.3648061354921177,
+      "grad_norm": 19.5,
+      "learning_rate": 5.767583140673156e-07,
+      "loss": 1.0446,
+      "step": 16016
+    },
+    {
+      "epoch": 1.3648913506604177,
+      "grad_norm": 16.625,
+      "learning_rate": 5.766167391350048e-07,
+      "loss": 0.6855,
+      "step": 16017
+    },
+    {
+      "epoch": 1.3649765658287176,
+      "grad_norm": 16.75,
+      "learning_rate": 5.764751763714072e-07,
+      "loss": 0.6523,
+      "step": 16018
+    },
+    {
+      "epoch": 1.3650617809970176,
+      "grad_norm": 16.875,
+      "learning_rate": 5.763336257790798e-07,
+      "loss": 0.7014,
+      "step": 16019
+    },
+    {
+      "epoch": 1.3651469961653175,
+      "grad_norm": 12.25,
+      "learning_rate": 5.761920873605818e-07,
+      "loss": 0.469,
+      "step": 16020
+    },
+    {
+      "epoch": 1.3652322113336175,
+      "grad_norm": 19.0,
+      "learning_rate": 5.760505611184698e-07,
+      "loss": 0.9983,
+      "step": 16021
+    },
+    {
+      "epoch": 1.3653174265019175,
+      "grad_norm": 12.5625,
+      "learning_rate": 5.759090470553019e-07,
+      "loss": 0.4177,
+      "step": 16022
+    },
+    {
+      "epoch": 1.3654026416702174,
+      "grad_norm": 15.0,
+      "learning_rate": 5.757675451736358e-07,
+      "loss": 0.6135,
+      "step": 16023
+    },
+    {
+      "epoch": 1.3654878568385174,
+      "grad_norm": 16.625,
+      "learning_rate": 5.756260554760278e-07,
+      "loss": 0.6252,
+      "step": 16024
+    },
+    {
+      "epoch": 1.3655730720068173,
+      "grad_norm": 17.5,
+      "learning_rate": 5.75484577965035e-07,
+      "loss": 0.8448,
+      "step": 16025
+    },
+    {
+      "epoch": 1.3656582871751173,
+      "grad_norm": 16.25,
+      "learning_rate": 5.753431126432147e-07,
+      "loss": 0.6765,
+      "step": 16026
+    },
+    {
+      "epoch": 1.3657435023434172,
+      "grad_norm": 21.875,
+      "learning_rate": 5.752016595131223e-07,
+      "loss": 0.6313,
+      "step": 16027
+    },
+    {
+      "epoch": 1.3658287175117172,
+      "grad_norm": 16.5,
+      "learning_rate": 5.750602185773149e-07,
+      "loss": 0.5846,
+      "step": 16028
+    },
+    {
+      "epoch": 1.3659139326800172,
+      "grad_norm": 22.25,
+      "learning_rate": 5.749187898383476e-07,
+      "loss": 0.7186,
+      "step": 16029
+    },
+    {
+      "epoch": 1.365999147848317,
+      "grad_norm": 18.25,
+      "learning_rate": 5.747773732987765e-07,
+      "loss": 0.9485,
+      "step": 16030
+    },
+    {
+      "epoch": 1.366084363016617,
+      "grad_norm": 17.25,
+      "learning_rate": 5.74635968961158e-07,
+      "loss": 0.58,
+      "step": 16031
+    },
+    {
+      "epoch": 1.366169578184917,
+      "grad_norm": 17.25,
+      "learning_rate": 5.74494576828046e-07,
+      "loss": 0.6079,
+      "step": 16032
+    },
+    {
+      "epoch": 1.366254793353217,
+      "grad_norm": 15.5,
+      "learning_rate": 5.743531969019964e-07,
+      "loss": 0.6933,
+      "step": 16033
+    },
+    {
+      "epoch": 1.366340008521517,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.742118291855637e-07,
+      "loss": 0.7072,
+      "step": 16034
+    },
+    {
+      "epoch": 1.366425223689817,
+      "grad_norm": 18.25,
+      "learning_rate": 5.740704736813035e-07,
+      "loss": 0.6852,
+      "step": 16035
+    },
+    {
+      "epoch": 1.3665104388581168,
+      "grad_norm": 15.5,
+      "learning_rate": 5.739291303917695e-07,
+      "loss": 0.5727,
+      "step": 16036
+    },
+    {
+      "epoch": 1.3665956540264168,
+      "grad_norm": 17.875,
+      "learning_rate": 5.737877993195153e-07,
+      "loss": 1.0102,
+      "step": 16037
+    },
+    {
+      "epoch": 1.3666808691947168,
+      "grad_norm": 12.0625,
+      "learning_rate": 5.736464804670959e-07,
+      "loss": 0.3029,
+      "step": 16038
+    },
+    {
+      "epoch": 1.3667660843630167,
+      "grad_norm": 23.625,
+      "learning_rate": 5.735051738370646e-07,
+      "loss": 0.7543,
+      "step": 16039
+    },
+    {
+      "epoch": 1.3668512995313167,
+      "grad_norm": 10.0625,
+      "learning_rate": 5.733638794319754e-07,
+      "loss": 0.2312,
+      "step": 16040
+    },
+    {
+      "epoch": 1.3669365146996166,
+      "grad_norm": 13.4375,
+      "learning_rate": 5.732225972543809e-07,
+      "loss": 0.6068,
+      "step": 16041
+    },
+    {
+      "epoch": 1.3670217298679166,
+      "grad_norm": 18.5,
+      "learning_rate": 5.730813273068346e-07,
+      "loss": 0.8657,
+      "step": 16042
+    },
+    {
+      "epoch": 1.3671069450362165,
+      "grad_norm": 13.0,
+      "learning_rate": 5.729400695918899e-07,
+      "loss": 0.4295,
+      "step": 16043
+    },
+    {
+      "epoch": 1.3671921602045165,
+      "grad_norm": 12.375,
+      "learning_rate": 5.727988241120986e-07,
+      "loss": 0.4244,
+      "step": 16044
+    },
+    {
+      "epoch": 1.3672773753728165,
+      "grad_norm": 17.25,
+      "learning_rate": 5.726575908700139e-07,
+      "loss": 0.5043,
+      "step": 16045
+    },
+    {
+      "epoch": 1.3673625905411164,
+      "grad_norm": 21.75,
+      "learning_rate": 5.725163698681873e-07,
+      "loss": 0.7034,
+      "step": 16046
+    },
+    {
+      "epoch": 1.3674478057094164,
+      "grad_norm": 16.25,
+      "learning_rate": 5.72375161109171e-07,
+      "loss": 0.5909,
+      "step": 16047
+    },
+    {
+      "epoch": 1.3675330208777163,
+      "grad_norm": 11.25,
+      "learning_rate": 5.722339645955175e-07,
+      "loss": 0.5964,
+      "step": 16048
+    },
+    {
+      "epoch": 1.3676182360460163,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.720927803297772e-07,
+      "loss": 0.3673,
+      "step": 16049
+    },
+    {
+      "epoch": 1.3677034512143162,
+      "grad_norm": 22.375,
+      "learning_rate": 5.719516083145023e-07,
+      "loss": 0.777,
+      "step": 16050
+    },
+    {
+      "epoch": 1.3677886663826162,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.718104485522435e-07,
+      "loss": 0.3528,
+      "step": 16051
+    },
+    {
+      "epoch": 1.3678738815509162,
+      "grad_norm": 14.625,
+      "learning_rate": 5.716693010455524e-07,
+      "loss": 0.5524,
+      "step": 16052
+    },
+    {
+      "epoch": 1.3679590967192161,
+      "grad_norm": 22.75,
+      "learning_rate": 5.71528165796979e-07,
+      "loss": 1.1691,
+      "step": 16053
+    },
+    {
+      "epoch": 1.368044311887516,
+      "grad_norm": 13.0,
+      "learning_rate": 5.713870428090735e-07,
+      "loss": 0.3076,
+      "step": 16054
+    },
+    {
+      "epoch": 1.368129527055816,
+      "grad_norm": 21.375,
+      "learning_rate": 5.712459320843867e-07,
+      "loss": 0.7869,
+      "step": 16055
+    },
+    {
+      "epoch": 1.368214742224116,
+      "grad_norm": 18.25,
+      "learning_rate": 5.711048336254688e-07,
+      "loss": 0.8944,
+      "step": 16056
+    },
+    {
+      "epoch": 1.368299957392416,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.709637474348688e-07,
+      "loss": 0.4312,
+      "step": 16057
+    },
+    {
+      "epoch": 1.368385172560716,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.708226735151366e-07,
+      "loss": 0.3191,
+      "step": 16058
+    },
+    {
+      "epoch": 1.3684703877290159,
+      "grad_norm": 18.5,
+      "learning_rate": 5.706816118688217e-07,
+      "loss": 0.7212,
+      "step": 16059
+    },
+    {
+      "epoch": 1.3685556028973158,
+      "grad_norm": 32.75,
+      "learning_rate": 5.705405624984738e-07,
+      "loss": 0.6661,
+      "step": 16060
+    },
+    {
+      "epoch": 1.3686408180656158,
+      "grad_norm": 19.0,
+      "learning_rate": 5.703995254066409e-07,
+      "loss": 0.6797,
+      "step": 16061
+    },
+    {
+      "epoch": 1.3687260332339157,
+      "grad_norm": 33.5,
+      "learning_rate": 5.702585005958718e-07,
+      "loss": 1.162,
+      "step": 16062
+    },
+    {
+      "epoch": 1.3688112484022157,
+      "grad_norm": 13.625,
+      "learning_rate": 5.701174880687149e-07,
+      "loss": 0.3124,
+      "step": 16063
+    },
+    {
+      "epoch": 1.3688964635705156,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.699764878277188e-07,
+      "loss": 0.7422,
+      "step": 16064
+    },
+    {
+      "epoch": 1.3689816787388156,
+      "grad_norm": 17.75,
+      "learning_rate": 5.698354998754319e-07,
+      "loss": 0.7739,
+      "step": 16065
+    },
+    {
+      "epoch": 1.3690668939071156,
+      "grad_norm": 11.75,
+      "learning_rate": 5.696945242144008e-07,
+      "loss": 0.4205,
+      "step": 16066
+    },
+    {
+      "epoch": 1.3691521090754155,
+      "grad_norm": 16.0,
+      "learning_rate": 5.695535608471739e-07,
+      "loss": 0.3947,
+      "step": 16067
+    },
+    {
+      "epoch": 1.3692373242437155,
+      "grad_norm": 9.625,
+      "learning_rate": 5.694126097762984e-07,
+      "loss": 0.4074,
+      "step": 16068
+    },
+    {
+      "epoch": 1.3693225394120154,
+      "grad_norm": 16.0,
+      "learning_rate": 5.692716710043218e-07,
+      "loss": 0.5328,
+      "step": 16069
+    },
+    {
+      "epoch": 1.3694077545803154,
+      "grad_norm": 18.25,
+      "learning_rate": 5.691307445337906e-07,
+      "loss": 0.8244,
+      "step": 16070
+    },
+    {
+      "epoch": 1.3694929697486153,
+      "grad_norm": 16.375,
+      "learning_rate": 5.689898303672512e-07,
+      "loss": 0.6688,
+      "step": 16071
+    },
+    {
+      "epoch": 1.3695781849169153,
+      "grad_norm": 19.625,
+      "learning_rate": 5.688489285072503e-07,
+      "loss": 1.0184,
+      "step": 16072
+    },
+    {
+      "epoch": 1.3696634000852153,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.687080389563346e-07,
+      "loss": 0.6495,
+      "step": 16073
+    },
+    {
+      "epoch": 1.3697486152535152,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.685671617170493e-07,
+      "loss": 0.4662,
+      "step": 16074
+    },
+    {
+      "epoch": 1.3698338304218152,
+      "grad_norm": 19.375,
+      "learning_rate": 5.684262967919407e-07,
+      "loss": 0.9025,
+      "step": 16075
+    },
+    {
+      "epoch": 1.3699190455901151,
+      "grad_norm": 19.25,
+      "learning_rate": 5.682854441835543e-07,
+      "loss": 0.9998,
+      "step": 16076
+    },
+    {
+      "epoch": 1.370004260758415,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.681446038944359e-07,
+      "loss": 0.6481,
+      "step": 16077
+    },
+    {
+      "epoch": 1.370089475926715,
+      "grad_norm": 15.125,
+      "learning_rate": 5.680037759271301e-07,
+      "loss": 0.2297,
+      "step": 16078
+    },
+    {
+      "epoch": 1.370174691095015,
+      "grad_norm": 19.75,
+      "learning_rate": 5.678629602841815e-07,
+      "loss": 0.5858,
+      "step": 16079
+    },
+    {
+      "epoch": 1.370259906263315,
+      "grad_norm": 22.25,
+      "learning_rate": 5.67722156968135e-07,
+      "loss": 0.8521,
+      "step": 16080
+    },
+    {
+      "epoch": 1.370345121431615,
+      "grad_norm": 33.5,
+      "learning_rate": 5.675813659815354e-07,
+      "loss": 0.9786,
+      "step": 16081
+    },
+    {
+      "epoch": 1.3704303365999149,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.67440587326927e-07,
+      "loss": 0.6121,
+      "step": 16082
+    },
+    {
+      "epoch": 1.3705155517682148,
+      "grad_norm": 16.75,
+      "learning_rate": 5.672998210068531e-07,
+      "loss": 0.6145,
+      "step": 16083
+    },
+    {
+      "epoch": 1.3706007669365148,
+      "grad_norm": 12.25,
+      "learning_rate": 5.671590670238581e-07,
+      "loss": 0.2772,
+      "step": 16084
+    },
+    {
+      "epoch": 1.3706859821048147,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.670183253804857e-07,
+      "loss": 0.5794,
+      "step": 16085
+    },
+    {
+      "epoch": 1.3707711972731147,
+      "grad_norm": 17.375,
+      "learning_rate": 5.668775960792791e-07,
+      "loss": 0.7444,
+      "step": 16086
+    },
+    {
+      "epoch": 1.3708564124414147,
+      "grad_norm": 18.125,
+      "learning_rate": 5.667368791227808e-07,
+      "loss": 0.8679,
+      "step": 16087
+    },
+    {
+      "epoch": 1.3709416276097146,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.665961745135342e-07,
+      "loss": 0.6098,
+      "step": 16088
+    },
+    {
+      "epoch": 1.3710268427780146,
+      "grad_norm": 16.375,
+      "learning_rate": 5.664554822540817e-07,
+      "loss": 0.6882,
+      "step": 16089
+    },
+    {
+      "epoch": 1.3711120579463145,
+      "grad_norm": 14.0,
+      "learning_rate": 5.663148023469667e-07,
+      "loss": 0.3834,
+      "step": 16090
+    },
+    {
+      "epoch": 1.3711972731146145,
+      "grad_norm": 18.125,
+      "learning_rate": 5.661741347947301e-07,
+      "loss": 0.7943,
+      "step": 16091
+    },
+    {
+      "epoch": 1.3712824882829144,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.660334795999146e-07,
+      "loss": 0.8472,
+      "step": 16092
+    },
+    {
+      "epoch": 1.3713677034512144,
+      "grad_norm": 13.0,
+      "learning_rate": 5.658928367650625e-07,
+      "loss": 0.4627,
+      "step": 16093
+    },
+    {
+      "epoch": 1.3714529186195143,
+      "grad_norm": 14.75,
+      "learning_rate": 5.657522062927142e-07,
+      "loss": 0.5775,
+      "step": 16094
+    },
+    {
+      "epoch": 1.3715381337878143,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.65611588185412e-07,
+      "loss": 0.8127,
+      "step": 16095
+    },
+    {
+      "epoch": 1.3716233489561143,
+      "grad_norm": 14.75,
+      "learning_rate": 5.65470982445696e-07,
+      "loss": 0.4273,
+      "step": 16096
+    },
+    {
+      "epoch": 1.3717085641244142,
+      "grad_norm": 18.25,
+      "learning_rate": 5.653303890761081e-07,
+      "loss": 0.7115,
+      "step": 16097
+    },
+    {
+      "epoch": 1.3717937792927142,
+      "grad_norm": 17.25,
+      "learning_rate": 5.651898080791881e-07,
+      "loss": 0.4268,
+      "step": 16098
+    },
+    {
+      "epoch": 1.3718789944610141,
+      "grad_norm": 12.75,
+      "learning_rate": 5.650492394574777e-07,
+      "loss": 0.4857,
+      "step": 16099
+    },
+    {
+      "epoch": 1.371964209629314,
+      "grad_norm": 16.75,
+      "learning_rate": 5.649086832135158e-07,
+      "loss": 0.6239,
+      "step": 16100
+    },
+    {
+      "epoch": 1.372049424797614,
+      "grad_norm": 12.0625,
+      "learning_rate": 5.647681393498433e-07,
+      "loss": 0.4281,
+      "step": 16101
+    },
+    {
+      "epoch": 1.372134639965914,
+      "grad_norm": 15.25,
+      "learning_rate": 5.646276078689991e-07,
+      "loss": 0.6254,
+      "step": 16102
+    },
+    {
+      "epoch": 1.372219855134214,
+      "grad_norm": 15.5,
+      "learning_rate": 5.644870887735236e-07,
+      "loss": 0.8104,
+      "step": 16103
+    },
+    {
+      "epoch": 1.372305070302514,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.643465820659555e-07,
+      "loss": 0.4898,
+      "step": 16104
+    },
+    {
+      "epoch": 1.3723902854708139,
+      "grad_norm": 16.0,
+      "learning_rate": 5.64206087748834e-07,
+      "loss": 0.7035,
+      "step": 16105
+    },
+    {
+      "epoch": 1.3724755006391138,
+      "grad_norm": 18.25,
+      "learning_rate": 5.64065605824698e-07,
+      "loss": 0.7884,
+      "step": 16106
+    },
+    {
+      "epoch": 1.3725607158074138,
+      "grad_norm": 15.625,
+      "learning_rate": 5.63925136296087e-07,
+      "loss": 0.471,
+      "step": 16107
+    },
+    {
+      "epoch": 1.3726459309757137,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.637846791655381e-07,
+      "loss": 0.7667,
+      "step": 16108
+    },
+    {
+      "epoch": 1.3727311461440137,
+      "grad_norm": 14.25,
+      "learning_rate": 5.636442344355901e-07,
+      "loss": 0.6779,
+      "step": 16109
+    },
+    {
+      "epoch": 1.3728163613123137,
+      "grad_norm": 13.0,
+      "learning_rate": 5.635038021087815e-07,
+      "loss": 0.3917,
+      "step": 16110
+    },
+    {
+      "epoch": 1.3729015764806136,
+      "grad_norm": 12.375,
+      "learning_rate": 5.633633821876488e-07,
+      "loss": 0.4201,
+      "step": 16111
+    },
+    {
+      "epoch": 1.3729867916489136,
+      "grad_norm": 12.4375,
+      "learning_rate": 5.63222974674731e-07,
+      "loss": 0.5646,
+      "step": 16112
+    },
+    {
+      "epoch": 1.3730720068172135,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.630825795725641e-07,
+      "loss": 0.6594,
+      "step": 16113
+    },
+    {
+      "epoch": 1.3731572219855135,
+      "grad_norm": 17.125,
+      "learning_rate": 5.629421968836857e-07,
+      "loss": 0.6535,
+      "step": 16114
+    },
+    {
+      "epoch": 1.3732424371538134,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.628018266106326e-07,
+      "loss": 0.5179,
+      "step": 16115
+    },
+    {
+      "epoch": 1.3733276523221134,
+      "grad_norm": 12.25,
+      "learning_rate": 5.626614687559422e-07,
+      "loss": 0.4165,
+      "step": 16116
+    },
+    {
+      "epoch": 1.3734128674904134,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.625211233221496e-07,
+      "loss": 0.4319,
+      "step": 16117
+    },
+    {
+      "epoch": 1.3734980826587133,
+      "grad_norm": 16.625,
+      "learning_rate": 5.623807903117921e-07,
+      "loss": 0.7363,
+      "step": 16118
+    },
+    {
+      "epoch": 1.3735832978270133,
+      "grad_norm": 17.0,
+      "learning_rate": 5.622404697274047e-07,
+      "loss": 0.4946,
+      "step": 16119
+    },
+    {
+      "epoch": 1.3736685129953132,
+      "grad_norm": 12.1875,
+      "learning_rate": 5.621001615715242e-07,
+      "loss": 0.3595,
+      "step": 16120
+    },
+    {
+      "epoch": 1.3737537281636132,
+      "grad_norm": 15.875,
+      "learning_rate": 5.61959865846685e-07,
+      "loss": 0.7856,
+      "step": 16121
+    },
+    {
+      "epoch": 1.3738389433319131,
+      "grad_norm": 15.4375,
+      "learning_rate": 5.618195825554229e-07,
+      "loss": 0.5068,
+      "step": 16122
+    },
+    {
+      "epoch": 1.373924158500213,
+      "grad_norm": 11.125,
+      "learning_rate": 5.616793117002729e-07,
+      "loss": 0.2673,
+      "step": 16123
+    },
+    {
+      "epoch": 1.374009373668513,
+      "grad_norm": 12.125,
+      "learning_rate": 5.615390532837704e-07,
+      "loss": 0.5927,
+      "step": 16124
+    },
+    {
+      "epoch": 1.374094588836813,
+      "grad_norm": 20.75,
+      "learning_rate": 5.613988073084491e-07,
+      "loss": 0.8598,
+      "step": 16125
+    },
+    {
+      "epoch": 1.374179804005113,
+      "grad_norm": 22.75,
+      "learning_rate": 5.612585737768442e-07,
+      "loss": 0.9524,
+      "step": 16126
+    },
+    {
+      "epoch": 1.374265019173413,
+      "grad_norm": 23.75,
+      "learning_rate": 5.611183526914889e-07,
+      "loss": 1.0647,
+      "step": 16127
+    },
+    {
+      "epoch": 1.3743502343417129,
+      "grad_norm": 12.4375,
+      "learning_rate": 5.609781440549178e-07,
+      "loss": 0.5867,
+      "step": 16128
+    },
+    {
+      "epoch": 1.3744354495100128,
+      "grad_norm": 15.0,
+      "learning_rate": 5.608379478696649e-07,
+      "loss": 0.5602,
+      "step": 16129
+    },
+    {
+      "epoch": 1.3745206646783128,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.60697764138263e-07,
+      "loss": 0.5048,
+      "step": 16130
+    },
+    {
+      "epoch": 1.3746058798466128,
+      "grad_norm": 18.625,
+      "learning_rate": 5.605575928632455e-07,
+      "loss": 0.6622,
+      "step": 16131
+    },
+    {
+      "epoch": 1.3746910950149127,
+      "grad_norm": 16.75,
+      "learning_rate": 5.604174340471455e-07,
+      "loss": 0.6081,
+      "step": 16132
+    },
+    {
+      "epoch": 1.3747763101832127,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.602772876924965e-07,
+      "loss": 0.7895,
+      "step": 16133
+    },
+    {
+      "epoch": 1.3748615253515126,
+      "grad_norm": 12.875,
+      "learning_rate": 5.601371538018304e-07,
+      "loss": 0.3368,
+      "step": 16134
+    },
+    {
+      "epoch": 1.3749467405198126,
+      "grad_norm": 18.75,
+      "learning_rate": 5.599970323776794e-07,
+      "loss": 0.7931,
+      "step": 16135
+    },
+    {
+      "epoch": 1.3750319556881125,
+      "grad_norm": 15.4375,
+      "learning_rate": 5.598569234225757e-07,
+      "loss": 0.8102,
+      "step": 16136
+    },
+    {
+      "epoch": 1.3751171708564125,
+      "grad_norm": 10.5,
+      "learning_rate": 5.597168269390517e-07,
+      "loss": 0.2217,
+      "step": 16137
+    },
+    {
+      "epoch": 1.3752023860247125,
+      "grad_norm": 9.375,
+      "learning_rate": 5.595767429296385e-07,
+      "loss": 0.3478,
+      "step": 16138
+    },
+    {
+      "epoch": 1.3752876011930124,
+      "grad_norm": 15.75,
+      "learning_rate": 5.594366713968678e-07,
+      "loss": 0.3109,
+      "step": 16139
+    },
+    {
+      "epoch": 1.3753728163613124,
+      "grad_norm": 19.5,
+      "learning_rate": 5.592966123432708e-07,
+      "loss": 0.8687,
+      "step": 16140
+    },
+    {
+      "epoch": 1.3754580315296123,
+      "grad_norm": 17.125,
+      "learning_rate": 5.591565657713792e-07,
+      "loss": 0.7723,
+      "step": 16141
+    },
+    {
+      "epoch": 1.3755432466979123,
+      "grad_norm": 16.25,
+      "learning_rate": 5.59016531683723e-07,
+      "loss": 0.8457,
+      "step": 16142
+    },
+    {
+      "epoch": 1.3756284618662122,
+      "grad_norm": 20.0,
+      "learning_rate": 5.588765100828325e-07,
+      "loss": 0.8141,
+      "step": 16143
+    },
+    {
+      "epoch": 1.3757136770345122,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.587365009712384e-07,
+      "loss": 0.6031,
+      "step": 16144
+    },
+    {
+      "epoch": 1.3757988922028122,
+      "grad_norm": 11.5625,
+      "learning_rate": 5.585965043514709e-07,
+      "loss": 0.2805,
+      "step": 16145
+    },
+    {
+      "epoch": 1.375884107371112,
+      "grad_norm": 14.625,
+      "learning_rate": 5.584565202260602e-07,
+      "loss": 0.8765,
+      "step": 16146
+    },
+    {
+      "epoch": 1.375969322539412,
+      "grad_norm": 18.25,
+      "learning_rate": 5.583165485975352e-07,
+      "loss": 0.7144,
+      "step": 16147
+    },
+    {
+      "epoch": 1.376054537707712,
+      "grad_norm": 9.8125,
+      "learning_rate": 5.581765894684257e-07,
+      "loss": 0.2466,
+      "step": 16148
+    },
+    {
+      "epoch": 1.376139752876012,
+      "grad_norm": 16.5,
+      "learning_rate": 5.580366428412613e-07,
+      "loss": 0.7653,
+      "step": 16149
+    },
+    {
+      "epoch": 1.376224968044312,
+      "grad_norm": 17.5,
+      "learning_rate": 5.578967087185708e-07,
+      "loss": 0.8808,
+      "step": 16150
+    },
+    {
+      "epoch": 1.376310183212612,
+      "grad_norm": 16.875,
+      "learning_rate": 5.57756787102882e-07,
+      "loss": 0.6259,
+      "step": 16151
+    },
+    {
+      "epoch": 1.3763953983809118,
+      "grad_norm": 11.4375,
+      "learning_rate": 5.576168779967243e-07,
+      "loss": 0.3255,
+      "step": 16152
+    },
+    {
+      "epoch": 1.3764806135492118,
+      "grad_norm": 13.375,
+      "learning_rate": 5.574769814026258e-07,
+      "loss": 0.5066,
+      "step": 16153
+    },
+    {
+      "epoch": 1.3765658287175118,
+      "grad_norm": 15.875,
+      "learning_rate": 5.573370973231153e-07,
+      "loss": 0.5777,
+      "step": 16154
+    },
+    {
+      "epoch": 1.3766510438858117,
+      "grad_norm": 14.25,
+      "learning_rate": 5.571972257607193e-07,
+      "loss": 0.7687,
+      "step": 16155
+    },
+    {
+      "epoch": 1.3767362590541117,
+      "grad_norm": 12.0,
+      "learning_rate": 5.570573667179662e-07,
+      "loss": 0.5043,
+      "step": 16156
+    },
+    {
+      "epoch": 1.3768214742224116,
+      "grad_norm": 17.25,
+      "learning_rate": 5.569175201973831e-07,
+      "loss": 0.6726,
+      "step": 16157
+    },
+    {
+      "epoch": 1.3769066893907116,
+      "grad_norm": 17.0,
+      "learning_rate": 5.56777686201498e-07,
+      "loss": 0.7493,
+      "step": 16158
+    },
+    {
+      "epoch": 1.3769919045590115,
+      "grad_norm": 13.625,
+      "learning_rate": 5.566378647328372e-07,
+      "loss": 0.4753,
+      "step": 16159
+    },
+    {
+      "epoch": 1.3770771197273115,
+      "grad_norm": 14.75,
+      "learning_rate": 5.564980557939269e-07,
+      "loss": 0.5677,
+      "step": 16160
+    },
+    {
+      "epoch": 1.3771623348956115,
+      "grad_norm": 17.125,
+      "learning_rate": 5.563582593872941e-07,
+      "loss": 0.4901,
+      "step": 16161
+    },
+    {
+      "epoch": 1.3772475500639114,
+      "grad_norm": 23.375,
+      "learning_rate": 5.562184755154651e-07,
+      "loss": 0.7359,
+      "step": 16162
+    },
+    {
+      "epoch": 1.3773327652322114,
+      "grad_norm": 22.25,
+      "learning_rate": 5.560787041809664e-07,
+      "loss": 0.5983,
+      "step": 16163
+    },
+    {
+      "epoch": 1.3774179804005113,
+      "grad_norm": 12.9375,
+      "learning_rate": 5.559389453863227e-07,
+      "loss": 0.3806,
+      "step": 16164
+    },
+    {
+      "epoch": 1.3775031955688113,
+      "grad_norm": 15.375,
+      "learning_rate": 5.557991991340603e-07,
+      "loss": 0.6937,
+      "step": 16165
+    },
+    {
+      "epoch": 1.3775884107371112,
+      "grad_norm": 17.625,
+      "learning_rate": 5.556594654267049e-07,
+      "loss": 0.7008,
+      "step": 16166
+    },
+    {
+      "epoch": 1.3776736259054112,
+      "grad_norm": 15.625,
+      "learning_rate": 5.555197442667811e-07,
+      "loss": 0.6989,
+      "step": 16167
+    },
+    {
+      "epoch": 1.3777588410737112,
+      "grad_norm": 14.75,
+      "learning_rate": 5.553800356568136e-07,
+      "loss": 0.5194,
+      "step": 16168
+    },
+    {
+      "epoch": 1.3778440562420111,
+      "grad_norm": 31.5,
+      "learning_rate": 5.552403395993271e-07,
+      "loss": 1.1434,
+      "step": 16169
+    },
+    {
+      "epoch": 1.377929271410311,
+      "grad_norm": 17.125,
+      "learning_rate": 5.551006560968464e-07,
+      "loss": 1.2713,
+      "step": 16170
+    },
+    {
+      "epoch": 1.378014486578611,
+      "grad_norm": 17.625,
+      "learning_rate": 5.549609851518961e-07,
+      "loss": 0.5922,
+      "step": 16171
+    },
+    {
+      "epoch": 1.378099701746911,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.548213267669993e-07,
+      "loss": 0.2707,
+      "step": 16172
+    },
+    {
+      "epoch": 1.378184916915211,
+      "grad_norm": 25.25,
+      "learning_rate": 5.546816809446802e-07,
+      "loss": 1.1922,
+      "step": 16173
+    },
+    {
+      "epoch": 1.378270132083511,
+      "grad_norm": 12.3125,
+      "learning_rate": 5.545420476874626e-07,
+      "loss": 0.3915,
+      "step": 16174
+    },
+    {
+      "epoch": 1.3783553472518109,
+      "grad_norm": 21.75,
+      "learning_rate": 5.544024269978692e-07,
+      "loss": 0.5879,
+      "step": 16175
+    },
+    {
+      "epoch": 1.3784405624201108,
+      "grad_norm": 17.25,
+      "learning_rate": 5.542628188784239e-07,
+      "loss": 0.6716,
+      "step": 16176
+    },
+    {
+      "epoch": 1.3785257775884108,
+      "grad_norm": 15.6875,
+      "learning_rate": 5.541232233316486e-07,
+      "loss": 0.4344,
+      "step": 16177
+    },
+    {
+      "epoch": 1.3786109927567107,
+      "grad_norm": 15.625,
+      "learning_rate": 5.539836403600664e-07,
+      "loss": 0.5461,
+      "step": 16178
+    },
+    {
+      "epoch": 1.3786962079250107,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.538440699662002e-07,
+      "loss": 0.6723,
+      "step": 16179
+    },
+    {
+      "epoch": 1.3787814230933106,
+      "grad_norm": 15.375,
+      "learning_rate": 5.537045121525712e-07,
+      "loss": 0.6039,
+      "step": 16180
+    },
+    {
+      "epoch": 1.3788666382616106,
+      "grad_norm": 17.25,
+      "learning_rate": 5.535649669217019e-07,
+      "loss": 0.8407,
+      "step": 16181
+    },
+    {
+      "epoch": 1.3789518534299106,
+      "grad_norm": 27.375,
+      "learning_rate": 5.534254342761139e-07,
+      "loss": 0.9365,
+      "step": 16182
+    },
+    {
+      "epoch": 1.3790370685982105,
+      "grad_norm": 12.0625,
+      "learning_rate": 5.532859142183293e-07,
+      "loss": 0.4117,
+      "step": 16183
+    },
+    {
+      "epoch": 1.3791222837665105,
+      "grad_norm": 17.5,
+      "learning_rate": 5.531464067508688e-07,
+      "loss": 0.5665,
+      "step": 16184
+    },
+    {
+      "epoch": 1.3792074989348104,
+      "grad_norm": 16.25,
+      "learning_rate": 5.53006911876253e-07,
+      "loss": 0.8722,
+      "step": 16185
+    },
+    {
+      "epoch": 1.3792927141031104,
+      "grad_norm": 15.625,
+      "learning_rate": 5.528674295970031e-07,
+      "loss": 0.5691,
+      "step": 16186
+    },
+    {
+      "epoch": 1.3793779292714103,
+      "grad_norm": 17.375,
+      "learning_rate": 5.527279599156399e-07,
+      "loss": 0.8627,
+      "step": 16187
+    },
+    {
+      "epoch": 1.3794631444397103,
+      "grad_norm": 16.75,
+      "learning_rate": 5.525885028346839e-07,
+      "loss": 0.624,
+      "step": 16188
+    },
+    {
+      "epoch": 1.3795483596080103,
+      "grad_norm": 11.1875,
+      "learning_rate": 5.524490583566547e-07,
+      "loss": 0.2264,
+      "step": 16189
+    },
+    {
+      "epoch": 1.3796335747763102,
+      "grad_norm": 16.875,
+      "learning_rate": 5.523096264840724e-07,
+      "loss": 0.5036,
+      "step": 16190
+    },
+    {
+      "epoch": 1.3797187899446102,
+      "grad_norm": 15.375,
+      "learning_rate": 5.52170207219457e-07,
+      "loss": 0.5264,
+      "step": 16191
+    },
+    {
+      "epoch": 1.3798040051129101,
+      "grad_norm": 12.3125,
+      "learning_rate": 5.520308005653276e-07,
+      "loss": 0.379,
+      "step": 16192
+    },
+    {
+      "epoch": 1.37988922028121,
+      "grad_norm": 15.6875,
+      "learning_rate": 5.518914065242037e-07,
+      "loss": 0.5936,
+      "step": 16193
+    },
+    {
+      "epoch": 1.37997443544951,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.517520250986036e-07,
+      "loss": 0.7415,
+      "step": 16194
+    },
+    {
+      "epoch": 1.38005965061781,
+      "grad_norm": 11.5,
+      "learning_rate": 5.516126562910467e-07,
+      "loss": 0.267,
+      "step": 16195
+    },
+    {
+      "epoch": 1.38014486578611,
+      "grad_norm": 21.375,
+      "learning_rate": 5.514733001040517e-07,
+      "loss": 0.6606,
+      "step": 16196
+    },
+    {
+      "epoch": 1.38023008095441,
+      "grad_norm": 18.0,
+      "learning_rate": 5.513339565401362e-07,
+      "loss": 0.7756,
+      "step": 16197
+    },
+    {
+      "epoch": 1.3803152961227099,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.511946256018187e-07,
+      "loss": 0.6675,
+      "step": 16198
+    },
+    {
+      "epoch": 1.3804005112910098,
+      "grad_norm": 16.75,
+      "learning_rate": 5.510553072916175e-07,
+      "loss": 0.6884,
+      "step": 16199
+    },
+    {
+      "epoch": 1.3804857264593098,
+      "grad_norm": 17.5,
+      "learning_rate": 5.509160016120492e-07,
+      "loss": 0.5537,
+      "step": 16200
+    },
+    {
+      "epoch": 1.3805709416276097,
+      "grad_norm": 18.25,
+      "learning_rate": 5.507767085656323e-07,
+      "loss": 0.8159,
+      "step": 16201
+    },
+    {
+      "epoch": 1.3806561567959097,
+      "grad_norm": 9.6875,
+      "learning_rate": 5.506374281548827e-07,
+      "loss": 0.2083,
+      "step": 16202
+    },
+    {
+      "epoch": 1.3807413719642097,
+      "grad_norm": 22.0,
+      "learning_rate": 5.504981603823182e-07,
+      "loss": 0.9635,
+      "step": 16203
+    },
+    {
+      "epoch": 1.3808265871325096,
+      "grad_norm": 11.0625,
+      "learning_rate": 5.503589052504554e-07,
+      "loss": 0.3829,
+      "step": 16204
+    },
+    {
+      "epoch": 1.3809118023008096,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.502196627618112e-07,
+      "loss": 0.4774,
+      "step": 16205
+    },
+    {
+      "epoch": 1.3809970174691095,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.500804329189008e-07,
+      "loss": 0.6848,
+      "step": 16206
+    },
+    {
+      "epoch": 1.3810822326374095,
+      "grad_norm": 16.5,
+      "learning_rate": 5.499412157242411e-07,
+      "loss": 0.6415,
+      "step": 16207
+    },
+    {
+      "epoch": 1.3811674478057094,
+      "grad_norm": 31.0,
+      "learning_rate": 5.498020111803475e-07,
+      "loss": 0.8314,
+      "step": 16208
+    },
+    {
+      "epoch": 1.3812526629740094,
+      "grad_norm": 12.375,
+      "learning_rate": 5.496628192897359e-07,
+      "loss": 0.4089,
+      "step": 16209
+    },
+    {
+      "epoch": 1.3813378781423093,
+      "grad_norm": 16.25,
+      "learning_rate": 5.49523640054921e-07,
+      "loss": 0.7087,
+      "step": 16210
+    },
+    {
+      "epoch": 1.3814230933106093,
+      "grad_norm": 18.125,
+      "learning_rate": 5.493844734784183e-07,
+      "loss": 0.6688,
+      "step": 16211
+    },
+    {
+      "epoch": 1.3815083084789093,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.492453195627427e-07,
+      "loss": 0.4449,
+      "step": 16212
+    },
+    {
+      "epoch": 1.3815935236472092,
+      "grad_norm": 13.0,
+      "learning_rate": 5.491061783104093e-07,
+      "loss": 0.6191,
+      "step": 16213
+    },
+    {
+      "epoch": 1.3816787388155092,
+      "grad_norm": 16.375,
+      "learning_rate": 5.489670497239316e-07,
+      "loss": 0.7047,
+      "step": 16214
+    },
+    {
+      "epoch": 1.3817639539838091,
+      "grad_norm": 9.6875,
+      "learning_rate": 5.488279338058247e-07,
+      "loss": 0.2489,
+      "step": 16215
+    },
+    {
+      "epoch": 1.381849169152109,
+      "grad_norm": 12.25,
+      "learning_rate": 5.486888305586018e-07,
+      "loss": 0.4164,
+      "step": 16216
+    },
+    {
+      "epoch": 1.381934384320409,
+      "grad_norm": 17.875,
+      "learning_rate": 5.485497399847766e-07,
+      "loss": 0.7807,
+      "step": 16217
+    },
+    {
+      "epoch": 1.382019599488709,
+      "grad_norm": 26.125,
+      "learning_rate": 5.484106620868637e-07,
+      "loss": 0.3848,
+      "step": 16218
+    },
+    {
+      "epoch": 1.382104814657009,
+      "grad_norm": 12.3125,
+      "learning_rate": 5.482715968673752e-07,
+      "loss": 0.4489,
+      "step": 16219
+    },
+    {
+      "epoch": 1.382190029825309,
+      "grad_norm": 21.125,
+      "learning_rate": 5.481325443288245e-07,
+      "loss": 0.5143,
+      "step": 16220
+    },
+    {
+      "epoch": 1.3822752449936089,
+      "grad_norm": 16.375,
+      "learning_rate": 5.479935044737245e-07,
+      "loss": 0.8863,
+      "step": 16221
+    },
+    {
+      "epoch": 1.3823604601619088,
+      "grad_norm": 12.9375,
+      "learning_rate": 5.478544773045882e-07,
+      "loss": 0.4683,
+      "step": 16222
+    },
+    {
+      "epoch": 1.3824456753302088,
+      "grad_norm": 16.625,
+      "learning_rate": 5.477154628239277e-07,
+      "loss": 0.4664,
+      "step": 16223
+    },
+    {
+      "epoch": 1.3825308904985087,
+      "grad_norm": 13.0,
+      "learning_rate": 5.475764610342544e-07,
+      "loss": 0.3818,
+      "step": 16224
+    },
+    {
+      "epoch": 1.3826161056668087,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.47437471938081e-07,
+      "loss": 0.6835,
+      "step": 16225
+    },
+    {
+      "epoch": 1.3827013208351087,
+      "grad_norm": 14.875,
+      "learning_rate": 5.472984955379192e-07,
+      "loss": 0.5074,
+      "step": 16226
+    },
+    {
+      "epoch": 1.3827865360034086,
+      "grad_norm": 13.4375,
+      "learning_rate": 5.471595318362797e-07,
+      "loss": 0.4608,
+      "step": 16227
+    },
+    {
+      "epoch": 1.3828717511717086,
+      "grad_norm": 16.75,
+      "learning_rate": 5.470205808356745e-07,
+      "loss": 0.5839,
+      "step": 16228
+    },
+    {
+      "epoch": 1.3829569663400085,
+      "grad_norm": 11.875,
+      "learning_rate": 5.468816425386141e-07,
+      "loss": 0.4228,
+      "step": 16229
+    },
+    {
+      "epoch": 1.3830421815083085,
+      "grad_norm": 13.0,
+      "learning_rate": 5.4674271694761e-07,
+      "loss": 0.4436,
+      "step": 16230
+    },
+    {
+      "epoch": 1.3831273966766084,
+      "grad_norm": 13.25,
+      "learning_rate": 5.466038040651721e-07,
+      "loss": 0.4287,
+      "step": 16231
+    },
+    {
+      "epoch": 1.3832126118449084,
+      "grad_norm": 14.0,
+      "learning_rate": 5.464649038938104e-07,
+      "loss": 0.6583,
+      "step": 16232
+    },
+    {
+      "epoch": 1.3832978270132084,
+      "grad_norm": 19.0,
+      "learning_rate": 5.463260164360353e-07,
+      "loss": 0.7777,
+      "step": 16233
+    },
+    {
+      "epoch": 1.3833830421815083,
+      "grad_norm": 29.5,
+      "learning_rate": 5.461871416943566e-07,
+      "loss": 1.0111,
+      "step": 16234
+    },
+    {
+      "epoch": 1.3834682573498083,
+      "grad_norm": 20.25,
+      "learning_rate": 5.460482796712846e-07,
+      "loss": 0.9076,
+      "step": 16235
+    },
+    {
+      "epoch": 1.3835534725181082,
+      "grad_norm": 12.1875,
+      "learning_rate": 5.459094303693276e-07,
+      "loss": 0.311,
+      "step": 16236
+    },
+    {
+      "epoch": 1.3836386876864082,
+      "grad_norm": 21.75,
+      "learning_rate": 5.457705937909951e-07,
+      "loss": 0.5241,
+      "step": 16237
+    },
+    {
+      "epoch": 1.3837239028547081,
+      "grad_norm": 19.625,
+      "learning_rate": 5.456317699387962e-07,
+      "loss": 0.7599,
+      "step": 16238
+    },
+    {
+      "epoch": 1.383809118023008,
+      "grad_norm": 13.625,
+      "learning_rate": 5.4549295881524e-07,
+      "loss": 0.465,
+      "step": 16239
+    },
+    {
+      "epoch": 1.383894333191308,
+      "grad_norm": 21.375,
+      "learning_rate": 5.453541604228343e-07,
+      "loss": 0.7384,
+      "step": 16240
+    },
+    {
+      "epoch": 1.383979548359608,
+      "grad_norm": 20.875,
+      "learning_rate": 5.452153747640873e-07,
+      "loss": 0.7375,
+      "step": 16241
+    },
+    {
+      "epoch": 1.384064763527908,
+      "grad_norm": 18.625,
+      "learning_rate": 5.45076601841507e-07,
+      "loss": 0.7463,
+      "step": 16242
+    },
+    {
+      "epoch": 1.384149978696208,
+      "grad_norm": 18.875,
+      "learning_rate": 5.449378416576019e-07,
+      "loss": 0.9104,
+      "step": 16243
+    },
+    {
+      "epoch": 1.3842351938645079,
+      "grad_norm": 15.75,
+      "learning_rate": 5.447990942148786e-07,
+      "loss": 0.685,
+      "step": 16244
+    },
+    {
+      "epoch": 1.3843204090328078,
+      "grad_norm": 14.5,
+      "learning_rate": 5.446603595158447e-07,
+      "loss": 0.5088,
+      "step": 16245
+    },
+    {
+      "epoch": 1.3844056242011078,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.445216375630073e-07,
+      "loss": 0.5055,
+      "step": 16246
+    },
+    {
+      "epoch": 1.3844908393694078,
+      "grad_norm": 17.125,
+      "learning_rate": 5.443829283588739e-07,
+      "loss": 0.6934,
+      "step": 16247
+    },
+    {
+      "epoch": 1.3845760545377077,
+      "grad_norm": 21.125,
+      "learning_rate": 5.442442319059506e-07,
+      "loss": 0.8992,
+      "step": 16248
+    },
+    {
+      "epoch": 1.3846612697060077,
+      "grad_norm": 21.0,
+      "learning_rate": 5.441055482067431e-07,
+      "loss": 0.6825,
+      "step": 16249
+    },
+    {
+      "epoch": 1.3847464848743076,
+      "grad_norm": 12.375,
+      "learning_rate": 5.439668772637582e-07,
+      "loss": 0.392,
+      "step": 16250
+    },
+    {
+      "epoch": 1.3848317000426076,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.43828219079502e-07,
+      "loss": 0.4674,
+      "step": 16251
+    },
+    {
+      "epoch": 1.3849169152109075,
+      "grad_norm": 20.625,
+      "learning_rate": 5.436895736564802e-07,
+      "loss": 0.384,
+      "step": 16252
+    },
+    {
+      "epoch": 1.3850021303792075,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.435509409971977e-07,
+      "loss": 0.8089,
+      "step": 16253
+    },
+    {
+      "epoch": 1.3850873455475075,
+      "grad_norm": 12.875,
+      "learning_rate": 5.434123211041601e-07,
+      "loss": 0.4132,
+      "step": 16254
+    },
+    {
+      "epoch": 1.3851725607158074,
+      "grad_norm": 17.25,
+      "learning_rate": 5.432737139798725e-07,
+      "loss": 0.6034,
+      "step": 16255
+    },
+    {
+      "epoch": 1.3852577758841074,
+      "grad_norm": 17.125,
+      "learning_rate": 5.431351196268403e-07,
+      "loss": 0.4918,
+      "step": 16256
+    },
+    {
+      "epoch": 1.3853429910524073,
+      "grad_norm": 12.5,
+      "learning_rate": 5.429965380475665e-07,
+      "loss": 0.3789,
+      "step": 16257
+    },
+    {
+      "epoch": 1.3854282062207073,
+      "grad_norm": 9.1875,
+      "learning_rate": 5.42857969244556e-07,
+      "loss": 0.1938,
+      "step": 16258
+    },
+    {
+      "epoch": 1.3855134213890072,
+      "grad_norm": 18.375,
+      "learning_rate": 5.427194132203133e-07,
+      "loss": 0.8812,
+      "step": 16259
+    },
+    {
+      "epoch": 1.3855986365573072,
+      "grad_norm": 21.75,
+      "learning_rate": 5.425808699773423e-07,
+      "loss": 0.684,
+      "step": 16260
+    },
+    {
+      "epoch": 1.3856838517256072,
+      "grad_norm": 16.25,
+      "learning_rate": 5.424423395181461e-07,
+      "loss": 0.6042,
+      "step": 16261
+    },
+    {
+      "epoch": 1.385769066893907,
+      "grad_norm": 16.5,
+      "learning_rate": 5.423038218452281e-07,
+      "loss": 0.7886,
+      "step": 16262
+    },
+    {
+      "epoch": 1.385854282062207,
+      "grad_norm": 17.625,
+      "learning_rate": 5.421653169610919e-07,
+      "loss": 0.6448,
+      "step": 16263
+    },
+    {
+      "epoch": 1.385939497230507,
+      "grad_norm": 20.75,
+      "learning_rate": 5.420268248682405e-07,
+      "loss": 0.7995,
+      "step": 16264
+    },
+    {
+      "epoch": 1.386024712398807,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.418883455691763e-07,
+      "loss": 0.7082,
+      "step": 16265
+    },
+    {
+      "epoch": 1.386109927567107,
+      "grad_norm": 13.9375,
+      "learning_rate": 5.417498790664014e-07,
+      "loss": 0.4509,
+      "step": 16266
+    },
+    {
+      "epoch": 1.386195142735407,
+      "grad_norm": 19.25,
+      "learning_rate": 5.416114253624183e-07,
+      "loss": 0.7585,
+      "step": 16267
+    },
+    {
+      "epoch": 1.3862803579037069,
+      "grad_norm": 20.5,
+      "learning_rate": 5.414729844597291e-07,
+      "loss": 0.7591,
+      "step": 16268
+    },
+    {
+      "epoch": 1.3863655730720068,
+      "grad_norm": 19.625,
+      "learning_rate": 5.41334556360836e-07,
+      "loss": 0.9383,
+      "step": 16269
+    },
+    {
+      "epoch": 1.3864507882403068,
+      "grad_norm": 16.625,
+      "learning_rate": 5.411961410682398e-07,
+      "loss": 0.7127,
+      "step": 16270
+    },
+    {
+      "epoch": 1.3865360034086067,
+      "grad_norm": 14.875,
+      "learning_rate": 5.410577385844419e-07,
+      "loss": 0.7662,
+      "step": 16271
+    },
+    {
+      "epoch": 1.3866212185769067,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.409193489119441e-07,
+      "loss": 0.7139,
+      "step": 16272
+    },
+    {
+      "epoch": 1.3867064337452066,
+      "grad_norm": 19.0,
+      "learning_rate": 5.407809720532466e-07,
+      "loss": 0.7459,
+      "step": 16273
+    },
+    {
+      "epoch": 1.3867916489135066,
+      "grad_norm": 12.625,
+      "learning_rate": 5.406426080108497e-07,
+      "loss": 0.4128,
+      "step": 16274
+    },
+    {
+      "epoch": 1.3868768640818065,
+      "grad_norm": 16.5,
+      "learning_rate": 5.405042567872543e-07,
+      "loss": 0.7258,
+      "step": 16275
+    },
+    {
+      "epoch": 1.3869620792501065,
+      "grad_norm": 20.75,
+      "learning_rate": 5.403659183849604e-07,
+      "loss": 0.5366,
+      "step": 16276
+    },
+    {
+      "epoch": 1.3870472944184065,
+      "grad_norm": 9.5,
+      "learning_rate": 5.402275928064682e-07,
+      "loss": 0.3374,
+      "step": 16277
+    },
+    {
+      "epoch": 1.3871325095867064,
+      "grad_norm": 28.5,
+      "learning_rate": 5.400892800542769e-07,
+      "loss": 0.7199,
+      "step": 16278
+    },
+    {
+      "epoch": 1.3872177247550064,
+      "grad_norm": 12.5625,
+      "learning_rate": 5.39950980130886e-07,
+      "loss": 0.5541,
+      "step": 16279
+    },
+    {
+      "epoch": 1.3873029399233063,
+      "grad_norm": 16.875,
+      "learning_rate": 5.398126930387955e-07,
+      "loss": 0.3705,
+      "step": 16280
+    },
+    {
+      "epoch": 1.3873881550916063,
+      "grad_norm": 17.625,
+      "learning_rate": 5.396744187805032e-07,
+      "loss": 0.564,
+      "step": 16281
+    },
+    {
+      "epoch": 1.3874733702599062,
+      "grad_norm": 20.625,
+      "learning_rate": 5.395361573585091e-07,
+      "loss": 0.8389,
+      "step": 16282
+    },
+    {
+      "epoch": 1.3875585854282062,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.393979087753104e-07,
+      "loss": 0.5423,
+      "step": 16283
+    },
+    {
+      "epoch": 1.3876438005965062,
+      "grad_norm": 18.125,
+      "learning_rate": 5.392596730334061e-07,
+      "loss": 0.7961,
+      "step": 16284
+    },
+    {
+      "epoch": 1.3877290157648061,
+      "grad_norm": 27.125,
+      "learning_rate": 5.391214501352943e-07,
+      "loss": 0.3258,
+      "step": 16285
+    },
+    {
+      "epoch": 1.387814230933106,
+      "grad_norm": 12.25,
+      "learning_rate": 5.389832400834732e-07,
+      "loss": 0.3986,
+      "step": 16286
+    },
+    {
+      "epoch": 1.387899446101406,
+      "grad_norm": 11.125,
+      "learning_rate": 5.388450428804394e-07,
+      "loss": 0.3016,
+      "step": 16287
+    },
+    {
+      "epoch": 1.387984661269706,
+      "grad_norm": 14.0,
+      "learning_rate": 5.387068585286913e-07,
+      "loss": 0.6687,
+      "step": 16288
+    },
+    {
+      "epoch": 1.388069876438006,
+      "grad_norm": 19.875,
+      "learning_rate": 5.385686870307251e-07,
+      "loss": 0.691,
+      "step": 16289
+    },
+    {
+      "epoch": 1.388155091606306,
+      "grad_norm": 16.875,
+      "learning_rate": 5.384305283890384e-07,
+      "loss": 0.4801,
+      "step": 16290
+    },
+    {
+      "epoch": 1.3882403067746059,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.382923826061274e-07,
+      "loss": 0.7809,
+      "step": 16291
+    },
+    {
+      "epoch": 1.3883255219429058,
+      "grad_norm": 21.25,
+      "learning_rate": 5.381542496844885e-07,
+      "loss": 0.8043,
+      "step": 16292
+    },
+    {
+      "epoch": 1.3884107371112058,
+      "grad_norm": 16.0,
+      "learning_rate": 5.380161296266183e-07,
+      "loss": 0.609,
+      "step": 16293
+    },
+    {
+      "epoch": 1.3884959522795057,
+      "grad_norm": 21.125,
+      "learning_rate": 5.378780224350129e-07,
+      "loss": 1.0666,
+      "step": 16294
+    },
+    {
+      "epoch": 1.3885811674478057,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.377399281121673e-07,
+      "loss": 0.7965,
+      "step": 16295
+    },
+    {
+      "epoch": 1.3886663826161056,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.37601846660578e-07,
+      "loss": 0.4348,
+      "step": 16296
+    },
+    {
+      "epoch": 1.3887515977844056,
+      "grad_norm": 11.6875,
+      "learning_rate": 5.374637780827391e-07,
+      "loss": 0.2844,
+      "step": 16297
+    },
+    {
+      "epoch": 1.3888368129527056,
+      "grad_norm": 25.625,
+      "learning_rate": 5.373257223811464e-07,
+      "loss": 0.3522,
+      "step": 16298
+    },
+    {
+      "epoch": 1.3889220281210055,
+      "grad_norm": 11.4375,
+      "learning_rate": 5.371876795582948e-07,
+      "loss": 0.3121,
+      "step": 16299
+    },
+    {
+      "epoch": 1.3890072432893055,
+      "grad_norm": 14.5,
+      "learning_rate": 5.370496496166783e-07,
+      "loss": 0.5959,
+      "step": 16300
+    },
+    {
+      "epoch": 1.3890924584576054,
+      "grad_norm": 23.125,
+      "learning_rate": 5.369116325587916e-07,
+      "loss": 0.9148,
+      "step": 16301
+    },
+    {
+      "epoch": 1.3891776736259054,
+      "grad_norm": 17.25,
+      "learning_rate": 5.367736283871286e-07,
+      "loss": 0.596,
+      "step": 16302
+    },
+    {
+      "epoch": 1.3892628887942053,
+      "grad_norm": 17.375,
+      "learning_rate": 5.36635637104184e-07,
+      "loss": 0.7201,
+      "step": 16303
+    },
+    {
+      "epoch": 1.3893481039625053,
+      "grad_norm": 21.5,
+      "learning_rate": 5.364976587124506e-07,
+      "loss": 0.5346,
+      "step": 16304
+    },
+    {
+      "epoch": 1.3894333191308053,
+      "grad_norm": 12.25,
+      "learning_rate": 5.363596932144216e-07,
+      "loss": 0.2148,
+      "step": 16305
+    },
+    {
+      "epoch": 1.3895185342991052,
+      "grad_norm": 15.75,
+      "learning_rate": 5.362217406125906e-07,
+      "loss": 0.6497,
+      "step": 16306
+    },
+    {
+      "epoch": 1.3896037494674052,
+      "grad_norm": 15.5,
+      "learning_rate": 5.360838009094508e-07,
+      "loss": 0.4809,
+      "step": 16307
+    },
+    {
+      "epoch": 1.3896889646357051,
+      "grad_norm": 18.625,
+      "learning_rate": 5.359458741074944e-07,
+      "loss": 0.7103,
+      "step": 16308
+    },
+    {
+      "epoch": 1.389774179804005,
+      "grad_norm": 32.0,
+      "learning_rate": 5.35807960209214e-07,
+      "loss": 0.7895,
+      "step": 16309
+    },
+    {
+      "epoch": 1.389859394972305,
+      "grad_norm": 26.0,
+      "learning_rate": 5.356700592171019e-07,
+      "loss": 1.0872,
+      "step": 16310
+    },
+    {
+      "epoch": 1.389944610140605,
+      "grad_norm": 12.125,
+      "learning_rate": 5.355321711336506e-07,
+      "loss": 0.4133,
+      "step": 16311
+    },
+    {
+      "epoch": 1.390029825308905,
+      "grad_norm": 15.625,
+      "learning_rate": 5.35394295961351e-07,
+      "loss": 0.7255,
+      "step": 16312
+    },
+    {
+      "epoch": 1.390115040477205,
+      "grad_norm": 24.625,
+      "learning_rate": 5.352564337026955e-07,
+      "loss": 1.0113,
+      "step": 16313
+    },
+    {
+      "epoch": 1.3902002556455049,
+      "grad_norm": 14.625,
+      "learning_rate": 5.351185843601745e-07,
+      "loss": 0.5893,
+      "step": 16314
+    },
+    {
+      "epoch": 1.3902854708138048,
+      "grad_norm": 11.4375,
+      "learning_rate": 5.349807479362796e-07,
+      "loss": 0.3435,
+      "step": 16315
+    },
+    {
+      "epoch": 1.3903706859821048,
+      "grad_norm": 14.125,
+      "learning_rate": 5.348429244335019e-07,
+      "loss": 0.6399,
+      "step": 16316
+    },
+    {
+      "epoch": 1.3904559011504047,
+      "grad_norm": 12.25,
+      "learning_rate": 5.347051138543315e-07,
+      "loss": 0.3775,
+      "step": 16317
+    },
+    {
+      "epoch": 1.3905411163187047,
+      "grad_norm": 17.0,
+      "learning_rate": 5.345673162012588e-07,
+      "loss": 0.8836,
+      "step": 16318
+    },
+    {
+      "epoch": 1.3906263314870047,
+      "grad_norm": 15.4375,
+      "learning_rate": 5.344295314767746e-07,
+      "loss": 0.491,
+      "step": 16319
+    },
+    {
+      "epoch": 1.3907115466553046,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.34291759683368e-07,
+      "loss": 0.5502,
+      "step": 16320
+    },
+    {
+      "epoch": 1.3907967618236046,
+      "grad_norm": 15.0,
+      "learning_rate": 5.341540008235293e-07,
+      "loss": 0.7206,
+      "step": 16321
+    },
+    {
+      "epoch": 1.3908819769919045,
+      "grad_norm": 20.375,
+      "learning_rate": 5.340162548997473e-07,
+      "loss": 0.8401,
+      "step": 16322
+    },
+    {
+      "epoch": 1.3909671921602045,
+      "grad_norm": 17.25,
+      "learning_rate": 5.338785219145114e-07,
+      "loss": 0.7415,
+      "step": 16323
+    },
+    {
+      "epoch": 1.3910524073285044,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.337408018703112e-07,
+      "loss": 0.6126,
+      "step": 16324
+    },
+    {
+      "epoch": 1.3911376224968044,
+      "grad_norm": 16.25,
+      "learning_rate": 5.336030947696346e-07,
+      "loss": 0.4307,
+      "step": 16325
+    },
+    {
+      "epoch": 1.3912228376651044,
+      "grad_norm": 11.3125,
+      "learning_rate": 5.334654006149705e-07,
+      "loss": 0.8282,
+      "step": 16326
+    },
+    {
+      "epoch": 1.3913080528334043,
+      "grad_norm": 14.0625,
+      "learning_rate": 5.333277194088071e-07,
+      "loss": 0.5582,
+      "step": 16327
+    },
+    {
+      "epoch": 1.3913932680017043,
+      "grad_norm": 15.75,
+      "learning_rate": 5.33190051153633e-07,
+      "loss": 0.5498,
+      "step": 16328
+    },
+    {
+      "epoch": 1.3914784831700042,
+      "grad_norm": 18.625,
+      "learning_rate": 5.330523958519353e-07,
+      "loss": 0.5722,
+      "step": 16329
+    },
+    {
+      "epoch": 1.3915636983383042,
+      "grad_norm": 22.75,
+      "learning_rate": 5.329147535062014e-07,
+      "loss": 0.7093,
+      "step": 16330
+    },
+    {
+      "epoch": 1.3916489135066041,
+      "grad_norm": 12.5625,
+      "learning_rate": 5.32777124118919e-07,
+      "loss": 0.4678,
+      "step": 16331
+    },
+    {
+      "epoch": 1.391734128674904,
+      "grad_norm": 16.375,
+      "learning_rate": 5.326395076925753e-07,
+      "loss": 0.7793,
+      "step": 16332
+    },
+    {
+      "epoch": 1.391819343843204,
+      "grad_norm": 14.75,
+      "learning_rate": 5.325019042296572e-07,
+      "loss": 0.4395,
+      "step": 16333
+    },
+    {
+      "epoch": 1.391904559011504,
+      "grad_norm": 17.5,
+      "learning_rate": 5.32364313732651e-07,
+      "loss": 0.785,
+      "step": 16334
+    },
+    {
+      "epoch": 1.391989774179804,
+      "grad_norm": 16.0,
+      "learning_rate": 5.322267362040433e-07,
+      "loss": 0.7196,
+      "step": 16335
+    },
+    {
+      "epoch": 1.392074989348104,
+      "grad_norm": 9.875,
+      "learning_rate": 5.320891716463206e-07,
+      "loss": 0.5907,
+      "step": 16336
+    },
+    {
+      "epoch": 1.3921602045164039,
+      "grad_norm": 17.75,
+      "learning_rate": 5.319516200619686e-07,
+      "loss": 0.688,
+      "step": 16337
+    },
+    {
+      "epoch": 1.3922454196847038,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.318140814534722e-07,
+      "loss": 0.3144,
+      "step": 16338
+    },
+    {
+      "epoch": 1.3923306348530038,
+      "grad_norm": 16.25,
+      "learning_rate": 5.316765558233178e-07,
+      "loss": 0.4942,
+      "step": 16339
+    },
+    {
+      "epoch": 1.3924158500213037,
+      "grad_norm": 19.375,
+      "learning_rate": 5.315390431739902e-07,
+      "loss": 0.6378,
+      "step": 16340
+    },
+    {
+      "epoch": 1.3925010651896037,
+      "grad_norm": 17.375,
+      "learning_rate": 5.31401543507975e-07,
+      "loss": 0.7395,
+      "step": 16341
+    },
+    {
+      "epoch": 1.3925862803579037,
+      "grad_norm": 16.0,
+      "learning_rate": 5.312640568277562e-07,
+      "loss": 0.7385,
+      "step": 16342
+    },
+    {
+      "epoch": 1.3926714955262036,
+      "grad_norm": 18.0,
+      "learning_rate": 5.311265831358185e-07,
+      "loss": 0.5627,
+      "step": 16343
+    },
+    {
+      "epoch": 1.3927567106945036,
+      "grad_norm": 11.5625,
+      "learning_rate": 5.309891224346464e-07,
+      "loss": 0.2779,
+      "step": 16344
+    },
+    {
+      "epoch": 1.3928419258628035,
+      "grad_norm": 16.375,
+      "learning_rate": 5.308516747267243e-07,
+      "loss": 0.6792,
+      "step": 16345
+    },
+    {
+      "epoch": 1.3929271410311035,
+      "grad_norm": 25.375,
+      "learning_rate": 5.307142400145357e-07,
+      "loss": 0.7731,
+      "step": 16346
+    },
+    {
+      "epoch": 1.3930123561994034,
+      "grad_norm": 13.375,
+      "learning_rate": 5.305768183005634e-07,
+      "loss": 0.6184,
+      "step": 16347
+    },
+    {
+      "epoch": 1.3930975713677034,
+      "grad_norm": 16.75,
+      "learning_rate": 5.304394095872917e-07,
+      "loss": 0.5288,
+      "step": 16348
+    },
+    {
+      "epoch": 1.3931827865360034,
+      "grad_norm": 14.875,
+      "learning_rate": 5.303020138772038e-07,
+      "loss": 0.6564,
+      "step": 16349
+    },
+    {
+      "epoch": 1.3932680017043033,
+      "grad_norm": 13.625,
+      "learning_rate": 5.301646311727816e-07,
+      "loss": 0.5452,
+      "step": 16350
+    },
+    {
+      "epoch": 1.3933532168726033,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.300272614765086e-07,
+      "loss": 0.3012,
+      "step": 16351
+    },
+    {
+      "epoch": 1.3934384320409032,
+      "grad_norm": 15.5625,
+      "learning_rate": 5.298899047908667e-07,
+      "loss": 0.5805,
+      "step": 16352
+    },
+    {
+      "epoch": 1.3935236472092032,
+      "grad_norm": 15.375,
+      "learning_rate": 5.29752561118339e-07,
+      "loss": 0.5796,
+      "step": 16353
+    },
+    {
+      "epoch": 1.3936088623775031,
+      "grad_norm": 18.625,
+      "learning_rate": 5.296152304614069e-07,
+      "loss": 0.8866,
+      "step": 16354
+    },
+    {
+      "epoch": 1.393694077545803,
+      "grad_norm": 17.75,
+      "learning_rate": 5.294779128225514e-07,
+      "loss": 0.7419,
+      "step": 16355
+    },
+    {
+      "epoch": 1.393779292714103,
+      "grad_norm": 19.25,
+      "learning_rate": 5.293406082042545e-07,
+      "loss": 0.9076,
+      "step": 16356
+    },
+    {
+      "epoch": 1.393864507882403,
+      "grad_norm": 17.0,
+      "learning_rate": 5.292033166089975e-07,
+      "loss": 0.8882,
+      "step": 16357
+    },
+    {
+      "epoch": 1.393949723050703,
+      "grad_norm": 18.375,
+      "learning_rate": 5.290660380392619e-07,
+      "loss": 0.843,
+      "step": 16358
+    },
+    {
+      "epoch": 1.394034938219003,
+      "grad_norm": 13.4375,
+      "learning_rate": 5.289287724975276e-07,
+      "loss": 0.5389,
+      "step": 16359
+    },
+    {
+      "epoch": 1.3941201533873029,
+      "grad_norm": 15.875,
+      "learning_rate": 5.287915199862753e-07,
+      "loss": 0.7784,
+      "step": 16360
+    },
+    {
+      "epoch": 1.3942053685556028,
+      "grad_norm": 17.375,
+      "learning_rate": 5.28654280507986e-07,
+      "loss": 0.8619,
+      "step": 16361
+    },
+    {
+      "epoch": 1.3942905837239028,
+      "grad_norm": 13.0625,
+      "learning_rate": 5.285170540651387e-07,
+      "loss": 0.5326,
+      "step": 16362
+    },
+    {
+      "epoch": 1.3943757988922028,
+      "grad_norm": 16.125,
+      "learning_rate": 5.283798406602142e-07,
+      "loss": 0.8422,
+      "step": 16363
+    },
+    {
+      "epoch": 1.3944610140605027,
+      "grad_norm": 18.375,
+      "learning_rate": 5.282426402956911e-07,
+      "loss": 0.6701,
+      "step": 16364
+    },
+    {
+      "epoch": 1.3945462292288027,
+      "grad_norm": 17.0,
+      "learning_rate": 5.281054529740494e-07,
+      "loss": 0.9546,
+      "step": 16365
+    },
+    {
+      "epoch": 1.3946314443971026,
+      "grad_norm": 15.5,
+      "learning_rate": 5.279682786977686e-07,
+      "loss": 0.549,
+      "step": 16366
+    },
+    {
+      "epoch": 1.3947166595654026,
+      "grad_norm": 18.875,
+      "learning_rate": 5.278311174693265e-07,
+      "loss": 0.808,
+      "step": 16367
+    },
+    {
+      "epoch": 1.3948018747337025,
+      "grad_norm": 18.875,
+      "learning_rate": 5.276939692912023e-07,
+      "loss": 0.6242,
+      "step": 16368
+    },
+    {
+      "epoch": 1.3948870899020025,
+      "grad_norm": 15.625,
+      "learning_rate": 5.275568341658749e-07,
+      "loss": 0.4896,
+      "step": 16369
+    },
+    {
+      "epoch": 1.3949723050703025,
+      "grad_norm": 15.125,
+      "learning_rate": 5.274197120958215e-07,
+      "loss": 0.7505,
+      "step": 16370
+    },
+    {
+      "epoch": 1.3950575202386024,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.27282603083521e-07,
+      "loss": 0.4665,
+      "step": 16371
+    },
+    {
+      "epoch": 1.3951427354069024,
+      "grad_norm": 16.125,
+      "learning_rate": 5.271455071314502e-07,
+      "loss": 0.8909,
+      "step": 16372
+    },
+    {
+      "epoch": 1.3952279505752023,
+      "grad_norm": 13.375,
+      "learning_rate": 5.270084242420868e-07,
+      "loss": 0.438,
+      "step": 16373
+    },
+    {
+      "epoch": 1.3953131657435023,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.268713544179084e-07,
+      "loss": 0.4716,
+      "step": 16374
+    },
+    {
+      "epoch": 1.3953983809118022,
+      "grad_norm": 17.0,
+      "learning_rate": 5.267342976613921e-07,
+      "loss": 0.8214,
+      "step": 16375
+    },
+    {
+      "epoch": 1.3954835960801022,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.265972539750139e-07,
+      "loss": 0.5241,
+      "step": 16376
+    },
+    {
+      "epoch": 1.3955688112484022,
+      "grad_norm": 16.375,
+      "learning_rate": 5.264602233612514e-07,
+      "loss": 0.8926,
+      "step": 16377
+    },
+    {
+      "epoch": 1.395654026416702,
+      "grad_norm": 14.375,
+      "learning_rate": 5.263232058225795e-07,
+      "loss": 0.6483,
+      "step": 16378
+    },
+    {
+      "epoch": 1.395739241585002,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.261862013614756e-07,
+      "loss": 0.4542,
+      "step": 16379
+    },
+    {
+      "epoch": 1.395824456753302,
+      "grad_norm": 18.0,
+      "learning_rate": 5.260492099804144e-07,
+      "loss": 0.6637,
+      "step": 16380
+    },
+    {
+      "epoch": 1.395909671921602,
+      "grad_norm": 17.5,
+      "learning_rate": 5.259122316818719e-07,
+      "loss": 0.5686,
+      "step": 16381
+    },
+    {
+      "epoch": 1.395994887089902,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.257752664683237e-07,
+      "loss": 0.5135,
+      "step": 16382
+    },
+    {
+      "epoch": 1.396080102258202,
+      "grad_norm": 17.625,
+      "learning_rate": 5.25638314342245e-07,
+      "loss": 1.0912,
+      "step": 16383
+    },
+    {
+      "epoch": 1.3961653174265019,
+      "grad_norm": 17.875,
+      "learning_rate": 5.255013753061099e-07,
+      "loss": 0.6753,
+      "step": 16384
+    },
+    {
+      "epoch": 1.3962505325948018,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.253644493623936e-07,
+      "loss": 0.4023,
+      "step": 16385
+    },
+    {
+      "epoch": 1.3963357477631018,
+      "grad_norm": 12.0,
+      "learning_rate": 5.252275365135708e-07,
+      "loss": 0.3412,
+      "step": 16386
+    },
+    {
+      "epoch": 1.3964209629314017,
+      "grad_norm": 18.625,
+      "learning_rate": 5.250906367621146e-07,
+      "loss": 0.9212,
+      "step": 16387
+    },
+    {
+      "epoch": 1.3965061780997017,
+      "grad_norm": 13.9375,
+      "learning_rate": 5.249537501105001e-07,
+      "loss": 0.6369,
+      "step": 16388
+    },
+    {
+      "epoch": 1.3965913932680016,
+      "grad_norm": 23.875,
+      "learning_rate": 5.248168765612e-07,
+      "loss": 1.1704,
+      "step": 16389
+    },
+    {
+      "epoch": 1.3966766084363016,
+      "grad_norm": 13.625,
+      "learning_rate": 5.246800161166882e-07,
+      "loss": 0.3576,
+      "step": 16390
+    },
+    {
+      "epoch": 1.3967618236046015,
+      "grad_norm": 19.25,
+      "learning_rate": 5.245431687794379e-07,
+      "loss": 0.692,
+      "step": 16391
+    },
+    {
+      "epoch": 1.3968470387729015,
+      "grad_norm": 13.1875,
+      "learning_rate": 5.244063345519223e-07,
+      "loss": 0.3199,
+      "step": 16392
+    },
+    {
+      "epoch": 1.3969322539412015,
+      "grad_norm": 16.25,
+      "learning_rate": 5.242695134366137e-07,
+      "loss": 0.3489,
+      "step": 16393
+    },
+    {
+      "epoch": 1.3970174691095014,
+      "grad_norm": 12.875,
+      "learning_rate": 5.24132705435985e-07,
+      "loss": 0.5171,
+      "step": 16394
+    },
+    {
+      "epoch": 1.3971026842778014,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.239959105525079e-07,
+      "loss": 0.6713,
+      "step": 16395
+    },
+    {
+      "epoch": 1.3971878994461013,
+      "grad_norm": 13.25,
+      "learning_rate": 5.238591287886552e-07,
+      "loss": 0.4684,
+      "step": 16396
+    },
+    {
+      "epoch": 1.3972731146144013,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.237223601468977e-07,
+      "loss": 0.2055,
+      "step": 16397
+    },
+    {
+      "epoch": 1.3973583297827012,
+      "grad_norm": 13.75,
+      "learning_rate": 5.235856046297076e-07,
+      "loss": 0.399,
+      "step": 16398
+    },
+    {
+      "epoch": 1.3974435449510012,
+      "grad_norm": 14.75,
+      "learning_rate": 5.23448862239556e-07,
+      "loss": 0.5487,
+      "step": 16399
+    },
+    {
+      "epoch": 1.3975287601193012,
+      "grad_norm": 27.125,
+      "learning_rate": 5.233121329789145e-07,
+      "loss": 0.4427,
+      "step": 16400
+    },
+    {
+      "epoch": 1.3976139752876011,
+      "grad_norm": 25.125,
+      "learning_rate": 5.231754168502532e-07,
+      "loss": 0.8255,
+      "step": 16401
+    },
+    {
+      "epoch": 1.397699190455901,
+      "grad_norm": 20.75,
+      "learning_rate": 5.230387138560433e-07,
+      "loss": 1.1072,
+      "step": 16402
+    },
+    {
+      "epoch": 1.397784405624201,
+      "grad_norm": 19.625,
+      "learning_rate": 5.229020239987544e-07,
+      "loss": 0.8599,
+      "step": 16403
+    },
+    {
+      "epoch": 1.397869620792501,
+      "grad_norm": 14.75,
+      "learning_rate": 5.22765347280857e-07,
+      "loss": 0.5858,
+      "step": 16404
+    },
+    {
+      "epoch": 1.397954835960801,
+      "grad_norm": 17.625,
+      "learning_rate": 5.226286837048218e-07,
+      "loss": 0.8556,
+      "step": 16405
+    },
+    {
+      "epoch": 1.398040051129101,
+      "grad_norm": 33.25,
+      "learning_rate": 5.224920332731171e-07,
+      "loss": 1.0935,
+      "step": 16406
+    },
+    {
+      "epoch": 1.3981252662974009,
+      "grad_norm": 13.0625,
+      "learning_rate": 5.223553959882129e-07,
+      "loss": 0.4323,
+      "step": 16407
+    },
+    {
+      "epoch": 1.3982104814657008,
+      "grad_norm": 13.25,
+      "learning_rate": 5.222187718525784e-07,
+      "loss": 0.5871,
+      "step": 16408
+    },
+    {
+      "epoch": 1.3982956966340008,
+      "grad_norm": 14.125,
+      "learning_rate": 5.220821608686829e-07,
+      "loss": 0.5311,
+      "step": 16409
+    },
+    {
+      "epoch": 1.3983809118023007,
+      "grad_norm": 26.625,
+      "learning_rate": 5.21945563038995e-07,
+      "loss": 0.7321,
+      "step": 16410
+    },
+    {
+      "epoch": 1.3984661269706007,
+      "grad_norm": 12.875,
+      "learning_rate": 5.218089783659822e-07,
+      "loss": 0.2922,
+      "step": 16411
+    },
+    {
+      "epoch": 1.3985513421389006,
+      "grad_norm": 17.25,
+      "learning_rate": 5.216724068521135e-07,
+      "loss": 0.7021,
+      "step": 16412
+    },
+    {
+      "epoch": 1.3986365573072006,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.215358484998572e-07,
+      "loss": 0.6954,
+      "step": 16413
+    },
+    {
+      "epoch": 1.3987217724755006,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.213993033116803e-07,
+      "loss": 0.4491,
+      "step": 16414
+    },
+    {
+      "epoch": 1.3988069876438005,
+      "grad_norm": 18.625,
+      "learning_rate": 5.212627712900505e-07,
+      "loss": 0.5117,
+      "step": 16415
+    },
+    {
+      "epoch": 1.3988922028121005,
+      "grad_norm": 17.25,
+      "learning_rate": 5.211262524374354e-07,
+      "loss": 0.8865,
+      "step": 16416
+    },
+    {
+      "epoch": 1.3989774179804004,
+      "grad_norm": 12.75,
+      "learning_rate": 5.209897467563021e-07,
+      "loss": 0.4027,
+      "step": 16417
+    },
+    {
+      "epoch": 1.3990626331487004,
+      "grad_norm": 13.625,
+      "learning_rate": 5.208532542491174e-07,
+      "loss": 0.5065,
+      "step": 16418
+    },
+    {
+      "epoch": 1.3991478483170003,
+      "grad_norm": 15.5,
+      "learning_rate": 5.20716774918347e-07,
+      "loss": 0.5038,
+      "step": 16419
+    },
+    {
+      "epoch": 1.3992330634853003,
+      "grad_norm": 12.6875,
+      "learning_rate": 5.205803087664577e-07,
+      "loss": 0.4637,
+      "step": 16420
+    },
+    {
+      "epoch": 1.3993182786536003,
+      "grad_norm": 16.125,
+      "learning_rate": 5.20443855795916e-07,
+      "loss": 0.5285,
+      "step": 16421
+    },
+    {
+      "epoch": 1.3994034938219002,
+      "grad_norm": 22.125,
+      "learning_rate": 5.203074160091877e-07,
+      "loss": 0.6923,
+      "step": 16422
+    },
+    {
+      "epoch": 1.3994887089902002,
+      "grad_norm": 18.125,
+      "learning_rate": 5.201709894087379e-07,
+      "loss": 0.5117,
+      "step": 16423
+    },
+    {
+      "epoch": 1.3995739241585001,
+      "grad_norm": 17.125,
+      "learning_rate": 5.20034575997032e-07,
+      "loss": 0.681,
+      "step": 16424
+    },
+    {
+      "epoch": 1.3996591393268,
+      "grad_norm": 10.3125,
+      "learning_rate": 5.198981757765355e-07,
+      "loss": 0.3168,
+      "step": 16425
+    },
+    {
+      "epoch": 1.3997443544951,
+      "grad_norm": 17.125,
+      "learning_rate": 5.197617887497139e-07,
+      "loss": 0.7524,
+      "step": 16426
+    },
+    {
+      "epoch": 1.3998295696634,
+      "grad_norm": 13.25,
+      "learning_rate": 5.196254149190303e-07,
+      "loss": 0.4455,
+      "step": 16427
+    },
+    {
+      "epoch": 1.3999147848317,
+      "grad_norm": 13.375,
+      "learning_rate": 5.194890542869499e-07,
+      "loss": 0.6791,
+      "step": 16428
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 12.75,
+      "learning_rate": 5.193527068559368e-07,
+      "loss": 0.4361,
+      "step": 16429
+    },
+    {
+      "epoch": 1.4000852151682999,
+      "grad_norm": 12.375,
+      "learning_rate": 5.192163726284555e-07,
+      "loss": 0.3726,
+      "step": 16430
+    },
+    {
+      "epoch": 1.4001704303365998,
+      "grad_norm": 20.375,
+      "learning_rate": 5.190800516069687e-07,
+      "loss": 0.8526,
+      "step": 16431
+    },
+    {
+      "epoch": 1.4002556455048998,
+      "grad_norm": 16.625,
+      "learning_rate": 5.189437437939406e-07,
+      "loss": 0.664,
+      "step": 16432
+    },
+    {
+      "epoch": 1.4003408606731997,
+      "grad_norm": 21.375,
+      "learning_rate": 5.188074491918339e-07,
+      "loss": 1.0777,
+      "step": 16433
+    },
+    {
+      "epoch": 1.4004260758414997,
+      "grad_norm": 19.25,
+      "learning_rate": 5.186711678031124e-07,
+      "loss": 0.7175,
+      "step": 16434
+    },
+    {
+      "epoch": 1.4005112910097997,
+      "grad_norm": 18.875,
+      "learning_rate": 5.185348996302383e-07,
+      "loss": 1.1691,
+      "step": 16435
+    },
+    {
+      "epoch": 1.4005965061780996,
+      "grad_norm": 15.875,
+      "learning_rate": 5.183986446756738e-07,
+      "loss": 0.4921,
+      "step": 16436
+    },
+    {
+      "epoch": 1.4006817213463996,
+      "grad_norm": 13.3125,
+      "learning_rate": 5.182624029418813e-07,
+      "loss": 0.58,
+      "step": 16437
+    },
+    {
+      "epoch": 1.4007669365146995,
+      "grad_norm": 11.0625,
+      "learning_rate": 5.181261744313231e-07,
+      "loss": 0.3165,
+      "step": 16438
+    },
+    {
+      "epoch": 1.4008521516829995,
+      "grad_norm": 20.125,
+      "learning_rate": 5.179899591464613e-07,
+      "loss": 0.7726,
+      "step": 16439
+    },
+    {
+      "epoch": 1.4009373668512994,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.178537570897565e-07,
+      "loss": 0.4885,
+      "step": 16440
+    },
+    {
+      "epoch": 1.4010225820195994,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.177175682636707e-07,
+      "loss": 0.6742,
+      "step": 16441
+    },
+    {
+      "epoch": 1.4011077971878994,
+      "grad_norm": 13.9375,
+      "learning_rate": 5.175813926706651e-07,
+      "loss": 0.4466,
+      "step": 16442
+    },
+    {
+      "epoch": 1.4011930123561993,
+      "grad_norm": 16.5,
+      "learning_rate": 5.174452303132004e-07,
+      "loss": 0.5636,
+      "step": 16443
+    },
+    {
+      "epoch": 1.4012782275244993,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.173090811937364e-07,
+      "loss": 0.7403,
+      "step": 16444
+    },
+    {
+      "epoch": 1.4013634426927992,
+      "grad_norm": 17.25,
+      "learning_rate": 5.171729453147343e-07,
+      "loss": 0.6367,
+      "step": 16445
+    },
+    {
+      "epoch": 1.4014486578610992,
+      "grad_norm": 14.75,
+      "learning_rate": 5.170368226786537e-07,
+      "loss": 0.5159,
+      "step": 16446
+    },
+    {
+      "epoch": 1.4015338730293991,
+      "grad_norm": 12.875,
+      "learning_rate": 5.169007132879554e-07,
+      "loss": 0.6055,
+      "step": 16447
+    },
+    {
+      "epoch": 1.401619088197699,
+      "grad_norm": 15.5,
+      "learning_rate": 5.16764617145098e-07,
+      "loss": 0.5841,
+      "step": 16448
+    },
+    {
+      "epoch": 1.401704303365999,
+      "grad_norm": 14.5,
+      "learning_rate": 5.166285342525411e-07,
+      "loss": 0.572,
+      "step": 16449
+    },
+    {
+      "epoch": 1.401789518534299,
+      "grad_norm": 12.125,
+      "learning_rate": 5.164924646127442e-07,
+      "loss": 0.5027,
+      "step": 16450
+    },
+    {
+      "epoch": 1.401874733702599,
+      "grad_norm": 19.0,
+      "learning_rate": 5.163564082281665e-07,
+      "loss": 0.8262,
+      "step": 16451
+    },
+    {
+      "epoch": 1.401959948870899,
+      "grad_norm": 10.875,
+      "learning_rate": 5.162203651012662e-07,
+      "loss": 0.2286,
+      "step": 16452
+    },
+    {
+      "epoch": 1.4020451640391989,
+      "grad_norm": 14.125,
+      "learning_rate": 5.160843352345013e-07,
+      "loss": 0.4557,
+      "step": 16453
+    },
+    {
+      "epoch": 1.4021303792074988,
+      "grad_norm": 17.5,
+      "learning_rate": 5.159483186303306e-07,
+      "loss": 0.698,
+      "step": 16454
+    },
+    {
+      "epoch": 1.4022155943757988,
+      "grad_norm": 17.25,
+      "learning_rate": 5.158123152912118e-07,
+      "loss": 0.5511,
+      "step": 16455
+    },
+    {
+      "epoch": 1.4023008095440987,
+      "grad_norm": 16.125,
+      "learning_rate": 5.156763252196033e-07,
+      "loss": 0.5065,
+      "step": 16456
+    },
+    {
+      "epoch": 1.4023860247123987,
+      "grad_norm": 23.0,
+      "learning_rate": 5.155403484179615e-07,
+      "loss": 0.5213,
+      "step": 16457
+    },
+    {
+      "epoch": 1.4024712398806987,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.154043848887441e-07,
+      "loss": 0.4562,
+      "step": 16458
+    },
+    {
+      "epoch": 1.4025564550489986,
+      "grad_norm": 11.6875,
+      "learning_rate": 5.152684346344087e-07,
+      "loss": 0.3337,
+      "step": 16459
+    },
+    {
+      "epoch": 1.4026416702172986,
+      "grad_norm": 15.25,
+      "learning_rate": 5.151324976574115e-07,
+      "loss": 0.6697,
+      "step": 16460
+    },
+    {
+      "epoch": 1.4027268853855985,
+      "grad_norm": 14.625,
+      "learning_rate": 5.149965739602086e-07,
+      "loss": 0.5686,
+      "step": 16461
+    },
+    {
+      "epoch": 1.4028121005538985,
+      "grad_norm": 11.5625,
+      "learning_rate": 5.148606635452568e-07,
+      "loss": 0.453,
+      "step": 16462
+    },
+    {
+      "epoch": 1.4028973157221984,
+      "grad_norm": 29.125,
+      "learning_rate": 5.147247664150118e-07,
+      "loss": 0.6343,
+      "step": 16463
+    },
+    {
+      "epoch": 1.4029825308904984,
+      "grad_norm": 14.625,
+      "learning_rate": 5.145888825719303e-07,
+      "loss": 0.5851,
+      "step": 16464
+    },
+    {
+      "epoch": 1.4030677460587984,
+      "grad_norm": 17.125,
+      "learning_rate": 5.144530120184667e-07,
+      "loss": 0.461,
+      "step": 16465
+    },
+    {
+      "epoch": 1.4031529612270983,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.143171547570767e-07,
+      "loss": 0.3404,
+      "step": 16466
+    },
+    {
+      "epoch": 1.4032381763953983,
+      "grad_norm": 19.375,
+      "learning_rate": 5.141813107902162e-07,
+      "loss": 1.0303,
+      "step": 16467
+    },
+    {
+      "epoch": 1.4033233915636982,
+      "grad_norm": 17.75,
+      "learning_rate": 5.140454801203387e-07,
+      "loss": 0.6771,
+      "step": 16468
+    },
+    {
+      "epoch": 1.4034086067319982,
+      "grad_norm": 14.625,
+      "learning_rate": 5.139096627498999e-07,
+      "loss": 0.4882,
+      "step": 16469
+    },
+    {
+      "epoch": 1.4034938219002981,
+      "grad_norm": 17.375,
+      "learning_rate": 5.137738586813533e-07,
+      "loss": 0.7031,
+      "step": 16470
+    },
+    {
+      "epoch": 1.403579037068598,
+      "grad_norm": 40.5,
+      "learning_rate": 5.136380679171534e-07,
+      "loss": 1.2443,
+      "step": 16471
+    },
+    {
+      "epoch": 1.403664252236898,
+      "grad_norm": 13.625,
+      "learning_rate": 5.135022904597544e-07,
+      "loss": 0.4532,
+      "step": 16472
+    },
+    {
+      "epoch": 1.403749467405198,
+      "grad_norm": 19.875,
+      "learning_rate": 5.133665263116094e-07,
+      "loss": 0.4926,
+      "step": 16473
+    },
+    {
+      "epoch": 1.403834682573498,
+      "grad_norm": 25.25,
+      "learning_rate": 5.132307754751719e-07,
+      "loss": 0.9242,
+      "step": 16474
+    },
+    {
+      "epoch": 1.403919897741798,
+      "grad_norm": 16.125,
+      "learning_rate": 5.130950379528956e-07,
+      "loss": 0.8087,
+      "step": 16475
+    },
+    {
+      "epoch": 1.4040051129100979,
+      "grad_norm": 19.125,
+      "learning_rate": 5.129593137472326e-07,
+      "loss": 1.1997,
+      "step": 16476
+    },
+    {
+      "epoch": 1.4040903280783978,
+      "grad_norm": 17.75,
+      "learning_rate": 5.128236028606363e-07,
+      "loss": 0.9101,
+      "step": 16477
+    },
+    {
+      "epoch": 1.4041755432466978,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.126879052955584e-07,
+      "loss": 0.6631,
+      "step": 16478
+    },
+    {
+      "epoch": 1.4042607584149978,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.125522210544515e-07,
+      "loss": 0.5251,
+      "step": 16479
+    },
+    {
+      "epoch": 1.4043459735832977,
+      "grad_norm": 16.5,
+      "learning_rate": 5.124165501397674e-07,
+      "loss": 0.5776,
+      "step": 16480
+    },
+    {
+      "epoch": 1.4044311887515977,
+      "grad_norm": 21.75,
+      "learning_rate": 5.122808925539585e-07,
+      "loss": 0.772,
+      "step": 16481
+    },
+    {
+      "epoch": 1.4045164039198976,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.121452482994751e-07,
+      "loss": 0.5172,
+      "step": 16482
+    },
+    {
+      "epoch": 1.4046016190881976,
+      "grad_norm": 12.1875,
+      "learning_rate": 5.120096173787696e-07,
+      "loss": 0.7046,
+      "step": 16483
+    },
+    {
+      "epoch": 1.4046868342564975,
+      "grad_norm": 19.375,
+      "learning_rate": 5.118739997942921e-07,
+      "loss": 1.0487,
+      "step": 16484
+    },
+    {
+      "epoch": 1.4047720494247975,
+      "grad_norm": 15.5,
+      "learning_rate": 5.117383955484935e-07,
+      "loss": 0.7468,
+      "step": 16485
+    },
+    {
+      "epoch": 1.4048572645930975,
+      "grad_norm": 13.625,
+      "learning_rate": 5.11602804643825e-07,
+      "loss": 0.6507,
+      "step": 16486
+    },
+    {
+      "epoch": 1.4049424797613974,
+      "grad_norm": 12.4375,
+      "learning_rate": 5.114672270827358e-07,
+      "loss": 0.4388,
+      "step": 16487
+    },
+    {
+      "epoch": 1.4050276949296974,
+      "grad_norm": 16.25,
+      "learning_rate": 5.113316628676765e-07,
+      "loss": 0.5757,
+      "step": 16488
+    },
+    {
+      "epoch": 1.4051129100979973,
+      "grad_norm": 19.25,
+      "learning_rate": 5.111961120010974e-07,
+      "loss": 0.7725,
+      "step": 16489
+    },
+    {
+      "epoch": 1.4051981252662973,
+      "grad_norm": 18.875,
+      "learning_rate": 5.110605744854469e-07,
+      "loss": 0.7043,
+      "step": 16490
+    },
+    {
+      "epoch": 1.4052833404345972,
+      "grad_norm": 17.125,
+      "learning_rate": 5.109250503231752e-07,
+      "loss": 0.8088,
+      "step": 16491
+    },
+    {
+      "epoch": 1.4053685556028972,
+      "grad_norm": 18.125,
+      "learning_rate": 5.107895395167309e-07,
+      "loss": 0.7808,
+      "step": 16492
+    },
+    {
+      "epoch": 1.4054537707711972,
+      "grad_norm": 18.75,
+      "learning_rate": 5.106540420685629e-07,
+      "loss": 0.8131,
+      "step": 16493
+    },
+    {
+      "epoch": 1.405538985939497,
+      "grad_norm": 15.75,
+      "learning_rate": 5.105185579811201e-07,
+      "loss": 0.7294,
+      "step": 16494
+    },
+    {
+      "epoch": 1.405624201107797,
+      "grad_norm": 15.0,
+      "learning_rate": 5.103830872568501e-07,
+      "loss": 0.7818,
+      "step": 16495
+    },
+    {
+      "epoch": 1.405709416276097,
+      "grad_norm": 12.9375,
+      "learning_rate": 5.102476298982017e-07,
+      "loss": 0.4041,
+      "step": 16496
+    },
+    {
+      "epoch": 1.405794631444397,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.101121859076222e-07,
+      "loss": 0.7404,
+      "step": 16497
+    },
+    {
+      "epoch": 1.405879846612697,
+      "grad_norm": 17.5,
+      "learning_rate": 5.099767552875602e-07,
+      "loss": 0.6422,
+      "step": 16498
+    },
+    {
+      "epoch": 1.405965061780997,
+      "grad_norm": 15.625,
+      "learning_rate": 5.098413380404622e-07,
+      "loss": 0.5735,
+      "step": 16499
+    },
+    {
+      "epoch": 1.4060502769492969,
+      "grad_norm": 17.125,
+      "learning_rate": 5.097059341687751e-07,
+      "loss": 0.5003,
+      "step": 16500
+    },
+    {
+      "epoch": 1.4061354921175968,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.095705436749464e-07,
+      "loss": 0.4601,
+      "step": 16501
+    },
+    {
+      "epoch": 1.4062207072858968,
+      "grad_norm": 16.625,
+      "learning_rate": 5.094351665614228e-07,
+      "loss": 0.8413,
+      "step": 16502
+    },
+    {
+      "epoch": 1.4063059224541967,
+      "grad_norm": 16.125,
+      "learning_rate": 5.0929980283065e-07,
+      "loss": 0.8693,
+      "step": 16503
+    },
+    {
+      "epoch": 1.4063911376224967,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.091644524850747e-07,
+      "loss": 0.2907,
+      "step": 16504
+    },
+    {
+      "epoch": 1.4064763527907966,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.090291155271427e-07,
+      "loss": 0.4796,
+      "step": 16505
+    },
+    {
+      "epoch": 1.4065615679590966,
+      "grad_norm": 17.875,
+      "learning_rate": 5.088937919593e-07,
+      "loss": 0.7473,
+      "step": 16506
+    },
+    {
+      "epoch": 1.4066467831273965,
+      "grad_norm": 33.5,
+      "learning_rate": 5.087584817839917e-07,
+      "loss": 0.7585,
+      "step": 16507
+    },
+    {
+      "epoch": 1.4067319982956965,
+      "grad_norm": 15.0,
+      "learning_rate": 5.086231850036627e-07,
+      "loss": 0.6283,
+      "step": 16508
+    },
+    {
+      "epoch": 1.4068172134639965,
+      "grad_norm": 16.75,
+      "learning_rate": 5.084879016207582e-07,
+      "loss": 0.5981,
+      "step": 16509
+    },
+    {
+      "epoch": 1.4069024286322964,
+      "grad_norm": 10.5,
+      "learning_rate": 5.083526316377229e-07,
+      "loss": 0.8238,
+      "step": 16510
+    },
+    {
+      "epoch": 1.4069876438005964,
+      "grad_norm": 18.0,
+      "learning_rate": 5.08217375057002e-07,
+      "loss": 0.6106,
+      "step": 16511
+    },
+    {
+      "epoch": 1.4070728589688963,
+      "grad_norm": 16.5,
+      "learning_rate": 5.080821318810385e-07,
+      "loss": 0.8085,
+      "step": 16512
+    },
+    {
+      "epoch": 1.4071580741371963,
+      "grad_norm": 16.5,
+      "learning_rate": 5.079469021122768e-07,
+      "loss": 0.7634,
+      "step": 16513
+    },
+    {
+      "epoch": 1.4072432893054962,
+      "grad_norm": 23.375,
+      "learning_rate": 5.078116857531609e-07,
+      "loss": 0.711,
+      "step": 16514
+    },
+    {
+      "epoch": 1.4073285044737962,
+      "grad_norm": 14.125,
+      "learning_rate": 5.076764828061345e-07,
+      "loss": 0.5395,
+      "step": 16515
+    },
+    {
+      "epoch": 1.4074137196420962,
+      "grad_norm": 17.125,
+      "learning_rate": 5.075412932736406e-07,
+      "loss": 0.4234,
+      "step": 16516
+    },
+    {
+      "epoch": 1.4074989348103961,
+      "grad_norm": 12.375,
+      "learning_rate": 5.074061171581216e-07,
+      "loss": 0.4247,
+      "step": 16517
+    },
+    {
+      "epoch": 1.407584149978696,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.072709544620211e-07,
+      "loss": 0.2638,
+      "step": 16518
+    },
+    {
+      "epoch": 1.407669365146996,
+      "grad_norm": 21.875,
+      "learning_rate": 5.071358051877814e-07,
+      "loss": 1.008,
+      "step": 16519
+    },
+    {
+      "epoch": 1.407754580315296,
+      "grad_norm": 35.5,
+      "learning_rate": 5.070006693378445e-07,
+      "loss": 1.0408,
+      "step": 16520
+    },
+    {
+      "epoch": 1.407839795483596,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.068655469146526e-07,
+      "loss": 0.5524,
+      "step": 16521
+    },
+    {
+      "epoch": 1.407925010651896,
+      "grad_norm": 15.5625,
+      "learning_rate": 5.067304379206473e-07,
+      "loss": 0.3218,
+      "step": 16522
+    },
+    {
+      "epoch": 1.4080102258201959,
+      "grad_norm": 17.875,
+      "learning_rate": 5.065953423582712e-07,
+      "loss": 0.8556,
+      "step": 16523
+    },
+    {
+      "epoch": 1.4080954409884958,
+      "grad_norm": 22.75,
+      "learning_rate": 5.064602602299647e-07,
+      "loss": 0.9158,
+      "step": 16524
+    },
+    {
+      "epoch": 1.4081806561567958,
+      "grad_norm": 15.375,
+      "learning_rate": 5.063251915381683e-07,
+      "loss": 0.4169,
+      "step": 16525
+    },
+    {
+      "epoch": 1.4082658713250957,
+      "grad_norm": 14.75,
+      "learning_rate": 5.061901362853239e-07,
+      "loss": 0.988,
+      "step": 16526
+    },
+    {
+      "epoch": 1.4083510864933957,
+      "grad_norm": 18.625,
+      "learning_rate": 5.060550944738715e-07,
+      "loss": 0.7488,
+      "step": 16527
+    },
+    {
+      "epoch": 1.4084363016616956,
+      "grad_norm": 11.125,
+      "learning_rate": 5.059200661062522e-07,
+      "loss": 0.3206,
+      "step": 16528
+    },
+    {
+      "epoch": 1.4085215168299956,
+      "grad_norm": 19.125,
+      "learning_rate": 5.057850511849049e-07,
+      "loss": 0.882,
+      "step": 16529
+    },
+    {
+      "epoch": 1.4086067319982956,
+      "grad_norm": 16.75,
+      "learning_rate": 5.056500497122702e-07,
+      "loss": 0.6264,
+      "step": 16530
+    },
+    {
+      "epoch": 1.4086919471665957,
+      "grad_norm": 28.375,
+      "learning_rate": 5.055150616907875e-07,
+      "loss": 0.8639,
+      "step": 16531
+    },
+    {
+      "epoch": 1.4087771623348957,
+      "grad_norm": 10.8125,
+      "learning_rate": 5.053800871228969e-07,
+      "loss": 0.3065,
+      "step": 16532
+    },
+    {
+      "epoch": 1.4088623775031957,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.052451260110368e-07,
+      "loss": 0.3775,
+      "step": 16533
+    },
+    {
+      "epoch": 1.4089475926714956,
+      "grad_norm": 21.0,
+      "learning_rate": 5.051101783576456e-07,
+      "loss": 0.4998,
+      "step": 16534
+    },
+    {
+      "epoch": 1.4090328078397956,
+      "grad_norm": 12.3125,
+      "learning_rate": 5.049752441651628e-07,
+      "loss": 0.4299,
+      "step": 16535
+    },
+    {
+      "epoch": 1.4091180230080955,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.048403234360268e-07,
+      "loss": 0.4584,
+      "step": 16536
+    },
+    {
+      "epoch": 1.4092032381763955,
+      "grad_norm": 13.25,
+      "learning_rate": 5.047054161726749e-07,
+      "loss": 0.4749,
+      "step": 16537
+    },
+    {
+      "epoch": 1.4092884533446954,
+      "grad_norm": 22.375,
+      "learning_rate": 5.045705223775459e-07,
+      "loss": 0.7882,
+      "step": 16538
+    },
+    {
+      "epoch": 1.4093736685129954,
+      "grad_norm": 19.875,
+      "learning_rate": 5.044356420530769e-07,
+      "loss": 0.787,
+      "step": 16539
+    },
+    {
+      "epoch": 1.4094588836812953,
+      "grad_norm": 10.8125,
+      "learning_rate": 5.043007752017061e-07,
+      "loss": 0.2653,
+      "step": 16540
+    },
+    {
+      "epoch": 1.4095440988495953,
+      "grad_norm": 17.875,
+      "learning_rate": 5.041659218258702e-07,
+      "loss": 0.6112,
+      "step": 16541
+    },
+    {
+      "epoch": 1.4096293140178953,
+      "grad_norm": 13.25,
+      "learning_rate": 5.040310819280056e-07,
+      "loss": 0.6303,
+      "step": 16542
+    },
+    {
+      "epoch": 1.4097145291861952,
+      "grad_norm": 12.875,
+      "learning_rate": 5.038962555105493e-07,
+      "loss": 0.3604,
+      "step": 16543
+    },
+    {
+      "epoch": 1.4097997443544952,
+      "grad_norm": 12.25,
+      "learning_rate": 5.037614425759382e-07,
+      "loss": 0.3688,
+      "step": 16544
+    },
+    {
+      "epoch": 1.4098849595227951,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.036266431266086e-07,
+      "loss": 0.3841,
+      "step": 16545
+    },
+    {
+      "epoch": 1.409970174691095,
+      "grad_norm": 15.6875,
+      "learning_rate": 5.034918571649956e-07,
+      "loss": 0.7882,
+      "step": 16546
+    },
+    {
+      "epoch": 1.410055389859395,
+      "grad_norm": 66.0,
+      "learning_rate": 5.033570846935354e-07,
+      "loss": 0.6905,
+      "step": 16547
+    },
+    {
+      "epoch": 1.410140605027695,
+      "grad_norm": 12.625,
+      "learning_rate": 5.03222325714664e-07,
+      "loss": 0.4332,
+      "step": 16548
+    },
+    {
+      "epoch": 1.410225820195995,
+      "grad_norm": 10.625,
+      "learning_rate": 5.030875802308161e-07,
+      "loss": 0.3606,
+      "step": 16549
+    },
+    {
+      "epoch": 1.410311035364295,
+      "grad_norm": 19.375,
+      "learning_rate": 5.029528482444262e-07,
+      "loss": 0.6726,
+      "step": 16550
+    },
+    {
+      "epoch": 1.4103962505325949,
+      "grad_norm": 29.25,
+      "learning_rate": 5.028181297579296e-07,
+      "loss": 0.946,
+      "step": 16551
+    },
+    {
+      "epoch": 1.4104814657008948,
+      "grad_norm": 17.0,
+      "learning_rate": 5.026834247737608e-07,
+      "loss": 0.7243,
+      "step": 16552
+    },
+    {
+      "epoch": 1.4105666808691948,
+      "grad_norm": 22.875,
+      "learning_rate": 5.025487332943543e-07,
+      "loss": 0.6982,
+      "step": 16553
+    },
+    {
+      "epoch": 1.4106518960374947,
+      "grad_norm": 15.125,
+      "learning_rate": 5.024140553221435e-07,
+      "loss": 0.7655,
+      "step": 16554
+    },
+    {
+      "epoch": 1.4107371112057947,
+      "grad_norm": 15.75,
+      "learning_rate": 5.022793908595625e-07,
+      "loss": 0.6237,
+      "step": 16555
+    },
+    {
+      "epoch": 1.4108223263740947,
+      "grad_norm": 13.25,
+      "learning_rate": 5.021447399090451e-07,
+      "loss": 0.5613,
+      "step": 16556
+    },
+    {
+      "epoch": 1.4109075415423946,
+      "grad_norm": 19.375,
+      "learning_rate": 5.020101024730238e-07,
+      "loss": 1.0481,
+      "step": 16557
+    },
+    {
+      "epoch": 1.4109927567106946,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.018754785539328e-07,
+      "loss": 0.4068,
+      "step": 16558
+    },
+    {
+      "epoch": 1.4110779718789945,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.017408681542035e-07,
+      "loss": 0.5206,
+      "step": 16559
+    },
+    {
+      "epoch": 1.4111631870472945,
+      "grad_norm": 21.625,
+      "learning_rate": 5.016062712762693e-07,
+      "loss": 0.8158,
+      "step": 16560
+    },
+    {
+      "epoch": 1.4112484022155944,
+      "grad_norm": 24.125,
+      "learning_rate": 5.014716879225623e-07,
+      "loss": 0.6583,
+      "step": 16561
+    },
+    {
+      "epoch": 1.4113336173838944,
+      "grad_norm": 18.75,
+      "learning_rate": 5.01337118095515e-07,
+      "loss": 0.6697,
+      "step": 16562
+    },
+    {
+      "epoch": 1.4114188325521944,
+      "grad_norm": 16.5,
+      "learning_rate": 5.012025617975584e-07,
+      "loss": 0.4441,
+      "step": 16563
+    },
+    {
+      "epoch": 1.4115040477204943,
+      "grad_norm": 11.5,
+      "learning_rate": 5.01068019031125e-07,
+      "loss": 1.2402,
+      "step": 16564
+    },
+    {
+      "epoch": 1.4115892628887943,
+      "grad_norm": 16.0,
+      "learning_rate": 5.009334897986451e-07,
+      "loss": 0.5753,
+      "step": 16565
+    },
+    {
+      "epoch": 1.4116744780570942,
+      "grad_norm": 16.375,
+      "learning_rate": 5.007989741025507e-07,
+      "loss": 0.9699,
+      "step": 16566
+    },
+    {
+      "epoch": 1.4117596932253942,
+      "grad_norm": 13.5,
+      "learning_rate": 5.006644719452717e-07,
+      "loss": 0.5311,
+      "step": 16567
+    },
+    {
+      "epoch": 1.4118449083936941,
+      "grad_norm": 20.5,
+      "learning_rate": 5.005299833292394e-07,
+      "loss": 0.8108,
+      "step": 16568
+    },
+    {
+      "epoch": 1.411930123561994,
+      "grad_norm": 16.25,
+      "learning_rate": 5.003955082568838e-07,
+      "loss": 0.7582,
+      "step": 16569
+    },
+    {
+      "epoch": 1.412015338730294,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.002610467306357e-07,
+      "loss": 0.5584,
+      "step": 16570
+    },
+    {
+      "epoch": 1.412100553898594,
+      "grad_norm": 23.125,
+      "learning_rate": 5.00126598752924e-07,
+      "loss": 0.489,
+      "step": 16571
+    },
+    {
+      "epoch": 1.412185769066894,
+      "grad_norm": 18.25,
+      "learning_rate": 4.999921643261791e-07,
+      "loss": 0.8342,
+      "step": 16572
+    },
+    {
+      "epoch": 1.412270984235194,
+      "grad_norm": 13.25,
+      "learning_rate": 4.998577434528294e-07,
+      "loss": 0.2673,
+      "step": 16573
+    },
+    {
+      "epoch": 1.4123561994034939,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.997233361353047e-07,
+      "loss": 0.7502,
+      "step": 16574
+    },
+    {
+      "epoch": 1.4124414145717938,
+      "grad_norm": 19.5,
+      "learning_rate": 4.995889423760343e-07,
+      "loss": 0.5668,
+      "step": 16575
+    },
+    {
+      "epoch": 1.4125266297400938,
+      "grad_norm": 18.625,
+      "learning_rate": 4.994545621774457e-07,
+      "loss": 0.6787,
+      "step": 16576
+    },
+    {
+      "epoch": 1.4126118449083938,
+      "grad_norm": 20.625,
+      "learning_rate": 4.993201955419681e-07,
+      "loss": 1.0089,
+      "step": 16577
+    },
+    {
+      "epoch": 1.4126970600766937,
+      "grad_norm": 12.5,
+      "learning_rate": 4.991858424720293e-07,
+      "loss": 0.4901,
+      "step": 16578
+    },
+    {
+      "epoch": 1.4127822752449937,
+      "grad_norm": 16.875,
+      "learning_rate": 4.990515029700578e-07,
+      "loss": 0.5484,
+      "step": 16579
+    },
+    {
+      "epoch": 1.4128674904132936,
+      "grad_norm": 15.375,
+      "learning_rate": 4.989171770384809e-07,
+      "loss": 0.4544,
+      "step": 16580
+    },
+    {
+      "epoch": 1.4129527055815936,
+      "grad_norm": 11.3125,
+      "learning_rate": 4.987828646797252e-07,
+      "loss": 0.2709,
+      "step": 16581
+    },
+    {
+      "epoch": 1.4130379207498935,
+      "grad_norm": 15.5,
+      "learning_rate": 4.986485658962186e-07,
+      "loss": 0.6004,
+      "step": 16582
+    },
+    {
+      "epoch": 1.4131231359181935,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.985142806903885e-07,
+      "loss": 0.3834,
+      "step": 16583
+    },
+    {
+      "epoch": 1.4132083510864935,
+      "grad_norm": 17.375,
+      "learning_rate": 4.983800090646607e-07,
+      "loss": 0.5642,
+      "step": 16584
+    },
+    {
+      "epoch": 1.4132935662547934,
+      "grad_norm": 10.875,
+      "learning_rate": 4.982457510214617e-07,
+      "loss": 0.4225,
+      "step": 16585
+    },
+    {
+      "epoch": 1.4133787814230934,
+      "grad_norm": 19.5,
+      "learning_rate": 4.98111506563218e-07,
+      "loss": 0.7194,
+      "step": 16586
+    },
+    {
+      "epoch": 1.4134639965913933,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.979772756923559e-07,
+      "loss": 0.6912,
+      "step": 16587
+    },
+    {
+      "epoch": 1.4135492117596933,
+      "grad_norm": 16.5,
+      "learning_rate": 4.978430584113003e-07,
+      "loss": 0.963,
+      "step": 16588
+    },
+    {
+      "epoch": 1.4136344269279932,
+      "grad_norm": 18.625,
+      "learning_rate": 4.977088547224773e-07,
+      "loss": 0.6714,
+      "step": 16589
+    },
+    {
+      "epoch": 1.4137196420962932,
+      "grad_norm": 13.25,
+      "learning_rate": 4.975746646283114e-07,
+      "loss": 0.5975,
+      "step": 16590
+    },
+    {
+      "epoch": 1.4138048572645932,
+      "grad_norm": 14.625,
+      "learning_rate": 4.97440488131228e-07,
+      "loss": 0.3642,
+      "step": 16591
+    },
+    {
+      "epoch": 1.413890072432893,
+      "grad_norm": 25.0,
+      "learning_rate": 4.973063252336521e-07,
+      "loss": 1.146,
+      "step": 16592
+    },
+    {
+      "epoch": 1.413975287601193,
+      "grad_norm": 10.75,
+      "learning_rate": 4.971721759380073e-07,
+      "loss": 0.416,
+      "step": 16593
+    },
+    {
+      "epoch": 1.414060502769493,
+      "grad_norm": 21.25,
+      "learning_rate": 4.970380402467185e-07,
+      "loss": 1.0392,
+      "step": 16594
+    },
+    {
+      "epoch": 1.414145717937793,
+      "grad_norm": 23.375,
+      "learning_rate": 4.969039181622093e-07,
+      "loss": 0.6082,
+      "step": 16595
+    },
+    {
+      "epoch": 1.414230933106093,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.96769809686904e-07,
+      "loss": 0.9213,
+      "step": 16596
+    },
+    {
+      "epoch": 1.414316148274393,
+      "grad_norm": 10.5,
+      "learning_rate": 4.966357148232258e-07,
+      "loss": 0.241,
+      "step": 16597
+    },
+    {
+      "epoch": 1.4144013634426928,
+      "grad_norm": 24.5,
+      "learning_rate": 4.965016335735973e-07,
+      "loss": 0.4489,
+      "step": 16598
+    },
+    {
+      "epoch": 1.4144865786109928,
+      "grad_norm": 10.9375,
+      "learning_rate": 4.963675659404419e-07,
+      "loss": 0.2269,
+      "step": 16599
+    },
+    {
+      "epoch": 1.4145717937792928,
+      "grad_norm": 18.125,
+      "learning_rate": 4.962335119261831e-07,
+      "loss": 0.7874,
+      "step": 16600
+    },
+    {
+      "epoch": 1.4146570089475927,
+      "grad_norm": 17.625,
+      "learning_rate": 4.960994715332421e-07,
+      "loss": 0.6893,
+      "step": 16601
+    },
+    {
+      "epoch": 1.4147422241158927,
+      "grad_norm": 22.125,
+      "learning_rate": 4.959654447640417e-07,
+      "loss": 0.4085,
+      "step": 16602
+    },
+    {
+      "epoch": 1.4148274392841926,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.958314316210041e-07,
+      "loss": 0.4995,
+      "step": 16603
+    },
+    {
+      "epoch": 1.4149126544524926,
+      "grad_norm": 18.625,
+      "learning_rate": 4.956974321065512e-07,
+      "loss": 0.9185,
+      "step": 16604
+    },
+    {
+      "epoch": 1.4149978696207925,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.955634462231043e-07,
+      "loss": 0.4122,
+      "step": 16605
+    },
+    {
+      "epoch": 1.4150830847890925,
+      "grad_norm": 18.0,
+      "learning_rate": 4.954294739730842e-07,
+      "loss": 0.8476,
+      "step": 16606
+    },
+    {
+      "epoch": 1.4151682999573925,
+      "grad_norm": 23.0,
+      "learning_rate": 4.952955153589122e-07,
+      "loss": 0.7513,
+      "step": 16607
+    },
+    {
+      "epoch": 1.4152535151256924,
+      "grad_norm": 14.375,
+      "learning_rate": 4.951615703830091e-07,
+      "loss": 0.4624,
+      "step": 16608
+    },
+    {
+      "epoch": 1.4153387302939924,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.95027639047796e-07,
+      "loss": 0.4218,
+      "step": 16609
+    },
+    {
+      "epoch": 1.4154239454622923,
+      "grad_norm": 17.125,
+      "learning_rate": 4.948937213556921e-07,
+      "loss": 0.8306,
+      "step": 16610
+    },
+    {
+      "epoch": 1.4155091606305923,
+      "grad_norm": 23.25,
+      "learning_rate": 4.94759817309118e-07,
+      "loss": 0.7987,
+      "step": 16611
+    },
+    {
+      "epoch": 1.4155943757988922,
+      "grad_norm": 11.375,
+      "learning_rate": 4.946259269104941e-07,
+      "loss": 0.2166,
+      "step": 16612
+    },
+    {
+      "epoch": 1.4156795909671922,
+      "grad_norm": 17.75,
+      "learning_rate": 4.944920501622392e-07,
+      "loss": 0.7787,
+      "step": 16613
+    },
+    {
+      "epoch": 1.4157648061354922,
+      "grad_norm": 12.3125,
+      "learning_rate": 4.94358187066772e-07,
+      "loss": 0.3062,
+      "step": 16614
+    },
+    {
+      "epoch": 1.4158500213037921,
+      "grad_norm": 11.5625,
+      "learning_rate": 4.942243376265124e-07,
+      "loss": 0.2991,
+      "step": 16615
+    },
+    {
+      "epoch": 1.415935236472092,
+      "grad_norm": 34.75,
+      "learning_rate": 4.940905018438789e-07,
+      "loss": 0.7591,
+      "step": 16616
+    },
+    {
+      "epoch": 1.416020451640392,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.939566797212908e-07,
+      "loss": 0.3228,
+      "step": 16617
+    },
+    {
+      "epoch": 1.416105666808692,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.938228712611652e-07,
+      "loss": 0.5307,
+      "step": 16618
+    },
+    {
+      "epoch": 1.416190881976992,
+      "grad_norm": 12.0,
+      "learning_rate": 4.936890764659206e-07,
+      "loss": 0.3378,
+      "step": 16619
+    },
+    {
+      "epoch": 1.416276097145292,
+      "grad_norm": 12.625,
+      "learning_rate": 4.935552953379751e-07,
+      "loss": 0.3132,
+      "step": 16620
+    },
+    {
+      "epoch": 1.4163613123135919,
+      "grad_norm": 14.875,
+      "learning_rate": 4.934215278797465e-07,
+      "loss": 0.5478,
+      "step": 16621
+    },
+    {
+      "epoch": 1.4164465274818918,
+      "grad_norm": 13.25,
+      "learning_rate": 4.932877740936516e-07,
+      "loss": 0.2866,
+      "step": 16622
+    },
+    {
+      "epoch": 1.4165317426501918,
+      "grad_norm": 18.625,
+      "learning_rate": 4.931540339821071e-07,
+      "loss": 0.7415,
+      "step": 16623
+    },
+    {
+      "epoch": 1.4166169578184917,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.930203075475304e-07,
+      "loss": 0.7254,
+      "step": 16624
+    },
+    {
+      "epoch": 1.4167021729867917,
+      "grad_norm": 18.25,
+      "learning_rate": 4.928865947923379e-07,
+      "loss": 0.9204,
+      "step": 16625
+    },
+    {
+      "epoch": 1.4167873881550916,
+      "grad_norm": 20.625,
+      "learning_rate": 4.927528957189463e-07,
+      "loss": 1.0934,
+      "step": 16626
+    },
+    {
+      "epoch": 1.4168726033233916,
+      "grad_norm": 16.5,
+      "learning_rate": 4.92619210329771e-07,
+      "loss": 0.6441,
+      "step": 16627
+    },
+    {
+      "epoch": 1.4169578184916916,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.924855386272281e-07,
+      "loss": 0.4072,
+      "step": 16628
+    },
+    {
+      "epoch": 1.4170430336599915,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.923518806137338e-07,
+      "loss": 0.6813,
+      "step": 16629
+    },
+    {
+      "epoch": 1.4171282488282915,
+      "grad_norm": 16.75,
+      "learning_rate": 4.922182362917029e-07,
+      "loss": 0.6055,
+      "step": 16630
+    },
+    {
+      "epoch": 1.4172134639965914,
+      "grad_norm": 16.125,
+      "learning_rate": 4.9208460566355e-07,
+      "loss": 0.6539,
+      "step": 16631
+    },
+    {
+      "epoch": 1.4172986791648914,
+      "grad_norm": 22.25,
+      "learning_rate": 4.919509887316903e-07,
+      "loss": 0.866,
+      "step": 16632
+    },
+    {
+      "epoch": 1.4173838943331913,
+      "grad_norm": 16.5,
+      "learning_rate": 4.918173854985388e-07,
+      "loss": 0.5245,
+      "step": 16633
+    },
+    {
+      "epoch": 1.4174691095014913,
+      "grad_norm": 15.25,
+      "learning_rate": 4.916837959665097e-07,
+      "loss": 0.6661,
+      "step": 16634
+    },
+    {
+      "epoch": 1.4175543246697913,
+      "grad_norm": 17.0,
+      "learning_rate": 4.915502201380166e-07,
+      "loss": 0.4548,
+      "step": 16635
+    },
+    {
+      "epoch": 1.4176395398380912,
+      "grad_norm": 10.75,
+      "learning_rate": 4.914166580154737e-07,
+      "loss": 0.7864,
+      "step": 16636
+    },
+    {
+      "epoch": 1.4177247550063912,
+      "grad_norm": 11.75,
+      "learning_rate": 4.912831096012951e-07,
+      "loss": 0.2432,
+      "step": 16637
+    },
+    {
+      "epoch": 1.4178099701746911,
+      "grad_norm": 19.0,
+      "learning_rate": 4.911495748978932e-07,
+      "loss": 0.8257,
+      "step": 16638
+    },
+    {
+      "epoch": 1.417895185342991,
+      "grad_norm": 14.6875,
+      "learning_rate": 4.91016053907682e-07,
+      "loss": 0.7638,
+      "step": 16639
+    },
+    {
+      "epoch": 1.417980400511291,
+      "grad_norm": 12.375,
+      "learning_rate": 4.908825466330735e-07,
+      "loss": 0.3615,
+      "step": 16640
+    },
+    {
+      "epoch": 1.418065615679591,
+      "grad_norm": 14.25,
+      "learning_rate": 4.907490530764808e-07,
+      "loss": 0.3604,
+      "step": 16641
+    },
+    {
+      "epoch": 1.418150830847891,
+      "grad_norm": 17.5,
+      "learning_rate": 4.906155732403164e-07,
+      "loss": 0.5031,
+      "step": 16642
+    },
+    {
+      "epoch": 1.418236046016191,
+      "grad_norm": 11.125,
+      "learning_rate": 4.90482107126992e-07,
+      "loss": 0.2802,
+      "step": 16643
+    },
+    {
+      "epoch": 1.4183212611844909,
+      "grad_norm": 21.25,
+      "learning_rate": 4.903486547389195e-07,
+      "loss": 0.5741,
+      "step": 16644
+    },
+    {
+      "epoch": 1.4184064763527908,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.902152160785113e-07,
+      "loss": 0.513,
+      "step": 16645
+    },
+    {
+      "epoch": 1.4184916915210908,
+      "grad_norm": 17.625,
+      "learning_rate": 4.900817911481777e-07,
+      "loss": 0.6645,
+      "step": 16646
+    },
+    {
+      "epoch": 1.4185769066893907,
+      "grad_norm": 40.25,
+      "learning_rate": 4.899483799503307e-07,
+      "loss": 0.9492,
+      "step": 16647
+    },
+    {
+      "epoch": 1.4186621218576907,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.898149824873803e-07,
+      "loss": 0.4754,
+      "step": 16648
+    },
+    {
+      "epoch": 1.4187473370259907,
+      "grad_norm": 17.875,
+      "learning_rate": 4.896815987617376e-07,
+      "loss": 0.8589,
+      "step": 16649
+    },
+    {
+      "epoch": 1.4188325521942906,
+      "grad_norm": 19.125,
+      "learning_rate": 4.89548228775813e-07,
+      "loss": 0.8617,
+      "step": 16650
+    },
+    {
+      "epoch": 1.4189177673625906,
+      "grad_norm": 14.125,
+      "learning_rate": 4.89414872532017e-07,
+      "loss": 0.4204,
+      "step": 16651
+    },
+    {
+      "epoch": 1.4190029825308905,
+      "grad_norm": 14.875,
+      "learning_rate": 4.892815300327587e-07,
+      "loss": 0.4971,
+      "step": 16652
+    },
+    {
+      "epoch": 1.4190881976991905,
+      "grad_norm": 14.875,
+      "learning_rate": 4.891482012804481e-07,
+      "loss": 0.4805,
+      "step": 16653
+    },
+    {
+      "epoch": 1.4191734128674904,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.89014886277495e-07,
+      "loss": 0.6677,
+      "step": 16654
+    },
+    {
+      "epoch": 1.4192586280357904,
+      "grad_norm": 12.375,
+      "learning_rate": 4.888815850263078e-07,
+      "loss": 0.2897,
+      "step": 16655
+    },
+    {
+      "epoch": 1.4193438432040903,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.887482975292959e-07,
+      "loss": 0.2136,
+      "step": 16656
+    },
+    {
+      "epoch": 1.4194290583723903,
+      "grad_norm": 18.25,
+      "learning_rate": 4.886150237888676e-07,
+      "loss": 0.6075,
+      "step": 16657
+    },
+    {
+      "epoch": 1.4195142735406903,
+      "grad_norm": 24.875,
+      "learning_rate": 4.884817638074313e-07,
+      "loss": 1.0217,
+      "step": 16658
+    },
+    {
+      "epoch": 1.4195994887089902,
+      "grad_norm": 24.375,
+      "learning_rate": 4.883485175873957e-07,
+      "loss": 0.6955,
+      "step": 16659
+    },
+    {
+      "epoch": 1.4196847038772902,
+      "grad_norm": 21.0,
+      "learning_rate": 4.88215285131168e-07,
+      "loss": 0.6843,
+      "step": 16660
+    },
+    {
+      "epoch": 1.4197699190455901,
+      "grad_norm": 19.75,
+      "learning_rate": 4.88082066441156e-07,
+      "loss": 1.1276,
+      "step": 16661
+    },
+    {
+      "epoch": 1.41985513421389,
+      "grad_norm": 11.0,
+      "learning_rate": 4.879488615197678e-07,
+      "loss": 0.4128,
+      "step": 16662
+    },
+    {
+      "epoch": 1.41994034938219,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.878156703694095e-07,
+      "loss": 0.4682,
+      "step": 16663
+    },
+    {
+      "epoch": 1.42002556455049,
+      "grad_norm": 16.75,
+      "learning_rate": 4.876824929924888e-07,
+      "loss": 0.7233,
+      "step": 16664
+    },
+    {
+      "epoch": 1.42011077971879,
+      "grad_norm": 15.125,
+      "learning_rate": 4.875493293914117e-07,
+      "loss": 0.9003,
+      "step": 16665
+    },
+    {
+      "epoch": 1.42019599488709,
+      "grad_norm": 16.75,
+      "learning_rate": 4.874161795685849e-07,
+      "loss": 0.6798,
+      "step": 16666
+    },
+    {
+      "epoch": 1.4202812100553899,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.872830435264144e-07,
+      "loss": 0.7864,
+      "step": 16667
+    },
+    {
+      "epoch": 1.4203664252236898,
+      "grad_norm": 15.375,
+      "learning_rate": 4.87149921267307e-07,
+      "loss": 0.7661,
+      "step": 16668
+    },
+    {
+      "epoch": 1.4204516403919898,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.87016812793667e-07,
+      "loss": 0.4782,
+      "step": 16669
+    },
+    {
+      "epoch": 1.4205368555602897,
+      "grad_norm": 13.5,
+      "learning_rate": 4.86883718107901e-07,
+      "loss": 0.6022,
+      "step": 16670
+    },
+    {
+      "epoch": 1.4206220707285897,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.86750637212413e-07,
+      "loss": 0.9416,
+      "step": 16671
+    },
+    {
+      "epoch": 1.4207072858968897,
+      "grad_norm": 17.75,
+      "learning_rate": 4.866175701096091e-07,
+      "loss": 0.7532,
+      "step": 16672
+    },
+    {
+      "epoch": 1.4207925010651896,
+      "grad_norm": 27.125,
+      "learning_rate": 4.864845168018926e-07,
+      "loss": 0.8238,
+      "step": 16673
+    },
+    {
+      "epoch": 1.4208777162334896,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.86351477291669e-07,
+      "loss": 0.4847,
+      "step": 16674
+    },
+    {
+      "epoch": 1.4209629314017895,
+      "grad_norm": 11.0625,
+      "learning_rate": 4.86218451581342e-07,
+      "loss": 0.3577,
+      "step": 16675
+    },
+    {
+      "epoch": 1.4210481465700895,
+      "grad_norm": 16.875,
+      "learning_rate": 4.86085439673316e-07,
+      "loss": 0.8356,
+      "step": 16676
+    },
+    {
+      "epoch": 1.4211333617383894,
+      "grad_norm": 16.0,
+      "learning_rate": 4.859524415699938e-07,
+      "loss": 0.5045,
+      "step": 16677
+    },
+    {
+      "epoch": 1.4212185769066894,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.858194572737798e-07,
+      "loss": 0.7373,
+      "step": 16678
+    },
+    {
+      "epoch": 1.4213037920749894,
+      "grad_norm": 15.0,
+      "learning_rate": 4.856864867870763e-07,
+      "loss": 0.6136,
+      "step": 16679
+    },
+    {
+      "epoch": 1.4213890072432893,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.855535301122864e-07,
+      "loss": 0.2688,
+      "step": 16680
+    },
+    {
+      "epoch": 1.4214742224115893,
+      "grad_norm": 13.0,
+      "learning_rate": 4.854205872518134e-07,
+      "loss": 0.402,
+      "step": 16681
+    },
+    {
+      "epoch": 1.4215594375798892,
+      "grad_norm": 15.125,
+      "learning_rate": 4.85287658208059e-07,
+      "loss": 0.6026,
+      "step": 16682
+    },
+    {
+      "epoch": 1.4216446527481892,
+      "grad_norm": 13.125,
+      "learning_rate": 4.851547429834253e-07,
+      "loss": 0.4469,
+      "step": 16683
+    },
+    {
+      "epoch": 1.4217298679164891,
+      "grad_norm": 12.5,
+      "learning_rate": 4.850218415803147e-07,
+      "loss": 0.4459,
+      "step": 16684
+    },
+    {
+      "epoch": 1.421815083084789,
+      "grad_norm": 17.625,
+      "learning_rate": 4.848889540011291e-07,
+      "loss": 0.9097,
+      "step": 16685
+    },
+    {
+      "epoch": 1.421900298253089,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.847560802482694e-07,
+      "loss": 0.4992,
+      "step": 16686
+    },
+    {
+      "epoch": 1.421985513421389,
+      "grad_norm": 22.25,
+      "learning_rate": 4.846232203241364e-07,
+      "loss": 0.8185,
+      "step": 16687
+    },
+    {
+      "epoch": 1.422070728589689,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.844903742311315e-07,
+      "loss": 0.4158,
+      "step": 16688
+    },
+    {
+      "epoch": 1.422155943757989,
+      "grad_norm": 17.75,
+      "learning_rate": 4.843575419716556e-07,
+      "loss": 0.4013,
+      "step": 16689
+    },
+    {
+      "epoch": 1.4222411589262889,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.842247235481086e-07,
+      "loss": 0.7037,
+      "step": 16690
+    },
+    {
+      "epoch": 1.4223263740945888,
+      "grad_norm": 17.375,
+      "learning_rate": 4.840919189628908e-07,
+      "loss": 0.4933,
+      "step": 16691
+    },
+    {
+      "epoch": 1.4224115892628888,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.839591282184023e-07,
+      "loss": 0.5539,
+      "step": 16692
+    },
+    {
+      "epoch": 1.4224968044311888,
+      "grad_norm": 14.75,
+      "learning_rate": 4.83826351317043e-07,
+      "loss": 0.4446,
+      "step": 16693
+    },
+    {
+      "epoch": 1.4225820195994887,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.83693588261212e-07,
+      "loss": 0.4061,
+      "step": 16694
+    },
+    {
+      "epoch": 1.4226672347677887,
+      "grad_norm": 16.375,
+      "learning_rate": 4.83560839053308e-07,
+      "loss": 0.556,
+      "step": 16695
+    },
+    {
+      "epoch": 1.4227524499360886,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.834281036957303e-07,
+      "loss": 0.532,
+      "step": 16696
+    },
+    {
+      "epoch": 1.4228376651043886,
+      "grad_norm": 17.25,
+      "learning_rate": 4.832953821908777e-07,
+      "loss": 0.5811,
+      "step": 16697
+    },
+    {
+      "epoch": 1.4229228802726885,
+      "grad_norm": 18.5,
+      "learning_rate": 4.83162674541149e-07,
+      "loss": 0.634,
+      "step": 16698
+    },
+    {
+      "epoch": 1.4230080954409885,
+      "grad_norm": 16.375,
+      "learning_rate": 4.830299807489413e-07,
+      "loss": 0.4792,
+      "step": 16699
+    },
+    {
+      "epoch": 1.4230933106092885,
+      "grad_norm": 12.75,
+      "learning_rate": 4.828973008166532e-07,
+      "loss": 0.3463,
+      "step": 16700
+    },
+    {
+      "epoch": 1.4231785257775884,
+      "grad_norm": 42.5,
+      "learning_rate": 4.827646347466823e-07,
+      "loss": 0.6968,
+      "step": 16701
+    },
+    {
+      "epoch": 1.4232637409458884,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.826319825414268e-07,
+      "loss": 0.6929,
+      "step": 16702
+    },
+    {
+      "epoch": 1.4233489561141883,
+      "grad_norm": 10.4375,
+      "learning_rate": 4.82499344203282e-07,
+      "loss": 0.3771,
+      "step": 16703
+    },
+    {
+      "epoch": 1.4234341712824883,
+      "grad_norm": 15.125,
+      "learning_rate": 4.82366719734646e-07,
+      "loss": 0.5272,
+      "step": 16704
+    },
+    {
+      "epoch": 1.4235193864507882,
+      "grad_norm": 18.5,
+      "learning_rate": 4.822341091379153e-07,
+      "loss": 0.9869,
+      "step": 16705
+    },
+    {
+      "epoch": 1.4236046016190882,
+      "grad_norm": 26.125,
+      "learning_rate": 4.821015124154867e-07,
+      "loss": 0.7637,
+      "step": 16706
+    },
+    {
+      "epoch": 1.4236898167873882,
+      "grad_norm": 17.0,
+      "learning_rate": 4.819689295697555e-07,
+      "loss": 0.4694,
+      "step": 16707
+    },
+    {
+      "epoch": 1.423775031955688,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.81836360603118e-07,
+      "loss": 0.4784,
+      "step": 16708
+    },
+    {
+      "epoch": 1.423860247123988,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.8170380551797e-07,
+      "loss": 0.6485,
+      "step": 16709
+    },
+    {
+      "epoch": 1.423945462292288,
+      "grad_norm": 18.25,
+      "learning_rate": 4.815712643167071e-07,
+      "loss": 0.5694,
+      "step": 16710
+    },
+    {
+      "epoch": 1.424030677460588,
+      "grad_norm": 19.375,
+      "learning_rate": 4.814387370017244e-07,
+      "loss": 0.8724,
+      "step": 16711
+    },
+    {
+      "epoch": 1.424115892628888,
+      "grad_norm": 17.0,
+      "learning_rate": 4.813062235754159e-07,
+      "loss": 0.4765,
+      "step": 16712
+    },
+    {
+      "epoch": 1.424201107797188,
+      "grad_norm": 14.125,
+      "learning_rate": 4.811737240401771e-07,
+      "loss": 0.8526,
+      "step": 16713
+    },
+    {
+      "epoch": 1.4242863229654878,
+      "grad_norm": 20.625,
+      "learning_rate": 4.810412383984022e-07,
+      "loss": 0.91,
+      "step": 16714
+    },
+    {
+      "epoch": 1.4243715381337878,
+      "grad_norm": 21.75,
+      "learning_rate": 4.809087666524857e-07,
+      "loss": 0.4954,
+      "step": 16715
+    },
+    {
+      "epoch": 1.4244567533020878,
+      "grad_norm": 19.0,
+      "learning_rate": 4.807763088048209e-07,
+      "loss": 0.5744,
+      "step": 16716
+    },
+    {
+      "epoch": 1.4245419684703877,
+      "grad_norm": 24.625,
+      "learning_rate": 4.806438648578016e-07,
+      "loss": 0.8589,
+      "step": 16717
+    },
+    {
+      "epoch": 1.4246271836386877,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.805114348138213e-07,
+      "loss": 0.4083,
+      "step": 16718
+    },
+    {
+      "epoch": 1.4247123988069876,
+      "grad_norm": 18.0,
+      "learning_rate": 4.80379018675274e-07,
+      "loss": 0.886,
+      "step": 16719
+    },
+    {
+      "epoch": 1.4247976139752876,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.80246616444551e-07,
+      "loss": 0.6228,
+      "step": 16720
+    },
+    {
+      "epoch": 1.4248828291435875,
+      "grad_norm": 11.0,
+      "learning_rate": 4.801142281240454e-07,
+      "loss": 0.2413,
+      "step": 16721
+    },
+    {
+      "epoch": 1.4249680443118875,
+      "grad_norm": 17.875,
+      "learning_rate": 4.799818537161501e-07,
+      "loss": 0.9458,
+      "step": 16722
+    },
+    {
+      "epoch": 1.4250532594801875,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.798494932232576e-07,
+      "loss": 0.5363,
+      "step": 16723
+    },
+    {
+      "epoch": 1.4251384746484874,
+      "grad_norm": 20.0,
+      "learning_rate": 4.797171466477586e-07,
+      "loss": 1.0386,
+      "step": 16724
+    },
+    {
+      "epoch": 1.4252236898167874,
+      "grad_norm": 15.875,
+      "learning_rate": 4.795848139920453e-07,
+      "loss": 0.5526,
+      "step": 16725
+    },
+    {
+      "epoch": 1.4253089049850873,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.794524952585091e-07,
+      "loss": 0.8001,
+      "step": 16726
+    },
+    {
+      "epoch": 1.4253941201533873,
+      "grad_norm": 19.625,
+      "learning_rate": 4.793201904495418e-07,
+      "loss": 0.6321,
+      "step": 16727
+    },
+    {
+      "epoch": 1.4254793353216872,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.791878995675335e-07,
+      "loss": 0.7873,
+      "step": 16728
+    },
+    {
+      "epoch": 1.4255645504899872,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.790556226148744e-07,
+      "loss": 0.4344,
+      "step": 16729
+    },
+    {
+      "epoch": 1.4256497656582872,
+      "grad_norm": 16.125,
+      "learning_rate": 4.789233595939556e-07,
+      "loss": 0.5117,
+      "step": 16730
+    },
+    {
+      "epoch": 1.4257349808265871,
+      "grad_norm": 15.875,
+      "learning_rate": 4.787911105071671e-07,
+      "loss": 0.5396,
+      "step": 16731
+    },
+    {
+      "epoch": 1.425820195994887,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.786588753568991e-07,
+      "loss": 0.6739,
+      "step": 16732
+    },
+    {
+      "epoch": 1.425905411163187,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.785266541455405e-07,
+      "loss": 0.6102,
+      "step": 16733
+    },
+    {
+      "epoch": 1.425990626331487,
+      "grad_norm": 25.375,
+      "learning_rate": 4.78394446875481e-07,
+      "loss": 0.7135,
+      "step": 16734
+    },
+    {
+      "epoch": 1.426075841499787,
+      "grad_norm": 10.6875,
+      "learning_rate": 4.782622535491102e-07,
+      "loss": 0.3138,
+      "step": 16735
+    },
+    {
+      "epoch": 1.426161056668087,
+      "grad_norm": 22.25,
+      "learning_rate": 4.781300741688165e-07,
+      "loss": 0.9088,
+      "step": 16736
+    },
+    {
+      "epoch": 1.4262462718363869,
+      "grad_norm": 14.875,
+      "learning_rate": 4.779979087369881e-07,
+      "loss": 0.5635,
+      "step": 16737
+    },
+    {
+      "epoch": 1.4263314870046868,
+      "grad_norm": 21.625,
+      "learning_rate": 4.778657572560137e-07,
+      "loss": 0.7276,
+      "step": 16738
+    },
+    {
+      "epoch": 1.4264167021729868,
+      "grad_norm": 11.875,
+      "learning_rate": 4.777336197282817e-07,
+      "loss": 0.3042,
+      "step": 16739
+    },
+    {
+      "epoch": 1.4265019173412867,
+      "grad_norm": 16.375,
+      "learning_rate": 4.776014961561801e-07,
+      "loss": 0.6561,
+      "step": 16740
+    },
+    {
+      "epoch": 1.4265871325095867,
+      "grad_norm": 14.625,
+      "learning_rate": 4.774693865420957e-07,
+      "loss": 0.3912,
+      "step": 16741
+    },
+    {
+      "epoch": 1.4266723476778866,
+      "grad_norm": 14.375,
+      "learning_rate": 4.773372908884164e-07,
+      "loss": 0.756,
+      "step": 16742
+    },
+    {
+      "epoch": 1.4267575628461866,
+      "grad_norm": 20.125,
+      "learning_rate": 4.772052091975296e-07,
+      "loss": 0.7665,
+      "step": 16743
+    },
+    {
+      "epoch": 1.4268427780144866,
+      "grad_norm": 20.25,
+      "learning_rate": 4.770731414718213e-07,
+      "loss": 0.5692,
+      "step": 16744
+    },
+    {
+      "epoch": 1.4269279931827865,
+      "grad_norm": 13.0,
+      "learning_rate": 4.769410877136792e-07,
+      "loss": 0.453,
+      "step": 16745
+    },
+    {
+      "epoch": 1.4270132083510865,
+      "grad_norm": 15.625,
+      "learning_rate": 4.7680904792548845e-07,
+      "loss": 0.7236,
+      "step": 16746
+    },
+    {
+      "epoch": 1.4270984235193864,
+      "grad_norm": 15.625,
+      "learning_rate": 4.766770221096356e-07,
+      "loss": 0.8154,
+      "step": 16747
+    },
+    {
+      "epoch": 1.4271836386876864,
+      "grad_norm": 17.5,
+      "learning_rate": 4.765450102685068e-07,
+      "loss": 0.5076,
+      "step": 16748
+    },
+    {
+      "epoch": 1.4272688538559863,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.764130124044877e-07,
+      "loss": 0.3557,
+      "step": 16749
+    },
+    {
+      "epoch": 1.4273540690242863,
+      "grad_norm": 15.125,
+      "learning_rate": 4.7628102851996303e-07,
+      "loss": 0.4774,
+      "step": 16750
+    },
+    {
+      "epoch": 1.4274392841925863,
+      "grad_norm": 17.875,
+      "learning_rate": 4.761490586173187e-07,
+      "loss": 0.7747,
+      "step": 16751
+    },
+    {
+      "epoch": 1.4275244993608862,
+      "grad_norm": 22.75,
+      "learning_rate": 4.760171026989385e-07,
+      "loss": 1.089,
+      "step": 16752
+    },
+    {
+      "epoch": 1.4276097145291862,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.7588516076720795e-07,
+      "loss": 0.8067,
+      "step": 16753
+    },
+    {
+      "epoch": 1.4276949296974861,
+      "grad_norm": 14.0,
+      "learning_rate": 4.757532328245107e-07,
+      "loss": 0.6167,
+      "step": 16754
+    },
+    {
+      "epoch": 1.427780144865786,
+      "grad_norm": 17.0,
+      "learning_rate": 4.75621318873231e-07,
+      "loss": 0.682,
+      "step": 16755
+    },
+    {
+      "epoch": 1.427865360034086,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.754894189157527e-07,
+      "loss": 0.6443,
+      "step": 16756
+    },
+    {
+      "epoch": 1.427950575202386,
+      "grad_norm": 19.0,
+      "learning_rate": 4.7535753295446006e-07,
+      "loss": 0.6721,
+      "step": 16757
+    },
+    {
+      "epoch": 1.428035790370686,
+      "grad_norm": 17.0,
+      "learning_rate": 4.752256609917352e-07,
+      "loss": 1.028,
+      "step": 16758
+    },
+    {
+      "epoch": 1.428121005538986,
+      "grad_norm": 15.75,
+      "learning_rate": 4.750938030299622e-07,
+      "loss": 0.7033,
+      "step": 16759
+    },
+    {
+      "epoch": 1.4282062207072859,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.749619590715229e-07,
+      "loss": 0.4882,
+      "step": 16760
+    },
+    {
+      "epoch": 1.4282914358755858,
+      "grad_norm": 15.25,
+      "learning_rate": 4.748301291188005e-07,
+      "loss": 0.6189,
+      "step": 16761
+    },
+    {
+      "epoch": 1.4283766510438858,
+      "grad_norm": 14.625,
+      "learning_rate": 4.746983131741775e-07,
+      "loss": 0.5514,
+      "step": 16762
+    },
+    {
+      "epoch": 1.4284618662121857,
+      "grad_norm": 13.375,
+      "learning_rate": 4.745665112400352e-07,
+      "loss": 0.6223,
+      "step": 16763
+    },
+    {
+      "epoch": 1.4285470813804857,
+      "grad_norm": 19.625,
+      "learning_rate": 4.7443472331875573e-07,
+      "loss": 0.6374,
+      "step": 16764
+    },
+    {
+      "epoch": 1.4286322965487857,
+      "grad_norm": 14.625,
+      "learning_rate": 4.743029494127213e-07,
+      "loss": 0.7396,
+      "step": 16765
+    },
+    {
+      "epoch": 1.4287175117170856,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.74171189524312e-07,
+      "loss": 0.4851,
+      "step": 16766
+    },
+    {
+      "epoch": 1.4288027268853856,
+      "grad_norm": 20.875,
+      "learning_rate": 4.7403944365590993e-07,
+      "loss": 0.6922,
+      "step": 16767
+    },
+    {
+      "epoch": 1.4288879420536855,
+      "grad_norm": 15.875,
+      "learning_rate": 4.7390771180989505e-07,
+      "loss": 0.5071,
+      "step": 16768
+    },
+    {
+      "epoch": 1.4289731572219855,
+      "grad_norm": 14.75,
+      "learning_rate": 4.7377599398864814e-07,
+      "loss": 0.5815,
+      "step": 16769
+    },
+    {
+      "epoch": 1.4290583723902854,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.736442901945501e-07,
+      "loss": 0.64,
+      "step": 16770
+    },
+    {
+      "epoch": 1.4291435875585854,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.735126004299799e-07,
+      "loss": 0.4494,
+      "step": 16771
+    },
+    {
+      "epoch": 1.4292288027268854,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.7338092469731795e-07,
+      "loss": 0.5254,
+      "step": 16772
+    },
+    {
+      "epoch": 1.4293140178951853,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.7324926299894354e-07,
+      "loss": 0.5048,
+      "step": 16773
+    },
+    {
+      "epoch": 1.4293992330634853,
+      "grad_norm": 12.5,
+      "learning_rate": 4.7311761533723645e-07,
+      "loss": 0.374,
+      "step": 16774
+    },
+    {
+      "epoch": 1.4294844482317852,
+      "grad_norm": 15.875,
+      "learning_rate": 4.729859817145753e-07,
+      "loss": 0.2672,
+      "step": 16775
+    },
+    {
+      "epoch": 1.4295696634000852,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.728543621333384e-07,
+      "loss": 0.4818,
+      "step": 16776
+    },
+    {
+      "epoch": 1.4296548785683851,
+      "grad_norm": 21.625,
+      "learning_rate": 4.7272275659590457e-07,
+      "loss": 1.0221,
+      "step": 16777
+    },
+    {
+      "epoch": 1.429740093736685,
+      "grad_norm": 14.9375,
+      "learning_rate": 4.725911651046522e-07,
+      "loss": 0.6897,
+      "step": 16778
+    },
+    {
+      "epoch": 1.429825308904985,
+      "grad_norm": 17.75,
+      "learning_rate": 4.7245958766195976e-07,
+      "loss": 0.6093,
+      "step": 16779
+    },
+    {
+      "epoch": 1.429910524073285,
+      "grad_norm": 16.0,
+      "learning_rate": 4.7232802427020385e-07,
+      "loss": 0.2735,
+      "step": 16780
+    },
+    {
+      "epoch": 1.429995739241585,
+      "grad_norm": 14.5,
+      "learning_rate": 4.721964749317626e-07,
+      "loss": 0.8889,
+      "step": 16781
+    },
+    {
+      "epoch": 1.430080954409885,
+      "grad_norm": 13.5,
+      "learning_rate": 4.720649396490137e-07,
+      "loss": 0.5683,
+      "step": 16782
+    },
+    {
+      "epoch": 1.4301661695781849,
+      "grad_norm": 16.75,
+      "learning_rate": 4.7193341842433347e-07,
+      "loss": 0.6365,
+      "step": 16783
+    },
+    {
+      "epoch": 1.4302513847464848,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.7180191126009843e-07,
+      "loss": 0.4008,
+      "step": 16784
+    },
+    {
+      "epoch": 1.4303365999147848,
+      "grad_norm": 13.25,
+      "learning_rate": 4.716704181586855e-07,
+      "loss": 0.4825,
+      "step": 16785
+    },
+    {
+      "epoch": 1.4304218150830847,
+      "grad_norm": 17.5,
+      "learning_rate": 4.715389391224706e-07,
+      "loss": 0.4019,
+      "step": 16786
+    },
+    {
+      "epoch": 1.4305070302513847,
+      "grad_norm": 13.625,
+      "learning_rate": 4.7140747415383035e-07,
+      "loss": 0.2482,
+      "step": 16787
+    },
+    {
+      "epoch": 1.4305922454196847,
+      "grad_norm": 16.75,
+      "learning_rate": 4.7127602325513955e-07,
+      "loss": 0.6586,
+      "step": 16788
+    },
+    {
+      "epoch": 1.4306774605879846,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.711445864287742e-07,
+      "loss": 0.5883,
+      "step": 16789
+    },
+    {
+      "epoch": 1.4307626757562846,
+      "grad_norm": 17.25,
+      "learning_rate": 4.7101316367710926e-07,
+      "loss": 0.9237,
+      "step": 16790
+    },
+    {
+      "epoch": 1.4308478909245845,
+      "grad_norm": 22.125,
+      "learning_rate": 4.708817550025202e-07,
+      "loss": 0.8415,
+      "step": 16791
+    },
+    {
+      "epoch": 1.4309331060928845,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.707503604073814e-07,
+      "loss": 0.4571,
+      "step": 16792
+    },
+    {
+      "epoch": 1.4310183212611844,
+      "grad_norm": 12.8125,
+      "learning_rate": 4.7061897989406664e-07,
+      "loss": 0.4654,
+      "step": 16793
+    },
+    {
+      "epoch": 1.4311035364294844,
+      "grad_norm": 18.375,
+      "learning_rate": 4.704876134649507e-07,
+      "loss": 0.917,
+      "step": 16794
+    },
+    {
+      "epoch": 1.4311887515977844,
+      "grad_norm": 11.375,
+      "learning_rate": 4.703562611224073e-07,
+      "loss": 0.2416,
+      "step": 16795
+    },
+    {
+      "epoch": 1.4312739667660843,
+      "grad_norm": 16.75,
+      "learning_rate": 4.702249228688109e-07,
+      "loss": 0.6096,
+      "step": 16796
+    },
+    {
+      "epoch": 1.4313591819343843,
+      "grad_norm": 19.0,
+      "learning_rate": 4.700935987065337e-07,
+      "loss": 1.0781,
+      "step": 16797
+    },
+    {
+      "epoch": 1.4314443971026842,
+      "grad_norm": 11.4375,
+      "learning_rate": 4.699622886379494e-07,
+      "loss": 0.4531,
+      "step": 16798
+    },
+    {
+      "epoch": 1.4315296122709842,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.6983099266543154e-07,
+      "loss": 0.3972,
+      "step": 16799
+    },
+    {
+      "epoch": 1.4316148274392841,
+      "grad_norm": 12.625,
+      "learning_rate": 4.69699710791352e-07,
+      "loss": 0.6589,
+      "step": 16800
+    },
+    {
+      "epoch": 1.431700042607584,
+      "grad_norm": 18.375,
+      "learning_rate": 4.69568443018083e-07,
+      "loss": 0.6045,
+      "step": 16801
+    },
+    {
+      "epoch": 1.431785257775884,
+      "grad_norm": 14.9375,
+      "learning_rate": 4.6943718934799694e-07,
+      "loss": 0.4343,
+      "step": 16802
+    },
+    {
+      "epoch": 1.431870472944184,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.693059497834658e-07,
+      "loss": 0.9234,
+      "step": 16803
+    },
+    {
+      "epoch": 1.431955688112484,
+      "grad_norm": 19.125,
+      "learning_rate": 4.691747243268617e-07,
+      "loss": 0.7311,
+      "step": 16804
+    },
+    {
+      "epoch": 1.432040903280784,
+      "grad_norm": 12.6875,
+      "learning_rate": 4.69043512980555e-07,
+      "loss": 0.4026,
+      "step": 16805
+    },
+    {
+      "epoch": 1.4321261184490839,
+      "grad_norm": 20.125,
+      "learning_rate": 4.689123157469175e-07,
+      "loss": 0.509,
+      "step": 16806
+    },
+    {
+      "epoch": 1.4322113336173838,
+      "grad_norm": 15.5,
+      "learning_rate": 4.687811326283197e-07,
+      "loss": 0.6349,
+      "step": 16807
+    },
+    {
+      "epoch": 1.4322965487856838,
+      "grad_norm": 16.25,
+      "learning_rate": 4.6864996362713285e-07,
+      "loss": 0.358,
+      "step": 16808
+    },
+    {
+      "epoch": 1.4323817639539838,
+      "grad_norm": 17.625,
+      "learning_rate": 4.6851880874572694e-07,
+      "loss": 0.7008,
+      "step": 16809
+    },
+    {
+      "epoch": 1.4324669791222837,
+      "grad_norm": 19.375,
+      "learning_rate": 4.6838766798647155e-07,
+      "loss": 0.7913,
+      "step": 16810
+    },
+    {
+      "epoch": 1.4325521942905837,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.6825654135173703e-07,
+      "loss": 0.5885,
+      "step": 16811
+    },
+    {
+      "epoch": 1.4326374094588836,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.681254288438933e-07,
+      "loss": 0.4588,
+      "step": 16812
+    },
+    {
+      "epoch": 1.4327226246271836,
+      "grad_norm": 11.75,
+      "learning_rate": 4.679943304653088e-07,
+      "loss": 0.4323,
+      "step": 16813
+    },
+    {
+      "epoch": 1.4328078397954835,
+      "grad_norm": 17.0,
+      "learning_rate": 4.6786324621835316e-07,
+      "loss": 0.7268,
+      "step": 16814
+    },
+    {
+      "epoch": 1.4328930549637835,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.6773217610539517e-07,
+      "loss": 0.4369,
+      "step": 16815
+    },
+    {
+      "epoch": 1.4329782701320835,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.6760112012880387e-07,
+      "loss": 0.3471,
+      "step": 16816
+    },
+    {
+      "epoch": 1.4330634853003834,
+      "grad_norm": 14.9375,
+      "learning_rate": 4.674700782909471e-07,
+      "loss": 0.5582,
+      "step": 16817
+    },
+    {
+      "epoch": 1.4331487004686834,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.673390505941923e-07,
+      "loss": 0.5054,
+      "step": 16818
+    },
+    {
+      "epoch": 1.4332339156369833,
+      "grad_norm": 15.375,
+      "learning_rate": 4.672080370409081e-07,
+      "loss": 0.4612,
+      "step": 16819
+    },
+    {
+      "epoch": 1.4333191308052833,
+      "grad_norm": 13.5,
+      "learning_rate": 4.670770376334617e-07,
+      "loss": 0.5002,
+      "step": 16820
+    },
+    {
+      "epoch": 1.4334043459735832,
+      "grad_norm": 16.875,
+      "learning_rate": 4.66946052374221e-07,
+      "loss": 0.4621,
+      "step": 16821
+    },
+    {
+      "epoch": 1.4334895611418832,
+      "grad_norm": 13.0,
+      "learning_rate": 4.668150812655521e-07,
+      "loss": 0.327,
+      "step": 16822
+    },
+    {
+      "epoch": 1.4335747763101832,
+      "grad_norm": 14.125,
+      "learning_rate": 4.666841243098222e-07,
+      "loss": 0.5494,
+      "step": 16823
+    },
+    {
+      "epoch": 1.433659991478483,
+      "grad_norm": 20.0,
+      "learning_rate": 4.6655318150939837e-07,
+      "loss": 0.833,
+      "step": 16824
+    },
+    {
+      "epoch": 1.433745206646783,
+      "grad_norm": 18.625,
+      "learning_rate": 4.664222528666459e-07,
+      "loss": 0.7251,
+      "step": 16825
+    },
+    {
+      "epoch": 1.433830421815083,
+      "grad_norm": 17.0,
+      "learning_rate": 4.6629133838393167e-07,
+      "loss": 0.6115,
+      "step": 16826
+    },
+    {
+      "epoch": 1.433915636983383,
+      "grad_norm": 11.875,
+      "learning_rate": 4.6616043806362074e-07,
+      "loss": 0.3541,
+      "step": 16827
+    },
+    {
+      "epoch": 1.434000852151683,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.6602955190807873e-07,
+      "loss": 0.3607,
+      "step": 16828
+    },
+    {
+      "epoch": 1.434086067319983,
+      "grad_norm": 15.0,
+      "learning_rate": 4.6589867991967156e-07,
+      "loss": 0.6337,
+      "step": 16829
+    },
+    {
+      "epoch": 1.4341712824882829,
+      "grad_norm": 22.5,
+      "learning_rate": 4.657678221007633e-07,
+      "loss": 0.6134,
+      "step": 16830
+    },
+    {
+      "epoch": 1.4342564976565828,
+      "grad_norm": 17.5,
+      "learning_rate": 4.6563697845371916e-07,
+      "loss": 0.753,
+      "step": 16831
+    },
+    {
+      "epoch": 1.4343417128248828,
+      "grad_norm": 16.25,
+      "learning_rate": 4.6550614898090385e-07,
+      "loss": 0.5584,
+      "step": 16832
+    },
+    {
+      "epoch": 1.4344269279931827,
+      "grad_norm": 24.5,
+      "learning_rate": 4.65375333684681e-07,
+      "loss": 0.6933,
+      "step": 16833
+    },
+    {
+      "epoch": 1.4345121431614827,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.652445325674151e-07,
+      "loss": 0.2988,
+      "step": 16834
+    },
+    {
+      "epoch": 1.4345973583297826,
+      "grad_norm": 15.875,
+      "learning_rate": 4.6511374563146946e-07,
+      "loss": 0.8899,
+      "step": 16835
+    },
+    {
+      "epoch": 1.4346825734980826,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.6498297287920755e-07,
+      "loss": 0.629,
+      "step": 16836
+    },
+    {
+      "epoch": 1.4347677886663825,
+      "grad_norm": 12.1875,
+      "learning_rate": 4.648522143129927e-07,
+      "loss": 0.4961,
+      "step": 16837
+    },
+    {
+      "epoch": 1.4348530038346825,
+      "grad_norm": 14.125,
+      "learning_rate": 4.647214699351883e-07,
+      "loss": 0.5789,
+      "step": 16838
+    },
+    {
+      "epoch": 1.4349382190029825,
+      "grad_norm": 17.125,
+      "learning_rate": 4.6459073974815624e-07,
+      "loss": 0.8912,
+      "step": 16839
+    },
+    {
+      "epoch": 1.4350234341712824,
+      "grad_norm": 16.0,
+      "learning_rate": 4.644600237542597e-07,
+      "loss": 0.8127,
+      "step": 16840
+    },
+    {
+      "epoch": 1.4351086493395824,
+      "grad_norm": 11.125,
+      "learning_rate": 4.6432932195586006e-07,
+      "loss": 0.3616,
+      "step": 16841
+    },
+    {
+      "epoch": 1.4351938645078823,
+      "grad_norm": 13.5,
+      "learning_rate": 4.641986343553201e-07,
+      "loss": 0.4179,
+      "step": 16842
+    },
+    {
+      "epoch": 1.4352790796761823,
+      "grad_norm": 17.625,
+      "learning_rate": 4.6406796095500044e-07,
+      "loss": 0.5292,
+      "step": 16843
+    },
+    {
+      "epoch": 1.4353642948444822,
+      "grad_norm": 17.25,
+      "learning_rate": 4.639373017572631e-07,
+      "loss": 0.7788,
+      "step": 16844
+    },
+    {
+      "epoch": 1.4354495100127822,
+      "grad_norm": 13.625,
+      "learning_rate": 4.6380665676446917e-07,
+      "loss": 0.6699,
+      "step": 16845
+    },
+    {
+      "epoch": 1.4355347251810822,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.6367602597897994e-07,
+      "loss": 0.5409,
+      "step": 16846
+    },
+    {
+      "epoch": 1.4356199403493821,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.635454094031551e-07,
+      "loss": 0.5046,
+      "step": 16847
+    },
+    {
+      "epoch": 1.435705155517682,
+      "grad_norm": 14.0,
+      "learning_rate": 4.6341480703935613e-07,
+      "loss": 0.4164,
+      "step": 16848
+    },
+    {
+      "epoch": 1.435790370685982,
+      "grad_norm": 18.5,
+      "learning_rate": 4.632842188899421e-07,
+      "loss": 0.9258,
+      "step": 16849
+    },
+    {
+      "epoch": 1.435875585854282,
+      "grad_norm": 17.0,
+      "learning_rate": 4.631536449572734e-07,
+      "loss": 0.4308,
+      "step": 16850
+    },
+    {
+      "epoch": 1.435960801022582,
+      "grad_norm": 11.75,
+      "learning_rate": 4.630230852437098e-07,
+      "loss": 0.3904,
+      "step": 16851
+    },
+    {
+      "epoch": 1.436046016190882,
+      "grad_norm": 14.75,
+      "learning_rate": 4.628925397516101e-07,
+      "loss": 0.7777,
+      "step": 16852
+    },
+    {
+      "epoch": 1.4361312313591819,
+      "grad_norm": 12.6875,
+      "learning_rate": 4.627620084833338e-07,
+      "loss": 0.4798,
+      "step": 16853
+    },
+    {
+      "epoch": 1.4362164465274818,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.6263149144123954e-07,
+      "loss": 0.401,
+      "step": 16854
+    },
+    {
+      "epoch": 1.4363016616957818,
+      "grad_norm": 15.625,
+      "learning_rate": 4.6250098862768657e-07,
+      "loss": 0.9116,
+      "step": 16855
+    },
+    {
+      "epoch": 1.4363868768640817,
+      "grad_norm": 15.75,
+      "learning_rate": 4.6237050004503197e-07,
+      "loss": 0.6061,
+      "step": 16856
+    },
+    {
+      "epoch": 1.4364720920323817,
+      "grad_norm": 16.125,
+      "learning_rate": 4.62240025695635e-07,
+      "loss": 0.5902,
+      "step": 16857
+    },
+    {
+      "epoch": 1.4365573072006816,
+      "grad_norm": 18.125,
+      "learning_rate": 4.621095655818525e-07,
+      "loss": 0.7063,
+      "step": 16858
+    },
+    {
+      "epoch": 1.4366425223689816,
+      "grad_norm": 16.75,
+      "learning_rate": 4.619791197060429e-07,
+      "loss": 0.6096,
+      "step": 16859
+    },
+    {
+      "epoch": 1.4367277375372816,
+      "grad_norm": 16.125,
+      "learning_rate": 4.618486880705625e-07,
+      "loss": 0.7369,
+      "step": 16860
+    },
+    {
+      "epoch": 1.4368129527055815,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.617182706777688e-07,
+      "loss": 0.5422,
+      "step": 16861
+    },
+    {
+      "epoch": 1.4368981678738815,
+      "grad_norm": 23.75,
+      "learning_rate": 4.615878675300188e-07,
+      "loss": 0.6886,
+      "step": 16862
+    },
+    {
+      "epoch": 1.4369833830421814,
+      "grad_norm": 58.5,
+      "learning_rate": 4.614574786296691e-07,
+      "loss": 0.9406,
+      "step": 16863
+    },
+    {
+      "epoch": 1.4370685982104814,
+      "grad_norm": 21.5,
+      "learning_rate": 4.6132710397907537e-07,
+      "loss": 0.8374,
+      "step": 16864
+    },
+    {
+      "epoch": 1.4371538133787813,
+      "grad_norm": 14.0,
+      "learning_rate": 4.6119674358059446e-07,
+      "loss": 0.4439,
+      "step": 16865
+    },
+    {
+      "epoch": 1.4372390285470813,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.61066397436581e-07,
+      "loss": 0.5845,
+      "step": 16866
+    },
+    {
+      "epoch": 1.4373242437153813,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.609360655493912e-07,
+      "loss": 0.6767,
+      "step": 16867
+    },
+    {
+      "epoch": 1.4374094588836812,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.608057479213805e-07,
+      "loss": 0.6349,
+      "step": 16868
+    },
+    {
+      "epoch": 1.4374946740519812,
+      "grad_norm": 23.375,
+      "learning_rate": 4.6067544455490315e-07,
+      "loss": 0.8008,
+      "step": 16869
+    },
+    {
+      "epoch": 1.4375798892202813,
+      "grad_norm": 20.375,
+      "learning_rate": 4.6054515545231425e-07,
+      "loss": 0.9809,
+      "step": 16870
+    },
+    {
+      "epoch": 1.4376651043885813,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.604148806159682e-07,
+      "loss": 0.6414,
+      "step": 16871
+    },
+    {
+      "epoch": 1.4377503195568813,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.602846200482197e-07,
+      "loss": 0.3736,
+      "step": 16872
+    },
+    {
+      "epoch": 1.4378355347251812,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.601543737514222e-07,
+      "loss": 0.351,
+      "step": 16873
+    },
+    {
+      "epoch": 1.4379207498934812,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.6002414172792916e-07,
+      "loss": 0.6214,
+      "step": 16874
+    },
+    {
+      "epoch": 1.4380059650617811,
+      "grad_norm": 16.875,
+      "learning_rate": 4.598939239800941e-07,
+      "loss": 0.7619,
+      "step": 16875
+    },
+    {
+      "epoch": 1.438091180230081,
+      "grad_norm": 18.375,
+      "learning_rate": 4.597637205102708e-07,
+      "loss": 0.5886,
+      "step": 16876
+    },
+    {
+      "epoch": 1.438176395398381,
+      "grad_norm": 16.125,
+      "learning_rate": 4.596335313208113e-07,
+      "loss": 0.7868,
+      "step": 16877
+    },
+    {
+      "epoch": 1.438261610566681,
+      "grad_norm": 31.125,
+      "learning_rate": 4.5950335641406857e-07,
+      "loss": 0.9351,
+      "step": 16878
+    },
+    {
+      "epoch": 1.438346825734981,
+      "grad_norm": 15.0,
+      "learning_rate": 4.5937319579239526e-07,
+      "loss": 0.6491,
+      "step": 16879
+    },
+    {
+      "epoch": 1.438432040903281,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.5924304945814363e-07,
+      "loss": 0.4664,
+      "step": 16880
+    },
+    {
+      "epoch": 1.4385172560715809,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.5911291741366535e-07,
+      "loss": 0.4907,
+      "step": 16881
+    },
+    {
+      "epoch": 1.4386024712398808,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.5898279966131154e-07,
+      "loss": 0.3438,
+      "step": 16882
+    },
+    {
+      "epoch": 1.4386876864081808,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.5885269620343386e-07,
+      "loss": 0.3246,
+      "step": 16883
+    },
+    {
+      "epoch": 1.4387729015764807,
+      "grad_norm": 13.25,
+      "learning_rate": 4.587226070423836e-07,
+      "loss": 0.406,
+      "step": 16884
+    },
+    {
+      "epoch": 1.4388581167447807,
+      "grad_norm": 17.5,
+      "learning_rate": 4.5859253218051194e-07,
+      "loss": 0.7256,
+      "step": 16885
+    },
+    {
+      "epoch": 1.4389433319130807,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.5846247162016857e-07,
+      "loss": 0.44,
+      "step": 16886
+    },
+    {
+      "epoch": 1.4390285470813806,
+      "grad_norm": 29.875,
+      "learning_rate": 4.5833242536370436e-07,
+      "loss": 0.7225,
+      "step": 16887
+    },
+    {
+      "epoch": 1.4391137622496806,
+      "grad_norm": 78.0,
+      "learning_rate": 4.5820239341346925e-07,
+      "loss": 0.4877,
+      "step": 16888
+    },
+    {
+      "epoch": 1.4391989774179805,
+      "grad_norm": 14.25,
+      "learning_rate": 4.5807237577181405e-07,
+      "loss": 0.4459,
+      "step": 16889
+    },
+    {
+      "epoch": 1.4392841925862805,
+      "grad_norm": 16.25,
+      "learning_rate": 4.5794237244108644e-07,
+      "loss": 0.6603,
+      "step": 16890
+    },
+    {
+      "epoch": 1.4393694077545804,
+      "grad_norm": 15.625,
+      "learning_rate": 4.5781238342363655e-07,
+      "loss": 0.5501,
+      "step": 16891
+    },
+    {
+      "epoch": 1.4394546229228804,
+      "grad_norm": 20.0,
+      "learning_rate": 4.576824087218136e-07,
+      "loss": 0.9311,
+      "step": 16892
+    },
+    {
+      "epoch": 1.4395398380911804,
+      "grad_norm": 13.75,
+      "learning_rate": 4.5755244833796673e-07,
+      "loss": 0.531,
+      "step": 16893
+    },
+    {
+      "epoch": 1.4396250532594803,
+      "grad_norm": 14.0,
+      "learning_rate": 4.5742250227444354e-07,
+      "loss": 0.6114,
+      "step": 16894
+    },
+    {
+      "epoch": 1.4397102684277803,
+      "grad_norm": 20.25,
+      "learning_rate": 4.572925705335926e-07,
+      "loss": 0.6188,
+      "step": 16895
+    },
+    {
+      "epoch": 1.4397954835960802,
+      "grad_norm": 18.0,
+      "learning_rate": 4.5716265311776224e-07,
+      "loss": 0.812,
+      "step": 16896
+    },
+    {
+      "epoch": 1.4398806987643802,
+      "grad_norm": 14.75,
+      "learning_rate": 4.570327500293002e-07,
+      "loss": 0.4128,
+      "step": 16897
+    },
+    {
+      "epoch": 1.4399659139326801,
+      "grad_norm": 16.125,
+      "learning_rate": 4.5690286127055394e-07,
+      "loss": 0.588,
+      "step": 16898
+    },
+    {
+      "epoch": 1.44005112910098,
+      "grad_norm": 17.75,
+      "learning_rate": 4.5677298684387e-07,
+      "loss": 0.7883,
+      "step": 16899
+    },
+    {
+      "epoch": 1.44013634426928,
+      "grad_norm": 17.0,
+      "learning_rate": 4.5664312675159584e-07,
+      "loss": 0.576,
+      "step": 16900
+    },
+    {
+      "epoch": 1.44022155943758,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.565132809960782e-07,
+      "loss": 0.2517,
+      "step": 16901
+    },
+    {
+      "epoch": 1.44030677460588,
+      "grad_norm": 11.75,
+      "learning_rate": 4.5638344957966385e-07,
+      "loss": 0.4348,
+      "step": 16902
+    },
+    {
+      "epoch": 1.44039198977418,
+      "grad_norm": 17.625,
+      "learning_rate": 4.562536325046983e-07,
+      "loss": 0.4361,
+      "step": 16903
+    },
+    {
+      "epoch": 1.4404772049424799,
+      "grad_norm": 17.125,
+      "learning_rate": 4.5612382977352764e-07,
+      "loss": 0.4496,
+      "step": 16904
+    },
+    {
+      "epoch": 1.4405624201107798,
+      "grad_norm": 11.9375,
+      "learning_rate": 4.559940413884982e-07,
+      "loss": 0.4086,
+      "step": 16905
+    },
+    {
+      "epoch": 1.4406476352790798,
+      "grad_norm": 18.5,
+      "learning_rate": 4.558642673519549e-07,
+      "loss": 0.4921,
+      "step": 16906
+    },
+    {
+      "epoch": 1.4407328504473798,
+      "grad_norm": 13.75,
+      "learning_rate": 4.557345076662423e-07,
+      "loss": 0.4418,
+      "step": 16907
+    },
+    {
+      "epoch": 1.4408180656156797,
+      "grad_norm": 16.625,
+      "learning_rate": 4.5560476233370593e-07,
+      "loss": 0.3985,
+      "step": 16908
+    },
+    {
+      "epoch": 1.4409032807839797,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.5547503135669026e-07,
+      "loss": 0.7148,
+      "step": 16909
+    },
+    {
+      "epoch": 1.4409884959522796,
+      "grad_norm": 14.875,
+      "learning_rate": 4.553453147375403e-07,
+      "loss": 0.7468,
+      "step": 16910
+    },
+    {
+      "epoch": 1.4410737111205796,
+      "grad_norm": 17.5,
+      "learning_rate": 4.5521561247859907e-07,
+      "loss": 0.566,
+      "step": 16911
+    },
+    {
+      "epoch": 1.4411589262888795,
+      "grad_norm": 14.5,
+      "learning_rate": 4.550859245822109e-07,
+      "loss": 0.6,
+      "step": 16912
+    },
+    {
+      "epoch": 1.4412441414571795,
+      "grad_norm": 22.75,
+      "learning_rate": 4.549562510507198e-07,
+      "loss": 0.7223,
+      "step": 16913
+    },
+    {
+      "epoch": 1.4413293566254795,
+      "grad_norm": 24.0,
+      "learning_rate": 4.5482659188646844e-07,
+      "loss": 0.7964,
+      "step": 16914
+    },
+    {
+      "epoch": 1.4414145717937794,
+      "grad_norm": 18.375,
+      "learning_rate": 4.546969470918006e-07,
+      "loss": 0.7927,
+      "step": 16915
+    },
+    {
+      "epoch": 1.4414997869620794,
+      "grad_norm": 21.625,
+      "learning_rate": 4.545673166690581e-07,
+      "loss": 0.9297,
+      "step": 16916
+    },
+    {
+      "epoch": 1.4415850021303793,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.544377006205841e-07,
+      "loss": 0.7559,
+      "step": 16917
+    },
+    {
+      "epoch": 1.4416702172986793,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.54308098948721e-07,
+      "loss": 0.585,
+      "step": 16918
+    },
+    {
+      "epoch": 1.4417554324669792,
+      "grad_norm": 17.5,
+      "learning_rate": 4.5417851165581087e-07,
+      "loss": 0.9007,
+      "step": 16919
+    },
+    {
+      "epoch": 1.4418406476352792,
+      "grad_norm": 11.5,
+      "learning_rate": 4.54048938744195e-07,
+      "loss": 0.5278,
+      "step": 16920
+    },
+    {
+      "epoch": 1.4419258628035792,
+      "grad_norm": 20.875,
+      "learning_rate": 4.539193802162152e-07,
+      "loss": 0.5611,
+      "step": 16921
+    },
+    {
+      "epoch": 1.442011077971879,
+      "grad_norm": 14.25,
+      "learning_rate": 4.5378983607421324e-07,
+      "loss": 0.4949,
+      "step": 16922
+    },
+    {
+      "epoch": 1.442096293140179,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.5366030632052944e-07,
+      "loss": 0.4909,
+      "step": 16923
+    },
+    {
+      "epoch": 1.442181508308479,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.535307909575043e-07,
+      "loss": 0.5904,
+      "step": 16924
+    },
+    {
+      "epoch": 1.442266723476779,
+      "grad_norm": 16.25,
+      "learning_rate": 4.5340128998747873e-07,
+      "loss": 0.8062,
+      "step": 16925
+    },
+    {
+      "epoch": 1.442351938645079,
+      "grad_norm": 20.25,
+      "learning_rate": 4.5327180341279284e-07,
+      "loss": 0.6103,
+      "step": 16926
+    },
+    {
+      "epoch": 1.442437153813379,
+      "grad_norm": 14.375,
+      "learning_rate": 4.53142331235787e-07,
+      "loss": 0.5257,
+      "step": 16927
+    },
+    {
+      "epoch": 1.4425223689816788,
+      "grad_norm": 22.125,
+      "learning_rate": 4.5301287345880016e-07,
+      "loss": 0.8956,
+      "step": 16928
+    },
+    {
+      "epoch": 1.4426075841499788,
+      "grad_norm": 17.375,
+      "learning_rate": 4.528834300841722e-07,
+      "loss": 0.6664,
+      "step": 16929
+    },
+    {
+      "epoch": 1.4426927993182788,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.5275400111424255e-07,
+      "loss": 0.4301,
+      "step": 16930
+    },
+    {
+      "epoch": 1.4427780144865787,
+      "grad_norm": 16.875,
+      "learning_rate": 4.526245865513494e-07,
+      "loss": 0.7455,
+      "step": 16931
+    },
+    {
+      "epoch": 1.4428632296548787,
+      "grad_norm": 15.25,
+      "learning_rate": 4.5249518639783233e-07,
+      "loss": 0.7453,
+      "step": 16932
+    },
+    {
+      "epoch": 1.4429484448231786,
+      "grad_norm": 16.875,
+      "learning_rate": 4.523658006560286e-07,
+      "loss": 0.6358,
+      "step": 16933
+    },
+    {
+      "epoch": 1.4430336599914786,
+      "grad_norm": 14.875,
+      "learning_rate": 4.5223642932827704e-07,
+      "loss": 0.6633,
+      "step": 16934
+    },
+    {
+      "epoch": 1.4431188751597785,
+      "grad_norm": 22.0,
+      "learning_rate": 4.5210707241691577e-07,
+      "loss": 0.846,
+      "step": 16935
+    },
+    {
+      "epoch": 1.4432040903280785,
+      "grad_norm": 26.5,
+      "learning_rate": 4.519777299242817e-07,
+      "loss": 0.8921,
+      "step": 16936
+    },
+    {
+      "epoch": 1.4432893054963785,
+      "grad_norm": 14.0,
+      "learning_rate": 4.518484018527125e-07,
+      "loss": 0.6387,
+      "step": 16937
+    },
+    {
+      "epoch": 1.4433745206646784,
+      "grad_norm": 20.5,
+      "learning_rate": 4.5171908820454564e-07,
+      "loss": 0.8986,
+      "step": 16938
+    },
+    {
+      "epoch": 1.4434597358329784,
+      "grad_norm": 16.5,
+      "learning_rate": 4.515897889821172e-07,
+      "loss": 0.7503,
+      "step": 16939
+    },
+    {
+      "epoch": 1.4435449510012783,
+      "grad_norm": 11.5625,
+      "learning_rate": 4.514605041877647e-07,
+      "loss": 0.3722,
+      "step": 16940
+    },
+    {
+      "epoch": 1.4436301661695783,
+      "grad_norm": 18.5,
+      "learning_rate": 4.513312338238232e-07,
+      "loss": 0.4259,
+      "step": 16941
+    },
+    {
+      "epoch": 1.4437153813378782,
+      "grad_norm": 19.25,
+      "learning_rate": 4.5120197789262967e-07,
+      "loss": 0.8735,
+      "step": 16942
+    },
+    {
+      "epoch": 1.4438005965061782,
+      "grad_norm": 12.25,
+      "learning_rate": 4.5107273639651947e-07,
+      "loss": 0.3001,
+      "step": 16943
+    },
+    {
+      "epoch": 1.4438858116744782,
+      "grad_norm": 12.375,
+      "learning_rate": 4.509435093378288e-07,
+      "loss": 0.4209,
+      "step": 16944
+    },
+    {
+      "epoch": 1.4439710268427781,
+      "grad_norm": 19.25,
+      "learning_rate": 4.508142967188921e-07,
+      "loss": 0.8443,
+      "step": 16945
+    },
+    {
+      "epoch": 1.444056242011078,
+      "grad_norm": 11.5625,
+      "learning_rate": 4.5068509854204494e-07,
+      "loss": 0.4116,
+      "step": 16946
+    },
+    {
+      "epoch": 1.444141457179378,
+      "grad_norm": 15.0,
+      "learning_rate": 4.5055591480962155e-07,
+      "loss": 0.5546,
+      "step": 16947
+    },
+    {
+      "epoch": 1.444226672347678,
+      "grad_norm": 15.875,
+      "learning_rate": 4.504267455239567e-07,
+      "loss": 0.5437,
+      "step": 16948
+    },
+    {
+      "epoch": 1.444311887515978,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.502975906873848e-07,
+      "loss": 0.673,
+      "step": 16949
+    },
+    {
+      "epoch": 1.444397102684278,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.5016845030223936e-07,
+      "loss": 0.6639,
+      "step": 16950
+    },
+    {
+      "epoch": 1.4444823178525779,
+      "grad_norm": 32.5,
+      "learning_rate": 4.500393243708542e-07,
+      "loss": 0.9389,
+      "step": 16951
+    },
+    {
+      "epoch": 1.4445675330208778,
+      "grad_norm": 17.875,
+      "learning_rate": 4.4991021289556345e-07,
+      "loss": 0.7824,
+      "step": 16952
+    },
+    {
+      "epoch": 1.4446527481891778,
+      "grad_norm": 16.25,
+      "learning_rate": 4.4978111587869924e-07,
+      "loss": 0.6268,
+      "step": 16953
+    },
+    {
+      "epoch": 1.4447379633574777,
+      "grad_norm": 11.9375,
+      "learning_rate": 4.496520333225954e-07,
+      "loss": 0.2983,
+      "step": 16954
+    },
+    {
+      "epoch": 1.4448231785257777,
+      "grad_norm": 15.25,
+      "learning_rate": 4.495229652295835e-07,
+      "loss": 0.5117,
+      "step": 16955
+    },
+    {
+      "epoch": 1.4449083936940776,
+      "grad_norm": 14.25,
+      "learning_rate": 4.493939116019967e-07,
+      "loss": 0.5626,
+      "step": 16956
+    },
+    {
+      "epoch": 1.4449936088623776,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.492648724421675e-07,
+      "loss": 0.923,
+      "step": 16957
+    },
+    {
+      "epoch": 1.4450788240306776,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.491358477524267e-07,
+      "loss": 0.5398,
+      "step": 16958
+    },
+    {
+      "epoch": 1.4451640391989775,
+      "grad_norm": 12.75,
+      "learning_rate": 4.490068375351066e-07,
+      "loss": 0.2582,
+      "step": 16959
+    },
+    {
+      "epoch": 1.4452492543672775,
+      "grad_norm": 26.375,
+      "learning_rate": 4.488778417925382e-07,
+      "loss": 1.026,
+      "step": 16960
+    },
+    {
+      "epoch": 1.4453344695355774,
+      "grad_norm": 11.75,
+      "learning_rate": 4.487488605270533e-07,
+      "loss": 0.4216,
+      "step": 16961
+    },
+    {
+      "epoch": 1.4454196847038774,
+      "grad_norm": 14.125,
+      "learning_rate": 4.4861989374098234e-07,
+      "loss": 0.5933,
+      "step": 16962
+    },
+    {
+      "epoch": 1.4455048998721773,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.484909414366552e-07,
+      "loss": 0.4097,
+      "step": 16963
+    },
+    {
+      "epoch": 1.4455901150404773,
+      "grad_norm": 17.5,
+      "learning_rate": 4.483620036164027e-07,
+      "loss": 0.6403,
+      "step": 16964
+    },
+    {
+      "epoch": 1.4456753302087773,
+      "grad_norm": 14.625,
+      "learning_rate": 4.4823308028255546e-07,
+      "loss": 0.5542,
+      "step": 16965
+    },
+    {
+      "epoch": 1.4457605453770772,
+      "grad_norm": 20.875,
+      "learning_rate": 4.481041714374422e-07,
+      "loss": 0.7441,
+      "step": 16966
+    },
+    {
+      "epoch": 1.4458457605453772,
+      "grad_norm": 11.625,
+      "learning_rate": 4.479752770833928e-07,
+      "loss": 0.4468,
+      "step": 16967
+    },
+    {
+      "epoch": 1.4459309757136771,
+      "grad_norm": 17.375,
+      "learning_rate": 4.478463972227369e-07,
+      "loss": 0.7209,
+      "step": 16968
+    },
+    {
+      "epoch": 1.446016190881977,
+      "grad_norm": 15.625,
+      "learning_rate": 4.477175318578035e-07,
+      "loss": 0.63,
+      "step": 16969
+    },
+    {
+      "epoch": 1.446101406050277,
+      "grad_norm": 17.375,
+      "learning_rate": 4.4758868099092106e-07,
+      "loss": 0.8403,
+      "step": 16970
+    },
+    {
+      "epoch": 1.446186621218577,
+      "grad_norm": 18.875,
+      "learning_rate": 4.474598446244176e-07,
+      "loss": 0.8098,
+      "step": 16971
+    },
+    {
+      "epoch": 1.446271836386877,
+      "grad_norm": 17.0,
+      "learning_rate": 4.473310227606219e-07,
+      "loss": 0.7286,
+      "step": 16972
+    },
+    {
+      "epoch": 1.446357051555177,
+      "grad_norm": 15.625,
+      "learning_rate": 4.472022154018615e-07,
+      "loss": 0.6459,
+      "step": 16973
+    },
+    {
+      "epoch": 1.4464422667234769,
+      "grad_norm": 16.875,
+      "learning_rate": 4.4707342255046497e-07,
+      "loss": 0.6156,
+      "step": 16974
+    },
+    {
+      "epoch": 1.4465274818917768,
+      "grad_norm": 15.5,
+      "learning_rate": 4.469446442087587e-07,
+      "loss": 0.4726,
+      "step": 16975
+    },
+    {
+      "epoch": 1.4466126970600768,
+      "grad_norm": 19.25,
+      "learning_rate": 4.4681588037907023e-07,
+      "loss": 0.7002,
+      "step": 16976
+    },
+    {
+      "epoch": 1.4466979122283767,
+      "grad_norm": 29.125,
+      "learning_rate": 4.4668713106372637e-07,
+      "loss": 0.7807,
+      "step": 16977
+    },
+    {
+      "epoch": 1.4467831273966767,
+      "grad_norm": 17.0,
+      "learning_rate": 4.4655839626505424e-07,
+      "loss": 0.7902,
+      "step": 16978
+    },
+    {
+      "epoch": 1.4468683425649767,
+      "grad_norm": 13.75,
+      "learning_rate": 4.4642967598537994e-07,
+      "loss": 0.6579,
+      "step": 16979
+    },
+    {
+      "epoch": 1.4469535577332766,
+      "grad_norm": 17.375,
+      "learning_rate": 4.463009702270289e-07,
+      "loss": 0.8732,
+      "step": 16980
+    },
+    {
+      "epoch": 1.4470387729015766,
+      "grad_norm": 14.25,
+      "learning_rate": 4.461722789923277e-07,
+      "loss": 0.7309,
+      "step": 16981
+    },
+    {
+      "epoch": 1.4471239880698765,
+      "grad_norm": 16.625,
+      "learning_rate": 4.4604360228360207e-07,
+      "loss": 0.8281,
+      "step": 16982
+    },
+    {
+      "epoch": 1.4472092032381765,
+      "grad_norm": 15.875,
+      "learning_rate": 4.459149401031765e-07,
+      "loss": 0.7871,
+      "step": 16983
+    },
+    {
+      "epoch": 1.4472944184064764,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.457862924533765e-07,
+      "loss": 0.5367,
+      "step": 16984
+    },
+    {
+      "epoch": 1.4473796335747764,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.4565765933652694e-07,
+      "loss": 0.4747,
+      "step": 16985
+    },
+    {
+      "epoch": 1.4474648487430763,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.4552904075495285e-07,
+      "loss": 0.5666,
+      "step": 16986
+    },
+    {
+      "epoch": 1.4475500639113763,
+      "grad_norm": 10.9375,
+      "learning_rate": 4.454004367109778e-07,
+      "loss": 0.2091,
+      "step": 16987
+    },
+    {
+      "epoch": 1.4476352790796763,
+      "grad_norm": 16.875,
+      "learning_rate": 4.452718472069255e-07,
+      "loss": 0.6444,
+      "step": 16988
+    },
+    {
+      "epoch": 1.4477204942479762,
+      "grad_norm": 19.0,
+      "learning_rate": 4.451432722451203e-07,
+      "loss": 0.5903,
+      "step": 16989
+    },
+    {
+      "epoch": 1.4478057094162762,
+      "grad_norm": 17.125,
+      "learning_rate": 4.4501471182788537e-07,
+      "loss": 0.6806,
+      "step": 16990
+    },
+    {
+      "epoch": 1.4478909245845761,
+      "grad_norm": 17.5,
+      "learning_rate": 4.4488616595754456e-07,
+      "loss": 0.5342,
+      "step": 16991
+    },
+    {
+      "epoch": 1.447976139752876,
+      "grad_norm": 16.25,
+      "learning_rate": 4.447576346364199e-07,
+      "loss": 0.4602,
+      "step": 16992
+    },
+    {
+      "epoch": 1.448061354921176,
+      "grad_norm": 19.5,
+      "learning_rate": 4.446291178668345e-07,
+      "loss": 0.8305,
+      "step": 16993
+    },
+    {
+      "epoch": 1.448146570089476,
+      "grad_norm": 16.125,
+      "learning_rate": 4.4450061565111084e-07,
+      "loss": 0.775,
+      "step": 16994
+    },
+    {
+      "epoch": 1.448231785257776,
+      "grad_norm": 17.5,
+      "learning_rate": 4.443721279915719e-07,
+      "loss": 0.7017,
+      "step": 16995
+    },
+    {
+      "epoch": 1.448317000426076,
+      "grad_norm": 14.625,
+      "learning_rate": 4.442436548905378e-07,
+      "loss": 0.5601,
+      "step": 16996
+    },
+    {
+      "epoch": 1.4484022155943759,
+      "grad_norm": 15.625,
+      "learning_rate": 4.441151963503312e-07,
+      "loss": 0.6108,
+      "step": 16997
+    },
+    {
+      "epoch": 1.4484874307626758,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.4398675237327325e-07,
+      "loss": 0.5857,
+      "step": 16998
+    },
+    {
+      "epoch": 1.4485726459309758,
+      "grad_norm": 19.0,
+      "learning_rate": 4.438583229616858e-07,
+      "loss": 0.938,
+      "step": 16999
+    },
+    {
+      "epoch": 1.4486578610992757,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.437299081178886e-07,
+      "loss": 0.5075,
+      "step": 17000
+    },
+    {
+      "epoch": 1.4487430762675757,
+      "grad_norm": 15.625,
+      "learning_rate": 4.436015078442027e-07,
+      "loss": 0.5078,
+      "step": 17001
+    },
+    {
+      "epoch": 1.4488282914358757,
+      "grad_norm": 16.375,
+      "learning_rate": 4.434731221429485e-07,
+      "loss": 0.4683,
+      "step": 17002
+    },
+    {
+      "epoch": 1.4489135066041756,
+      "grad_norm": 21.375,
+      "learning_rate": 4.433447510164464e-07,
+      "loss": 1.0,
+      "step": 17003
+    },
+    {
+      "epoch": 1.4489987217724756,
+      "grad_norm": 18.0,
+      "learning_rate": 4.4321639446701576e-07,
+      "loss": 0.7153,
+      "step": 17004
+    },
+    {
+      "epoch": 1.4490839369407755,
+      "grad_norm": 18.0,
+      "learning_rate": 4.430880524969758e-07,
+      "loss": 0.5156,
+      "step": 17005
+    },
+    {
+      "epoch": 1.4491691521090755,
+      "grad_norm": 16.375,
+      "learning_rate": 4.4295972510864606e-07,
+      "loss": 0.7852,
+      "step": 17006
+    },
+    {
+      "epoch": 1.4492543672773754,
+      "grad_norm": 19.375,
+      "learning_rate": 4.428314123043456e-07,
+      "loss": 0.7611,
+      "step": 17007
+    },
+    {
+      "epoch": 1.4493395824456754,
+      "grad_norm": 21.75,
+      "learning_rate": 4.427031140863937e-07,
+      "loss": 0.6999,
+      "step": 17008
+    },
+    {
+      "epoch": 1.4494247976139754,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.42574830457108e-07,
+      "loss": 0.4359,
+      "step": 17009
+    },
+    {
+      "epoch": 1.4495100127822753,
+      "grad_norm": 14.25,
+      "learning_rate": 4.424465614188068e-07,
+      "loss": 0.6387,
+      "step": 17010
+    },
+    {
+      "epoch": 1.4495952279505753,
+      "grad_norm": 14.0,
+      "learning_rate": 4.423183069738089e-07,
+      "loss": 0.3323,
+      "step": 17011
+    },
+    {
+      "epoch": 1.4496804431188752,
+      "grad_norm": 18.125,
+      "learning_rate": 4.421900671244314e-07,
+      "loss": 0.8071,
+      "step": 17012
+    },
+    {
+      "epoch": 1.4497656582871752,
+      "grad_norm": 17.25,
+      "learning_rate": 4.420618418729912e-07,
+      "loss": 0.7116,
+      "step": 17013
+    },
+    {
+      "epoch": 1.4498508734554751,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.4193363122180604e-07,
+      "loss": 0.7531,
+      "step": 17014
+    },
+    {
+      "epoch": 1.449936088623775,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.4180543517319264e-07,
+      "loss": 0.5508,
+      "step": 17015
+    },
+    {
+      "epoch": 1.450021303792075,
+      "grad_norm": 15.75,
+      "learning_rate": 4.416772537294682e-07,
+      "loss": 0.767,
+      "step": 17016
+    },
+    {
+      "epoch": 1.450106518960375,
+      "grad_norm": 19.5,
+      "learning_rate": 4.415490868929481e-07,
+      "loss": 0.6647,
+      "step": 17017
+    },
+    {
+      "epoch": 1.450191734128675,
+      "grad_norm": 12.625,
+      "learning_rate": 4.414209346659491e-07,
+      "loss": 0.3735,
+      "step": 17018
+    },
+    {
+      "epoch": 1.450276949296975,
+      "grad_norm": 14.25,
+      "learning_rate": 4.412927970507873e-07,
+      "loss": 0.7287,
+      "step": 17019
+    },
+    {
+      "epoch": 1.4503621644652749,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.411646740497773e-07,
+      "loss": 0.712,
+      "step": 17020
+    },
+    {
+      "epoch": 1.4504473796335748,
+      "grad_norm": 18.375,
+      "learning_rate": 4.410365656652356e-07,
+      "loss": 1.0275,
+      "step": 17021
+    },
+    {
+      "epoch": 1.4505325948018748,
+      "grad_norm": 13.25,
+      "learning_rate": 4.4090847189947605e-07,
+      "loss": 0.4065,
+      "step": 17022
+    },
+    {
+      "epoch": 1.4506178099701748,
+      "grad_norm": 32.25,
+      "learning_rate": 4.407803927548142e-07,
+      "loss": 0.4734,
+      "step": 17023
+    },
+    {
+      "epoch": 1.4507030251384747,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.4065232823356425e-07,
+      "loss": 0.6846,
+      "step": 17024
+    },
+    {
+      "epoch": 1.4507882403067747,
+      "grad_norm": 19.125,
+      "learning_rate": 4.4052427833804104e-07,
+      "loss": 0.8041,
+      "step": 17025
+    },
+    {
+      "epoch": 1.4508734554750746,
+      "grad_norm": 16.125,
+      "learning_rate": 4.403962430705577e-07,
+      "loss": 0.7057,
+      "step": 17026
+    },
+    {
+      "epoch": 1.4509586706433746,
+      "grad_norm": 16.875,
+      "learning_rate": 4.4026822243342884e-07,
+      "loss": 0.4326,
+      "step": 17027
+    },
+    {
+      "epoch": 1.4510438858116745,
+      "grad_norm": 10.9375,
+      "learning_rate": 4.40140216428967e-07,
+      "loss": 0.2751,
+      "step": 17028
+    },
+    {
+      "epoch": 1.4511291009799745,
+      "grad_norm": 14.625,
+      "learning_rate": 4.400122250594861e-07,
+      "loss": 0.31,
+      "step": 17029
+    },
+    {
+      "epoch": 1.4512143161482745,
+      "grad_norm": 17.5,
+      "learning_rate": 4.398842483272986e-07,
+      "loss": 0.5284,
+      "step": 17030
+    },
+    {
+      "epoch": 1.4512995313165744,
+      "grad_norm": 28.875,
+      "learning_rate": 4.3975628623471727e-07,
+      "loss": 0.7022,
+      "step": 17031
+    },
+    {
+      "epoch": 1.4513847464848744,
+      "grad_norm": 17.875,
+      "learning_rate": 4.3962833878405455e-07,
+      "loss": 0.4179,
+      "step": 17032
+    },
+    {
+      "epoch": 1.4514699616531743,
+      "grad_norm": 15.625,
+      "learning_rate": 4.395004059776231e-07,
+      "loss": 0.5578,
+      "step": 17033
+    },
+    {
+      "epoch": 1.4515551768214743,
+      "grad_norm": 25.0,
+      "learning_rate": 4.39372487817734e-07,
+      "loss": 0.6071,
+      "step": 17034
+    },
+    {
+      "epoch": 1.4516403919897742,
+      "grad_norm": 22.75,
+      "learning_rate": 4.3924458430669947e-07,
+      "loss": 0.5828,
+      "step": 17035
+    },
+    {
+      "epoch": 1.4517256071580742,
+      "grad_norm": 31.0,
+      "learning_rate": 4.3911669544683027e-07,
+      "loss": 0.8433,
+      "step": 17036
+    },
+    {
+      "epoch": 1.4518108223263742,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.3898882124043777e-07,
+      "loss": 0.5233,
+      "step": 17037
+    },
+    {
+      "epoch": 1.451896037494674,
+      "grad_norm": 16.5,
+      "learning_rate": 4.388609616898333e-07,
+      "loss": 0.6398,
+      "step": 17038
+    },
+    {
+      "epoch": 1.451981252662974,
+      "grad_norm": 12.875,
+      "learning_rate": 4.3873311679732634e-07,
+      "loss": 0.4237,
+      "step": 17039
+    },
+    {
+      "epoch": 1.452066467831274,
+      "grad_norm": 20.0,
+      "learning_rate": 4.386052865652278e-07,
+      "loss": 0.604,
+      "step": 17040
+    },
+    {
+      "epoch": 1.452151682999574,
+      "grad_norm": 18.0,
+      "learning_rate": 4.3847747099584766e-07,
+      "loss": 0.8829,
+      "step": 17041
+    },
+    {
+      "epoch": 1.452236898167874,
+      "grad_norm": 21.5,
+      "learning_rate": 4.3834967009149615e-07,
+      "loss": 0.714,
+      "step": 17042
+    },
+    {
+      "epoch": 1.452322113336174,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.3822188385448214e-07,
+      "loss": 0.6461,
+      "step": 17043
+    },
+    {
+      "epoch": 1.4524073285044738,
+      "grad_norm": 16.125,
+      "learning_rate": 4.3809411228711475e-07,
+      "loss": 0.7615,
+      "step": 17044
+    },
+    {
+      "epoch": 1.4524925436727738,
+      "grad_norm": 14.9375,
+      "learning_rate": 4.379663553917031e-07,
+      "loss": 0.6451,
+      "step": 17045
+    },
+    {
+      "epoch": 1.4525777588410738,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.3783861317055644e-07,
+      "loss": 0.9955,
+      "step": 17046
+    },
+    {
+      "epoch": 1.4526629740093737,
+      "grad_norm": 11.25,
+      "learning_rate": 4.3771088562598223e-07,
+      "loss": 0.2094,
+      "step": 17047
+    },
+    {
+      "epoch": 1.4527481891776737,
+      "grad_norm": 16.375,
+      "learning_rate": 4.3758317276028915e-07,
+      "loss": 0.6751,
+      "step": 17048
+    },
+    {
+      "epoch": 1.4528334043459736,
+      "grad_norm": 13.5,
+      "learning_rate": 4.3745547457578515e-07,
+      "loss": 0.3459,
+      "step": 17049
+    },
+    {
+      "epoch": 1.4529186195142736,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.3732779107477807e-07,
+      "loss": 0.2961,
+      "step": 17050
+    },
+    {
+      "epoch": 1.4530038346825735,
+      "grad_norm": 16.375,
+      "learning_rate": 4.3720012225957525e-07,
+      "loss": 0.7253,
+      "step": 17051
+    },
+    {
+      "epoch": 1.4530890498508735,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.37072468132483e-07,
+      "loss": 0.7138,
+      "step": 17052
+    },
+    {
+      "epoch": 1.4531742650191735,
+      "grad_norm": 11.5,
+      "learning_rate": 4.3694482869580865e-07,
+      "loss": 0.292,
+      "step": 17053
+    },
+    {
+      "epoch": 1.4532594801874734,
+      "grad_norm": 16.375,
+      "learning_rate": 4.36817203951859e-07,
+      "loss": 0.5646,
+      "step": 17054
+    },
+    {
+      "epoch": 1.4533446953557734,
+      "grad_norm": 16.875,
+      "learning_rate": 4.3668959390294044e-07,
+      "loss": 0.7833,
+      "step": 17055
+    },
+    {
+      "epoch": 1.4534299105240733,
+      "grad_norm": 13.375,
+      "learning_rate": 4.3656199855135834e-07,
+      "loss": 0.4672,
+      "step": 17056
+    },
+    {
+      "epoch": 1.4535151256923733,
+      "grad_norm": 14.375,
+      "learning_rate": 4.36434417899419e-07,
+      "loss": 0.6292,
+      "step": 17057
+    },
+    {
+      "epoch": 1.4536003408606732,
+      "grad_norm": 19.5,
+      "learning_rate": 4.363068519494283e-07,
+      "loss": 0.774,
+      "step": 17058
+    },
+    {
+      "epoch": 1.4536855560289732,
+      "grad_norm": 16.875,
+      "learning_rate": 4.3617930070369044e-07,
+      "loss": 0.6635,
+      "step": 17059
+    },
+    {
+      "epoch": 1.4537707711972732,
+      "grad_norm": 16.5,
+      "learning_rate": 4.360517641645115e-07,
+      "loss": 0.5252,
+      "step": 17060
+    },
+    {
+      "epoch": 1.4538559863655731,
+      "grad_norm": 14.625,
+      "learning_rate": 4.359242423341953e-07,
+      "loss": 0.4916,
+      "step": 17061
+    },
+    {
+      "epoch": 1.453941201533873,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.357967352150466e-07,
+      "loss": 0.4099,
+      "step": 17062
+    },
+    {
+      "epoch": 1.454026416702173,
+      "grad_norm": 18.375,
+      "learning_rate": 4.3566924280937004e-07,
+      "loss": 0.8215,
+      "step": 17063
+    },
+    {
+      "epoch": 1.454111631870473,
+      "grad_norm": 15.25,
+      "learning_rate": 4.3554176511946864e-07,
+      "loss": 0.554,
+      "step": 17064
+    },
+    {
+      "epoch": 1.454196847038773,
+      "grad_norm": 11.5,
+      "learning_rate": 4.3541430214764673e-07,
+      "loss": 0.3404,
+      "step": 17065
+    },
+    {
+      "epoch": 1.454282062207073,
+      "grad_norm": 11.75,
+      "learning_rate": 4.352868538962074e-07,
+      "loss": 0.2362,
+      "step": 17066
+    },
+    {
+      "epoch": 1.4543672773753729,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.3515942036745433e-07,
+      "loss": 0.6827,
+      "step": 17067
+    },
+    {
+      "epoch": 1.4544524925436728,
+      "grad_norm": 17.875,
+      "learning_rate": 4.3503200156368984e-07,
+      "loss": 0.404,
+      "step": 17068
+    },
+    {
+      "epoch": 1.4545377077119728,
+      "grad_norm": 18.625,
+      "learning_rate": 4.3490459748721627e-07,
+      "loss": 0.7319,
+      "step": 17069
+    },
+    {
+      "epoch": 1.4546229228802727,
+      "grad_norm": 15.0,
+      "learning_rate": 4.3477720814033617e-07,
+      "loss": 0.5502,
+      "step": 17070
+    },
+    {
+      "epoch": 1.4547081380485727,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.346498335253517e-07,
+      "loss": 0.4936,
+      "step": 17071
+    },
+    {
+      "epoch": 1.4547933532168726,
+      "grad_norm": 10.125,
+      "learning_rate": 4.3452247364456526e-07,
+      "loss": 0.1563,
+      "step": 17072
+    },
+    {
+      "epoch": 1.4548785683851726,
+      "grad_norm": 36.5,
+      "learning_rate": 4.34395128500277e-07,
+      "loss": 0.7963,
+      "step": 17073
+    },
+    {
+      "epoch": 1.4549637835534726,
+      "grad_norm": 12.1875,
+      "learning_rate": 4.3426779809478917e-07,
+      "loss": 0.3361,
+      "step": 17074
+    },
+    {
+      "epoch": 1.4550489987217725,
+      "grad_norm": 10.875,
+      "learning_rate": 4.3414048243040285e-07,
+      "loss": 0.2579,
+      "step": 17075
+    },
+    {
+      "epoch": 1.4551342138900725,
+      "grad_norm": 17.125,
+      "learning_rate": 4.340131815094184e-07,
+      "loss": 0.7359,
+      "step": 17076
+    },
+    {
+      "epoch": 1.4552194290583724,
+      "grad_norm": 15.75,
+      "learning_rate": 4.33885895334136e-07,
+      "loss": 0.816,
+      "step": 17077
+    },
+    {
+      "epoch": 1.4553046442266724,
+      "grad_norm": 19.5,
+      "learning_rate": 4.33758623906856e-07,
+      "loss": 0.6487,
+      "step": 17078
+    },
+    {
+      "epoch": 1.4553898593949723,
+      "grad_norm": 11.4375,
+      "learning_rate": 4.336313672298785e-07,
+      "loss": 0.3345,
+      "step": 17079
+    },
+    {
+      "epoch": 1.4554750745632723,
+      "grad_norm": 16.625,
+      "learning_rate": 4.335041253055035e-07,
+      "loss": 0.5754,
+      "step": 17080
+    },
+    {
+      "epoch": 1.4555602897315723,
+      "grad_norm": 20.125,
+      "learning_rate": 4.3337689813602963e-07,
+      "loss": 1.0234,
+      "step": 17081
+    },
+    {
+      "epoch": 1.4556455048998722,
+      "grad_norm": 13.25,
+      "learning_rate": 4.332496857237564e-07,
+      "loss": 0.4606,
+      "step": 17082
+    },
+    {
+      "epoch": 1.4557307200681722,
+      "grad_norm": 21.375,
+      "learning_rate": 4.3312248807098266e-07,
+      "loss": 0.4684,
+      "step": 17083
+    },
+    {
+      "epoch": 1.4558159352364721,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.329953051800073e-07,
+      "loss": 0.5979,
+      "step": 17084
+    },
+    {
+      "epoch": 1.455901150404772,
+      "grad_norm": 19.75,
+      "learning_rate": 4.3286813705312845e-07,
+      "loss": 0.6487,
+      "step": 17085
+    },
+    {
+      "epoch": 1.455986365573072,
+      "grad_norm": 18.0,
+      "learning_rate": 4.327409836926437e-07,
+      "loss": 0.8636,
+      "step": 17086
+    },
+    {
+      "epoch": 1.456071580741372,
+      "grad_norm": 15.125,
+      "learning_rate": 4.32613845100851e-07,
+      "loss": 0.5902,
+      "step": 17087
+    },
+    {
+      "epoch": 1.456156795909672,
+      "grad_norm": 20.625,
+      "learning_rate": 4.324867212800482e-07,
+      "loss": 0.9098,
+      "step": 17088
+    },
+    {
+      "epoch": 1.456242011077972,
+      "grad_norm": 16.625,
+      "learning_rate": 4.3235961223253284e-07,
+      "loss": 0.917,
+      "step": 17089
+    },
+    {
+      "epoch": 1.4563272262462719,
+      "grad_norm": 15.875,
+      "learning_rate": 4.3223251796060103e-07,
+      "loss": 0.3108,
+      "step": 17090
+    },
+    {
+      "epoch": 1.4564124414145718,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.321054384665499e-07,
+      "loss": 0.2126,
+      "step": 17091
+    },
+    {
+      "epoch": 1.4564976565828718,
+      "grad_norm": 21.125,
+      "learning_rate": 4.3197837375267645e-07,
+      "loss": 1.1543,
+      "step": 17092
+    },
+    {
+      "epoch": 1.4565828717511717,
+      "grad_norm": 19.625,
+      "learning_rate": 4.318513238212764e-07,
+      "loss": 0.7058,
+      "step": 17093
+    },
+    {
+      "epoch": 1.4566680869194717,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.3172428867464517e-07,
+      "loss": 0.6422,
+      "step": 17094
+    },
+    {
+      "epoch": 1.4567533020877717,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.315972683150789e-07,
+      "loss": 0.3985,
+      "step": 17095
+    },
+    {
+      "epoch": 1.4568385172560716,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.3147026274487287e-07,
+      "loss": 0.6573,
+      "step": 17096
+    },
+    {
+      "epoch": 1.4569237324243716,
+      "grad_norm": 16.875,
+      "learning_rate": 4.313432719663228e-07,
+      "loss": 0.9987,
+      "step": 17097
+    },
+    {
+      "epoch": 1.4570089475926715,
+      "grad_norm": 20.25,
+      "learning_rate": 4.312162959817226e-07,
+      "loss": 0.8918,
+      "step": 17098
+    },
+    {
+      "epoch": 1.4570941627609715,
+      "grad_norm": 11.375,
+      "learning_rate": 4.3108933479336727e-07,
+      "loss": 0.3491,
+      "step": 17099
+    },
+    {
+      "epoch": 1.4571793779292714,
+      "grad_norm": 18.0,
+      "learning_rate": 4.309623884035516e-07,
+      "loss": 0.6478,
+      "step": 17100
+    },
+    {
+      "epoch": 1.4572645930975714,
+      "grad_norm": 19.5,
+      "learning_rate": 4.3083545681456866e-07,
+      "loss": 0.7967,
+      "step": 17101
+    },
+    {
+      "epoch": 1.4573498082658713,
+      "grad_norm": 14.875,
+      "learning_rate": 4.307085400287132e-07,
+      "loss": 0.756,
+      "step": 17102
+    },
+    {
+      "epoch": 1.4574350234341713,
+      "grad_norm": 19.5,
+      "learning_rate": 4.30581638048278e-07,
+      "loss": 0.3875,
+      "step": 17103
+    },
+    {
+      "epoch": 1.4575202386024713,
+      "grad_norm": 12.5625,
+      "learning_rate": 4.304547508755565e-07,
+      "loss": 0.2657,
+      "step": 17104
+    },
+    {
+      "epoch": 1.4576054537707712,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.303278785128422e-07,
+      "loss": 0.7003,
+      "step": 17105
+    },
+    {
+      "epoch": 1.4576906689390712,
+      "grad_norm": 16.0,
+      "learning_rate": 4.3020102096242686e-07,
+      "loss": 0.5702,
+      "step": 17106
+    },
+    {
+      "epoch": 1.4577758841073711,
+      "grad_norm": 17.125,
+      "learning_rate": 4.3007417822660355e-07,
+      "loss": 0.8113,
+      "step": 17107
+    },
+    {
+      "epoch": 1.457861099275671,
+      "grad_norm": 23.125,
+      "learning_rate": 4.299473503076647e-07,
+      "loss": 0.6691,
+      "step": 17108
+    },
+    {
+      "epoch": 1.457946314443971,
+      "grad_norm": 15.875,
+      "learning_rate": 4.2982053720790144e-07,
+      "loss": 0.6064,
+      "step": 17109
+    },
+    {
+      "epoch": 1.458031529612271,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.2969373892960615e-07,
+      "loss": 0.4852,
+      "step": 17110
+    },
+    {
+      "epoch": 1.458116744780571,
+      "grad_norm": 12.0625,
+      "learning_rate": 4.295669554750696e-07,
+      "loss": 0.4579,
+      "step": 17111
+    },
+    {
+      "epoch": 1.458201959948871,
+      "grad_norm": 15.375,
+      "learning_rate": 4.294401868465832e-07,
+      "loss": 0.5323,
+      "step": 17112
+    },
+    {
+      "epoch": 1.4582871751171709,
+      "grad_norm": 16.625,
+      "learning_rate": 4.2931343304643766e-07,
+      "loss": 0.513,
+      "step": 17113
+    },
+    {
+      "epoch": 1.4583723902854708,
+      "grad_norm": 24.375,
+      "learning_rate": 4.2918669407692416e-07,
+      "loss": 0.7668,
+      "step": 17114
+    },
+    {
+      "epoch": 1.4584576054537708,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.290599699403321e-07,
+      "loss": 0.5011,
+      "step": 17115
+    },
+    {
+      "epoch": 1.4585428206220707,
+      "grad_norm": 16.5,
+      "learning_rate": 4.289332606389522e-07,
+      "loss": 0.8973,
+      "step": 17116
+    },
+    {
+      "epoch": 1.4586280357903707,
+      "grad_norm": 23.0,
+      "learning_rate": 4.288065661750736e-07,
+      "loss": 0.5071,
+      "step": 17117
+    },
+    {
+      "epoch": 1.4587132509586707,
+      "grad_norm": 11.3125,
+      "learning_rate": 4.28679886550986e-07,
+      "loss": 0.2919,
+      "step": 17118
+    },
+    {
+      "epoch": 1.4587984661269706,
+      "grad_norm": 28.625,
+      "learning_rate": 4.2855322176897917e-07,
+      "loss": 0.9846,
+      "step": 17119
+    },
+    {
+      "epoch": 1.4588836812952706,
+      "grad_norm": 16.25,
+      "learning_rate": 4.2842657183134143e-07,
+      "loss": 0.8774,
+      "step": 17120
+    },
+    {
+      "epoch": 1.4589688964635705,
+      "grad_norm": 13.0,
+      "learning_rate": 4.2829993674036145e-07,
+      "loss": 0.4633,
+      "step": 17121
+    },
+    {
+      "epoch": 1.4590541116318705,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.2817331649832845e-07,
+      "loss": 0.5004,
+      "step": 17122
+    },
+    {
+      "epoch": 1.4591393268001704,
+      "grad_norm": 17.25,
+      "learning_rate": 4.2804671110752967e-07,
+      "loss": 0.7934,
+      "step": 17123
+    },
+    {
+      "epoch": 1.4592245419684704,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.2792012057025323e-07,
+      "loss": 0.668,
+      "step": 17124
+    },
+    {
+      "epoch": 1.4593097571367704,
+      "grad_norm": 13.125,
+      "learning_rate": 4.2779354488878737e-07,
+      "loss": 0.4754,
+      "step": 17125
+    },
+    {
+      "epoch": 1.4593949723050703,
+      "grad_norm": 16.125,
+      "learning_rate": 4.276669840654185e-07,
+      "loss": 0.5229,
+      "step": 17126
+    },
+    {
+      "epoch": 1.4594801874733703,
+      "grad_norm": 19.875,
+      "learning_rate": 4.2754043810243464e-07,
+      "loss": 0.8423,
+      "step": 17127
+    },
+    {
+      "epoch": 1.4595654026416702,
+      "grad_norm": 20.5,
+      "learning_rate": 4.274139070021216e-07,
+      "loss": 0.4509,
+      "step": 17128
+    },
+    {
+      "epoch": 1.4596506178099702,
+      "grad_norm": 34.0,
+      "learning_rate": 4.2728739076676643e-07,
+      "loss": 1.3164,
+      "step": 17129
+    },
+    {
+      "epoch": 1.4597358329782701,
+      "grad_norm": 10.75,
+      "learning_rate": 4.271608893986555e-07,
+      "loss": 0.3478,
+      "step": 17130
+    },
+    {
+      "epoch": 1.45982104814657,
+      "grad_norm": 35.0,
+      "learning_rate": 4.27034402900075e-07,
+      "loss": 0.9832,
+      "step": 17131
+    },
+    {
+      "epoch": 1.45990626331487,
+      "grad_norm": 19.625,
+      "learning_rate": 4.269079312733101e-07,
+      "loss": 0.6734,
+      "step": 17132
+    },
+    {
+      "epoch": 1.45999147848317,
+      "grad_norm": 18.5,
+      "learning_rate": 4.267814745206469e-07,
+      "loss": 0.8575,
+      "step": 17133
+    },
+    {
+      "epoch": 1.46007669365147,
+      "grad_norm": 37.75,
+      "learning_rate": 4.2665503264436995e-07,
+      "loss": 0.8007,
+      "step": 17134
+    },
+    {
+      "epoch": 1.46016190881977,
+      "grad_norm": 14.0,
+      "learning_rate": 4.265286056467649e-07,
+      "loss": 0.6154,
+      "step": 17135
+    },
+    {
+      "epoch": 1.4602471239880699,
+      "grad_norm": 16.25,
+      "learning_rate": 4.264021935301156e-07,
+      "loss": 0.5894,
+      "step": 17136
+    },
+    {
+      "epoch": 1.4603323391563698,
+      "grad_norm": 18.0,
+      "learning_rate": 4.2627579629670677e-07,
+      "loss": 0.8026,
+      "step": 17137
+    },
+    {
+      "epoch": 1.4604175543246698,
+      "grad_norm": 17.875,
+      "learning_rate": 4.2614941394882257e-07,
+      "loss": 0.7687,
+      "step": 17138
+    },
+    {
+      "epoch": 1.4605027694929698,
+      "grad_norm": 19.125,
+      "learning_rate": 4.260230464887474e-07,
+      "loss": 0.8157,
+      "step": 17139
+    },
+    {
+      "epoch": 1.4605879846612697,
+      "grad_norm": 25.0,
+      "learning_rate": 4.258966939187641e-07,
+      "loss": 1.0111,
+      "step": 17140
+    },
+    {
+      "epoch": 1.4606731998295697,
+      "grad_norm": 11.0,
+      "learning_rate": 4.2577035624115653e-07,
+      "loss": 0.328,
+      "step": 17141
+    },
+    {
+      "epoch": 1.4607584149978696,
+      "grad_norm": 18.375,
+      "learning_rate": 4.25644033458207e-07,
+      "loss": 0.4941,
+      "step": 17142
+    },
+    {
+      "epoch": 1.4608436301661696,
+      "grad_norm": 15.75,
+      "learning_rate": 4.2551772557219873e-07,
+      "loss": 0.342,
+      "step": 17143
+    },
+    {
+      "epoch": 1.4609288453344695,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.2539143258541457e-07,
+      "loss": 0.3734,
+      "step": 17144
+    },
+    {
+      "epoch": 1.4610140605027695,
+      "grad_norm": 16.125,
+      "learning_rate": 4.2526515450013616e-07,
+      "loss": 0.584,
+      "step": 17145
+    },
+    {
+      "epoch": 1.4610992756710695,
+      "grad_norm": 17.5,
+      "learning_rate": 4.2513889131864576e-07,
+      "loss": 0.9631,
+      "step": 17146
+    },
+    {
+      "epoch": 1.4611844908393694,
+      "grad_norm": 21.375,
+      "learning_rate": 4.2501264304322496e-07,
+      "loss": 0.7708,
+      "step": 17147
+    },
+    {
+      "epoch": 1.4612697060076694,
+      "grad_norm": 18.25,
+      "learning_rate": 4.248864096761557e-07,
+      "loss": 0.8378,
+      "step": 17148
+    },
+    {
+      "epoch": 1.4613549211759693,
+      "grad_norm": 11.125,
+      "learning_rate": 4.2476019121971887e-07,
+      "loss": 0.2903,
+      "step": 17149
+    },
+    {
+      "epoch": 1.4614401363442693,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.2463398767619473e-07,
+      "loss": 0.4683,
+      "step": 17150
+    },
+    {
+      "epoch": 1.4615253515125692,
+      "grad_norm": 12.875,
+      "learning_rate": 4.245077990478644e-07,
+      "loss": 0.5628,
+      "step": 17151
+    },
+    {
+      "epoch": 1.4616105666808692,
+      "grad_norm": 17.0,
+      "learning_rate": 4.2438162533700853e-07,
+      "loss": 0.8071,
+      "step": 17152
+    },
+    {
+      "epoch": 1.4616957818491692,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.2425546654590673e-07,
+      "loss": 0.4636,
+      "step": 17153
+    },
+    {
+      "epoch": 1.461780997017469,
+      "grad_norm": 15.25,
+      "learning_rate": 4.241293226768388e-07,
+      "loss": 0.4963,
+      "step": 17154
+    },
+    {
+      "epoch": 1.461866212185769,
+      "grad_norm": 11.5,
+      "learning_rate": 4.240031937320846e-07,
+      "loss": 0.2781,
+      "step": 17155
+    },
+    {
+      "epoch": 1.461951427354069,
+      "grad_norm": 22.5,
+      "learning_rate": 4.238770797139237e-07,
+      "loss": 0.5648,
+      "step": 17156
+    },
+    {
+      "epoch": 1.462036642522369,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.237509806246347e-07,
+      "loss": 0.6565,
+      "step": 17157
+    },
+    {
+      "epoch": 1.462121857690669,
+      "grad_norm": 13.875,
+      "learning_rate": 4.236248964664959e-07,
+      "loss": 0.3532,
+      "step": 17158
+    },
+    {
+      "epoch": 1.462207072858969,
+      "grad_norm": 15.875,
+      "learning_rate": 4.2349882724178617e-07,
+      "loss": 0.6163,
+      "step": 17159
+    },
+    {
+      "epoch": 1.4622922880272688,
+      "grad_norm": 14.125,
+      "learning_rate": 4.233727729527837e-07,
+      "loss": 0.4131,
+      "step": 17160
+    },
+    {
+      "epoch": 1.4623775031955688,
+      "grad_norm": 16.375,
+      "learning_rate": 4.2324673360176703e-07,
+      "loss": 0.602,
+      "step": 17161
+    },
+    {
+      "epoch": 1.4624627183638688,
+      "grad_norm": 16.75,
+      "learning_rate": 4.231207091910128e-07,
+      "loss": 0.6405,
+      "step": 17162
+    },
+    {
+      "epoch": 1.4625479335321687,
+      "grad_norm": 12.875,
+      "learning_rate": 4.229946997227988e-07,
+      "loss": 0.6455,
+      "step": 17163
+    },
+    {
+      "epoch": 1.4626331487004687,
+      "grad_norm": 10.375,
+      "learning_rate": 4.228687051994021e-07,
+      "loss": 0.3049,
+      "step": 17164
+    },
+    {
+      "epoch": 1.4627183638687686,
+      "grad_norm": 25.625,
+      "learning_rate": 4.2274272562310065e-07,
+      "loss": 1.061,
+      "step": 17165
+    },
+    {
+      "epoch": 1.4628035790370686,
+      "grad_norm": 17.25,
+      "learning_rate": 4.22616760996169e-07,
+      "loss": 0.6431,
+      "step": 17166
+    },
+    {
+      "epoch": 1.4628887942053685,
+      "grad_norm": 14.5,
+      "learning_rate": 4.224908113208846e-07,
+      "loss": 0.6035,
+      "step": 17167
+    },
+    {
+      "epoch": 1.4629740093736685,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.2236487659952334e-07,
+      "loss": 0.4164,
+      "step": 17168
+    },
+    {
+      "epoch": 1.4630592245419685,
+      "grad_norm": 12.5625,
+      "learning_rate": 4.2223895683436157e-07,
+      "loss": 0.5536,
+      "step": 17169
+    },
+    {
+      "epoch": 1.4631444397102684,
+      "grad_norm": 27.125,
+      "learning_rate": 4.2211305202767364e-07,
+      "loss": 0.6174,
+      "step": 17170
+    },
+    {
+      "epoch": 1.4632296548785684,
+      "grad_norm": 10.875,
+      "learning_rate": 4.219871621817354e-07,
+      "loss": 0.4187,
+      "step": 17171
+    },
+    {
+      "epoch": 1.4633148700468683,
+      "grad_norm": 17.75,
+      "learning_rate": 4.2186128729882174e-07,
+      "loss": 0.6627,
+      "step": 17172
+    },
+    {
+      "epoch": 1.4634000852151683,
+      "grad_norm": 22.75,
+      "learning_rate": 4.2173542738120786e-07,
+      "loss": 0.3031,
+      "step": 17173
+    },
+    {
+      "epoch": 1.4634853003834682,
+      "grad_norm": 11.625,
+      "learning_rate": 4.2160958243116766e-07,
+      "loss": 0.3564,
+      "step": 17174
+    },
+    {
+      "epoch": 1.4635705155517682,
+      "grad_norm": 18.0,
+      "learning_rate": 4.214837524509749e-07,
+      "loss": 0.6727,
+      "step": 17175
+    },
+    {
+      "epoch": 1.4636557307200682,
+      "grad_norm": 17.75,
+      "learning_rate": 4.213579374429039e-07,
+      "loss": 0.4214,
+      "step": 17176
+    },
+    {
+      "epoch": 1.4637409458883681,
+      "grad_norm": 9.75,
+      "learning_rate": 4.212321374092282e-07,
+      "loss": 0.228,
+      "step": 17177
+    },
+    {
+      "epoch": 1.463826161056668,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.211063523522216e-07,
+      "loss": 0.1683,
+      "step": 17178
+    },
+    {
+      "epoch": 1.463911376224968,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.2098058227415624e-07,
+      "loss": 0.3877,
+      "step": 17179
+    },
+    {
+      "epoch": 1.463996591393268,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.2085482717730554e-07,
+      "loss": 0.449,
+      "step": 17180
+    },
+    {
+      "epoch": 1.464081806561568,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.2072908706394215e-07,
+      "loss": 0.5855,
+      "step": 17181
+    },
+    {
+      "epoch": 1.464167021729868,
+      "grad_norm": 12.5,
+      "learning_rate": 4.2060336193633813e-07,
+      "loss": 0.4783,
+      "step": 17182
+    },
+    {
+      "epoch": 1.4642522368981679,
+      "grad_norm": 24.75,
+      "learning_rate": 4.20477651796765e-07,
+      "loss": 0.8068,
+      "step": 17183
+    },
+    {
+      "epoch": 1.4643374520664678,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.203519566474948e-07,
+      "loss": 0.3929,
+      "step": 17184
+    },
+    {
+      "epoch": 1.4644226672347678,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.20226276490799e-07,
+      "loss": 0.257,
+      "step": 17185
+    },
+    {
+      "epoch": 1.4645078824030677,
+      "grad_norm": 20.25,
+      "learning_rate": 4.201006113289492e-07,
+      "loss": 0.6328,
+      "step": 17186
+    },
+    {
+      "epoch": 1.4645930975713677,
+      "grad_norm": 18.5,
+      "learning_rate": 4.199749611642155e-07,
+      "loss": 0.6912,
+      "step": 17187
+    },
+    {
+      "epoch": 1.4646783127396676,
+      "grad_norm": 13.0,
+      "learning_rate": 4.19849325998869e-07,
+      "loss": 0.4515,
+      "step": 17188
+    },
+    {
+      "epoch": 1.4647635279079676,
+      "grad_norm": 20.375,
+      "learning_rate": 4.1972370583517983e-07,
+      "loss": 0.4455,
+      "step": 17189
+    },
+    {
+      "epoch": 1.4648487430762676,
+      "grad_norm": 19.0,
+      "learning_rate": 4.1959810067541863e-07,
+      "loss": 0.6437,
+      "step": 17190
+    },
+    {
+      "epoch": 1.4649339582445675,
+      "grad_norm": 17.5,
+      "learning_rate": 4.1947251052185486e-07,
+      "loss": 0.7048,
+      "step": 17191
+    },
+    {
+      "epoch": 1.4650191734128675,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.193469353767575e-07,
+      "loss": 0.8161,
+      "step": 17192
+    },
+    {
+      "epoch": 1.4651043885811674,
+      "grad_norm": 16.125,
+      "learning_rate": 4.1922137524239624e-07,
+      "loss": 0.5937,
+      "step": 17193
+    },
+    {
+      "epoch": 1.4651896037494674,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.190958301210403e-07,
+      "loss": 0.5203,
+      "step": 17194
+    },
+    {
+      "epoch": 1.4652748189177673,
+      "grad_norm": 16.625,
+      "learning_rate": 4.1897030001495863e-07,
+      "loss": 0.5454,
+      "step": 17195
+    },
+    {
+      "epoch": 1.4653600340860673,
+      "grad_norm": 12.6875,
+      "learning_rate": 4.188447849264188e-07,
+      "loss": 0.4373,
+      "step": 17196
+    },
+    {
+      "epoch": 1.4654452492543673,
+      "grad_norm": 14.125,
+      "learning_rate": 4.187192848576896e-07,
+      "loss": 0.3678,
+      "step": 17197
+    },
+    {
+      "epoch": 1.4655304644226672,
+      "grad_norm": 16.75,
+      "learning_rate": 4.1859379981103914e-07,
+      "loss": 0.9469,
+      "step": 17198
+    },
+    {
+      "epoch": 1.4656156795909672,
+      "grad_norm": 15.25,
+      "learning_rate": 4.184683297887349e-07,
+      "loss": 0.5068,
+      "step": 17199
+    },
+    {
+      "epoch": 1.4657008947592671,
+      "grad_norm": 29.5,
+      "learning_rate": 4.1834287479304346e-07,
+      "loss": 0.6375,
+      "step": 17200
+    },
+    {
+      "epoch": 1.465786109927567,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.1821743482623284e-07,
+      "loss": 0.4469,
+      "step": 17201
+    },
+    {
+      "epoch": 1.465871325095867,
+      "grad_norm": 14.125,
+      "learning_rate": 4.180920098905694e-07,
+      "loss": 0.411,
+      "step": 17202
+    },
+    {
+      "epoch": 1.465956540264167,
+      "grad_norm": 21.875,
+      "learning_rate": 4.179665999883205e-07,
+      "loss": 0.7702,
+      "step": 17203
+    },
+    {
+      "epoch": 1.466041755432467,
+      "grad_norm": 16.625,
+      "learning_rate": 4.178412051217512e-07,
+      "loss": 0.5243,
+      "step": 17204
+    },
+    {
+      "epoch": 1.466126970600767,
+      "grad_norm": 9.125,
+      "learning_rate": 4.177158252931282e-07,
+      "loss": 0.2955,
+      "step": 17205
+    },
+    {
+      "epoch": 1.4662121857690669,
+      "grad_norm": 15.0,
+      "learning_rate": 4.1759046050471765e-07,
+      "loss": 0.4988,
+      "step": 17206
+    },
+    {
+      "epoch": 1.4662974009373668,
+      "grad_norm": 12.375,
+      "learning_rate": 4.1746511075878407e-07,
+      "loss": 0.4868,
+      "step": 17207
+    },
+    {
+      "epoch": 1.4663826161056668,
+      "grad_norm": 17.75,
+      "learning_rate": 4.173397760575934e-07,
+      "loss": 0.545,
+      "step": 17208
+    },
+    {
+      "epoch": 1.4664678312739667,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.1721445640341e-07,
+      "loss": 0.4494,
+      "step": 17209
+    },
+    {
+      "epoch": 1.4665530464422667,
+      "grad_norm": 19.25,
+      "learning_rate": 4.1708915179849873e-07,
+      "loss": 0.5919,
+      "step": 17210
+    },
+    {
+      "epoch": 1.4666382616105667,
+      "grad_norm": 15.75,
+      "learning_rate": 4.1696386224512416e-07,
+      "loss": 0.7615,
+      "step": 17211
+    },
+    {
+      "epoch": 1.4667234767788666,
+      "grad_norm": 17.75,
+      "learning_rate": 4.1683858774555065e-07,
+      "loss": 0.7595,
+      "step": 17212
+    },
+    {
+      "epoch": 1.4668086919471666,
+      "grad_norm": 17.375,
+      "learning_rate": 4.1671332830204123e-07,
+      "loss": 0.5004,
+      "step": 17213
+    },
+    {
+      "epoch": 1.4668939071154665,
+      "grad_norm": 17.375,
+      "learning_rate": 4.1658808391686026e-07,
+      "loss": 0.6223,
+      "step": 17214
+    },
+    {
+      "epoch": 1.4669791222837665,
+      "grad_norm": 18.375,
+      "learning_rate": 4.164628545922704e-07,
+      "loss": 0.8085,
+      "step": 17215
+    },
+    {
+      "epoch": 1.4670643374520664,
+      "grad_norm": 16.125,
+      "learning_rate": 4.1633764033053523e-07,
+      "loss": 0.7252,
+      "step": 17216
+    },
+    {
+      "epoch": 1.4671495526203664,
+      "grad_norm": 18.875,
+      "learning_rate": 4.162124411339169e-07,
+      "loss": 0.7536,
+      "step": 17217
+    },
+    {
+      "epoch": 1.4672347677886663,
+      "grad_norm": 18.25,
+      "learning_rate": 4.1608725700467813e-07,
+      "loss": 1.0706,
+      "step": 17218
+    },
+    {
+      "epoch": 1.4673199829569663,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.1596208794508127e-07,
+      "loss": 0.4806,
+      "step": 17219
+    },
+    {
+      "epoch": 1.4674051981252663,
+      "grad_norm": 15.5,
+      "learning_rate": 4.158369339573884e-07,
+      "loss": 0.5664,
+      "step": 17220
+    },
+    {
+      "epoch": 1.4674904132935662,
+      "grad_norm": 17.875,
+      "learning_rate": 4.1571179504386064e-07,
+      "loss": 0.5412,
+      "step": 17221
+    },
+    {
+      "epoch": 1.4675756284618662,
+      "grad_norm": 19.125,
+      "learning_rate": 4.1558667120676024e-07,
+      "loss": 0.7067,
+      "step": 17222
+    },
+    {
+      "epoch": 1.4676608436301661,
+      "grad_norm": 13.0,
+      "learning_rate": 4.1546156244834713e-07,
+      "loss": 0.1982,
+      "step": 17223
+    },
+    {
+      "epoch": 1.467746058798466,
+      "grad_norm": 21.625,
+      "learning_rate": 4.153364687708829e-07,
+      "loss": 0.6483,
+      "step": 17224
+    },
+    {
+      "epoch": 1.467831273966766,
+      "grad_norm": 16.75,
+      "learning_rate": 4.152113901766283e-07,
+      "loss": 0.7269,
+      "step": 17225
+    },
+    {
+      "epoch": 1.467916489135066,
+      "grad_norm": 13.25,
+      "learning_rate": 4.1508632666784303e-07,
+      "loss": 0.4097,
+      "step": 17226
+    },
+    {
+      "epoch": 1.468001704303366,
+      "grad_norm": 11.1875,
+      "learning_rate": 4.1496127824678737e-07,
+      "loss": 0.3734,
+      "step": 17227
+    },
+    {
+      "epoch": 1.468086919471666,
+      "grad_norm": 47.5,
+      "learning_rate": 4.1483624491572156e-07,
+      "loss": 0.8318,
+      "step": 17228
+    },
+    {
+      "epoch": 1.4681721346399659,
+      "grad_norm": 10.9375,
+      "learning_rate": 4.147112266769042e-07,
+      "loss": 0.3817,
+      "step": 17229
+    },
+    {
+      "epoch": 1.4682573498082658,
+      "grad_norm": 13.375,
+      "learning_rate": 4.145862235325952e-07,
+      "loss": 0.561,
+      "step": 17230
+    },
+    {
+      "epoch": 1.4683425649765658,
+      "grad_norm": 16.75,
+      "learning_rate": 4.14461235485053e-07,
+      "loss": 0.5147,
+      "step": 17231
+    },
+    {
+      "epoch": 1.4684277801448657,
+      "grad_norm": 18.625,
+      "learning_rate": 4.1433626253653633e-07,
+      "loss": 0.8197,
+      "step": 17232
+    },
+    {
+      "epoch": 1.4685129953131657,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.1421130468930426e-07,
+      "loss": 0.4453,
+      "step": 17233
+    },
+    {
+      "epoch": 1.4685982104814657,
+      "grad_norm": 21.125,
+      "learning_rate": 4.140863619456138e-07,
+      "loss": 0.9174,
+      "step": 17234
+    },
+    {
+      "epoch": 1.4686834256497656,
+      "grad_norm": 19.125,
+      "learning_rate": 4.1396143430772347e-07,
+      "loss": 0.8058,
+      "step": 17235
+    },
+    {
+      "epoch": 1.4687686408180656,
+      "grad_norm": 20.875,
+      "learning_rate": 4.138365217778907e-07,
+      "loss": 0.6684,
+      "step": 17236
+    },
+    {
+      "epoch": 1.4688538559863655,
+      "grad_norm": 14.9375,
+      "learning_rate": 4.1371162435837316e-07,
+      "loss": 0.5723,
+      "step": 17237
+    },
+    {
+      "epoch": 1.4689390711546655,
+      "grad_norm": 15.5,
+      "learning_rate": 4.1358674205142763e-07,
+      "loss": 0.511,
+      "step": 17238
+    },
+    {
+      "epoch": 1.4690242863229654,
+      "grad_norm": 19.0,
+      "learning_rate": 4.134618748593103e-07,
+      "loss": 0.4162,
+      "step": 17239
+    },
+    {
+      "epoch": 1.4691095014912654,
+      "grad_norm": 14.5,
+      "learning_rate": 4.13337022784278e-07,
+      "loss": 0.4359,
+      "step": 17240
+    },
+    {
+      "epoch": 1.4691947166595654,
+      "grad_norm": 12.875,
+      "learning_rate": 4.132121858285872e-07,
+      "loss": 0.6238,
+      "step": 17241
+    },
+    {
+      "epoch": 1.4692799318278653,
+      "grad_norm": 14.375,
+      "learning_rate": 4.1308736399449394e-07,
+      "loss": 0.7326,
+      "step": 17242
+    },
+    {
+      "epoch": 1.4693651469961653,
+      "grad_norm": 17.625,
+      "learning_rate": 4.129625572842531e-07,
+      "loss": 0.5964,
+      "step": 17243
+    },
+    {
+      "epoch": 1.4694503621644652,
+      "grad_norm": 14.375,
+      "learning_rate": 4.128377657001208e-07,
+      "loss": 0.477,
+      "step": 17244
+    },
+    {
+      "epoch": 1.4695355773327652,
+      "grad_norm": 15.25,
+      "learning_rate": 4.127129892443521e-07,
+      "loss": 0.6399,
+      "step": 17245
+    },
+    {
+      "epoch": 1.4696207925010651,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.1258822791920173e-07,
+      "loss": 0.3021,
+      "step": 17246
+    },
+    {
+      "epoch": 1.469706007669365,
+      "grad_norm": 14.75,
+      "learning_rate": 4.124634817269238e-07,
+      "loss": 0.4937,
+      "step": 17247
+    },
+    {
+      "epoch": 1.469791222837665,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.123387506697729e-07,
+      "loss": 0.5869,
+      "step": 17248
+    },
+    {
+      "epoch": 1.469876438005965,
+      "grad_norm": 19.0,
+      "learning_rate": 4.122140347500031e-07,
+      "loss": 0.8371,
+      "step": 17249
+    },
+    {
+      "epoch": 1.469961653174265,
+      "grad_norm": 12.6875,
+      "learning_rate": 4.1208933396986865e-07,
+      "loss": 0.4551,
+      "step": 17250
+    },
+    {
+      "epoch": 1.470046868342565,
+      "grad_norm": 13.875,
+      "learning_rate": 4.11964648331622e-07,
+      "loss": 0.4806,
+      "step": 17251
+    },
+    {
+      "epoch": 1.4701320835108649,
+      "grad_norm": 13.375,
+      "learning_rate": 4.1183997783751694e-07,
+      "loss": 0.4856,
+      "step": 17252
+    },
+    {
+      "epoch": 1.4702172986791648,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.117153224898063e-07,
+      "loss": 0.5222,
+      "step": 17253
+    },
+    {
+      "epoch": 1.4703025138474648,
+      "grad_norm": 17.75,
+      "learning_rate": 4.1159068229074314e-07,
+      "loss": 0.6354,
+      "step": 17254
+    },
+    {
+      "epoch": 1.4703877290157648,
+      "grad_norm": 14.4375,
+      "learning_rate": 4.114660572425794e-07,
+      "loss": 0.4421,
+      "step": 17255
+    },
+    {
+      "epoch": 1.4704729441840647,
+      "grad_norm": 12.5,
+      "learning_rate": 4.113414473475669e-07,
+      "loss": 0.3916,
+      "step": 17256
+    },
+    {
+      "epoch": 1.4705581593523647,
+      "grad_norm": 17.125,
+      "learning_rate": 4.112168526079577e-07,
+      "loss": 0.6804,
+      "step": 17257
+    },
+    {
+      "epoch": 1.4706433745206646,
+      "grad_norm": 15.25,
+      "learning_rate": 4.1109227302600375e-07,
+      "loss": 0.451,
+      "step": 17258
+    },
+    {
+      "epoch": 1.4707285896889646,
+      "grad_norm": 19.875,
+      "learning_rate": 4.109677086039558e-07,
+      "loss": 0.5363,
+      "step": 17259
+    },
+    {
+      "epoch": 1.4708138048572645,
+      "grad_norm": 19.25,
+      "learning_rate": 4.1084315934406496e-07,
+      "loss": 0.4129,
+      "step": 17260
+    },
+    {
+      "epoch": 1.4708990200255645,
+      "grad_norm": 19.0,
+      "learning_rate": 4.1071862524858215e-07,
+      "loss": 0.7315,
+      "step": 17261
+    },
+    {
+      "epoch": 1.4709842351938645,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.105941063197581e-07,
+      "loss": 0.7197,
+      "step": 17262
+    },
+    {
+      "epoch": 1.4710694503621644,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.104696025598426e-07,
+      "loss": 0.397,
+      "step": 17263
+    },
+    {
+      "epoch": 1.4711546655304644,
+      "grad_norm": 18.375,
+      "learning_rate": 4.1034511397108524e-07,
+      "loss": 0.7737,
+      "step": 17264
+    },
+    {
+      "epoch": 1.4712398806987643,
+      "grad_norm": 12.75,
+      "learning_rate": 4.10220640555736e-07,
+      "loss": 0.5017,
+      "step": 17265
+    },
+    {
+      "epoch": 1.4713250958670643,
+      "grad_norm": 12.8125,
+      "learning_rate": 4.1009618231604416e-07,
+      "loss": 0.4812,
+      "step": 17266
+    },
+    {
+      "epoch": 1.4714103110353642,
+      "grad_norm": 16.375,
+      "learning_rate": 4.0997173925425946e-07,
+      "loss": 0.6555,
+      "step": 17267
+    },
+    {
+      "epoch": 1.4714955262036642,
+      "grad_norm": 16.125,
+      "learning_rate": 4.0984731137262967e-07,
+      "loss": 0.4136,
+      "step": 17268
+    },
+    {
+      "epoch": 1.4715807413719642,
+      "grad_norm": 11.25,
+      "learning_rate": 4.0972289867340376e-07,
+      "loss": 0.2908,
+      "step": 17269
+    },
+    {
+      "epoch": 1.471665956540264,
+      "grad_norm": 24.5,
+      "learning_rate": 4.095985011588301e-07,
+      "loss": 0.523,
+      "step": 17270
+    },
+    {
+      "epoch": 1.471751171708564,
+      "grad_norm": 14.5,
+      "learning_rate": 4.09474118831157e-07,
+      "loss": 0.5015,
+      "step": 17271
+    },
+    {
+      "epoch": 1.471836386876864,
+      "grad_norm": 18.25,
+      "learning_rate": 4.0934975169263175e-07,
+      "loss": 0.7505,
+      "step": 17272
+    },
+    {
+      "epoch": 1.471921602045164,
+      "grad_norm": 13.5,
+      "learning_rate": 4.0922539974550144e-07,
+      "loss": 0.6751,
+      "step": 17273
+    },
+    {
+      "epoch": 1.472006817213464,
+      "grad_norm": 19.375,
+      "learning_rate": 4.091010629920136e-07,
+      "loss": 0.7152,
+      "step": 17274
+    },
+    {
+      "epoch": 1.472092032381764,
+      "grad_norm": 18.625,
+      "learning_rate": 4.089767414344156e-07,
+      "loss": 0.9618,
+      "step": 17275
+    },
+    {
+      "epoch": 1.4721772475500638,
+      "grad_norm": 21.0,
+      "learning_rate": 4.0885243507495315e-07,
+      "loss": 0.5924,
+      "step": 17276
+    },
+    {
+      "epoch": 1.4722624627183638,
+      "grad_norm": 18.625,
+      "learning_rate": 4.0872814391587303e-07,
+      "loss": 0.5115,
+      "step": 17277
+    },
+    {
+      "epoch": 1.4723476778866638,
+      "grad_norm": 14.5,
+      "learning_rate": 4.0860386795942135e-07,
+      "loss": 0.5079,
+      "step": 17278
+    },
+    {
+      "epoch": 1.4724328930549637,
+      "grad_norm": 19.875,
+      "learning_rate": 4.084796072078442e-07,
+      "loss": 0.6159,
+      "step": 17279
+    },
+    {
+      "epoch": 1.4725181082232637,
+      "grad_norm": 18.75,
+      "learning_rate": 4.083553616633869e-07,
+      "loss": 0.7739,
+      "step": 17280
+    },
+    {
+      "epoch": 1.4726033233915636,
+      "grad_norm": 14.375,
+      "learning_rate": 4.082311313282941e-07,
+      "loss": 0.4308,
+      "step": 17281
+    },
+    {
+      "epoch": 1.4726885385598636,
+      "grad_norm": 14.125,
+      "learning_rate": 4.081069162048111e-07,
+      "loss": 0.4191,
+      "step": 17282
+    },
+    {
+      "epoch": 1.4727737537281635,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.0798271629518287e-07,
+      "loss": 0.4707,
+      "step": 17283
+    },
+    {
+      "epoch": 1.4728589688964635,
+      "grad_norm": 17.25,
+      "learning_rate": 4.0785853160165393e-07,
+      "loss": 0.746,
+      "step": 17284
+    },
+    {
+      "epoch": 1.4729441840647635,
+      "grad_norm": 16.5,
+      "learning_rate": 4.077343621264679e-07,
+      "loss": 0.6376,
+      "step": 17285
+    },
+    {
+      "epoch": 1.4730293992330634,
+      "grad_norm": 16.5,
+      "learning_rate": 4.0761020787186884e-07,
+      "loss": 0.882,
+      "step": 17286
+    },
+    {
+      "epoch": 1.4731146144013634,
+      "grad_norm": 14.375,
+      "learning_rate": 4.074860688401008e-07,
+      "loss": 0.7019,
+      "step": 17287
+    },
+    {
+      "epoch": 1.4731998295696633,
+      "grad_norm": 17.25,
+      "learning_rate": 4.073619450334068e-07,
+      "loss": 0.7243,
+      "step": 17288
+    },
+    {
+      "epoch": 1.4732850447379633,
+      "grad_norm": 24.5,
+      "learning_rate": 4.0723783645402927e-07,
+      "loss": 0.8805,
+      "step": 17289
+    },
+    {
+      "epoch": 1.4733702599062632,
+      "grad_norm": 12.875,
+      "learning_rate": 4.071137431042116e-07,
+      "loss": 0.4146,
+      "step": 17290
+    },
+    {
+      "epoch": 1.4734554750745632,
+      "grad_norm": 20.625,
+      "learning_rate": 4.069896649861961e-07,
+      "loss": 0.7767,
+      "step": 17291
+    },
+    {
+      "epoch": 1.4735406902428632,
+      "grad_norm": 19.625,
+      "learning_rate": 4.0686560210222547e-07,
+      "loss": 0.7326,
+      "step": 17292
+    },
+    {
+      "epoch": 1.4736259054111631,
+      "grad_norm": 25.875,
+      "learning_rate": 4.0674155445454083e-07,
+      "loss": 0.8807,
+      "step": 17293
+    },
+    {
+      "epoch": 1.473711120579463,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.0661752204538427e-07,
+      "loss": 0.3739,
+      "step": 17294
+    },
+    {
+      "epoch": 1.473796335747763,
+      "grad_norm": 12.375,
+      "learning_rate": 4.0649350487699757e-07,
+      "loss": 0.3865,
+      "step": 17295
+    },
+    {
+      "epoch": 1.473881550916063,
+      "grad_norm": 16.625,
+      "learning_rate": 4.0636950295162113e-07,
+      "loss": 0.5387,
+      "step": 17296
+    },
+    {
+      "epoch": 1.473966766084363,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.062455162714965e-07,
+      "loss": 0.3619,
+      "step": 17297
+    },
+    {
+      "epoch": 1.474051981252663,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.0612154483886334e-07,
+      "loss": 0.3844,
+      "step": 17298
+    },
+    {
+      "epoch": 1.4741371964209629,
+      "grad_norm": 17.375,
+      "learning_rate": 4.0599758865596264e-07,
+      "loss": 0.7273,
+      "step": 17299
+    },
+    {
+      "epoch": 1.4742224115892628,
+      "grad_norm": 12.375,
+      "learning_rate": 4.05873647725034e-07,
+      "loss": 0.5801,
+      "step": 17300
+    },
+    {
+      "epoch": 1.4743076267575628,
+      "grad_norm": 11.875,
+      "learning_rate": 4.057497220483179e-07,
+      "loss": 0.4331,
+      "step": 17301
+    },
+    {
+      "epoch": 1.4743928419258627,
+      "grad_norm": 17.75,
+      "learning_rate": 4.0562581162805287e-07,
+      "loss": 0.5393,
+      "step": 17302
+    },
+    {
+      "epoch": 1.4744780570941627,
+      "grad_norm": 14.25,
+      "learning_rate": 4.055019164664789e-07,
+      "loss": 0.7151,
+      "step": 17303
+    },
+    {
+      "epoch": 1.4745632722624626,
+      "grad_norm": 20.125,
+      "learning_rate": 4.053780365658341e-07,
+      "loss": 0.7055,
+      "step": 17304
+    },
+    {
+      "epoch": 1.4746484874307626,
+      "grad_norm": 21.625,
+      "learning_rate": 4.05254171928358e-07,
+      "loss": 1.0267,
+      "step": 17305
+    },
+    {
+      "epoch": 1.4747337025990626,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.05130322556288e-07,
+      "loss": 0.4165,
+      "step": 17306
+    },
+    {
+      "epoch": 1.4748189177673625,
+      "grad_norm": 16.625,
+      "learning_rate": 4.0500648845186255e-07,
+      "loss": 0.7469,
+      "step": 17307
+    },
+    {
+      "epoch": 1.4749041329356625,
+      "grad_norm": 11.25,
+      "learning_rate": 4.048826696173197e-07,
+      "loss": 0.3703,
+      "step": 17308
+    },
+    {
+      "epoch": 1.4749893481039624,
+      "grad_norm": 13.5,
+      "learning_rate": 4.047588660548972e-07,
+      "loss": 0.4605,
+      "step": 17309
+    },
+    {
+      "epoch": 1.4750745632722624,
+      "grad_norm": 13.375,
+      "learning_rate": 4.046350777668316e-07,
+      "loss": 0.5429,
+      "step": 17310
+    },
+    {
+      "epoch": 1.4751597784405623,
+      "grad_norm": 23.125,
+      "learning_rate": 4.045113047553606e-07,
+      "loss": 0.8238,
+      "step": 17311
+    },
+    {
+      "epoch": 1.4752449936088623,
+      "grad_norm": 11.75,
+      "learning_rate": 4.0438754702272e-07,
+      "loss": 0.2821,
+      "step": 17312
+    },
+    {
+      "epoch": 1.4753302087771623,
+      "grad_norm": 12.5,
+      "learning_rate": 4.0426380457114693e-07,
+      "loss": 0.5182,
+      "step": 17313
+    },
+    {
+      "epoch": 1.4754154239454622,
+      "grad_norm": 16.625,
+      "learning_rate": 4.0414007740287765e-07,
+      "loss": 0.6123,
+      "step": 17314
+    },
+    {
+      "epoch": 1.4755006391137622,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.0401636552014746e-07,
+      "loss": 0.6747,
+      "step": 17315
+    },
+    {
+      "epoch": 1.4755858542820621,
+      "grad_norm": 13.625,
+      "learning_rate": 4.0389266892519205e-07,
+      "loss": 0.345,
+      "step": 17316
+    },
+    {
+      "epoch": 1.475671069450362,
+      "grad_norm": 27.0,
+      "learning_rate": 4.037689876202472e-07,
+      "loss": 0.3827,
+      "step": 17317
+    },
+    {
+      "epoch": 1.475756284618662,
+      "grad_norm": 19.25,
+      "learning_rate": 4.0364532160754795e-07,
+      "loss": 0.178,
+      "step": 17318
+    },
+    {
+      "epoch": 1.475841499786962,
+      "grad_norm": 12.5,
+      "learning_rate": 4.035216708893289e-07,
+      "loss": 0.4519,
+      "step": 17319
+    },
+    {
+      "epoch": 1.475926714955262,
+      "grad_norm": 12.0625,
+      "learning_rate": 4.0339803546782396e-07,
+      "loss": 0.3261,
+      "step": 17320
+    },
+    {
+      "epoch": 1.476011930123562,
+      "grad_norm": 21.75,
+      "learning_rate": 4.032744153452678e-07,
+      "loss": 0.6774,
+      "step": 17321
+    },
+    {
+      "epoch": 1.4760971452918619,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.031508105238949e-07,
+      "loss": 0.6829,
+      "step": 17322
+    },
+    {
+      "epoch": 1.4761823604601618,
+      "grad_norm": 19.75,
+      "learning_rate": 4.0302722100593797e-07,
+      "loss": 0.5042,
+      "step": 17323
+    },
+    {
+      "epoch": 1.4762675756284618,
+      "grad_norm": 20.25,
+      "learning_rate": 4.029036467936308e-07,
+      "loss": 0.6105,
+      "step": 17324
+    },
+    {
+      "epoch": 1.4763527907967617,
+      "grad_norm": 11.875,
+      "learning_rate": 4.027800878892067e-07,
+      "loss": 0.257,
+      "step": 17325
+    },
+    {
+      "epoch": 1.4764380059650617,
+      "grad_norm": 17.5,
+      "learning_rate": 4.0265654429489864e-07,
+      "loss": 0.5321,
+      "step": 17326
+    },
+    {
+      "epoch": 1.4765232211333617,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.0253301601293856e-07,
+      "loss": 0.4372,
+      "step": 17327
+    },
+    {
+      "epoch": 1.4766084363016616,
+      "grad_norm": 13.375,
+      "learning_rate": 4.0240950304555953e-07,
+      "loss": 0.4607,
+      "step": 17328
+    },
+    {
+      "epoch": 1.4766936514699616,
+      "grad_norm": 13.5,
+      "learning_rate": 4.022860053949926e-07,
+      "loss": 0.5283,
+      "step": 17329
+    },
+    {
+      "epoch": 1.4767788666382615,
+      "grad_norm": 17.5,
+      "learning_rate": 4.0216252306347e-07,
+      "loss": 0.6731,
+      "step": 17330
+    },
+    {
+      "epoch": 1.4768640818065615,
+      "grad_norm": 21.5,
+      "learning_rate": 4.020390560532236e-07,
+      "loss": 0.5663,
+      "step": 17331
+    },
+    {
+      "epoch": 1.4769492969748614,
+      "grad_norm": 11.5625,
+      "learning_rate": 4.0191560436648376e-07,
+      "loss": 0.3791,
+      "step": 17332
+    },
+    {
+      "epoch": 1.4770345121431614,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.0179216800548176e-07,
+      "loss": 0.4408,
+      "step": 17333
+    },
+    {
+      "epoch": 1.4771197273114614,
+      "grad_norm": 16.25,
+      "learning_rate": 4.016687469724481e-07,
+      "loss": 0.7263,
+      "step": 17334
+    },
+    {
+      "epoch": 1.4772049424797613,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.015453412696138e-07,
+      "loss": 0.6094,
+      "step": 17335
+    },
+    {
+      "epoch": 1.4772901576480613,
+      "grad_norm": 12.125,
+      "learning_rate": 4.014219508992083e-07,
+      "loss": 0.3427,
+      "step": 17336
+    },
+    {
+      "epoch": 1.4773753728163612,
+      "grad_norm": 16.0,
+      "learning_rate": 4.012985758634612e-07,
+      "loss": 0.7118,
+      "step": 17337
+    },
+    {
+      "epoch": 1.4774605879846612,
+      "grad_norm": 13.875,
+      "learning_rate": 4.0117521616460207e-07,
+      "loss": 0.4285,
+      "step": 17338
+    },
+    {
+      "epoch": 1.4775458031529611,
+      "grad_norm": 16.375,
+      "learning_rate": 4.010518718048609e-07,
+      "loss": 0.7916,
+      "step": 17339
+    },
+    {
+      "epoch": 1.477631018321261,
+      "grad_norm": 14.75,
+      "learning_rate": 4.009285427864655e-07,
+      "loss": 0.5064,
+      "step": 17340
+    },
+    {
+      "epoch": 1.477716233489561,
+      "grad_norm": 13.875,
+      "learning_rate": 4.008052291116453e-07,
+      "loss": 0.6246,
+      "step": 17341
+    },
+    {
+      "epoch": 1.477801448657861,
+      "grad_norm": 10.5,
+      "learning_rate": 4.0068193078262845e-07,
+      "loss": 0.3465,
+      "step": 17342
+    },
+    {
+      "epoch": 1.477886663826161,
+      "grad_norm": 14.75,
+      "learning_rate": 4.005586478016436e-07,
+      "loss": 0.3951,
+      "step": 17343
+    },
+    {
+      "epoch": 1.477971878994461,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.004353801709182e-07,
+      "loss": 0.2934,
+      "step": 17344
+    },
+    {
+      "epoch": 1.4780570941627609,
+      "grad_norm": 20.0,
+      "learning_rate": 4.0031212789267945e-07,
+      "loss": 0.6358,
+      "step": 17345
+    },
+    {
+      "epoch": 1.4781423093310608,
+      "grad_norm": 28.0,
+      "learning_rate": 4.001888909691548e-07,
+      "loss": 1.1585,
+      "step": 17346
+    },
+    {
+      "epoch": 1.4782275244993608,
+      "grad_norm": 17.25,
+      "learning_rate": 4.000656694025716e-07,
+      "loss": 0.6532,
+      "step": 17347
+    },
+    {
+      "epoch": 1.4783127396676607,
+      "grad_norm": 23.125,
+      "learning_rate": 3.9994246319515677e-07,
+      "loss": 1.0379,
+      "step": 17348
+    },
+    {
+      "epoch": 1.4783979548359607,
+      "grad_norm": 12.0,
+      "learning_rate": 3.998192723491359e-07,
+      "loss": 0.457,
+      "step": 17349
+    },
+    {
+      "epoch": 1.4784831700042607,
+      "grad_norm": 15.875,
+      "learning_rate": 3.9969609686673587e-07,
+      "loss": 0.4298,
+      "step": 17350
+    },
+    {
+      "epoch": 1.4785683851725606,
+      "grad_norm": 15.0,
+      "learning_rate": 3.9957293675018244e-07,
+      "loss": 0.3518,
+      "step": 17351
+    },
+    {
+      "epoch": 1.4786536003408606,
+      "grad_norm": 13.375,
+      "learning_rate": 3.9944979200170193e-07,
+      "loss": 0.3844,
+      "step": 17352
+    },
+    {
+      "epoch": 1.4787388155091605,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.9932666262351814e-07,
+      "loss": 0.3707,
+      "step": 17353
+    },
+    {
+      "epoch": 1.4788240306774605,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.9920354861785705e-07,
+      "loss": 0.6726,
+      "step": 17354
+    },
+    {
+      "epoch": 1.4789092458457604,
+      "grad_norm": 18.75,
+      "learning_rate": 3.990804499869433e-07,
+      "loss": 0.7034,
+      "step": 17355
+    },
+    {
+      "epoch": 1.4789944610140604,
+      "grad_norm": 16.5,
+      "learning_rate": 3.98957366733002e-07,
+      "loss": 0.4547,
+      "step": 17356
+    },
+    {
+      "epoch": 1.4790796761823604,
+      "grad_norm": 17.5,
+      "learning_rate": 3.9883429885825635e-07,
+      "loss": 0.7266,
+      "step": 17357
+    },
+    {
+      "epoch": 1.4791648913506603,
+      "grad_norm": 15.875,
+      "learning_rate": 3.9871124636493093e-07,
+      "loss": 0.4778,
+      "step": 17358
+    },
+    {
+      "epoch": 1.4792501065189603,
+      "grad_norm": 19.25,
+      "learning_rate": 3.985882092552493e-07,
+      "loss": 0.6309,
+      "step": 17359
+    },
+    {
+      "epoch": 1.4793353216872602,
+      "grad_norm": 17.125,
+      "learning_rate": 3.984651875314352e-07,
+      "loss": 0.7441,
+      "step": 17360
+    },
+    {
+      "epoch": 1.4794205368555602,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.9834218119571154e-07,
+      "loss": 0.5085,
+      "step": 17361
+    },
+    {
+      "epoch": 1.4795057520238601,
+      "grad_norm": 14.625,
+      "learning_rate": 3.982191902503006e-07,
+      "loss": 0.4947,
+      "step": 17362
+    },
+    {
+      "epoch": 1.47959096719216,
+      "grad_norm": 17.625,
+      "learning_rate": 3.980962146974254e-07,
+      "loss": 0.6811,
+      "step": 17363
+    },
+    {
+      "epoch": 1.47967618236046,
+      "grad_norm": 20.5,
+      "learning_rate": 3.9797325453930837e-07,
+      "loss": 0.732,
+      "step": 17364
+    },
+    {
+      "epoch": 1.47976139752876,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.9785030977817175e-07,
+      "loss": 0.3419,
+      "step": 17365
+    },
+    {
+      "epoch": 1.47984661269706,
+      "grad_norm": 14.375,
+      "learning_rate": 3.977273804162365e-07,
+      "loss": 0.575,
+      "step": 17366
+    },
+    {
+      "epoch": 1.47993182786536,
+      "grad_norm": 14.875,
+      "learning_rate": 3.9760446645572465e-07,
+      "loss": 0.7468,
+      "step": 17367
+    },
+    {
+      "epoch": 1.4800170430336599,
+      "grad_norm": 13.5,
+      "learning_rate": 3.9748156789885766e-07,
+      "loss": 0.3805,
+      "step": 17368
+    },
+    {
+      "epoch": 1.4801022582019598,
+      "grad_norm": 14.375,
+      "learning_rate": 3.973586847478558e-07,
+      "loss": 0.469,
+      "step": 17369
+    },
+    {
+      "epoch": 1.4801874733702598,
+      "grad_norm": 16.75,
+      "learning_rate": 3.972358170049397e-07,
+      "loss": 0.6803,
+      "step": 17370
+    },
+    {
+      "epoch": 1.4802726885385598,
+      "grad_norm": 18.875,
+      "learning_rate": 3.971129646723297e-07,
+      "loss": 0.9482,
+      "step": 17371
+    },
+    {
+      "epoch": 1.4803579037068597,
+      "grad_norm": 17.5,
+      "learning_rate": 3.969901277522463e-07,
+      "loss": 0.8263,
+      "step": 17372
+    },
+    {
+      "epoch": 1.4804431188751597,
+      "grad_norm": 51.75,
+      "learning_rate": 3.968673062469093e-07,
+      "loss": 1.3426,
+      "step": 17373
+    },
+    {
+      "epoch": 1.4805283340434596,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.967445001585377e-07,
+      "loss": 0.5326,
+      "step": 17374
+    },
+    {
+      "epoch": 1.4806135492117596,
+      "grad_norm": 16.125,
+      "learning_rate": 3.966217094893508e-07,
+      "loss": 0.3408,
+      "step": 17375
+    },
+    {
+      "epoch": 1.4806987643800595,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.9649893424156827e-07,
+      "loss": 0.5664,
+      "step": 17376
+    },
+    {
+      "epoch": 1.4807839795483595,
+      "grad_norm": 17.625,
+      "learning_rate": 3.963761744174079e-07,
+      "loss": 0.4569,
+      "step": 17377
+    },
+    {
+      "epoch": 1.4808691947166595,
+      "grad_norm": 16.25,
+      "learning_rate": 3.962534300190887e-07,
+      "loss": 0.8735,
+      "step": 17378
+    },
+    {
+      "epoch": 1.4809544098849594,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.9613070104882826e-07,
+      "loss": 0.5867,
+      "step": 17379
+    },
+    {
+      "epoch": 1.4810396250532594,
+      "grad_norm": 17.5,
+      "learning_rate": 3.9600798750884463e-07,
+      "loss": 0.6037,
+      "step": 17380
+    },
+    {
+      "epoch": 1.4811248402215593,
+      "grad_norm": 19.125,
+      "learning_rate": 3.9588528940135554e-07,
+      "loss": 0.6245,
+      "step": 17381
+    },
+    {
+      "epoch": 1.4812100553898593,
+      "grad_norm": 12.75,
+      "learning_rate": 3.9576260672857843e-07,
+      "loss": 0.2978,
+      "step": 17382
+    },
+    {
+      "epoch": 1.4812952705581592,
+      "grad_norm": 19.75,
+      "learning_rate": 3.956399394927296e-07,
+      "loss": 0.7487,
+      "step": 17383
+    },
+    {
+      "epoch": 1.4813804857264592,
+      "grad_norm": 17.0,
+      "learning_rate": 3.9551728769602655e-07,
+      "loss": 0.75,
+      "step": 17384
+    },
+    {
+      "epoch": 1.4814657008947592,
+      "grad_norm": 13.25,
+      "learning_rate": 3.9539465134068505e-07,
+      "loss": 0.5412,
+      "step": 17385
+    },
+    {
+      "epoch": 1.481550916063059,
+      "grad_norm": 16.375,
+      "learning_rate": 3.9527203042892205e-07,
+      "loss": 0.9773,
+      "step": 17386
+    },
+    {
+      "epoch": 1.481636131231359,
+      "grad_norm": 13.375,
+      "learning_rate": 3.9514942496295244e-07,
+      "loss": 0.6105,
+      "step": 17387
+    },
+    {
+      "epoch": 1.481721346399659,
+      "grad_norm": 12.3125,
+      "learning_rate": 3.950268349449923e-07,
+      "loss": 0.3127,
+      "step": 17388
+    },
+    {
+      "epoch": 1.481806561567959,
+      "grad_norm": 18.875,
+      "learning_rate": 3.949042603772571e-07,
+      "loss": 0.6285,
+      "step": 17389
+    },
+    {
+      "epoch": 1.481891776736259,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.9478170126196217e-07,
+      "loss": 0.6967,
+      "step": 17390
+    },
+    {
+      "epoch": 1.481976991904559,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.946591576013215e-07,
+      "loss": 0.7812,
+      "step": 17391
+    },
+    {
+      "epoch": 1.4820622070728589,
+      "grad_norm": 16.75,
+      "learning_rate": 3.9453662939754987e-07,
+      "loss": 0.4848,
+      "step": 17392
+    },
+    {
+      "epoch": 1.4821474222411588,
+      "grad_norm": 29.75,
+      "learning_rate": 3.94414116652862e-07,
+      "loss": 0.7147,
+      "step": 17393
+    },
+    {
+      "epoch": 1.4822326374094588,
+      "grad_norm": 15.875,
+      "learning_rate": 3.9429161936947113e-07,
+      "loss": 0.2916,
+      "step": 17394
+    },
+    {
+      "epoch": 1.4823178525777587,
+      "grad_norm": 12.9375,
+      "learning_rate": 3.941691375495914e-07,
+      "loss": 0.4166,
+      "step": 17395
+    },
+    {
+      "epoch": 1.4824030677460587,
+      "grad_norm": 18.75,
+      "learning_rate": 3.9404667119543565e-07,
+      "loss": 0.6084,
+      "step": 17396
+    },
+    {
+      "epoch": 1.4824882829143586,
+      "grad_norm": 12.5,
+      "learning_rate": 3.9392422030921724e-07,
+      "loss": 0.5985,
+      "step": 17397
+    },
+    {
+      "epoch": 1.4825734980826586,
+      "grad_norm": 13.9375,
+      "learning_rate": 3.938017848931494e-07,
+      "loss": 0.5318,
+      "step": 17398
+    },
+    {
+      "epoch": 1.4826587132509585,
+      "grad_norm": 15.625,
+      "learning_rate": 3.93679364949444e-07,
+      "loss": 0.4099,
+      "step": 17399
+    },
+    {
+      "epoch": 1.4827439284192585,
+      "grad_norm": 13.25,
+      "learning_rate": 3.935569604803134e-07,
+      "loss": 0.3588,
+      "step": 17400
+    },
+    {
+      "epoch": 1.4828291435875585,
+      "grad_norm": 17.75,
+      "learning_rate": 3.9343457148797034e-07,
+      "loss": 0.4108,
+      "step": 17401
+    },
+    {
+      "epoch": 1.4829143587558584,
+      "grad_norm": 16.125,
+      "learning_rate": 3.9331219797462543e-07,
+      "loss": 0.4985,
+      "step": 17402
+    },
+    {
+      "epoch": 1.4829995739241584,
+      "grad_norm": 20.875,
+      "learning_rate": 3.931898399424909e-07,
+      "loss": 0.7963,
+      "step": 17403
+    },
+    {
+      "epoch": 1.4830847890924583,
+      "grad_norm": 16.125,
+      "learning_rate": 3.930674973937773e-07,
+      "loss": 0.6314,
+      "step": 17404
+    },
+    {
+      "epoch": 1.4831700042607583,
+      "grad_norm": 16.875,
+      "learning_rate": 3.929451703306957e-07,
+      "loss": 0.9827,
+      "step": 17405
+    },
+    {
+      "epoch": 1.4832552194290582,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.9282285875545655e-07,
+      "loss": 0.6056,
+      "step": 17406
+    },
+    {
+      "epoch": 1.4833404345973582,
+      "grad_norm": 13.75,
+      "learning_rate": 3.927005626702708e-07,
+      "loss": 0.4519,
+      "step": 17407
+    },
+    {
+      "epoch": 1.4834256497656582,
+      "grad_norm": 31.0,
+      "learning_rate": 3.9257828207734756e-07,
+      "loss": 0.8683,
+      "step": 17408
+    },
+    {
+      "epoch": 1.4835108649339581,
+      "grad_norm": 17.875,
+      "learning_rate": 3.924560169788974e-07,
+      "loss": 0.5357,
+      "step": 17409
+    },
+    {
+      "epoch": 1.483596080102258,
+      "grad_norm": 17.0,
+      "learning_rate": 3.9233376737712895e-07,
+      "loss": 0.7233,
+      "step": 17410
+    },
+    {
+      "epoch": 1.483681295270558,
+      "grad_norm": 14.125,
+      "learning_rate": 3.922115332742517e-07,
+      "loss": 0.4108,
+      "step": 17411
+    },
+    {
+      "epoch": 1.483766510438858,
+      "grad_norm": 16.5,
+      "learning_rate": 3.920893146724751e-07,
+      "loss": 0.5881,
+      "step": 17412
+    },
+    {
+      "epoch": 1.483851725607158,
+      "grad_norm": 17.875,
+      "learning_rate": 3.9196711157400674e-07,
+      "loss": 0.6182,
+      "step": 17413
+    },
+    {
+      "epoch": 1.483936940775458,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.9184492398105565e-07,
+      "loss": 0.4522,
+      "step": 17414
+    },
+    {
+      "epoch": 1.4840221559437579,
+      "grad_norm": 13.4375,
+      "learning_rate": 3.9172275189583e-07,
+      "loss": 0.6417,
+      "step": 17415
+    },
+    {
+      "epoch": 1.4841073711120578,
+      "grad_norm": 16.0,
+      "learning_rate": 3.9160059532053693e-07,
+      "loss": 0.4736,
+      "step": 17416
+    },
+    {
+      "epoch": 1.4841925862803578,
+      "grad_norm": 18.125,
+      "learning_rate": 3.9147845425738474e-07,
+      "loss": 0.6005,
+      "step": 17417
+    },
+    {
+      "epoch": 1.4842778014486577,
+      "grad_norm": 19.125,
+      "learning_rate": 3.9135632870857975e-07,
+      "loss": 0.8761,
+      "step": 17418
+    },
+    {
+      "epoch": 1.4843630166169577,
+      "grad_norm": 16.75,
+      "learning_rate": 3.9123421867632933e-07,
+      "loss": 0.7302,
+      "step": 17419
+    },
+    {
+      "epoch": 1.4844482317852576,
+      "grad_norm": 13.375,
+      "learning_rate": 3.911121241628406e-07,
+      "loss": 0.3297,
+      "step": 17420
+    },
+    {
+      "epoch": 1.4845334469535576,
+      "grad_norm": 10.75,
+      "learning_rate": 3.90990045170319e-07,
+      "loss": 0.2252,
+      "step": 17421
+    },
+    {
+      "epoch": 1.4846186621218576,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.908679817009711e-07,
+      "loss": 0.5195,
+      "step": 17422
+    },
+    {
+      "epoch": 1.4847038772901575,
+      "grad_norm": 16.5,
+      "learning_rate": 3.907459337570027e-07,
+      "loss": 0.5478,
+      "step": 17423
+    },
+    {
+      "epoch": 1.4847890924584575,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.9062390134061974e-07,
+      "loss": 0.4883,
+      "step": 17424
+    },
+    {
+      "epoch": 1.4848743076267574,
+      "grad_norm": 22.0,
+      "learning_rate": 3.9050188445402704e-07,
+      "loss": 0.4488,
+      "step": 17425
+    },
+    {
+      "epoch": 1.4849595227950574,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.9037988309942915e-07,
+      "loss": 0.8827,
+      "step": 17426
+    },
+    {
+      "epoch": 1.4850447379633573,
+      "grad_norm": 10.875,
+      "learning_rate": 3.9025789727903123e-07,
+      "loss": 0.2778,
+      "step": 17427
+    },
+    {
+      "epoch": 1.4851299531316573,
+      "grad_norm": 13.875,
+      "learning_rate": 3.9013592699503805e-07,
+      "loss": 0.4557,
+      "step": 17428
+    },
+    {
+      "epoch": 1.4852151682999573,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.9001397224965295e-07,
+      "loss": 0.4385,
+      "step": 17429
+    },
+    {
+      "epoch": 1.4853003834682572,
+      "grad_norm": 16.375,
+      "learning_rate": 3.8989203304508007e-07,
+      "loss": 0.8618,
+      "step": 17430
+    },
+    {
+      "epoch": 1.4853855986365572,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.8977010938352307e-07,
+      "loss": 0.5813,
+      "step": 17431
+    },
+    {
+      "epoch": 1.4854708138048571,
+      "grad_norm": 14.5,
+      "learning_rate": 3.896482012671857e-07,
+      "loss": 0.6423,
+      "step": 17432
+    },
+    {
+      "epoch": 1.485556028973157,
+      "grad_norm": 15.125,
+      "learning_rate": 3.8952630869827037e-07,
+      "loss": 0.5752,
+      "step": 17433
+    },
+    {
+      "epoch": 1.4856412441414573,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.8940443167897956e-07,
+      "loss": 0.6695,
+      "step": 17434
+    },
+    {
+      "epoch": 1.4857264593097572,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.8928257021151605e-07,
+      "loss": 0.531,
+      "step": 17435
+    },
+    {
+      "epoch": 1.4858116744780572,
+      "grad_norm": 15.125,
+      "learning_rate": 3.89160724298082e-07,
+      "loss": 0.5529,
+      "step": 17436
+    },
+    {
+      "epoch": 1.4858968896463571,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.8903889394087986e-07,
+      "loss": 0.3,
+      "step": 17437
+    },
+    {
+      "epoch": 1.485982104814657,
+      "grad_norm": 15.75,
+      "learning_rate": 3.889170791421101e-07,
+      "loss": 0.4806,
+      "step": 17438
+    },
+    {
+      "epoch": 1.486067319982957,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.8879527990397475e-07,
+      "loss": 0.6601,
+      "step": 17439
+    },
+    {
+      "epoch": 1.486152535151257,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.886734962286745e-07,
+      "loss": 0.6116,
+      "step": 17440
+    },
+    {
+      "epoch": 1.486237750319557,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.885517281184109e-07,
+      "loss": 0.634,
+      "step": 17441
+    },
+    {
+      "epoch": 1.486322965487857,
+      "grad_norm": 18.25,
+      "learning_rate": 3.884299755753837e-07,
+      "loss": 0.8881,
+      "step": 17442
+    },
+    {
+      "epoch": 1.4864081806561569,
+      "grad_norm": 18.5,
+      "learning_rate": 3.883082386017928e-07,
+      "loss": 0.5247,
+      "step": 17443
+    },
+    {
+      "epoch": 1.4864933958244568,
+      "grad_norm": 14.625,
+      "learning_rate": 3.8818651719983863e-07,
+      "loss": 0.5927,
+      "step": 17444
+    },
+    {
+      "epoch": 1.4865786109927568,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.8806481137172113e-07,
+      "loss": 0.6384,
+      "step": 17445
+    },
+    {
+      "epoch": 1.4866638261610567,
+      "grad_norm": 19.5,
+      "learning_rate": 3.879431211196387e-07,
+      "loss": 0.8251,
+      "step": 17446
+    },
+    {
+      "epoch": 1.4867490413293567,
+      "grad_norm": 14.0,
+      "learning_rate": 3.8782144644579105e-07,
+      "loss": 0.5723,
+      "step": 17447
+    },
+    {
+      "epoch": 1.4868342564976567,
+      "grad_norm": 14.75,
+      "learning_rate": 3.8769978735237675e-07,
+      "loss": 0.6051,
+      "step": 17448
+    },
+    {
+      "epoch": 1.4869194716659566,
+      "grad_norm": 21.25,
+      "learning_rate": 3.8757814384159486e-07,
+      "loss": 0.7784,
+      "step": 17449
+    },
+    {
+      "epoch": 1.4870046868342566,
+      "grad_norm": 12.9375,
+      "learning_rate": 3.874565159156432e-07,
+      "loss": 0.8535,
+      "step": 17450
+    },
+    {
+      "epoch": 1.4870899020025565,
+      "grad_norm": 12.0,
+      "learning_rate": 3.873349035767192e-07,
+      "loss": 0.326,
+      "step": 17451
+    },
+    {
+      "epoch": 1.4871751171708565,
+      "grad_norm": 14.875,
+      "learning_rate": 3.872133068270209e-07,
+      "loss": 0.298,
+      "step": 17452
+    },
+    {
+      "epoch": 1.4872603323391564,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.8709172566874576e-07,
+      "loss": 0.7943,
+      "step": 17453
+    },
+    {
+      "epoch": 1.4873455475074564,
+      "grad_norm": 15.75,
+      "learning_rate": 3.869701601040914e-07,
+      "loss": 0.6357,
+      "step": 17454
+    },
+    {
+      "epoch": 1.4874307626757564,
+      "grad_norm": 16.5,
+      "learning_rate": 3.868486101352537e-07,
+      "loss": 0.5098,
+      "step": 17455
+    },
+    {
+      "epoch": 1.4875159778440563,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.867270757644294e-07,
+      "loss": 0.5143,
+      "step": 17456
+    },
+    {
+      "epoch": 1.4876011930123563,
+      "grad_norm": 17.25,
+      "learning_rate": 3.866055569938154e-07,
+      "loss": 0.8338,
+      "step": 17457
+    },
+    {
+      "epoch": 1.4876864081806562,
+      "grad_norm": 17.875,
+      "learning_rate": 3.864840538256072e-07,
+      "loss": 0.7286,
+      "step": 17458
+    },
+    {
+      "epoch": 1.4877716233489562,
+      "grad_norm": 20.375,
+      "learning_rate": 3.8636256626200007e-07,
+      "loss": 0.7703,
+      "step": 17459
+    },
+    {
+      "epoch": 1.4878568385172561,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.862410943051899e-07,
+      "loss": 0.2893,
+      "step": 17460
+    },
+    {
+      "epoch": 1.487942053685556,
+      "grad_norm": 22.5,
+      "learning_rate": 3.861196379573716e-07,
+      "loss": 0.7677,
+      "step": 17461
+    },
+    {
+      "epoch": 1.488027268853856,
+      "grad_norm": 18.125,
+      "learning_rate": 3.8599819722074066e-07,
+      "loss": 0.591,
+      "step": 17462
+    },
+    {
+      "epoch": 1.488112484022156,
+      "grad_norm": 17.5,
+      "learning_rate": 3.858767720974907e-07,
+      "loss": 0.7926,
+      "step": 17463
+    },
+    {
+      "epoch": 1.488197699190456,
+      "grad_norm": 16.75,
+      "learning_rate": 3.8575536258981624e-07,
+      "loss": 0.8462,
+      "step": 17464
+    },
+    {
+      "epoch": 1.488282914358756,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.856339686999117e-07,
+      "loss": 0.512,
+      "step": 17465
+    },
+    {
+      "epoch": 1.4883681295270559,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.855125904299707e-07,
+      "loss": 0.6381,
+      "step": 17466
+    },
+    {
+      "epoch": 1.4884533446953558,
+      "grad_norm": 17.375,
+      "learning_rate": 3.853912277821867e-07,
+      "loss": 0.7897,
+      "step": 17467
+    },
+    {
+      "epoch": 1.4885385598636558,
+      "grad_norm": 20.0,
+      "learning_rate": 3.852698807587521e-07,
+      "loss": 0.9579,
+      "step": 17468
+    },
+    {
+      "epoch": 1.4886237750319558,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.8514854936186037e-07,
+      "loss": 0.8539,
+      "step": 17469
+    },
+    {
+      "epoch": 1.4887089902002557,
+      "grad_norm": 14.625,
+      "learning_rate": 3.85027233593704e-07,
+      "loss": 0.7443,
+      "step": 17470
+    },
+    {
+      "epoch": 1.4887942053685557,
+      "grad_norm": 20.25,
+      "learning_rate": 3.849059334564759e-07,
+      "loss": 0.5436,
+      "step": 17471
+    },
+    {
+      "epoch": 1.4888794205368556,
+      "grad_norm": 35.5,
+      "learning_rate": 3.8478464895236685e-07,
+      "loss": 0.8486,
+      "step": 17472
+    },
+    {
+      "epoch": 1.4889646357051556,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.846633800835693e-07,
+      "loss": 0.6118,
+      "step": 17473
+    },
+    {
+      "epoch": 1.4890498508734555,
+      "grad_norm": 13.125,
+      "learning_rate": 3.8454212685227503e-07,
+      "loss": 0.4509,
+      "step": 17474
+    },
+    {
+      "epoch": 1.4891350660417555,
+      "grad_norm": 15.5,
+      "learning_rate": 3.844208892606749e-07,
+      "loss": 0.3828,
+      "step": 17475
+    },
+    {
+      "epoch": 1.4892202812100555,
+      "grad_norm": 19.125,
+      "learning_rate": 3.842996673109593e-07,
+      "loss": 1.0877,
+      "step": 17476
+    },
+    {
+      "epoch": 1.4893054963783554,
+      "grad_norm": 14.375,
+      "learning_rate": 3.841784610053191e-07,
+      "loss": 0.403,
+      "step": 17477
+    },
+    {
+      "epoch": 1.4893907115466554,
+      "grad_norm": 16.75,
+      "learning_rate": 3.84057270345945e-07,
+      "loss": 0.6259,
+      "step": 17478
+    },
+    {
+      "epoch": 1.4894759267149553,
+      "grad_norm": 15.625,
+      "learning_rate": 3.83936095335027e-07,
+      "loss": 0.4996,
+      "step": 17479
+    },
+    {
+      "epoch": 1.4895611418832553,
+      "grad_norm": 11.625,
+      "learning_rate": 3.8381493597475424e-07,
+      "loss": 0.4583,
+      "step": 17480
+    },
+    {
+      "epoch": 1.4896463570515552,
+      "grad_norm": 13.25,
+      "learning_rate": 3.836937922673167e-07,
+      "loss": 0.5134,
+      "step": 17481
+    },
+    {
+      "epoch": 1.4897315722198552,
+      "grad_norm": 21.0,
+      "learning_rate": 3.8357266421490387e-07,
+      "loss": 0.6506,
+      "step": 17482
+    },
+    {
+      "epoch": 1.4898167873881552,
+      "grad_norm": 17.25,
+      "learning_rate": 3.8345155181970377e-07,
+      "loss": 0.6902,
+      "step": 17483
+    },
+    {
+      "epoch": 1.489902002556455,
+      "grad_norm": 21.625,
+      "learning_rate": 3.8333045508390607e-07,
+      "loss": 1.1121,
+      "step": 17484
+    },
+    {
+      "epoch": 1.489987217724755,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.83209374009698e-07,
+      "loss": 0.5483,
+      "step": 17485
+    },
+    {
+      "epoch": 1.490072432893055,
+      "grad_norm": 15.125,
+      "learning_rate": 3.8308830859926816e-07,
+      "loss": 0.6629,
+      "step": 17486
+    },
+    {
+      "epoch": 1.490157648061355,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.829672588548045e-07,
+      "loss": 0.5512,
+      "step": 17487
+    },
+    {
+      "epoch": 1.490242863229655,
+      "grad_norm": 17.875,
+      "learning_rate": 3.828462247784946e-07,
+      "loss": 0.6615,
+      "step": 17488
+    },
+    {
+      "epoch": 1.490328078397955,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.8272520637252495e-07,
+      "loss": 0.3605,
+      "step": 17489
+    },
+    {
+      "epoch": 1.4904132935662548,
+      "grad_norm": 14.875,
+      "learning_rate": 3.8260420363908355e-07,
+      "loss": 0.693,
+      "step": 17490
+    },
+    {
+      "epoch": 1.4904985087345548,
+      "grad_norm": 22.625,
+      "learning_rate": 3.8248321658035573e-07,
+      "loss": 0.8082,
+      "step": 17491
+    },
+    {
+      "epoch": 1.4905837239028548,
+      "grad_norm": 31.75,
+      "learning_rate": 3.8236224519852915e-07,
+      "loss": 0.822,
+      "step": 17492
+    },
+    {
+      "epoch": 1.4906689390711547,
+      "grad_norm": 35.75,
+      "learning_rate": 3.8224128949578885e-07,
+      "loss": 0.765,
+      "step": 17493
+    },
+    {
+      "epoch": 1.4907541542394547,
+      "grad_norm": 12.5,
+      "learning_rate": 3.82120349474321e-07,
+      "loss": 0.6235,
+      "step": 17494
+    },
+    {
+      "epoch": 1.4908393694077546,
+      "grad_norm": 63.5,
+      "learning_rate": 3.8199942513631125e-07,
+      "loss": 0.544,
+      "step": 17495
+    },
+    {
+      "epoch": 1.4909245845760546,
+      "grad_norm": 9.875,
+      "learning_rate": 3.81878516483945e-07,
+      "loss": 0.4154,
+      "step": 17496
+    },
+    {
+      "epoch": 1.4910097997443545,
+      "grad_norm": 19.875,
+      "learning_rate": 3.8175762351940664e-07,
+      "loss": 0.3964,
+      "step": 17497
+    },
+    {
+      "epoch": 1.4910950149126545,
+      "grad_norm": 19.5,
+      "learning_rate": 3.8163674624488154e-07,
+      "loss": 0.7185,
+      "step": 17498
+    },
+    {
+      "epoch": 1.4911802300809545,
+      "grad_norm": 19.125,
+      "learning_rate": 3.8151588466255313e-07,
+      "loss": 0.8671,
+      "step": 17499
+    },
+    {
+      "epoch": 1.4912654452492544,
+      "grad_norm": 17.5,
+      "learning_rate": 3.81395038774606e-07,
+      "loss": 0.5918,
+      "step": 17500
+    },
+    {
+      "epoch": 1.4913506604175544,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.8127420858322437e-07,
+      "loss": 0.4712,
+      "step": 17501
+    },
+    {
+      "epoch": 1.4914358755858543,
+      "grad_norm": 12.875,
+      "learning_rate": 3.811533940905911e-07,
+      "loss": 0.479,
+      "step": 17502
+    },
+    {
+      "epoch": 1.4915210907541543,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.8103259529888956e-07,
+      "loss": 0.527,
+      "step": 17503
+    },
+    {
+      "epoch": 1.4916063059224542,
+      "grad_norm": 16.875,
+      "learning_rate": 3.809118122103028e-07,
+      "loss": 0.6575,
+      "step": 17504
+    },
+    {
+      "epoch": 1.4916915210907542,
+      "grad_norm": 11.4375,
+      "learning_rate": 3.80791044827014e-07,
+      "loss": 0.309,
+      "step": 17505
+    },
+    {
+      "epoch": 1.4917767362590542,
+      "grad_norm": 17.125,
+      "learning_rate": 3.8067029315120516e-07,
+      "loss": 0.2089,
+      "step": 17506
+    },
+    {
+      "epoch": 1.4918619514273541,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.8054955718505773e-07,
+      "loss": 0.5543,
+      "step": 17507
+    },
+    {
+      "epoch": 1.491947166595654,
+      "grad_norm": 20.125,
+      "learning_rate": 3.804288369307542e-07,
+      "loss": 0.9201,
+      "step": 17508
+    },
+    {
+      "epoch": 1.492032381763954,
+      "grad_norm": 11.0,
+      "learning_rate": 3.803081323904764e-07,
+      "loss": 0.1756,
+      "step": 17509
+    },
+    {
+      "epoch": 1.492117596932254,
+      "grad_norm": 9.125,
+      "learning_rate": 3.801874435664048e-07,
+      "loss": 0.2427,
+      "step": 17510
+    },
+    {
+      "epoch": 1.492202812100554,
+      "grad_norm": 14.125,
+      "learning_rate": 3.8006677046072083e-07,
+      "loss": 0.7541,
+      "step": 17511
+    },
+    {
+      "epoch": 1.492288027268854,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.7994611307560494e-07,
+      "loss": 0.719,
+      "step": 17512
+    },
+    {
+      "epoch": 1.4923732424371539,
+      "grad_norm": 13.375,
+      "learning_rate": 3.798254714132382e-07,
+      "loss": 0.6811,
+      "step": 17513
+    },
+    {
+      "epoch": 1.4924584576054538,
+      "grad_norm": 13.9375,
+      "learning_rate": 3.7970484547580026e-07,
+      "loss": 0.7262,
+      "step": 17514
+    },
+    {
+      "epoch": 1.4925436727737538,
+      "grad_norm": 13.125,
+      "learning_rate": 3.7958423526547037e-07,
+      "loss": 0.4447,
+      "step": 17515
+    },
+    {
+      "epoch": 1.4926288879420537,
+      "grad_norm": 12.75,
+      "learning_rate": 3.7946364078442864e-07,
+      "loss": 0.5503,
+      "step": 17516
+    },
+    {
+      "epoch": 1.4927141031103537,
+      "grad_norm": 20.5,
+      "learning_rate": 3.793430620348544e-07,
+      "loss": 0.5929,
+      "step": 17517
+    },
+    {
+      "epoch": 1.4927993182786536,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.7922249901892694e-07,
+      "loss": 0.5588,
+      "step": 17518
+    },
+    {
+      "epoch": 1.4928845334469536,
+      "grad_norm": 17.5,
+      "learning_rate": 3.7910195173882405e-07,
+      "loss": 0.5391,
+      "step": 17519
+    },
+    {
+      "epoch": 1.4929697486152536,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.7898142019672487e-07,
+      "loss": 0.3766,
+      "step": 17520
+    },
+    {
+      "epoch": 1.4930549637835535,
+      "grad_norm": 17.0,
+      "learning_rate": 3.7886090439480745e-07,
+      "loss": 0.5782,
+      "step": 17521
+    },
+    {
+      "epoch": 1.4931401789518535,
+      "grad_norm": 11.5625,
+      "learning_rate": 3.7874040433524966e-07,
+      "loss": 0.3422,
+      "step": 17522
+    },
+    {
+      "epoch": 1.4932253941201534,
+      "grad_norm": 22.75,
+      "learning_rate": 3.7861992002022843e-07,
+      "loss": 0.8153,
+      "step": 17523
+    },
+    {
+      "epoch": 1.4933106092884534,
+      "grad_norm": 21.0,
+      "learning_rate": 3.784994514519215e-07,
+      "loss": 0.7669,
+      "step": 17524
+    },
+    {
+      "epoch": 1.4933958244567533,
+      "grad_norm": 12.625,
+      "learning_rate": 3.783789986325058e-07,
+      "loss": 0.2496,
+      "step": 17525
+    },
+    {
+      "epoch": 1.4934810396250533,
+      "grad_norm": 14.0,
+      "learning_rate": 3.782585615641585e-07,
+      "loss": 0.4726,
+      "step": 17526
+    },
+    {
+      "epoch": 1.4935662547933533,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.781381402490551e-07,
+      "loss": 0.3446,
+      "step": 17527
+    },
+    {
+      "epoch": 1.4936514699616532,
+      "grad_norm": 12.4375,
+      "learning_rate": 3.7801773468937235e-07,
+      "loss": 0.3443,
+      "step": 17528
+    },
+    {
+      "epoch": 1.4937366851299532,
+      "grad_norm": 18.125,
+      "learning_rate": 3.7789734488728593e-07,
+      "loss": 0.5557,
+      "step": 17529
+    },
+    {
+      "epoch": 1.4938219002982531,
+      "grad_norm": 9.625,
+      "learning_rate": 3.777769708449719e-07,
+      "loss": 0.2419,
+      "step": 17530
+    },
+    {
+      "epoch": 1.493907115466553,
+      "grad_norm": 16.625,
+      "learning_rate": 3.77656612564605e-07,
+      "loss": 0.5952,
+      "step": 17531
+    },
+    {
+      "epoch": 1.493992330634853,
+      "grad_norm": 13.375,
+      "learning_rate": 3.775362700483601e-07,
+      "loss": 0.5698,
+      "step": 17532
+    },
+    {
+      "epoch": 1.494077545803153,
+      "grad_norm": 17.875,
+      "learning_rate": 3.7741594329841204e-07,
+      "loss": 0.743,
+      "step": 17533
+    },
+    {
+      "epoch": 1.494162760971453,
+      "grad_norm": 17.625,
+      "learning_rate": 3.772956323169352e-07,
+      "loss": 0.6814,
+      "step": 17534
+    },
+    {
+      "epoch": 1.494247976139753,
+      "grad_norm": 17.125,
+      "learning_rate": 3.7717533710610445e-07,
+      "loss": 0.667,
+      "step": 17535
+    },
+    {
+      "epoch": 1.4943331913080529,
+      "grad_norm": 15.875,
+      "learning_rate": 3.7705505766809256e-07,
+      "loss": 0.7306,
+      "step": 17536
+    },
+    {
+      "epoch": 1.4944184064763528,
+      "grad_norm": 13.6875,
+      "learning_rate": 3.7693479400507357e-07,
+      "loss": 0.6171,
+      "step": 17537
+    },
+    {
+      "epoch": 1.4945036216446528,
+      "grad_norm": 17.375,
+      "learning_rate": 3.768145461192213e-07,
+      "loss": 0.6451,
+      "step": 17538
+    },
+    {
+      "epoch": 1.4945888368129527,
+      "grad_norm": 15.0,
+      "learning_rate": 3.7669431401270814e-07,
+      "loss": 0.4427,
+      "step": 17539
+    },
+    {
+      "epoch": 1.4946740519812527,
+      "grad_norm": 18.5,
+      "learning_rate": 3.765740976877065e-07,
+      "loss": 0.7907,
+      "step": 17540
+    },
+    {
+      "epoch": 1.4947592671495527,
+      "grad_norm": 14.5,
+      "learning_rate": 3.764538971463892e-07,
+      "loss": 0.4297,
+      "step": 17541
+    },
+    {
+      "epoch": 1.4948444823178526,
+      "grad_norm": 15.75,
+      "learning_rate": 3.7633371239092847e-07,
+      "loss": 0.4595,
+      "step": 17542
+    },
+    {
+      "epoch": 1.4949296974861526,
+      "grad_norm": 12.9375,
+      "learning_rate": 3.7621354342349646e-07,
+      "loss": 0.4803,
+      "step": 17543
+    },
+    {
+      "epoch": 1.4950149126544525,
+      "grad_norm": 20.125,
+      "learning_rate": 3.760933902462639e-07,
+      "loss": 0.8995,
+      "step": 17544
+    },
+    {
+      "epoch": 1.4951001278227525,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.759732528614027e-07,
+      "loss": 0.5473,
+      "step": 17545
+    },
+    {
+      "epoch": 1.4951853429910524,
+      "grad_norm": 18.75,
+      "learning_rate": 3.758531312710836e-07,
+      "loss": 0.6203,
+      "step": 17546
+    },
+    {
+      "epoch": 1.4952705581593524,
+      "grad_norm": 17.5,
+      "learning_rate": 3.75733025477478e-07,
+      "loss": 0.741,
+      "step": 17547
+    },
+    {
+      "epoch": 1.4953557733276523,
+      "grad_norm": 15.0,
+      "learning_rate": 3.7561293548275565e-07,
+      "loss": 0.7313,
+      "step": 17548
+    },
+    {
+      "epoch": 1.4954409884959523,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.754928612890865e-07,
+      "loss": 0.4672,
+      "step": 17549
+    },
+    {
+      "epoch": 1.4955262036642523,
+      "grad_norm": 13.875,
+      "learning_rate": 3.753728028986406e-07,
+      "loss": 0.5798,
+      "step": 17550
+    },
+    {
+      "epoch": 1.4956114188325522,
+      "grad_norm": 13.25,
+      "learning_rate": 3.752527603135882e-07,
+      "loss": 0.5281,
+      "step": 17551
+    },
+    {
+      "epoch": 1.4956966340008522,
+      "grad_norm": 20.0,
+      "learning_rate": 3.7513273353609754e-07,
+      "loss": 0.6578,
+      "step": 17552
+    },
+    {
+      "epoch": 1.4957818491691521,
+      "grad_norm": 13.25,
+      "learning_rate": 3.7501272256833824e-07,
+      "loss": 0.5646,
+      "step": 17553
+    },
+    {
+      "epoch": 1.495867064337452,
+      "grad_norm": 13.4375,
+      "learning_rate": 3.7489272741247877e-07,
+      "loss": 0.4328,
+      "step": 17554
+    },
+    {
+      "epoch": 1.495952279505752,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.747727480706882e-07,
+      "loss": 0.5498,
+      "step": 17555
+    },
+    {
+      "epoch": 1.496037494674052,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.7465278454513423e-07,
+      "loss": 0.6597,
+      "step": 17556
+    },
+    {
+      "epoch": 1.496122709842352,
+      "grad_norm": 23.25,
+      "learning_rate": 3.745328368379841e-07,
+      "loss": 1.0158,
+      "step": 17557
+    },
+    {
+      "epoch": 1.496207925010652,
+      "grad_norm": 15.25,
+      "learning_rate": 3.744129049514061e-07,
+      "loss": 0.7415,
+      "step": 17558
+    },
+    {
+      "epoch": 1.4962931401789519,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.742929888875674e-07,
+      "loss": 0.7127,
+      "step": 17559
+    },
+    {
+      "epoch": 1.4963783553472518,
+      "grad_norm": 14.125,
+      "learning_rate": 3.741730886486353e-07,
+      "loss": 0.4559,
+      "step": 17560
+    },
+    {
+      "epoch": 1.4964635705155518,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.740532042367759e-07,
+      "loss": 0.7112,
+      "step": 17561
+    },
+    {
+      "epoch": 1.4965487856838517,
+      "grad_norm": 18.875,
+      "learning_rate": 3.739333356541558e-07,
+      "loss": 0.8795,
+      "step": 17562
+    },
+    {
+      "epoch": 1.4966340008521517,
+      "grad_norm": 16.125,
+      "learning_rate": 3.7381348290294174e-07,
+      "loss": 0.5574,
+      "step": 17563
+    },
+    {
+      "epoch": 1.4967192160204517,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.736936459852987e-07,
+      "loss": 0.397,
+      "step": 17564
+    },
+    {
+      "epoch": 1.4968044311887516,
+      "grad_norm": 17.125,
+      "learning_rate": 3.7357382490339307e-07,
+      "loss": 0.45,
+      "step": 17565
+    },
+    {
+      "epoch": 1.4968896463570516,
+      "grad_norm": 21.0,
+      "learning_rate": 3.734540196593893e-07,
+      "loss": 0.8355,
+      "step": 17566
+    },
+    {
+      "epoch": 1.4969748615253515,
+      "grad_norm": 16.875,
+      "learning_rate": 3.73334230255453e-07,
+      "loss": 0.6461,
+      "step": 17567
+    },
+    {
+      "epoch": 1.4970600766936515,
+      "grad_norm": 18.75,
+      "learning_rate": 3.7321445669374884e-07,
+      "loss": 0.5792,
+      "step": 17568
+    },
+    {
+      "epoch": 1.4971452918619514,
+      "grad_norm": 11.8125,
+      "learning_rate": 3.73094698976441e-07,
+      "loss": 0.3175,
+      "step": 17569
+    },
+    {
+      "epoch": 1.4972305070302514,
+      "grad_norm": 13.6875,
+      "learning_rate": 3.729749571056936e-07,
+      "loss": 0.5115,
+      "step": 17570
+    },
+    {
+      "epoch": 1.4973157221985514,
+      "grad_norm": 18.75,
+      "learning_rate": 3.728552310836711e-07,
+      "loss": 0.7201,
+      "step": 17571
+    },
+    {
+      "epoch": 1.4974009373668513,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.7273552091253606e-07,
+      "loss": 0.4319,
+      "step": 17572
+    },
+    {
+      "epoch": 1.4974861525351513,
+      "grad_norm": 13.0,
+      "learning_rate": 3.726158265944529e-07,
+      "loss": 0.4696,
+      "step": 17573
+    },
+    {
+      "epoch": 1.4975713677034512,
+      "grad_norm": 13.75,
+      "learning_rate": 3.724961481315835e-07,
+      "loss": 0.583,
+      "step": 17574
+    },
+    {
+      "epoch": 1.4976565828717512,
+      "grad_norm": 18.375,
+      "learning_rate": 3.723764855260913e-07,
+      "loss": 0.8521,
+      "step": 17575
+    },
+    {
+      "epoch": 1.4977417980400511,
+      "grad_norm": 16.125,
+      "learning_rate": 3.7225683878013825e-07,
+      "loss": 0.6204,
+      "step": 17576
+    },
+    {
+      "epoch": 1.497827013208351,
+      "grad_norm": 16.75,
+      "learning_rate": 3.7213720789588735e-07,
+      "loss": 0.7008,
+      "step": 17577
+    },
+    {
+      "epoch": 1.497912228376651,
+      "grad_norm": 15.125,
+      "learning_rate": 3.7201759287549947e-07,
+      "loss": 0.6444,
+      "step": 17578
+    },
+    {
+      "epoch": 1.497997443544951,
+      "grad_norm": 12.25,
+      "learning_rate": 3.7189799372113687e-07,
+      "loss": 0.4901,
+      "step": 17579
+    },
+    {
+      "epoch": 1.498082658713251,
+      "grad_norm": 17.25,
+      "learning_rate": 3.7177841043496017e-07,
+      "loss": 0.5006,
+      "step": 17580
+    },
+    {
+      "epoch": 1.498167873881551,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.716588430191312e-07,
+      "loss": 0.5075,
+      "step": 17581
+    },
+    {
+      "epoch": 1.4982530890498509,
+      "grad_norm": 13.75,
+      "learning_rate": 3.7153929147580967e-07,
+      "loss": 0.4614,
+      "step": 17582
+    },
+    {
+      "epoch": 1.4983383042181508,
+      "grad_norm": 19.0,
+      "learning_rate": 3.7141975580715653e-07,
+      "loss": 0.5827,
+      "step": 17583
+    },
+    {
+      "epoch": 1.4984235193864508,
+      "grad_norm": 17.375,
+      "learning_rate": 3.7130023601533183e-07,
+      "loss": 0.6694,
+      "step": 17584
+    },
+    {
+      "epoch": 1.4985087345547508,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.711807321024959e-07,
+      "loss": 0.5078,
+      "step": 17585
+    },
+    {
+      "epoch": 1.4985939497230507,
+      "grad_norm": 12.0,
+      "learning_rate": 3.710612440708075e-07,
+      "loss": 0.6483,
+      "step": 17586
+    },
+    {
+      "epoch": 1.4986791648913507,
+      "grad_norm": 12.4375,
+      "learning_rate": 3.7094177192242664e-07,
+      "loss": 0.4887,
+      "step": 17587
+    },
+    {
+      "epoch": 1.4987643800596506,
+      "grad_norm": 22.25,
+      "learning_rate": 3.708223156595114e-07,
+      "loss": 0.8873,
+      "step": 17588
+    },
+    {
+      "epoch": 1.4988495952279506,
+      "grad_norm": 21.75,
+      "learning_rate": 3.707028752842212e-07,
+      "loss": 1.1401,
+      "step": 17589
+    },
+    {
+      "epoch": 1.4989348103962505,
+      "grad_norm": 25.625,
+      "learning_rate": 3.7058345079871453e-07,
+      "loss": 0.8154,
+      "step": 17590
+    },
+    {
+      "epoch": 1.4990200255645505,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.7046404220514894e-07,
+      "loss": 0.4349,
+      "step": 17591
+    },
+    {
+      "epoch": 1.4991052407328505,
+      "grad_norm": 20.875,
+      "learning_rate": 3.703446495056824e-07,
+      "loss": 0.6151,
+      "step": 17592
+    },
+    {
+      "epoch": 1.4991904559011504,
+      "grad_norm": 16.5,
+      "learning_rate": 3.702252727024727e-07,
+      "loss": 0.4761,
+      "step": 17593
+    },
+    {
+      "epoch": 1.4992756710694504,
+      "grad_norm": 11.875,
+      "learning_rate": 3.7010591179767725e-07,
+      "loss": 0.9119,
+      "step": 17594
+    },
+    {
+      "epoch": 1.4993608862377503,
+      "grad_norm": 13.0,
+      "learning_rate": 3.6998656679345253e-07,
+      "loss": 0.379,
+      "step": 17595
+    },
+    {
+      "epoch": 1.4994461014060503,
+      "grad_norm": 31.875,
+      "learning_rate": 3.698672376919558e-07,
+      "loss": 0.657,
+      "step": 17596
+    },
+    {
+      "epoch": 1.4995313165743502,
+      "grad_norm": 17.75,
+      "learning_rate": 3.6974792449534283e-07,
+      "loss": 0.5086,
+      "step": 17597
+    },
+    {
+      "epoch": 1.4996165317426502,
+      "grad_norm": 16.375,
+      "learning_rate": 3.696286272057703e-07,
+      "loss": 0.4567,
+      "step": 17598
+    },
+    {
+      "epoch": 1.4997017469109502,
+      "grad_norm": 12.125,
+      "learning_rate": 3.6950934582539346e-07,
+      "loss": 0.3842,
+      "step": 17599
+    },
+    {
+      "epoch": 1.49978696207925,
+      "grad_norm": 10.25,
+      "learning_rate": 3.693900803563681e-07,
+      "loss": 0.3043,
+      "step": 17600
+    },
+    {
+      "epoch": 1.49987217724755,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.6927083080084955e-07,
+      "loss": 0.2383,
+      "step": 17601
+    },
+    {
+      "epoch": 1.49995739241585,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.6915159716099313e-07,
+      "loss": 0.4425,
+      "step": 17602
+    },
+    {
+      "epoch": 1.50004260758415,
+      "grad_norm": 16.375,
+      "learning_rate": 3.690323794389526e-07,
+      "loss": 0.158,
+      "step": 17603
+    },
+    {
+      "epoch": 1.50012782275245,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.689131776368833e-07,
+      "loss": 0.8201,
+      "step": 17604
+    },
+    {
+      "epoch": 1.50021303792075,
+      "grad_norm": 13.375,
+      "learning_rate": 3.687939917569386e-07,
+      "loss": 0.3662,
+      "step": 17605
+    },
+    {
+      "epoch": 1.5002982530890498,
+      "grad_norm": 20.75,
+      "learning_rate": 3.686748218012724e-07,
+      "loss": 1.1057,
+      "step": 17606
+    },
+    {
+      "epoch": 1.5003834682573498,
+      "grad_norm": 14.625,
+      "learning_rate": 3.6855566777203894e-07,
+      "loss": 0.4443,
+      "step": 17607
+    },
+    {
+      "epoch": 1.5004686834256498,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.6843652967139055e-07,
+      "loss": 0.4159,
+      "step": 17608
+    },
+    {
+      "epoch": 1.5005538985939497,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.683174075014805e-07,
+      "loss": 0.3446,
+      "step": 17609
+    },
+    {
+      "epoch": 1.5006391137622497,
+      "grad_norm": 16.0,
+      "learning_rate": 3.681983012644615e-07,
+      "loss": 0.3066,
+      "step": 17610
+    },
+    {
+      "epoch": 1.5007243289305496,
+      "grad_norm": 19.375,
+      "learning_rate": 3.680792109624863e-07,
+      "loss": 0.7687,
+      "step": 17611
+    },
+    {
+      "epoch": 1.5008095440988496,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.679601365977066e-07,
+      "loss": 0.3632,
+      "step": 17612
+    },
+    {
+      "epoch": 1.5008947592671495,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.6784107817227383e-07,
+      "loss": 0.6686,
+      "step": 17613
+    },
+    {
+      "epoch": 1.5009799744354495,
+      "grad_norm": 11.5,
+      "learning_rate": 3.677220356883397e-07,
+      "loss": 0.3584,
+      "step": 17614
+    },
+    {
+      "epoch": 1.5010651896037495,
+      "grad_norm": 14.625,
+      "learning_rate": 3.6760300914805616e-07,
+      "loss": 0.7623,
+      "step": 17615
+    },
+    {
+      "epoch": 1.5011504047720494,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.6748399855357314e-07,
+      "loss": 0.5739,
+      "step": 17616
+    },
+    {
+      "epoch": 1.5012356199403494,
+      "grad_norm": 19.5,
+      "learning_rate": 3.673650039070417e-07,
+      "loss": 0.6488,
+      "step": 17617
+    },
+    {
+      "epoch": 1.5013208351086493,
+      "grad_norm": 19.25,
+      "learning_rate": 3.672460252106122e-07,
+      "loss": 0.7729,
+      "step": 17618
+    },
+    {
+      "epoch": 1.5014060502769493,
+      "grad_norm": 22.875,
+      "learning_rate": 3.6712706246643496e-07,
+      "loss": 0.7672,
+      "step": 17619
+    },
+    {
+      "epoch": 1.5014912654452492,
+      "grad_norm": 13.0,
+      "learning_rate": 3.6700811567665957e-07,
+      "loss": 0.4725,
+      "step": 17620
+    },
+    {
+      "epoch": 1.5015764806135492,
+      "grad_norm": 15.5,
+      "learning_rate": 3.6688918484343515e-07,
+      "loss": 0.6762,
+      "step": 17621
+    },
+    {
+      "epoch": 1.5016616957818492,
+      "grad_norm": 17.875,
+      "learning_rate": 3.6677026996891116e-07,
+      "loss": 0.6464,
+      "step": 17622
+    },
+    {
+      "epoch": 1.5017469109501491,
+      "grad_norm": 21.25,
+      "learning_rate": 3.6665137105523646e-07,
+      "loss": 0.8062,
+      "step": 17623
+    },
+    {
+      "epoch": 1.501832126118449,
+      "grad_norm": 17.375,
+      "learning_rate": 3.6653248810456025e-07,
+      "loss": 0.6739,
+      "step": 17624
+    },
+    {
+      "epoch": 1.501917341286749,
+      "grad_norm": 19.75,
+      "learning_rate": 3.6641362111903e-07,
+      "loss": 0.6772,
+      "step": 17625
+    },
+    {
+      "epoch": 1.502002556455049,
+      "grad_norm": 14.125,
+      "learning_rate": 3.662947701007941e-07,
+      "loss": 0.4247,
+      "step": 17626
+    },
+    {
+      "epoch": 1.502087771623349,
+      "grad_norm": 20.875,
+      "learning_rate": 3.661759350520003e-07,
+      "loss": 0.7413,
+      "step": 17627
+    },
+    {
+      "epoch": 1.502172986791649,
+      "grad_norm": 12.75,
+      "learning_rate": 3.660571159747969e-07,
+      "loss": 0.4522,
+      "step": 17628
+    },
+    {
+      "epoch": 1.5022582019599489,
+      "grad_norm": 16.5,
+      "learning_rate": 3.6593831287132953e-07,
+      "loss": 0.5569,
+      "step": 17629
+    },
+    {
+      "epoch": 1.5023434171282488,
+      "grad_norm": 15.375,
+      "learning_rate": 3.658195257437458e-07,
+      "loss": 0.8198,
+      "step": 17630
+    },
+    {
+      "epoch": 1.5024286322965488,
+      "grad_norm": 16.0,
+      "learning_rate": 3.657007545941924e-07,
+      "loss": 0.3184,
+      "step": 17631
+    },
+    {
+      "epoch": 1.5025138474648487,
+      "grad_norm": 11.625,
+      "learning_rate": 3.6558199942481595e-07,
+      "loss": 0.6567,
+      "step": 17632
+    },
+    {
+      "epoch": 1.5025990626331487,
+      "grad_norm": 17.0,
+      "learning_rate": 3.6546326023776184e-07,
+      "loss": 0.7709,
+      "step": 17633
+    },
+    {
+      "epoch": 1.5026842778014486,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.653445370351759e-07,
+      "loss": 0.8675,
+      "step": 17634
+    },
+    {
+      "epoch": 1.5027694929697486,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.6522582981920395e-07,
+      "loss": 0.5696,
+      "step": 17635
+    },
+    {
+      "epoch": 1.5028547081380486,
+      "grad_norm": 13.875,
+      "learning_rate": 3.651071385919912e-07,
+      "loss": 0.3766,
+      "step": 17636
+    },
+    {
+      "epoch": 1.5029399233063485,
+      "grad_norm": 17.625,
+      "learning_rate": 3.649884633556823e-07,
+      "loss": 0.5326,
+      "step": 17637
+    },
+    {
+      "epoch": 1.5030251384746485,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.648698041124216e-07,
+      "loss": 0.5351,
+      "step": 17638
+    },
+    {
+      "epoch": 1.5031103536429484,
+      "grad_norm": 14.25,
+      "learning_rate": 3.6475116086435335e-07,
+      "loss": 0.6083,
+      "step": 17639
+    },
+    {
+      "epoch": 1.5031955688112484,
+      "grad_norm": 17.375,
+      "learning_rate": 3.6463253361362193e-07,
+      "loss": 0.619,
+      "step": 17640
+    },
+    {
+      "epoch": 1.5032807839795483,
+      "grad_norm": 25.375,
+      "learning_rate": 3.6451392236237145e-07,
+      "loss": 0.7584,
+      "step": 17641
+    },
+    {
+      "epoch": 1.5033659991478483,
+      "grad_norm": 14.5,
+      "learning_rate": 3.6439532711274426e-07,
+      "loss": 0.5143,
+      "step": 17642
+    },
+    {
+      "epoch": 1.5034512143161483,
+      "grad_norm": 13.875,
+      "learning_rate": 3.6427674786688405e-07,
+      "loss": 0.3843,
+      "step": 17643
+    },
+    {
+      "epoch": 1.5035364294844482,
+      "grad_norm": 12.1875,
+      "learning_rate": 3.6415818462693417e-07,
+      "loss": 0.4416,
+      "step": 17644
+    },
+    {
+      "epoch": 1.5036216446527482,
+      "grad_norm": 16.125,
+      "learning_rate": 3.6403963739503666e-07,
+      "loss": 0.7131,
+      "step": 17645
+    },
+    {
+      "epoch": 1.5037068598210481,
+      "grad_norm": 24.625,
+      "learning_rate": 3.639211061733333e-07,
+      "loss": 0.713,
+      "step": 17646
+    },
+    {
+      "epoch": 1.503792074989348,
+      "grad_norm": 13.5,
+      "learning_rate": 3.638025909639667e-07,
+      "loss": 0.5319,
+      "step": 17647
+    },
+    {
+      "epoch": 1.503877290157648,
+      "grad_norm": 16.625,
+      "learning_rate": 3.6368409176907844e-07,
+      "loss": 0.4675,
+      "step": 17648
+    },
+    {
+      "epoch": 1.503962505325948,
+      "grad_norm": 11.875,
+      "learning_rate": 3.6356560859081016e-07,
+      "loss": 0.4492,
+      "step": 17649
+    },
+    {
+      "epoch": 1.504047720494248,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.6344714143130234e-07,
+      "loss": 0.3993,
+      "step": 17650
+    },
+    {
+      "epoch": 1.504132935662548,
+      "grad_norm": 18.375,
+      "learning_rate": 3.6332869029269623e-07,
+      "loss": 0.7939,
+      "step": 17651
+    },
+    {
+      "epoch": 1.5042181508308479,
+      "grad_norm": 24.0,
+      "learning_rate": 3.632102551771327e-07,
+      "loss": 0.9664,
+      "step": 17652
+    },
+    {
+      "epoch": 1.5043033659991478,
+      "grad_norm": 14.0,
+      "learning_rate": 3.630918360867512e-07,
+      "loss": 0.673,
+      "step": 17653
+    },
+    {
+      "epoch": 1.5043885811674478,
+      "grad_norm": 12.75,
+      "learning_rate": 3.629734330236924e-07,
+      "loss": 0.4649,
+      "step": 17654
+    },
+    {
+      "epoch": 1.5044737963357477,
+      "grad_norm": 13.125,
+      "learning_rate": 3.6285504599009536e-07,
+      "loss": 0.4627,
+      "step": 17655
+    },
+    {
+      "epoch": 1.5045590115040477,
+      "grad_norm": 16.75,
+      "learning_rate": 3.6273667498809965e-07,
+      "loss": 0.5896,
+      "step": 17656
+    },
+    {
+      "epoch": 1.5046442266723477,
+      "grad_norm": 14.75,
+      "learning_rate": 3.626183200198445e-07,
+      "loss": 0.7516,
+      "step": 17657
+    },
+    {
+      "epoch": 1.5047294418406476,
+      "grad_norm": 22.375,
+      "learning_rate": 3.6249998108746897e-07,
+      "loss": 1.1351,
+      "step": 17658
+    },
+    {
+      "epoch": 1.5048146570089476,
+      "grad_norm": 12.125,
+      "learning_rate": 3.6238165819311083e-07,
+      "loss": 0.3761,
+      "step": 17659
+    },
+    {
+      "epoch": 1.5048998721772475,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.6226335133890915e-07,
+      "loss": 0.6368,
+      "step": 17660
+    },
+    {
+      "epoch": 1.5049850873455475,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.621450605270009e-07,
+      "loss": 0.4275,
+      "step": 17661
+    },
+    {
+      "epoch": 1.5050703025138474,
+      "grad_norm": 21.25,
+      "learning_rate": 3.620267857595247e-07,
+      "loss": 0.6888,
+      "step": 17662
+    },
+    {
+      "epoch": 1.5051555176821474,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.61908527038617e-07,
+      "loss": 0.4342,
+      "step": 17663
+    },
+    {
+      "epoch": 1.5052407328504473,
+      "grad_norm": 25.25,
+      "learning_rate": 3.617902843664153e-07,
+      "loss": 0.7437,
+      "step": 17664
+    },
+    {
+      "epoch": 1.5053259480187473,
+      "grad_norm": 17.0,
+      "learning_rate": 3.6167205774505625e-07,
+      "loss": 0.5054,
+      "step": 17665
+    },
+    {
+      "epoch": 1.5054111631870473,
+      "grad_norm": 18.625,
+      "learning_rate": 3.6155384717667676e-07,
+      "loss": 0.9845,
+      "step": 17666
+    },
+    {
+      "epoch": 1.5054963783553472,
+      "grad_norm": 17.625,
+      "learning_rate": 3.6143565266341234e-07,
+      "loss": 0.7661,
+      "step": 17667
+    },
+    {
+      "epoch": 1.5055815935236472,
+      "grad_norm": 16.625,
+      "learning_rate": 3.6131747420739925e-07,
+      "loss": 0.6002,
+      "step": 17668
+    },
+    {
+      "epoch": 1.5056668086919471,
+      "grad_norm": 26.0,
+      "learning_rate": 3.611993118107733e-07,
+      "loss": 0.5808,
+      "step": 17669
+    },
+    {
+      "epoch": 1.505752023860247,
+      "grad_norm": 17.625,
+      "learning_rate": 3.610811654756692e-07,
+      "loss": 0.9035,
+      "step": 17670
+    },
+    {
+      "epoch": 1.505837239028547,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.6096303520422274e-07,
+      "loss": 0.365,
+      "step": 17671
+    },
+    {
+      "epoch": 1.505922454196847,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.608449209985679e-07,
+      "loss": 0.4467,
+      "step": 17672
+    },
+    {
+      "epoch": 1.506007669365147,
+      "grad_norm": 15.875,
+      "learning_rate": 3.607268228608393e-07,
+      "loss": 0.5869,
+      "step": 17673
+    },
+    {
+      "epoch": 1.506092884533447,
+      "grad_norm": 18.0,
+      "learning_rate": 3.606087407931713e-07,
+      "loss": 0.5201,
+      "step": 17674
+    },
+    {
+      "epoch": 1.5061780997017469,
+      "grad_norm": 24.25,
+      "learning_rate": 3.604906747976981e-07,
+      "loss": 0.8088,
+      "step": 17675
+    },
+    {
+      "epoch": 1.5062633148700468,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.6037262487655255e-07,
+      "loss": 0.7864,
+      "step": 17676
+    },
+    {
+      "epoch": 1.5063485300383468,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.602545910318686e-07,
+      "loss": 0.6319,
+      "step": 17677
+    },
+    {
+      "epoch": 1.5064337452066467,
+      "grad_norm": 11.9375,
+      "learning_rate": 3.6013657326577856e-07,
+      "loss": 0.3707,
+      "step": 17678
+    },
+    {
+      "epoch": 1.5065189603749467,
+      "grad_norm": 16.625,
+      "learning_rate": 3.600185715804158e-07,
+      "loss": 0.6445,
+      "step": 17679
+    },
+    {
+      "epoch": 1.5066041755432467,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.59900585977912e-07,
+      "loss": 0.5241,
+      "step": 17680
+    },
+    {
+      "epoch": 1.5066893907115466,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.597826164603997e-07,
+      "loss": 0.5009,
+      "step": 17681
+    },
+    {
+      "epoch": 1.5067746058798466,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.596646630300106e-07,
+      "loss": 0.7159,
+      "step": 17682
+    },
+    {
+      "epoch": 1.5068598210481465,
+      "grad_norm": 15.25,
+      "learning_rate": 3.5954672568887685e-07,
+      "loss": 0.6728,
+      "step": 17683
+    },
+    {
+      "epoch": 1.5069450362164465,
+      "grad_norm": 25.5,
+      "learning_rate": 3.5942880443912876e-07,
+      "loss": 0.8292,
+      "step": 17684
+    },
+    {
+      "epoch": 1.5070302513847464,
+      "grad_norm": 20.375,
+      "learning_rate": 3.5931089928289824e-07,
+      "loss": 0.975,
+      "step": 17685
+    },
+    {
+      "epoch": 1.5071154665530464,
+      "grad_norm": 13.875,
+      "learning_rate": 3.5919301022231485e-07,
+      "loss": 0.5244,
+      "step": 17686
+    },
+    {
+      "epoch": 1.5072006817213464,
+      "grad_norm": 20.375,
+      "learning_rate": 3.5907513725950956e-07,
+      "loss": 0.8668,
+      "step": 17687
+    },
+    {
+      "epoch": 1.5072858968896463,
+      "grad_norm": 17.0,
+      "learning_rate": 3.589572803966129e-07,
+      "loss": 0.5268,
+      "step": 17688
+    },
+    {
+      "epoch": 1.5073711120579463,
+      "grad_norm": 15.625,
+      "learning_rate": 3.588394396357538e-07,
+      "loss": 0.6326,
+      "step": 17689
+    },
+    {
+      "epoch": 1.5074563272262462,
+      "grad_norm": 16.25,
+      "learning_rate": 3.5872161497906204e-07,
+      "loss": 0.7481,
+      "step": 17690
+    },
+    {
+      "epoch": 1.5075415423945462,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.586038064286675e-07,
+      "loss": 0.6844,
+      "step": 17691
+    },
+    {
+      "epoch": 1.5076267575628461,
+      "grad_norm": 14.0,
+      "learning_rate": 3.58486013986698e-07,
+      "loss": 0.5031,
+      "step": 17692
+    },
+    {
+      "epoch": 1.507711972731146,
+      "grad_norm": 16.125,
+      "learning_rate": 3.583682376552833e-07,
+      "loss": 0.5117,
+      "step": 17693
+    },
+    {
+      "epoch": 1.507797187899446,
+      "grad_norm": 25.375,
+      "learning_rate": 3.5825047743655057e-07,
+      "loss": 0.6086,
+      "step": 17694
+    },
+    {
+      "epoch": 1.507882403067746,
+      "grad_norm": 17.375,
+      "learning_rate": 3.581327333326284e-07,
+      "loss": 0.5559,
+      "step": 17695
+    },
+    {
+      "epoch": 1.507967618236046,
+      "grad_norm": 14.875,
+      "learning_rate": 3.580150053456452e-07,
+      "loss": 0.6285,
+      "step": 17696
+    },
+    {
+      "epoch": 1.508052833404346,
+      "grad_norm": 12.5,
+      "learning_rate": 3.5789729347772724e-07,
+      "loss": 0.4137,
+      "step": 17697
+    },
+    {
+      "epoch": 1.5081380485726459,
+      "grad_norm": 30.875,
+      "learning_rate": 3.577795977310024e-07,
+      "loss": 0.767,
+      "step": 17698
+    },
+    {
+      "epoch": 1.5082232637409458,
+      "grad_norm": 12.3125,
+      "learning_rate": 3.576619181075973e-07,
+      "loss": 0.4889,
+      "step": 17699
+    },
+    {
+      "epoch": 1.5083084789092458,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.5754425460963914e-07,
+      "loss": 0.5556,
+      "step": 17700
+    },
+    {
+      "epoch": 1.5083936940775458,
+      "grad_norm": 19.375,
+      "learning_rate": 3.5742660723925363e-07,
+      "loss": 0.6036,
+      "step": 17701
+    },
+    {
+      "epoch": 1.5084789092458457,
+      "grad_norm": 12.625,
+      "learning_rate": 3.5730897599856667e-07,
+      "loss": 0.5615,
+      "step": 17702
+    },
+    {
+      "epoch": 1.5085641244141457,
+      "grad_norm": 12.5,
+      "learning_rate": 3.57191360889704e-07,
+      "loss": 0.532,
+      "step": 17703
+    },
+    {
+      "epoch": 1.5086493395824456,
+      "grad_norm": 22.75,
+      "learning_rate": 3.570737619147913e-07,
+      "loss": 0.7152,
+      "step": 17704
+    },
+    {
+      "epoch": 1.5087345547507456,
+      "grad_norm": 20.375,
+      "learning_rate": 3.56956179075954e-07,
+      "loss": 0.5297,
+      "step": 17705
+    },
+    {
+      "epoch": 1.5088197699190455,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.568386123753162e-07,
+      "loss": 0.5827,
+      "step": 17706
+    },
+    {
+      "epoch": 1.5089049850873455,
+      "grad_norm": 17.625,
+      "learning_rate": 3.5672106181500284e-07,
+      "loss": 0.9004,
+      "step": 17707
+    },
+    {
+      "epoch": 1.5089902002556455,
+      "grad_norm": 14.125,
+      "learning_rate": 3.5660352739713853e-07,
+      "loss": 0.5621,
+      "step": 17708
+    },
+    {
+      "epoch": 1.5090754154239454,
+      "grad_norm": 12.875,
+      "learning_rate": 3.564860091238467e-07,
+      "loss": 0.6179,
+      "step": 17709
+    },
+    {
+      "epoch": 1.5091606305922454,
+      "grad_norm": 20.625,
+      "learning_rate": 3.56368506997251e-07,
+      "loss": 0.879,
+      "step": 17710
+    },
+    {
+      "epoch": 1.5092458457605453,
+      "grad_norm": 13.375,
+      "learning_rate": 3.56251021019475e-07,
+      "loss": 0.5925,
+      "step": 17711
+    },
+    {
+      "epoch": 1.5093310609288453,
+      "grad_norm": 17.5,
+      "learning_rate": 3.561335511926417e-07,
+      "loss": 0.7588,
+      "step": 17712
+    },
+    {
+      "epoch": 1.5094162760971452,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.5601609751887434e-07,
+      "loss": 0.5237,
+      "step": 17713
+    },
+    {
+      "epoch": 1.5095014912654452,
+      "grad_norm": 16.375,
+      "learning_rate": 3.558986600002948e-07,
+      "loss": 0.4084,
+      "step": 17714
+    },
+    {
+      "epoch": 1.5095867064337452,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.557812386390254e-07,
+      "loss": 0.5346,
+      "step": 17715
+    },
+    {
+      "epoch": 1.5096719216020453,
+      "grad_norm": 14.125,
+      "learning_rate": 3.556638334371884e-07,
+      "loss": 0.6737,
+      "step": 17716
+    },
+    {
+      "epoch": 1.5097571367703453,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.5554644439690545e-07,
+      "loss": 0.7111,
+      "step": 17717
+    },
+    {
+      "epoch": 1.5098423519386452,
+      "grad_norm": 16.0,
+      "learning_rate": 3.554290715202978e-07,
+      "loss": 0.6989,
+      "step": 17718
+    },
+    {
+      "epoch": 1.5099275671069452,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.5531171480948604e-07,
+      "loss": 0.8963,
+      "step": 17719
+    },
+    {
+      "epoch": 1.5100127822752452,
+      "grad_norm": 21.375,
+      "learning_rate": 3.5519437426659126e-07,
+      "loss": 0.6907,
+      "step": 17720
+    },
+    {
+      "epoch": 1.5100979974435451,
+      "grad_norm": 16.75,
+      "learning_rate": 3.5507704989373434e-07,
+      "loss": 0.6058,
+      "step": 17721
+    },
+    {
+      "epoch": 1.510183212611845,
+      "grad_norm": 12.625,
+      "learning_rate": 3.549597416930345e-07,
+      "loss": 0.4691,
+      "step": 17722
+    },
+    {
+      "epoch": 1.510268427780145,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.5484244966661224e-07,
+      "loss": 0.4355,
+      "step": 17723
+    },
+    {
+      "epoch": 1.510353642948445,
+      "grad_norm": 12.75,
+      "learning_rate": 3.5472517381658716e-07,
+      "loss": 0.3041,
+      "step": 17724
+    },
+    {
+      "epoch": 1.510438858116745,
+      "grad_norm": 19.0,
+      "learning_rate": 3.546079141450788e-07,
+      "loss": 0.6243,
+      "step": 17725
+    },
+    {
+      "epoch": 1.510524073285045,
+      "grad_norm": 17.0,
+      "learning_rate": 3.544906706542058e-07,
+      "loss": 0.6576,
+      "step": 17726
+    },
+    {
+      "epoch": 1.5106092884533449,
+      "grad_norm": 18.125,
+      "learning_rate": 3.5437344334608645e-07,
+      "loss": 0.9424,
+      "step": 17727
+    },
+    {
+      "epoch": 1.5106945036216448,
+      "grad_norm": 16.75,
+      "learning_rate": 3.5425623222283964e-07,
+      "loss": 0.8699,
+      "step": 17728
+    },
+    {
+      "epoch": 1.5107797187899448,
+      "grad_norm": 14.75,
+      "learning_rate": 3.541390372865834e-07,
+      "loss": 0.6933,
+      "step": 17729
+    },
+    {
+      "epoch": 1.5108649339582447,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.5402185853943617e-07,
+      "loss": 0.4274,
+      "step": 17730
+    },
+    {
+      "epoch": 1.5109501491265447,
+      "grad_norm": 16.5,
+      "learning_rate": 3.5390469598351436e-07,
+      "loss": 0.5428,
+      "step": 17731
+    },
+    {
+      "epoch": 1.5110353642948446,
+      "grad_norm": 13.5,
+      "learning_rate": 3.5378754962093587e-07,
+      "loss": 0.5076,
+      "step": 17732
+    },
+    {
+      "epoch": 1.5111205794631446,
+      "grad_norm": 11.5625,
+      "learning_rate": 3.5367041945381765e-07,
+      "loss": 0.2623,
+      "step": 17733
+    },
+    {
+      "epoch": 1.5112057946314446,
+      "grad_norm": 16.0,
+      "learning_rate": 3.535533054842766e-07,
+      "loss": 0.5259,
+      "step": 17734
+    },
+    {
+      "epoch": 1.5112910097997445,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.5343620771442864e-07,
+      "loss": 0.5355,
+      "step": 17735
+    },
+    {
+      "epoch": 1.5113762249680445,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.5331912614638974e-07,
+      "loss": 0.5876,
+      "step": 17736
+    },
+    {
+      "epoch": 1.5114614401363444,
+      "grad_norm": 13.375,
+      "learning_rate": 3.5320206078227587e-07,
+      "loss": 0.7021,
+      "step": 17737
+    },
+    {
+      "epoch": 1.5115466553046444,
+      "grad_norm": 22.875,
+      "learning_rate": 3.5308501162420293e-07,
+      "loss": 0.6281,
+      "step": 17738
+    },
+    {
+      "epoch": 1.5116318704729443,
+      "grad_norm": 41.25,
+      "learning_rate": 3.529679786742854e-07,
+      "loss": 1.3791,
+      "step": 17739
+    },
+    {
+      "epoch": 1.5117170856412443,
+      "grad_norm": 16.625,
+      "learning_rate": 3.5285096193463847e-07,
+      "loss": 0.7039,
+      "step": 17740
+    },
+    {
+      "epoch": 1.5118023008095443,
+      "grad_norm": 15.375,
+      "learning_rate": 3.527339614073769e-07,
+      "loss": 0.6886,
+      "step": 17741
+    },
+    {
+      "epoch": 1.5118875159778442,
+      "grad_norm": 15.5,
+      "learning_rate": 3.5261697709461515e-07,
+      "loss": 0.6373,
+      "step": 17742
+    },
+    {
+      "epoch": 1.5119727311461442,
+      "grad_norm": 16.0,
+      "learning_rate": 3.525000089984672e-07,
+      "loss": 0.4837,
+      "step": 17743
+    },
+    {
+      "epoch": 1.5120579463144441,
+      "grad_norm": 12.1875,
+      "learning_rate": 3.52383057121046e-07,
+      "loss": 0.4429,
+      "step": 17744
+    },
+    {
+      "epoch": 1.512143161482744,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.5226612146446573e-07,
+      "loss": 0.6118,
+      "step": 17745
+    },
+    {
+      "epoch": 1.512228376651044,
+      "grad_norm": 14.0,
+      "learning_rate": 3.5214920203083936e-07,
+      "loss": 0.3669,
+      "step": 17746
+    },
+    {
+      "epoch": 1.512313591819344,
+      "grad_norm": 16.0,
+      "learning_rate": 3.520322988222802e-07,
+      "loss": 0.6445,
+      "step": 17747
+    },
+    {
+      "epoch": 1.512398806987644,
+      "grad_norm": 14.875,
+      "learning_rate": 3.5191541184089987e-07,
+      "loss": 0.4992,
+      "step": 17748
+    },
+    {
+      "epoch": 1.512484022155944,
+      "grad_norm": 13.625,
+      "learning_rate": 3.5179854108881126e-07,
+      "loss": 0.2678,
+      "step": 17749
+    },
+    {
+      "epoch": 1.5125692373242439,
+      "grad_norm": 21.625,
+      "learning_rate": 3.516816865681265e-07,
+      "loss": 0.7409,
+      "step": 17750
+    },
+    {
+      "epoch": 1.5126544524925438,
+      "grad_norm": 17.0,
+      "learning_rate": 3.515648482809571e-07,
+      "loss": 0.4859,
+      "step": 17751
+    },
+    {
+      "epoch": 1.5127396676608438,
+      "grad_norm": 18.125,
+      "learning_rate": 3.5144802622941387e-07,
+      "loss": 0.6261,
+      "step": 17752
+    },
+    {
+      "epoch": 1.5128248828291437,
+      "grad_norm": 15.625,
+      "learning_rate": 3.5133122041560846e-07,
+      "loss": 0.5144,
+      "step": 17753
+    },
+    {
+      "epoch": 1.5129100979974437,
+      "grad_norm": 13.875,
+      "learning_rate": 3.5121443084165154e-07,
+      "loss": 0.4632,
+      "step": 17754
+    },
+    {
+      "epoch": 1.5129953131657436,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.510976575096539e-07,
+      "loss": 0.5195,
+      "step": 17755
+    },
+    {
+      "epoch": 1.5130805283340436,
+      "grad_norm": 16.25,
+      "learning_rate": 3.509809004217253e-07,
+      "loss": 0.7954,
+      "step": 17756
+    },
+    {
+      "epoch": 1.5131657435023436,
+      "grad_norm": 20.375,
+      "learning_rate": 3.5086415957997585e-07,
+      "loss": 1.0288,
+      "step": 17757
+    },
+    {
+      "epoch": 1.5132509586706435,
+      "grad_norm": 18.0,
+      "learning_rate": 3.5074743498651545e-07,
+      "loss": 0.6139,
+      "step": 17758
+    },
+    {
+      "epoch": 1.5133361738389435,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.506307266434529e-07,
+      "loss": 0.7012,
+      "step": 17759
+    },
+    {
+      "epoch": 1.5134213890072434,
+      "grad_norm": 12.125,
+      "learning_rate": 3.505140345528979e-07,
+      "loss": 0.4396,
+      "step": 17760
+    },
+    {
+      "epoch": 1.5135066041755434,
+      "grad_norm": 16.375,
+      "learning_rate": 3.5039735871695835e-07,
+      "loss": 0.5636,
+      "step": 17761
+    },
+    {
+      "epoch": 1.5135918193438433,
+      "grad_norm": 18.5,
+      "learning_rate": 3.502806991377432e-07,
+      "loss": 0.8029,
+      "step": 17762
+    },
+    {
+      "epoch": 1.5136770345121433,
+      "grad_norm": 20.5,
+      "learning_rate": 3.501640558173604e-07,
+      "loss": 0.9982,
+      "step": 17763
+    },
+    {
+      "epoch": 1.5137622496804433,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.500474287579185e-07,
+      "loss": 0.3789,
+      "step": 17764
+    },
+    {
+      "epoch": 1.5138474648487432,
+      "grad_norm": 21.0,
+      "learning_rate": 3.4993081796152407e-07,
+      "loss": 0.6428,
+      "step": 17765
+    },
+    {
+      "epoch": 1.5139326800170432,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.498142234302854e-07,
+      "loss": 0.5053,
+      "step": 17766
+    },
+    {
+      "epoch": 1.5140178951853431,
+      "grad_norm": 18.25,
+      "learning_rate": 3.496976451663084e-07,
+      "loss": 0.7665,
+      "step": 17767
+    },
+    {
+      "epoch": 1.514103110353643,
+      "grad_norm": 17.75,
+      "learning_rate": 3.4958108317170056e-07,
+      "loss": 0.6266,
+      "step": 17768
+    },
+    {
+      "epoch": 1.514188325521943,
+      "grad_norm": 14.5,
+      "learning_rate": 3.494645374485678e-07,
+      "loss": 0.7095,
+      "step": 17769
+    },
+    {
+      "epoch": 1.514273540690243,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.4934800799901635e-07,
+      "loss": 0.6648,
+      "step": 17770
+    },
+    {
+      "epoch": 1.514358755858543,
+      "grad_norm": 23.125,
+      "learning_rate": 3.49231494825152e-07,
+      "loss": 0.7907,
+      "step": 17771
+    },
+    {
+      "epoch": 1.514443971026843,
+      "grad_norm": 14.625,
+      "learning_rate": 3.4911499792908074e-07,
+      "loss": 0.7316,
+      "step": 17772
+    },
+    {
+      "epoch": 1.5145291861951429,
+      "grad_norm": 12.3125,
+      "learning_rate": 3.489985173129071e-07,
+      "loss": 0.3217,
+      "step": 17773
+    },
+    {
+      "epoch": 1.5146144013634428,
+      "grad_norm": 22.0,
+      "learning_rate": 3.4888205297873653e-07,
+      "loss": 0.8005,
+      "step": 17774
+    },
+    {
+      "epoch": 1.5146996165317428,
+      "grad_norm": 17.5,
+      "learning_rate": 3.4876560492867306e-07,
+      "loss": 0.602,
+      "step": 17775
+    },
+    {
+      "epoch": 1.5147848317000427,
+      "grad_norm": 18.125,
+      "learning_rate": 3.4864917316482145e-07,
+      "loss": 0.7116,
+      "step": 17776
+    },
+    {
+      "epoch": 1.5148700468683427,
+      "grad_norm": 15.5,
+      "learning_rate": 3.4853275768928593e-07,
+      "loss": 0.4481,
+      "step": 17777
+    },
+    {
+      "epoch": 1.5149552620366427,
+      "grad_norm": 18.25,
+      "learning_rate": 3.484163585041696e-07,
+      "loss": 0.7989,
+      "step": 17778
+    },
+    {
+      "epoch": 1.5150404772049426,
+      "grad_norm": 12.875,
+      "learning_rate": 3.482999756115764e-07,
+      "loss": 0.2885,
+      "step": 17779
+    },
+    {
+      "epoch": 1.5151256923732426,
+      "grad_norm": 33.5,
+      "learning_rate": 3.481836090136093e-07,
+      "loss": 1.0799,
+      "step": 17780
+    },
+    {
+      "epoch": 1.5152109075415425,
+      "grad_norm": 13.875,
+      "learning_rate": 3.480672587123715e-07,
+      "loss": 0.3723,
+      "step": 17781
+    },
+    {
+      "epoch": 1.5152961227098425,
+      "grad_norm": 17.5,
+      "learning_rate": 3.479509247099655e-07,
+      "loss": 0.7683,
+      "step": 17782
+    },
+    {
+      "epoch": 1.5153813378781424,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.478346070084929e-07,
+      "loss": 0.5179,
+      "step": 17783
+    },
+    {
+      "epoch": 1.5154665530464424,
+      "grad_norm": 23.75,
+      "learning_rate": 3.477183056100561e-07,
+      "loss": 0.6655,
+      "step": 17784
+    },
+    {
+      "epoch": 1.5155517682147424,
+      "grad_norm": 12.875,
+      "learning_rate": 3.4760202051675726e-07,
+      "loss": 0.4075,
+      "step": 17785
+    },
+    {
+      "epoch": 1.5156369833830423,
+      "grad_norm": 17.125,
+      "learning_rate": 3.474857517306969e-07,
+      "loss": 0.4393,
+      "step": 17786
+    },
+    {
+      "epoch": 1.5157221985513423,
+      "grad_norm": 16.125,
+      "learning_rate": 3.473694992539767e-07,
+      "loss": 0.936,
+      "step": 17787
+    },
+    {
+      "epoch": 1.5158074137196422,
+      "grad_norm": 25.125,
+      "learning_rate": 3.4725326308869713e-07,
+      "loss": 0.8175,
+      "step": 17788
+    },
+    {
+      "epoch": 1.5158926288879422,
+      "grad_norm": 19.0,
+      "learning_rate": 3.4713704323695947e-07,
+      "loss": 0.6298,
+      "step": 17789
+    },
+    {
+      "epoch": 1.5159778440562421,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.4702083970086314e-07,
+      "loss": 0.6322,
+      "step": 17790
+    },
+    {
+      "epoch": 1.516063059224542,
+      "grad_norm": 20.0,
+      "learning_rate": 3.4690465248250803e-07,
+      "loss": 0.7886,
+      "step": 17791
+    },
+    {
+      "epoch": 1.516148274392842,
+      "grad_norm": 14.5,
+      "learning_rate": 3.467884815839938e-07,
+      "loss": 0.3825,
+      "step": 17792
+    },
+    {
+      "epoch": 1.516233489561142,
+      "grad_norm": 17.25,
+      "learning_rate": 3.4667232700742016e-07,
+      "loss": 0.3798,
+      "step": 17793
+    },
+    {
+      "epoch": 1.516318704729442,
+      "grad_norm": 19.25,
+      "learning_rate": 3.4655618875488613e-07,
+      "loss": 0.6367,
+      "step": 17794
+    },
+    {
+      "epoch": 1.516403919897742,
+      "grad_norm": 11.9375,
+      "learning_rate": 3.464400668284899e-07,
+      "loss": 0.4919,
+      "step": 17795
+    },
+    {
+      "epoch": 1.5164891350660419,
+      "grad_norm": 17.375,
+      "learning_rate": 3.463239612303304e-07,
+      "loss": 0.9935,
+      "step": 17796
+    },
+    {
+      "epoch": 1.5165743502343418,
+      "grad_norm": 11.8125,
+      "learning_rate": 3.462078719625056e-07,
+      "loss": 0.5642,
+      "step": 17797
+    },
+    {
+      "epoch": 1.5166595654026418,
+      "grad_norm": 14.875,
+      "learning_rate": 3.460917990271137e-07,
+      "loss": 0.6166,
+      "step": 17798
+    },
+    {
+      "epoch": 1.5167447805709418,
+      "grad_norm": 12.6875,
+      "learning_rate": 3.459757424262519e-07,
+      "loss": 0.3854,
+      "step": 17799
+    },
+    {
+      "epoch": 1.5168299957392417,
+      "grad_norm": 14.25,
+      "learning_rate": 3.458597021620172e-07,
+      "loss": 0.5625,
+      "step": 17800
+    },
+    {
+      "epoch": 1.5169152109075417,
+      "grad_norm": 17.75,
+      "learning_rate": 3.4574367823650687e-07,
+      "loss": 0.5641,
+      "step": 17801
+    },
+    {
+      "epoch": 1.5170004260758416,
+      "grad_norm": 16.5,
+      "learning_rate": 3.4562767065181803e-07,
+      "loss": 0.6121,
+      "step": 17802
+    },
+    {
+      "epoch": 1.5170856412441416,
+      "grad_norm": 13.875,
+      "learning_rate": 3.4551167941004603e-07,
+      "loss": 0.478,
+      "step": 17803
+    },
+    {
+      "epoch": 1.5171708564124415,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.453957045132876e-07,
+      "loss": 0.681,
+      "step": 17804
+    },
+    {
+      "epoch": 1.5172560715807415,
+      "grad_norm": 20.375,
+      "learning_rate": 3.4527974596363853e-07,
+      "loss": 0.9487,
+      "step": 17805
+    },
+    {
+      "epoch": 1.5173412867490415,
+      "grad_norm": 16.375,
+      "learning_rate": 3.451638037631945e-07,
+      "loss": 0.4665,
+      "step": 17806
+    },
+    {
+      "epoch": 1.5174265019173414,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.4504787791405036e-07,
+      "loss": 0.7874,
+      "step": 17807
+    },
+    {
+      "epoch": 1.5175117170856414,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.4493196841830087e-07,
+      "loss": 0.5522,
+      "step": 17808
+    },
+    {
+      "epoch": 1.5175969322539413,
+      "grad_norm": 19.75,
+      "learning_rate": 3.4481607527804067e-07,
+      "loss": 0.5713,
+      "step": 17809
+    },
+    {
+      "epoch": 1.5176821474222413,
+      "grad_norm": 16.75,
+      "learning_rate": 3.4470019849536434e-07,
+      "loss": 0.8379,
+      "step": 17810
+    },
+    {
+      "epoch": 1.5177673625905412,
+      "grad_norm": 16.375,
+      "learning_rate": 3.44584338072366e-07,
+      "loss": 0.7655,
+      "step": 17811
+    },
+    {
+      "epoch": 1.5178525777588412,
+      "grad_norm": 18.125,
+      "learning_rate": 3.444684940111388e-07,
+      "loss": 0.5344,
+      "step": 17812
+    },
+    {
+      "epoch": 1.5179377929271411,
+      "grad_norm": 19.625,
+      "learning_rate": 3.4435266631377665e-07,
+      "loss": 0.6214,
+      "step": 17813
+    },
+    {
+      "epoch": 1.518023008095441,
+      "grad_norm": 18.0,
+      "learning_rate": 3.442368549823727e-07,
+      "loss": 0.5239,
+      "step": 17814
+    },
+    {
+      "epoch": 1.518108223263741,
+      "grad_norm": 11.875,
+      "learning_rate": 3.441210600190198e-07,
+      "loss": 0.2436,
+      "step": 17815
+    },
+    {
+      "epoch": 1.518193438432041,
+      "grad_norm": 14.375,
+      "learning_rate": 3.4400528142580976e-07,
+      "loss": 0.491,
+      "step": 17816
+    },
+    {
+      "epoch": 1.518278653600341,
+      "grad_norm": 24.375,
+      "learning_rate": 3.4388951920483527e-07,
+      "loss": 0.9086,
+      "step": 17817
+    },
+    {
+      "epoch": 1.518363868768641,
+      "grad_norm": 12.5,
+      "learning_rate": 3.4377377335818846e-07,
+      "loss": 0.4767,
+      "step": 17818
+    },
+    {
+      "epoch": 1.518449083936941,
+      "grad_norm": 12.1875,
+      "learning_rate": 3.4365804388796125e-07,
+      "loss": 0.3505,
+      "step": 17819
+    },
+    {
+      "epoch": 1.5185342991052408,
+      "grad_norm": 17.375,
+      "learning_rate": 3.4354233079624406e-07,
+      "loss": 0.7322,
+      "step": 17820
+    },
+    {
+      "epoch": 1.5186195142735408,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.4342663408512863e-07,
+      "loss": 0.6824,
+      "step": 17821
+    },
+    {
+      "epoch": 1.5187047294418408,
+      "grad_norm": 16.125,
+      "learning_rate": 3.4331095375670534e-07,
+      "loss": 0.633,
+      "step": 17822
+    },
+    {
+      "epoch": 1.5187899446101407,
+      "grad_norm": 16.75,
+      "learning_rate": 3.431952898130654e-07,
+      "loss": 0.4536,
+      "step": 17823
+    },
+    {
+      "epoch": 1.5188751597784407,
+      "grad_norm": 17.25,
+      "learning_rate": 3.4307964225629835e-07,
+      "loss": 0.4228,
+      "step": 17824
+    },
+    {
+      "epoch": 1.5189603749467406,
+      "grad_norm": 18.75,
+      "learning_rate": 3.429640110884938e-07,
+      "loss": 0.7009,
+      "step": 17825
+    },
+    {
+      "epoch": 1.5190455901150406,
+      "grad_norm": 26.0,
+      "learning_rate": 3.4284839631174167e-07,
+      "loss": 0.6222,
+      "step": 17826
+    },
+    {
+      "epoch": 1.5191308052833405,
+      "grad_norm": 11.75,
+      "learning_rate": 3.4273279792813125e-07,
+      "loss": 0.4099,
+      "step": 17827
+    },
+    {
+      "epoch": 1.5192160204516405,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.42617215939752e-07,
+      "loss": 0.3199,
+      "step": 17828
+    },
+    {
+      "epoch": 1.5193012356199405,
+      "grad_norm": 16.75,
+      "learning_rate": 3.425016503486915e-07,
+      "loss": 0.5446,
+      "step": 17829
+    },
+    {
+      "epoch": 1.5193864507882404,
+      "grad_norm": 15.75,
+      "learning_rate": 3.423861011570388e-07,
+      "loss": 0.3547,
+      "step": 17830
+    },
+    {
+      "epoch": 1.5194716659565404,
+      "grad_norm": 18.375,
+      "learning_rate": 3.422705683668824e-07,
+      "loss": 0.6353,
+      "step": 17831
+    },
+    {
+      "epoch": 1.5195568811248403,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.421550519803095e-07,
+      "loss": 0.5042,
+      "step": 17832
+    },
+    {
+      "epoch": 1.5196420962931403,
+      "grad_norm": 16.5,
+      "learning_rate": 3.420395519994074e-07,
+      "loss": 0.4955,
+      "step": 17833
+    },
+    {
+      "epoch": 1.5197273114614402,
+      "grad_norm": 12.4375,
+      "learning_rate": 3.419240684262637e-07,
+      "loss": 0.5612,
+      "step": 17834
+    },
+    {
+      "epoch": 1.5198125266297402,
+      "grad_norm": 18.75,
+      "learning_rate": 3.418086012629651e-07,
+      "loss": 0.2756,
+      "step": 17835
+    },
+    {
+      "epoch": 1.5198977417980402,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.4169315051159877e-07,
+      "loss": 0.4937,
+      "step": 17836
+    },
+    {
+      "epoch": 1.5199829569663401,
+      "grad_norm": 17.125,
+      "learning_rate": 3.415777161742502e-07,
+      "loss": 0.7251,
+      "step": 17837
+    },
+    {
+      "epoch": 1.52006817213464,
+      "grad_norm": 15.75,
+      "learning_rate": 3.414622982530058e-07,
+      "loss": 0.8248,
+      "step": 17838
+    },
+    {
+      "epoch": 1.52015338730294,
+      "grad_norm": 14.875,
+      "learning_rate": 3.413468967499516e-07,
+      "loss": 0.6812,
+      "step": 17839
+    },
+    {
+      "epoch": 1.52023860247124,
+      "grad_norm": 18.125,
+      "learning_rate": 3.412315116671723e-07,
+      "loss": 0.4737,
+      "step": 17840
+    },
+    {
+      "epoch": 1.52032381763954,
+      "grad_norm": 12.25,
+      "learning_rate": 3.411161430067539e-07,
+      "loss": 0.3894,
+      "step": 17841
+    },
+    {
+      "epoch": 1.52040903280784,
+      "grad_norm": 13.625,
+      "learning_rate": 3.4100079077078034e-07,
+      "loss": 0.5425,
+      "step": 17842
+    },
+    {
+      "epoch": 1.5204942479761399,
+      "grad_norm": 48.5,
+      "learning_rate": 3.408854549613365e-07,
+      "loss": 0.3435,
+      "step": 17843
+    },
+    {
+      "epoch": 1.5205794631444398,
+      "grad_norm": 67.0,
+      "learning_rate": 3.407701355805071e-07,
+      "loss": 0.6264,
+      "step": 17844
+    },
+    {
+      "epoch": 1.5206646783127398,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.4065483263037534e-07,
+      "loss": 0.4741,
+      "step": 17845
+    },
+    {
+      "epoch": 1.5207498934810397,
+      "grad_norm": 25.0,
+      "learning_rate": 3.405395461130251e-07,
+      "loss": 0.7344,
+      "step": 17846
+    },
+    {
+      "epoch": 1.5208351086493397,
+      "grad_norm": 17.375,
+      "learning_rate": 3.404242760305401e-07,
+      "loss": 0.5876,
+      "step": 17847
+    },
+    {
+      "epoch": 1.5209203238176396,
+      "grad_norm": 18.625,
+      "learning_rate": 3.403090223850028e-07,
+      "loss": 0.3864,
+      "step": 17848
+    },
+    {
+      "epoch": 1.5210055389859396,
+      "grad_norm": 19.75,
+      "learning_rate": 3.4019378517849657e-07,
+      "loss": 0.6581,
+      "step": 17849
+    },
+    {
+      "epoch": 1.5210907541542396,
+      "grad_norm": 19.75,
+      "learning_rate": 3.4007856441310315e-07,
+      "loss": 0.7212,
+      "step": 17850
+    },
+    {
+      "epoch": 1.5211759693225395,
+      "grad_norm": 24.0,
+      "learning_rate": 3.399633600909051e-07,
+      "loss": 0.8558,
+      "step": 17851
+    },
+    {
+      "epoch": 1.5212611844908395,
+      "grad_norm": 15.5,
+      "learning_rate": 3.398481722139842e-07,
+      "loss": 0.5803,
+      "step": 17852
+    },
+    {
+      "epoch": 1.5213463996591394,
+      "grad_norm": 14.625,
+      "learning_rate": 3.397330007844225e-07,
+      "loss": 0.8765,
+      "step": 17853
+    },
+    {
+      "epoch": 1.5214316148274394,
+      "grad_norm": 30.25,
+      "learning_rate": 3.3961784580430033e-07,
+      "loss": 0.7677,
+      "step": 17854
+    },
+    {
+      "epoch": 1.5215168299957393,
+      "grad_norm": 16.25,
+      "learning_rate": 3.395027072756997e-07,
+      "loss": 0.516,
+      "step": 17855
+    },
+    {
+      "epoch": 1.5216020451640393,
+      "grad_norm": 19.5,
+      "learning_rate": 3.3938758520070017e-07,
+      "loss": 0.6145,
+      "step": 17856
+    },
+    {
+      "epoch": 1.5216872603323393,
+      "grad_norm": 20.0,
+      "learning_rate": 3.3927247958138275e-07,
+      "loss": 0.8716,
+      "step": 17857
+    },
+    {
+      "epoch": 1.5217724755006392,
+      "grad_norm": 15.625,
+      "learning_rate": 3.3915739041982784e-07,
+      "loss": 0.7272,
+      "step": 17858
+    },
+    {
+      "epoch": 1.5218576906689392,
+      "grad_norm": 19.125,
+      "learning_rate": 3.390423177181143e-07,
+      "loss": 0.7388,
+      "step": 17859
+    },
+    {
+      "epoch": 1.5219429058372391,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.389272614783222e-07,
+      "loss": 0.3343,
+      "step": 17860
+    },
+    {
+      "epoch": 1.522028121005539,
+      "grad_norm": 17.625,
+      "learning_rate": 3.388122217025311e-07,
+      "loss": 0.5333,
+      "step": 17861
+    },
+    {
+      "epoch": 1.522113336173839,
+      "grad_norm": 18.0,
+      "learning_rate": 3.3869719839281887e-07,
+      "loss": 0.5761,
+      "step": 17862
+    },
+    {
+      "epoch": 1.522198551342139,
+      "grad_norm": 18.5,
+      "learning_rate": 3.385821915512652e-07,
+      "loss": 0.6957,
+      "step": 17863
+    },
+    {
+      "epoch": 1.522283766510439,
+      "grad_norm": 17.875,
+      "learning_rate": 3.384672011799474e-07,
+      "loss": 0.7111,
+      "step": 17864
+    },
+    {
+      "epoch": 1.522368981678739,
+      "grad_norm": 24.0,
+      "learning_rate": 3.3835222728094393e-07,
+      "loss": 0.5566,
+      "step": 17865
+    },
+    {
+      "epoch": 1.5224541968470389,
+      "grad_norm": 17.375,
+      "learning_rate": 3.382372698563327e-07,
+      "loss": 0.6995,
+      "step": 17866
+    },
+    {
+      "epoch": 1.5225394120153388,
+      "grad_norm": 18.875,
+      "learning_rate": 3.3812232890819057e-07,
+      "loss": 0.6477,
+      "step": 17867
+    },
+    {
+      "epoch": 1.5226246271836388,
+      "grad_norm": 18.875,
+      "learning_rate": 3.380074044385949e-07,
+      "loss": 0.7522,
+      "step": 17868
+    },
+    {
+      "epoch": 1.5227098423519387,
+      "grad_norm": 20.0,
+      "learning_rate": 3.3789249644962247e-07,
+      "loss": 0.5775,
+      "step": 17869
+    },
+    {
+      "epoch": 1.5227950575202387,
+      "grad_norm": 18.625,
+      "learning_rate": 3.377776049433502e-07,
+      "loss": 0.8255,
+      "step": 17870
+    },
+    {
+      "epoch": 1.5228802726885386,
+      "grad_norm": 15.125,
+      "learning_rate": 3.3766272992185366e-07,
+      "loss": 0.759,
+      "step": 17871
+    },
+    {
+      "epoch": 1.5229654878568386,
+      "grad_norm": 12.0,
+      "learning_rate": 3.375478713872092e-07,
+      "loss": 0.2591,
+      "step": 17872
+    },
+    {
+      "epoch": 1.5230507030251386,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.374330293414921e-07,
+      "loss": 0.3141,
+      "step": 17873
+    },
+    {
+      "epoch": 1.5231359181934385,
+      "grad_norm": 30.25,
+      "learning_rate": 3.3731820378677793e-07,
+      "loss": 1.1391,
+      "step": 17874
+    },
+    {
+      "epoch": 1.5232211333617385,
+      "grad_norm": 18.125,
+      "learning_rate": 3.372033947251414e-07,
+      "loss": 0.6465,
+      "step": 17875
+    },
+    {
+      "epoch": 1.5233063485300384,
+      "grad_norm": 16.25,
+      "learning_rate": 3.370886021586574e-07,
+      "loss": 0.9742,
+      "step": 17876
+    },
+    {
+      "epoch": 1.5233915636983384,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.3697382608940024e-07,
+      "loss": 0.3066,
+      "step": 17877
+    },
+    {
+      "epoch": 1.5234767788666383,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.3685906651944477e-07,
+      "loss": 0.3006,
+      "step": 17878
+    },
+    {
+      "epoch": 1.5235619940349383,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.3674432345086356e-07,
+      "loss": 0.6079,
+      "step": 17879
+    },
+    {
+      "epoch": 1.5236472092032383,
+      "grad_norm": 21.125,
+      "learning_rate": 3.366295968857314e-07,
+      "loss": 0.7518,
+      "step": 17880
+    },
+    {
+      "epoch": 1.5237324243715382,
+      "grad_norm": 13.875,
+      "learning_rate": 3.365148868261203e-07,
+      "loss": 0.5284,
+      "step": 17881
+    },
+    {
+      "epoch": 1.5238176395398382,
+      "grad_norm": 17.375,
+      "learning_rate": 3.3640019327410387e-07,
+      "loss": 0.7063,
+      "step": 17882
+    },
+    {
+      "epoch": 1.5239028547081381,
+      "grad_norm": 15.0,
+      "learning_rate": 3.362855162317551e-07,
+      "loss": 0.5995,
+      "step": 17883
+    },
+    {
+      "epoch": 1.523988069876438,
+      "grad_norm": 12.125,
+      "learning_rate": 3.3617085570114534e-07,
+      "loss": 0.3809,
+      "step": 17884
+    },
+    {
+      "epoch": 1.524073285044738,
+      "grad_norm": 17.125,
+      "learning_rate": 3.3605621168434717e-07,
+      "loss": 0.5204,
+      "step": 17885
+    },
+    {
+      "epoch": 1.524158500213038,
+      "grad_norm": 12.75,
+      "learning_rate": 3.3594158418343235e-07,
+      "loss": 0.2736,
+      "step": 17886
+    },
+    {
+      "epoch": 1.524243715381338,
+      "grad_norm": 20.375,
+      "learning_rate": 3.358269732004726e-07,
+      "loss": 0.6167,
+      "step": 17887
+    },
+    {
+      "epoch": 1.524328930549638,
+      "grad_norm": 11.5,
+      "learning_rate": 3.3571237873753883e-07,
+      "loss": 0.3267,
+      "step": 17888
+    },
+    {
+      "epoch": 1.5244141457179379,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.3559780079670117e-07,
+      "loss": 0.5807,
+      "step": 17889
+    },
+    {
+      "epoch": 1.5244993608862378,
+      "grad_norm": 10.875,
+      "learning_rate": 3.3548323938003084e-07,
+      "loss": 0.2132,
+      "step": 17890
+    },
+    {
+      "epoch": 1.5245845760545378,
+      "grad_norm": 18.125,
+      "learning_rate": 3.3536869448959843e-07,
+      "loss": 0.4648,
+      "step": 17891
+    },
+    {
+      "epoch": 1.5246697912228377,
+      "grad_norm": 14.375,
+      "learning_rate": 3.35254166127473e-07,
+      "loss": 0.471,
+      "step": 17892
+    },
+    {
+      "epoch": 1.5247550063911377,
+      "grad_norm": 19.75,
+      "learning_rate": 3.351396542957247e-07,
+      "loss": 0.8795,
+      "step": 17893
+    },
+    {
+      "epoch": 1.5248402215594377,
+      "grad_norm": 12.5,
+      "learning_rate": 3.350251589964229e-07,
+      "loss": 0.5426,
+      "step": 17894
+    },
+    {
+      "epoch": 1.5249254367277376,
+      "grad_norm": 18.5,
+      "learning_rate": 3.3491068023163683e-07,
+      "loss": 0.9085,
+      "step": 17895
+    },
+    {
+      "epoch": 1.5250106518960376,
+      "grad_norm": 18.75,
+      "learning_rate": 3.347962180034349e-07,
+      "loss": 0.7864,
+      "step": 17896
+    },
+    {
+      "epoch": 1.5250958670643375,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.3468177231388526e-07,
+      "loss": 0.403,
+      "step": 17897
+    },
+    {
+      "epoch": 1.5251810822326375,
+      "grad_norm": 19.75,
+      "learning_rate": 3.3456734316505645e-07,
+      "loss": 0.5928,
+      "step": 17898
+    },
+    {
+      "epoch": 1.5252662974009374,
+      "grad_norm": 17.375,
+      "learning_rate": 3.3445293055901636e-07,
+      "loss": 0.4978,
+      "step": 17899
+    },
+    {
+      "epoch": 1.5253515125692374,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.343385344978327e-07,
+      "loss": 0.5398,
+      "step": 17900
+    },
+    {
+      "epoch": 1.5254367277375374,
+      "grad_norm": 18.75,
+      "learning_rate": 3.342241549835722e-07,
+      "loss": 0.4468,
+      "step": 17901
+    },
+    {
+      "epoch": 1.5255219429058373,
+      "grad_norm": 26.625,
+      "learning_rate": 3.34109792018302e-07,
+      "loss": 0.6135,
+      "step": 17902
+    },
+    {
+      "epoch": 1.5256071580741373,
+      "grad_norm": 20.875,
+      "learning_rate": 3.3399544560408907e-07,
+      "loss": 0.8641,
+      "step": 17903
+    },
+    {
+      "epoch": 1.5256923732424372,
+      "grad_norm": 11.375,
+      "learning_rate": 3.338811157430001e-07,
+      "loss": 0.3021,
+      "step": 17904
+    },
+    {
+      "epoch": 1.5257775884107372,
+      "grad_norm": 14.125,
+      "learning_rate": 3.337668024370999e-07,
+      "loss": 0.5622,
+      "step": 17905
+    },
+    {
+      "epoch": 1.5258628035790371,
+      "grad_norm": 12.875,
+      "learning_rate": 3.33652505688455e-07,
+      "loss": 0.6561,
+      "step": 17906
+    },
+    {
+      "epoch": 1.525948018747337,
+      "grad_norm": 16.625,
+      "learning_rate": 3.3353822549913066e-07,
+      "loss": 0.4569,
+      "step": 17907
+    },
+    {
+      "epoch": 1.526033233915637,
+      "grad_norm": 27.125,
+      "learning_rate": 3.3342396187119266e-07,
+      "loss": 0.9026,
+      "step": 17908
+    },
+    {
+      "epoch": 1.526118449083937,
+      "grad_norm": 18.25,
+      "learning_rate": 3.3330971480670495e-07,
+      "loss": 0.781,
+      "step": 17909
+    },
+    {
+      "epoch": 1.526203664252237,
+      "grad_norm": 13.75,
+      "learning_rate": 3.3319548430773244e-07,
+      "loss": 0.6328,
+      "step": 17910
+    },
+    {
+      "epoch": 1.526288879420537,
+      "grad_norm": 15.25,
+      "learning_rate": 3.3308127037633945e-07,
+      "loss": 0.7338,
+      "step": 17911
+    },
+    {
+      "epoch": 1.5263740945888369,
+      "grad_norm": 23.125,
+      "learning_rate": 3.329670730145904e-07,
+      "loss": 0.9208,
+      "step": 17912
+    },
+    {
+      "epoch": 1.5264593097571368,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.3285289222454846e-07,
+      "loss": 0.5795,
+      "step": 17913
+    },
+    {
+      "epoch": 1.5265445249254368,
+      "grad_norm": 19.75,
+      "learning_rate": 3.327387280082765e-07,
+      "loss": 0.7229,
+      "step": 17914
+    },
+    {
+      "epoch": 1.5266297400937368,
+      "grad_norm": 14.5,
+      "learning_rate": 3.3262458036783837e-07,
+      "loss": 0.3938,
+      "step": 17915
+    },
+    {
+      "epoch": 1.5267149552620367,
+      "grad_norm": 18.625,
+      "learning_rate": 3.3251044930529634e-07,
+      "loss": 0.9254,
+      "step": 17916
+    },
+    {
+      "epoch": 1.5268001704303367,
+      "grad_norm": 15.375,
+      "learning_rate": 3.3239633482271355e-07,
+      "loss": 0.4848,
+      "step": 17917
+    },
+    {
+      "epoch": 1.5268853855986366,
+      "grad_norm": 18.375,
+      "learning_rate": 3.3228223692215134e-07,
+      "loss": 0.8506,
+      "step": 17918
+    },
+    {
+      "epoch": 1.5269706007669366,
+      "grad_norm": 12.625,
+      "learning_rate": 3.321681556056719e-07,
+      "loss": 0.3717,
+      "step": 17919
+    },
+    {
+      "epoch": 1.5270558159352365,
+      "grad_norm": 17.375,
+      "learning_rate": 3.3205409087533727e-07,
+      "loss": 0.4057,
+      "step": 17920
+    },
+    {
+      "epoch": 1.5271410311035365,
+      "grad_norm": 17.5,
+      "learning_rate": 3.319400427332083e-07,
+      "loss": 0.7842,
+      "step": 17921
+    },
+    {
+      "epoch": 1.5272262462718365,
+      "grad_norm": 18.875,
+      "learning_rate": 3.3182601118134553e-07,
+      "loss": 0.5639,
+      "step": 17922
+    },
+    {
+      "epoch": 1.5273114614401364,
+      "grad_norm": 17.375,
+      "learning_rate": 3.3171199622181e-07,
+      "loss": 0.782,
+      "step": 17923
+    },
+    {
+      "epoch": 1.5273966766084364,
+      "grad_norm": 32.75,
+      "learning_rate": 3.315979978566622e-07,
+      "loss": 1.1505,
+      "step": 17924
+    },
+    {
+      "epoch": 1.5274818917767363,
+      "grad_norm": 16.0,
+      "learning_rate": 3.314840160879623e-07,
+      "loss": 0.5953,
+      "step": 17925
+    },
+    {
+      "epoch": 1.5275671069450363,
+      "grad_norm": 15.125,
+      "learning_rate": 3.3137005091776973e-07,
+      "loss": 0.5799,
+      "step": 17926
+    },
+    {
+      "epoch": 1.5276523221133362,
+      "grad_norm": 17.0,
+      "learning_rate": 3.3125610234814395e-07,
+      "loss": 0.514,
+      "step": 17927
+    },
+    {
+      "epoch": 1.5277375372816362,
+      "grad_norm": 11.625,
+      "learning_rate": 3.311421703811446e-07,
+      "loss": 0.2853,
+      "step": 17928
+    },
+    {
+      "epoch": 1.5278227524499362,
+      "grad_norm": 13.9375,
+      "learning_rate": 3.3102825501883e-07,
+      "loss": 0.6798,
+      "step": 17929
+    },
+    {
+      "epoch": 1.527907967618236,
+      "grad_norm": 15.75,
+      "learning_rate": 3.309143562632591e-07,
+      "loss": 0.5874,
+      "step": 17930
+    },
+    {
+      "epoch": 1.527993182786536,
+      "grad_norm": 12.9375,
+      "learning_rate": 3.3080047411648984e-07,
+      "loss": 0.4893,
+      "step": 17931
+    },
+    {
+      "epoch": 1.528078397954836,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.3068660858058013e-07,
+      "loss": 0.6891,
+      "step": 17932
+    },
+    {
+      "epoch": 1.528163613123136,
+      "grad_norm": 19.0,
+      "learning_rate": 3.305727596575881e-07,
+      "loss": 0.8052,
+      "step": 17933
+    },
+    {
+      "epoch": 1.528248828291436,
+      "grad_norm": 12.125,
+      "learning_rate": 3.3045892734957106e-07,
+      "loss": 0.4717,
+      "step": 17934
+    },
+    {
+      "epoch": 1.528334043459736,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.303451116585855e-07,
+      "loss": 0.517,
+      "step": 17935
+    },
+    {
+      "epoch": 1.5284192586280358,
+      "grad_norm": 14.25,
+      "learning_rate": 3.302313125866885e-07,
+      "loss": 0.473,
+      "step": 17936
+    },
+    {
+      "epoch": 1.5285044737963358,
+      "grad_norm": 19.75,
+      "learning_rate": 3.30117530135937e-07,
+      "loss": 0.5991,
+      "step": 17937
+    },
+    {
+      "epoch": 1.5285896889646358,
+      "grad_norm": 25.75,
+      "learning_rate": 3.3000376430838687e-07,
+      "loss": 0.6874,
+      "step": 17938
+    },
+    {
+      "epoch": 1.5286749041329357,
+      "grad_norm": 17.375,
+      "learning_rate": 3.298900151060934e-07,
+      "loss": 0.588,
+      "step": 17939
+    },
+    {
+      "epoch": 1.5287601193012357,
+      "grad_norm": 14.25,
+      "learning_rate": 3.297762825311125e-07,
+      "loss": 0.5701,
+      "step": 17940
+    },
+    {
+      "epoch": 1.5288453344695356,
+      "grad_norm": 16.125,
+      "learning_rate": 3.2966256658549966e-07,
+      "loss": 0.4914,
+      "step": 17941
+    },
+    {
+      "epoch": 1.5289305496378356,
+      "grad_norm": 10.75,
+      "learning_rate": 3.2954886727131e-07,
+      "loss": 0.2853,
+      "step": 17942
+    },
+    {
+      "epoch": 1.5290157648061355,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.2943518459059755e-07,
+      "loss": 0.606,
+      "step": 17943
+    },
+    {
+      "epoch": 1.5291009799744355,
+      "grad_norm": 19.5,
+      "learning_rate": 3.293215185454171e-07,
+      "loss": 0.7778,
+      "step": 17944
+    },
+    {
+      "epoch": 1.5291861951427355,
+      "grad_norm": 18.25,
+      "learning_rate": 3.292078691378228e-07,
+      "loss": 0.7281,
+      "step": 17945
+    },
+    {
+      "epoch": 1.5292714103110354,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.2909423636986806e-07,
+      "loss": 0.963,
+      "step": 17946
+    },
+    {
+      "epoch": 1.5293566254793354,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.289806202436067e-07,
+      "loss": 0.3774,
+      "step": 17947
+    },
+    {
+      "epoch": 1.5294418406476353,
+      "grad_norm": 14.875,
+      "learning_rate": 3.2886702076109155e-07,
+      "loss": 0.5016,
+      "step": 17948
+    },
+    {
+      "epoch": 1.5295270558159353,
+      "grad_norm": 16.75,
+      "learning_rate": 3.2875343792437537e-07,
+      "loss": 0.6474,
+      "step": 17949
+    },
+    {
+      "epoch": 1.5296122709842352,
+      "grad_norm": 17.75,
+      "learning_rate": 3.28639871735511e-07,
+      "loss": 0.9139,
+      "step": 17950
+    },
+    {
+      "epoch": 1.5296974861525352,
+      "grad_norm": 24.5,
+      "learning_rate": 3.2852632219655123e-07,
+      "loss": 0.6568,
+      "step": 17951
+    },
+    {
+      "epoch": 1.5297827013208352,
+      "grad_norm": 13.5,
+      "learning_rate": 3.284127893095468e-07,
+      "loss": 0.315,
+      "step": 17952
+    },
+    {
+      "epoch": 1.5298679164891351,
+      "grad_norm": 12.5,
+      "learning_rate": 3.282992730765505e-07,
+      "loss": 0.2429,
+      "step": 17953
+    },
+    {
+      "epoch": 1.529953131657435,
+      "grad_norm": 22.0,
+      "learning_rate": 3.2818577349961274e-07,
+      "loss": 0.8803,
+      "step": 17954
+    },
+    {
+      "epoch": 1.530038346825735,
+      "grad_norm": 22.25,
+      "learning_rate": 3.280722905807854e-07,
+      "loss": 0.6947,
+      "step": 17955
+    },
+    {
+      "epoch": 1.530123561994035,
+      "grad_norm": 18.625,
+      "learning_rate": 3.2795882432211853e-07,
+      "loss": 0.8352,
+      "step": 17956
+    },
+    {
+      "epoch": 1.530208777162335,
+      "grad_norm": 24.0,
+      "learning_rate": 3.2784537472566283e-07,
+      "loss": 0.7527,
+      "step": 17957
+    },
+    {
+      "epoch": 1.530293992330635,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.277319417934685e-07,
+      "loss": 0.6465,
+      "step": 17958
+    },
+    {
+      "epoch": 1.5303792074989349,
+      "grad_norm": 17.625,
+      "learning_rate": 3.2761852552758587e-07,
+      "loss": 0.9668,
+      "step": 17959
+    },
+    {
+      "epoch": 1.5304644226672348,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.275051259300635e-07,
+      "loss": 0.4368,
+      "step": 17960
+    },
+    {
+      "epoch": 1.5305496378355348,
+      "grad_norm": 14.125,
+      "learning_rate": 3.273917430029515e-07,
+      "loss": 0.7976,
+      "step": 17961
+    },
+    {
+      "epoch": 1.5306348530038347,
+      "grad_norm": 19.0,
+      "learning_rate": 3.2727837674829817e-07,
+      "loss": 0.8612,
+      "step": 17962
+    },
+    {
+      "epoch": 1.5307200681721347,
+      "grad_norm": 16.0,
+      "learning_rate": 3.2716502716815245e-07,
+      "loss": 0.6077,
+      "step": 17963
+    },
+    {
+      "epoch": 1.5308052833404346,
+      "grad_norm": 36.0,
+      "learning_rate": 3.2705169426456297e-07,
+      "loss": 0.6216,
+      "step": 17964
+    },
+    {
+      "epoch": 1.5308904985087346,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.269383780395773e-07,
+      "loss": 0.481,
+      "step": 17965
+    },
+    {
+      "epoch": 1.5309757136770346,
+      "grad_norm": 19.375,
+      "learning_rate": 3.268250784952431e-07,
+      "loss": 0.4499,
+      "step": 17966
+    },
+    {
+      "epoch": 1.5310609288453345,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.267117956336081e-07,
+      "loss": 0.3486,
+      "step": 17967
+    },
+    {
+      "epoch": 1.5311461440136345,
+      "grad_norm": 13.9375,
+      "learning_rate": 3.265985294567199e-07,
+      "loss": 0.5763,
+      "step": 17968
+    },
+    {
+      "epoch": 1.5312313591819344,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.264852799666247e-07,
+      "loss": 0.5514,
+      "step": 17969
+    },
+    {
+      "epoch": 1.5313165743502344,
+      "grad_norm": 17.25,
+      "learning_rate": 3.263720471653688e-07,
+      "loss": 0.7823,
+      "step": 17970
+    },
+    {
+      "epoch": 1.5314017895185343,
+      "grad_norm": 15.75,
+      "learning_rate": 3.2625883105499886e-07,
+      "loss": 0.4922,
+      "step": 17971
+    },
+    {
+      "epoch": 1.5314870046868343,
+      "grad_norm": 16.375,
+      "learning_rate": 3.261456316375611e-07,
+      "loss": 0.5687,
+      "step": 17972
+    },
+    {
+      "epoch": 1.5315722198551343,
+      "grad_norm": 17.5,
+      "learning_rate": 3.2603244891510025e-07,
+      "loss": 0.8644,
+      "step": 17973
+    },
+    {
+      "epoch": 1.5316574350234342,
+      "grad_norm": 10.8125,
+      "learning_rate": 3.2591928288966235e-07,
+      "loss": 0.2395,
+      "step": 17974
+    },
+    {
+      "epoch": 1.5317426501917342,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.2580613356329214e-07,
+      "loss": 0.6047,
+      "step": 17975
+    },
+    {
+      "epoch": 1.5318278653600341,
+      "grad_norm": 24.0,
+      "learning_rate": 3.256930009380349e-07,
+      "loss": 0.8693,
+      "step": 17976
+    },
+    {
+      "epoch": 1.531913080528334,
+      "grad_norm": 21.375,
+      "learning_rate": 3.255798850159346e-07,
+      "loss": 0.8947,
+      "step": 17977
+    },
+    {
+      "epoch": 1.531998295696634,
+      "grad_norm": 21.5,
+      "learning_rate": 3.2546678579903487e-07,
+      "loss": 0.919,
+      "step": 17978
+    },
+    {
+      "epoch": 1.532083510864934,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.2535370328938015e-07,
+      "loss": 0.7489,
+      "step": 17979
+    },
+    {
+      "epoch": 1.532168726033234,
+      "grad_norm": 14.0,
+      "learning_rate": 3.2524063748901367e-07,
+      "loss": 0.4281,
+      "step": 17980
+    },
+    {
+      "epoch": 1.532253941201534,
+      "grad_norm": 17.375,
+      "learning_rate": 3.251275883999792e-07,
+      "loss": 0.8333,
+      "step": 17981
+    },
+    {
+      "epoch": 1.5323391563698339,
+      "grad_norm": 14.875,
+      "learning_rate": 3.25014556024319e-07,
+      "loss": 0.4825,
+      "step": 17982
+    },
+    {
+      "epoch": 1.5324243715381338,
+      "grad_norm": 16.75,
+      "learning_rate": 3.249015403640758e-07,
+      "loss": 0.6794,
+      "step": 17983
+    },
+    {
+      "epoch": 1.5325095867064338,
+      "grad_norm": 20.25,
+      "learning_rate": 3.247885414212924e-07,
+      "loss": 0.5464,
+      "step": 17984
+    },
+    {
+      "epoch": 1.5325948018747337,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.2467555919801037e-07,
+      "loss": 0.5698,
+      "step": 17985
+    },
+    {
+      "epoch": 1.5326800170430337,
+      "grad_norm": 12.5,
+      "learning_rate": 3.245625936962712e-07,
+      "loss": 0.267,
+      "step": 17986
+    },
+    {
+      "epoch": 1.5327652322113337,
+      "grad_norm": 16.125,
+      "learning_rate": 3.2444964491811654e-07,
+      "loss": 0.5837,
+      "step": 17987
+    },
+    {
+      "epoch": 1.5328504473796336,
+      "grad_norm": 12.9375,
+      "learning_rate": 3.243367128655875e-07,
+      "loss": 0.3568,
+      "step": 17988
+    },
+    {
+      "epoch": 1.5329356625479336,
+      "grad_norm": 16.25,
+      "learning_rate": 3.2422379754072533e-07,
+      "loss": 0.6231,
+      "step": 17989
+    },
+    {
+      "epoch": 1.5330208777162335,
+      "grad_norm": 12.75,
+      "learning_rate": 3.2411089894556965e-07,
+      "loss": 0.4442,
+      "step": 17990
+    },
+    {
+      "epoch": 1.5331060928845335,
+      "grad_norm": 18.25,
+      "learning_rate": 3.239980170821611e-07,
+      "loss": 0.9889,
+      "step": 17991
+    },
+    {
+      "epoch": 1.5331913080528334,
+      "grad_norm": 13.125,
+      "learning_rate": 3.238851519525395e-07,
+      "loss": 0.7207,
+      "step": 17992
+    },
+    {
+      "epoch": 1.5332765232211334,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.237723035587448e-07,
+      "loss": 0.4521,
+      "step": 17993
+    },
+    {
+      "epoch": 1.5333617383894333,
+      "grad_norm": 16.375,
+      "learning_rate": 3.2365947190281606e-07,
+      "loss": 0.4908,
+      "step": 17994
+    },
+    {
+      "epoch": 1.5334469535577333,
+      "grad_norm": 16.375,
+      "learning_rate": 3.235466569867918e-07,
+      "loss": 0.7526,
+      "step": 17995
+    },
+    {
+      "epoch": 1.5335321687260333,
+      "grad_norm": 13.125,
+      "learning_rate": 3.23433858812711e-07,
+      "loss": 0.5891,
+      "step": 17996
+    },
+    {
+      "epoch": 1.5336173838943332,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.233210773826122e-07,
+      "loss": 0.7766,
+      "step": 17997
+    },
+    {
+      "epoch": 1.5337025990626332,
+      "grad_norm": 17.5,
+      "learning_rate": 3.232083126985336e-07,
+      "loss": 0.7337,
+      "step": 17998
+    },
+    {
+      "epoch": 1.5337878142309331,
+      "grad_norm": 17.875,
+      "learning_rate": 3.2309556476251254e-07,
+      "loss": 0.6895,
+      "step": 17999
+    },
+    {
+      "epoch": 1.533873029399233,
+      "grad_norm": 11.375,
+      "learning_rate": 3.2298283357658666e-07,
+      "loss": 0.3635,
+      "step": 18000
+    },
+    {
+      "epoch": 1.533958244567533,
+      "grad_norm": 17.375,
+      "learning_rate": 3.2287011914279354e-07,
+      "loss": 0.486,
+      "step": 18001
+    },
+    {
+      "epoch": 1.534043459735833,
+      "grad_norm": 42.0,
+      "learning_rate": 3.2275742146316955e-07,
+      "loss": 0.7138,
+      "step": 18002
+    },
+    {
+      "epoch": 1.534128674904133,
+      "grad_norm": 17.625,
+      "learning_rate": 3.226447405397512e-07,
+      "loss": 0.9913,
+      "step": 18003
+    },
+    {
+      "epoch": 1.534213890072433,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.2253207637457473e-07,
+      "loss": 0.6176,
+      "step": 18004
+    },
+    {
+      "epoch": 1.5342991052407329,
+      "grad_norm": 20.25,
+      "learning_rate": 3.2241942896967637e-07,
+      "loss": 0.7447,
+      "step": 18005
+    },
+    {
+      "epoch": 1.5343843204090328,
+      "grad_norm": 20.0,
+      "learning_rate": 3.2230679832709205e-07,
+      "loss": 0.7433,
+      "step": 18006
+    },
+    {
+      "epoch": 1.5344695355773328,
+      "grad_norm": 16.875,
+      "learning_rate": 3.2219418444885646e-07,
+      "loss": 0.4164,
+      "step": 18007
+    },
+    {
+      "epoch": 1.5345547507456327,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.220815873370048e-07,
+      "loss": 0.728,
+      "step": 18008
+    },
+    {
+      "epoch": 1.5346399659139327,
+      "grad_norm": 13.25,
+      "learning_rate": 3.21969006993572e-07,
+      "loss": 0.327,
+      "step": 18009
+    },
+    {
+      "epoch": 1.5347251810822327,
+      "grad_norm": 18.5,
+      "learning_rate": 3.2185644342059285e-07,
+      "loss": 0.8874,
+      "step": 18010
+    },
+    {
+      "epoch": 1.5348103962505326,
+      "grad_norm": 18.0,
+      "learning_rate": 3.21743896620101e-07,
+      "loss": 1.0342,
+      "step": 18011
+    },
+    {
+      "epoch": 1.5348956114188326,
+      "grad_norm": 21.0,
+      "learning_rate": 3.2163136659413007e-07,
+      "loss": 0.7385,
+      "step": 18012
+    },
+    {
+      "epoch": 1.5349808265871325,
+      "grad_norm": 41.0,
+      "learning_rate": 3.215188533447139e-07,
+      "loss": 0.3905,
+      "step": 18013
+    },
+    {
+      "epoch": 1.5350660417554325,
+      "grad_norm": 15.875,
+      "learning_rate": 3.214063568738861e-07,
+      "loss": 1.0007,
+      "step": 18014
+    },
+    {
+      "epoch": 1.5351512569237324,
+      "grad_norm": 15.75,
+      "learning_rate": 3.2129387718367875e-07,
+      "loss": 0.5405,
+      "step": 18015
+    },
+    {
+      "epoch": 1.5352364720920324,
+      "grad_norm": 17.25,
+      "learning_rate": 3.211814142761249e-07,
+      "loss": 0.461,
+      "step": 18016
+    },
+    {
+      "epoch": 1.5353216872603324,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.210689681532571e-07,
+      "loss": 0.926,
+      "step": 18017
+    },
+    {
+      "epoch": 1.5354069024286323,
+      "grad_norm": 18.5,
+      "learning_rate": 3.2095653881710735e-07,
+      "loss": 0.8262,
+      "step": 18018
+    },
+    {
+      "epoch": 1.5354921175969323,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.208441262697072e-07,
+      "loss": 0.5203,
+      "step": 18019
+    },
+    {
+      "epoch": 1.5355773327652322,
+      "grad_norm": 12.5,
+      "learning_rate": 3.2073173051308774e-07,
+      "loss": 0.5053,
+      "step": 18020
+    },
+    {
+      "epoch": 1.5356625479335322,
+      "grad_norm": 18.875,
+      "learning_rate": 3.206193515492803e-07,
+      "loss": 0.5634,
+      "step": 18021
+    },
+    {
+      "epoch": 1.5357477631018321,
+      "grad_norm": 17.75,
+      "learning_rate": 3.2050698938031577e-07,
+      "loss": 0.7365,
+      "step": 18022
+    },
+    {
+      "epoch": 1.535832978270132,
+      "grad_norm": 15.625,
+      "learning_rate": 3.2039464400822494e-07,
+      "loss": 0.7466,
+      "step": 18023
+    },
+    {
+      "epoch": 1.535918193438432,
+      "grad_norm": 23.625,
+      "learning_rate": 3.202823154350372e-07,
+      "loss": 0.678,
+      "step": 18024
+    },
+    {
+      "epoch": 1.536003408606732,
+      "grad_norm": 31.875,
+      "learning_rate": 3.201700036627832e-07,
+      "loss": 0.9351,
+      "step": 18025
+    },
+    {
+      "epoch": 1.536088623775032,
+      "grad_norm": 13.75,
+      "learning_rate": 3.2005770869349236e-07,
+      "loss": 0.3708,
+      "step": 18026
+    },
+    {
+      "epoch": 1.536173838943332,
+      "grad_norm": 16.875,
+      "learning_rate": 3.199454305291937e-07,
+      "loss": 0.7273,
+      "step": 18027
+    },
+    {
+      "epoch": 1.5362590541116319,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.1983316917191654e-07,
+      "loss": 0.5334,
+      "step": 18028
+    },
+    {
+      "epoch": 1.5363442692799318,
+      "grad_norm": 11.625,
+      "learning_rate": 3.19720924623689e-07,
+      "loss": 0.3952,
+      "step": 18029
+    },
+    {
+      "epoch": 1.5364294844482318,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.1960869688654e-07,
+      "loss": 0.6862,
+      "step": 18030
+    },
+    {
+      "epoch": 1.5365146996165318,
+      "grad_norm": 14.25,
+      "learning_rate": 3.194964859624977e-07,
+      "loss": 0.3309,
+      "step": 18031
+    },
+    {
+      "epoch": 1.5365999147848317,
+      "grad_norm": 18.25,
+      "learning_rate": 3.1938429185358913e-07,
+      "loss": 1.137,
+      "step": 18032
+    },
+    {
+      "epoch": 1.5366851299531317,
+      "grad_norm": 13.0,
+      "learning_rate": 3.192721145618424e-07,
+      "loss": 0.3966,
+      "step": 18033
+    },
+    {
+      "epoch": 1.5367703451214316,
+      "grad_norm": 16.0,
+      "learning_rate": 3.191599540892848e-07,
+      "loss": 0.5956,
+      "step": 18034
+    },
+    {
+      "epoch": 1.5368555602897316,
+      "grad_norm": 17.375,
+      "learning_rate": 3.1904781043794244e-07,
+      "loss": 0.7526,
+      "step": 18035
+    },
+    {
+      "epoch": 1.5369407754580315,
+      "grad_norm": 17.625,
+      "learning_rate": 3.189356836098426e-07,
+      "loss": 0.5607,
+      "step": 18036
+    },
+    {
+      "epoch": 1.5370259906263315,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.1882357360701095e-07,
+      "loss": 0.6523,
+      "step": 18037
+    },
+    {
+      "epoch": 1.5371112057946315,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.1871148043147366e-07,
+      "loss": 0.2046,
+      "step": 18038
+    },
+    {
+      "epoch": 1.5371964209629314,
+      "grad_norm": 17.25,
+      "learning_rate": 3.1859940408525637e-07,
+      "loss": 0.5863,
+      "step": 18039
+    },
+    {
+      "epoch": 1.5372816361312314,
+      "grad_norm": 19.375,
+      "learning_rate": 3.1848734457038483e-07,
+      "loss": 0.4456,
+      "step": 18040
+    },
+    {
+      "epoch": 1.5373668512995313,
+      "grad_norm": 19.5,
+      "learning_rate": 3.183753018888833e-07,
+      "loss": 0.4274,
+      "step": 18041
+    },
+    {
+      "epoch": 1.5374520664678313,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.182632760427773e-07,
+      "loss": 0.3521,
+      "step": 18042
+    },
+    {
+      "epoch": 1.5375372816361312,
+      "grad_norm": 16.625,
+      "learning_rate": 3.181512670340904e-07,
+      "loss": 0.7194,
+      "step": 18043
+    },
+    {
+      "epoch": 1.5376224968044312,
+      "grad_norm": 14.125,
+      "learning_rate": 3.1803927486484755e-07,
+      "loss": 0.5185,
+      "step": 18044
+    },
+    {
+      "epoch": 1.5377077119727312,
+      "grad_norm": 13.875,
+      "learning_rate": 3.1792729953707176e-07,
+      "loss": 0.6,
+      "step": 18045
+    },
+    {
+      "epoch": 1.537792927141031,
+      "grad_norm": 17.25,
+      "learning_rate": 3.1781534105278705e-07,
+      "loss": 0.6751,
+      "step": 18046
+    },
+    {
+      "epoch": 1.537878142309331,
+      "grad_norm": 18.375,
+      "learning_rate": 3.1770339941401646e-07,
+      "loss": 0.7212,
+      "step": 18047
+    },
+    {
+      "epoch": 1.537963357477631,
+      "grad_norm": 15.375,
+      "learning_rate": 3.1759147462278333e-07,
+      "loss": 0.6568,
+      "step": 18048
+    },
+    {
+      "epoch": 1.538048572645931,
+      "grad_norm": 11.5,
+      "learning_rate": 3.174795666811094e-07,
+      "loss": 0.4066,
+      "step": 18049
+    },
+    {
+      "epoch": 1.538133787814231,
+      "grad_norm": 17.625,
+      "learning_rate": 3.173676755910178e-07,
+      "loss": 0.5937,
+      "step": 18050
+    },
+    {
+      "epoch": 1.538219002982531,
+      "grad_norm": 10.5625,
+      "learning_rate": 3.1725580135452986e-07,
+      "loss": 0.285,
+      "step": 18051
+    },
+    {
+      "epoch": 1.5383042181508308,
+      "grad_norm": 16.25,
+      "learning_rate": 3.1714394397366745e-07,
+      "loss": 0.3227,
+      "step": 18052
+    },
+    {
+      "epoch": 1.5383894333191308,
+      "grad_norm": 18.0,
+      "learning_rate": 3.170321034504525e-07,
+      "loss": 0.6471,
+      "step": 18053
+    },
+    {
+      "epoch": 1.5384746484874308,
+      "grad_norm": 18.75,
+      "learning_rate": 3.1692027978690516e-07,
+      "loss": 0.7491,
+      "step": 18054
+    },
+    {
+      "epoch": 1.5385598636557307,
+      "grad_norm": 23.25,
+      "learning_rate": 3.1680847298504665e-07,
+      "loss": 0.7991,
+      "step": 18055
+    },
+    {
+      "epoch": 1.5386450788240307,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.1669668304689747e-07,
+      "loss": 0.7255,
+      "step": 18056
+    },
+    {
+      "epoch": 1.5387302939923306,
+      "grad_norm": 22.875,
+      "learning_rate": 3.1658490997447807e-07,
+      "loss": 0.8376,
+      "step": 18057
+    },
+    {
+      "epoch": 1.5388155091606306,
+      "grad_norm": 16.875,
+      "learning_rate": 3.1647315376980803e-07,
+      "loss": 0.6293,
+      "step": 18058
+    },
+    {
+      "epoch": 1.5389007243289305,
+      "grad_norm": 21.75,
+      "learning_rate": 3.163614144349064e-07,
+      "loss": 0.9939,
+      "step": 18059
+    },
+    {
+      "epoch": 1.5389859394972305,
+      "grad_norm": 17.625,
+      "learning_rate": 3.1624969197179294e-07,
+      "loss": 0.5389,
+      "step": 18060
+    },
+    {
+      "epoch": 1.5390711546655305,
+      "grad_norm": 13.25,
+      "learning_rate": 3.1613798638248675e-07,
+      "loss": 0.3476,
+      "step": 18061
+    },
+    {
+      "epoch": 1.5391563698338304,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.1602629766900596e-07,
+      "loss": 0.2934,
+      "step": 18062
+    },
+    {
+      "epoch": 1.5392415850021304,
+      "grad_norm": 17.75,
+      "learning_rate": 3.159146258333691e-07,
+      "loss": 0.8337,
+      "step": 18063
+    },
+    {
+      "epoch": 1.5393268001704303,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.158029708775942e-07,
+      "loss": 0.6701,
+      "step": 18064
+    },
+    {
+      "epoch": 1.5394120153387303,
+      "grad_norm": 16.125,
+      "learning_rate": 3.156913328036995e-07,
+      "loss": 0.3361,
+      "step": 18065
+    },
+    {
+      "epoch": 1.5394972305070302,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.155797116137019e-07,
+      "loss": 0.5023,
+      "step": 18066
+    },
+    {
+      "epoch": 1.5395824456753302,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.1546810730961814e-07,
+      "loss": 0.3776,
+      "step": 18067
+    },
+    {
+      "epoch": 1.5396676608436302,
+      "grad_norm": 11.625,
+      "learning_rate": 3.1535651989346535e-07,
+      "loss": 0.3814,
+      "step": 18068
+    },
+    {
+      "epoch": 1.5397528760119301,
+      "grad_norm": 17.25,
+      "learning_rate": 3.152449493672603e-07,
+      "loss": 0.8499,
+      "step": 18069
+    },
+    {
+      "epoch": 1.53983809118023,
+      "grad_norm": 22.125,
+      "learning_rate": 3.151333957330192e-07,
+      "loss": 0.9117,
+      "step": 18070
+    },
+    {
+      "epoch": 1.53992330634853,
+      "grad_norm": 13.5,
+      "learning_rate": 3.1502185899275744e-07,
+      "loss": 0.6264,
+      "step": 18071
+    },
+    {
+      "epoch": 1.54000852151683,
+      "grad_norm": 17.125,
+      "learning_rate": 3.1491033914849077e-07,
+      "loss": 0.7371,
+      "step": 18072
+    },
+    {
+      "epoch": 1.54009373668513,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.1479883620223473e-07,
+      "loss": 0.6366,
+      "step": 18073
+    },
+    {
+      "epoch": 1.54017895185343,
+      "grad_norm": 15.375,
+      "learning_rate": 3.146873501560044e-07,
+      "loss": 0.5138,
+      "step": 18074
+    },
+    {
+      "epoch": 1.5402641670217299,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.145758810118142e-07,
+      "loss": 0.8285,
+      "step": 18075
+    },
+    {
+      "epoch": 1.5403493821900298,
+      "grad_norm": 13.375,
+      "learning_rate": 3.1446442877167803e-07,
+      "loss": 0.3618,
+      "step": 18076
+    },
+    {
+      "epoch": 1.5404345973583298,
+      "grad_norm": 11.5625,
+      "learning_rate": 3.143529934376105e-07,
+      "loss": 0.4399,
+      "step": 18077
+    },
+    {
+      "epoch": 1.5405198125266297,
+      "grad_norm": 16.875,
+      "learning_rate": 3.1424157501162567e-07,
+      "loss": 0.7838,
+      "step": 18078
+    },
+    {
+      "epoch": 1.5406050276949297,
+      "grad_norm": 14.5,
+      "learning_rate": 3.1413017349573595e-07,
+      "loss": 0.504,
+      "step": 18079
+    },
+    {
+      "epoch": 1.5406902428632296,
+      "grad_norm": 16.25,
+      "learning_rate": 3.1401878889195534e-07,
+      "loss": 0.7607,
+      "step": 18080
+    },
+    {
+      "epoch": 1.5407754580315296,
+      "grad_norm": 17.5,
+      "learning_rate": 3.1390742120229635e-07,
+      "loss": 0.638,
+      "step": 18081
+    },
+    {
+      "epoch": 1.5408606731998296,
+      "grad_norm": 21.25,
+      "learning_rate": 3.137960704287718e-07,
+      "loss": 0.7269,
+      "step": 18082
+    },
+    {
+      "epoch": 1.5409458883681295,
+      "grad_norm": 13.0,
+      "learning_rate": 3.1368473657339373e-07,
+      "loss": 0.5328,
+      "step": 18083
+    },
+    {
+      "epoch": 1.5410311035364295,
+      "grad_norm": 16.625,
+      "learning_rate": 3.135734196381737e-07,
+      "loss": 0.8019,
+      "step": 18084
+    },
+    {
+      "epoch": 1.5411163187047294,
+      "grad_norm": 14.875,
+      "learning_rate": 3.134621196251236e-07,
+      "loss": 0.8248,
+      "step": 18085
+    },
+    {
+      "epoch": 1.5412015338730294,
+      "grad_norm": 16.375,
+      "learning_rate": 3.133508365362549e-07,
+      "loss": 0.6918,
+      "step": 18086
+    },
+    {
+      "epoch": 1.5412867490413293,
+      "grad_norm": 14.0,
+      "learning_rate": 3.132395703735787e-07,
+      "loss": 0.5365,
+      "step": 18087
+    },
+    {
+      "epoch": 1.5413719642096293,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.131283211391051e-07,
+      "loss": 0.3397,
+      "step": 18088
+    },
+    {
+      "epoch": 1.5414571793779293,
+      "grad_norm": 17.25,
+      "learning_rate": 3.130170888348448e-07,
+      "loss": 0.6985,
+      "step": 18089
+    },
+    {
+      "epoch": 1.5415423945462292,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.129058734628079e-07,
+      "loss": 0.5279,
+      "step": 18090
+    },
+    {
+      "epoch": 1.5416276097145292,
+      "grad_norm": 13.875,
+      "learning_rate": 3.1279467502500494e-07,
+      "loss": 0.69,
+      "step": 18091
+    },
+    {
+      "epoch": 1.5417128248828291,
+      "grad_norm": 13.0,
+      "learning_rate": 3.1268349352344395e-07,
+      "loss": 0.5498,
+      "step": 18092
+    },
+    {
+      "epoch": 1.541798040051129,
+      "grad_norm": 18.875,
+      "learning_rate": 3.125723289601347e-07,
+      "loss": 0.7817,
+      "step": 18093
+    },
+    {
+      "epoch": 1.541883255219429,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.124611813370862e-07,
+      "loss": 0.6948,
+      "step": 18094
+    },
+    {
+      "epoch": 1.541968470387729,
+      "grad_norm": 15.5,
+      "learning_rate": 3.123500506563072e-07,
+      "loss": 0.5398,
+      "step": 18095
+    },
+    {
+      "epoch": 1.542053685556029,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.1223893691980536e-07,
+      "loss": 0.4303,
+      "step": 18096
+    },
+    {
+      "epoch": 1.542138900724329,
+      "grad_norm": 25.5,
+      "learning_rate": 3.121278401295889e-07,
+      "loss": 0.498,
+      "step": 18097
+    },
+    {
+      "epoch": 1.5422241158926289,
+      "grad_norm": 22.375,
+      "learning_rate": 3.120167602876656e-07,
+      "loss": 0.8328,
+      "step": 18098
+    },
+    {
+      "epoch": 1.5423093310609288,
+      "grad_norm": 14.875,
+      "learning_rate": 3.1190569739604285e-07,
+      "loss": 0.4627,
+      "step": 18099
+    },
+    {
+      "epoch": 1.5423945462292288,
+      "grad_norm": 17.25,
+      "learning_rate": 3.117946514567276e-07,
+      "loss": 0.9619,
+      "step": 18100
+    },
+    {
+      "epoch": 1.5424797613975287,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.1168362247172594e-07,
+      "loss": 0.4909,
+      "step": 18101
+    },
+    {
+      "epoch": 1.5425649765658287,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.1157261044304484e-07,
+      "loss": 0.8078,
+      "step": 18102
+    },
+    {
+      "epoch": 1.5426501917341287,
+      "grad_norm": 19.5,
+      "learning_rate": 3.114616153726903e-07,
+      "loss": 0.8505,
+      "step": 18103
+    },
+    {
+      "epoch": 1.5427354069024286,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.113506372626686e-07,
+      "loss": 0.6811,
+      "step": 18104
+    },
+    {
+      "epoch": 1.5428206220707286,
+      "grad_norm": 18.375,
+      "learning_rate": 3.1123967611498426e-07,
+      "loss": 0.826,
+      "step": 18105
+    },
+    {
+      "epoch": 1.5429058372390285,
+      "grad_norm": 19.0,
+      "learning_rate": 3.111287319316431e-07,
+      "loss": 0.4378,
+      "step": 18106
+    },
+    {
+      "epoch": 1.5429910524073285,
+      "grad_norm": 17.625,
+      "learning_rate": 3.110178047146502e-07,
+      "loss": 0.706,
+      "step": 18107
+    },
+    {
+      "epoch": 1.5430762675756284,
+      "grad_norm": 16.375,
+      "learning_rate": 3.109068944660097e-07,
+      "loss": 0.6275,
+      "step": 18108
+    },
+    {
+      "epoch": 1.5431614827439284,
+      "grad_norm": 21.75,
+      "learning_rate": 3.1079600118772564e-07,
+      "loss": 1.0648,
+      "step": 18109
+    },
+    {
+      "epoch": 1.5432466979122283,
+      "grad_norm": 21.25,
+      "learning_rate": 3.1068512488180225e-07,
+      "loss": 0.5861,
+      "step": 18110
+    },
+    {
+      "epoch": 1.5433319130805283,
+      "grad_norm": 16.375,
+      "learning_rate": 3.105742655502433e-07,
+      "loss": 0.4717,
+      "step": 18111
+    },
+    {
+      "epoch": 1.5434171282488283,
+      "grad_norm": 10.5625,
+      "learning_rate": 3.1046342319505233e-07,
+      "loss": 0.29,
+      "step": 18112
+    },
+    {
+      "epoch": 1.5435023434171282,
+      "grad_norm": 36.5,
+      "learning_rate": 3.1035259781823173e-07,
+      "loss": 1.2628,
+      "step": 18113
+    },
+    {
+      "epoch": 1.5435875585854282,
+      "grad_norm": 10.3125,
+      "learning_rate": 3.1024178942178464e-07,
+      "loss": 0.346,
+      "step": 18114
+    },
+    {
+      "epoch": 1.5436727737537281,
+      "grad_norm": 22.5,
+      "learning_rate": 3.101309980077137e-07,
+      "loss": 0.7548,
+      "step": 18115
+    },
+    {
+      "epoch": 1.543757988922028,
+      "grad_norm": 16.0,
+      "learning_rate": 3.100202235780203e-07,
+      "loss": 0.9395,
+      "step": 18116
+    },
+    {
+      "epoch": 1.543843204090328,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.0990946613470727e-07,
+      "loss": 0.5107,
+      "step": 18117
+    },
+    {
+      "epoch": 1.543928419258628,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.097987256797749e-07,
+      "loss": 0.3576,
+      "step": 18118
+    },
+    {
+      "epoch": 1.544013634426928,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.0968800221522516e-07,
+      "loss": 0.3867,
+      "step": 18119
+    },
+    {
+      "epoch": 1.544098849595228,
+      "grad_norm": 22.0,
+      "learning_rate": 3.0957729574305873e-07,
+      "loss": 0.748,
+      "step": 18120
+    },
+    {
+      "epoch": 1.5441840647635279,
+      "grad_norm": 14.625,
+      "learning_rate": 3.0946660626527664e-07,
+      "loss": 0.5932,
+      "step": 18121
+    },
+    {
+      "epoch": 1.5442692799318278,
+      "grad_norm": 22.75,
+      "learning_rate": 3.0935593378387836e-07,
+      "loss": 0.3873,
+      "step": 18122
+    },
+    {
+      "epoch": 1.5443544951001278,
+      "grad_norm": 15.75,
+      "learning_rate": 3.092452783008647e-07,
+      "loss": 0.6452,
+      "step": 18123
+    },
+    {
+      "epoch": 1.5444397102684277,
+      "grad_norm": 16.625,
+      "learning_rate": 3.091346398182343e-07,
+      "loss": 0.6708,
+      "step": 18124
+    },
+    {
+      "epoch": 1.5445249254367277,
+      "grad_norm": 18.5,
+      "learning_rate": 3.0902401833798765e-07,
+      "loss": 0.4371,
+      "step": 18125
+    },
+    {
+      "epoch": 1.5446101406050277,
+      "grad_norm": 25.5,
+      "learning_rate": 3.089134138621226e-07,
+      "loss": 1.1384,
+      "step": 18126
+    },
+    {
+      "epoch": 1.5446953557733276,
+      "grad_norm": 15.75,
+      "learning_rate": 3.088028263926388e-07,
+      "loss": 0.4547,
+      "step": 18127
+    },
+    {
+      "epoch": 1.5447805709416276,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.086922559315342e-07,
+      "loss": 0.8119,
+      "step": 18128
+    },
+    {
+      "epoch": 1.5448657861099275,
+      "grad_norm": 11.0,
+      "learning_rate": 3.085817024808074e-07,
+      "loss": 0.4128,
+      "step": 18129
+    },
+    {
+      "epoch": 1.5449510012782275,
+      "grad_norm": 12.625,
+      "learning_rate": 3.084711660424557e-07,
+      "loss": 0.5205,
+      "step": 18130
+    },
+    {
+      "epoch": 1.5450362164465274,
+      "grad_norm": 11.4375,
+      "learning_rate": 3.08360646618477e-07,
+      "loss": 0.3793,
+      "step": 18131
+    },
+    {
+      "epoch": 1.5451214316148274,
+      "grad_norm": 19.125,
+      "learning_rate": 3.08250144210868e-07,
+      "loss": 0.8055,
+      "step": 18132
+    },
+    {
+      "epoch": 1.5452066467831274,
+      "grad_norm": 21.375,
+      "learning_rate": 3.0813965882162596e-07,
+      "loss": 1.1525,
+      "step": 18133
+    },
+    {
+      "epoch": 1.5452918619514273,
+      "grad_norm": 10.1875,
+      "learning_rate": 3.080291904527477e-07,
+      "loss": 0.3026,
+      "step": 18134
+    },
+    {
+      "epoch": 1.5453770771197273,
+      "grad_norm": 19.875,
+      "learning_rate": 3.0791873910622894e-07,
+      "loss": 1.0445,
+      "step": 18135
+    },
+    {
+      "epoch": 1.5454622922880272,
+      "grad_norm": 16.0,
+      "learning_rate": 3.078083047840658e-07,
+      "loss": 0.7081,
+      "step": 18136
+    },
+    {
+      "epoch": 1.5455475074563272,
+      "grad_norm": 18.375,
+      "learning_rate": 3.076978874882544e-07,
+      "loss": 0.8116,
+      "step": 18137
+    },
+    {
+      "epoch": 1.5456327226246271,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.075874872207894e-07,
+      "loss": 0.4058,
+      "step": 18138
+    },
+    {
+      "epoch": 1.545717937792927,
+      "grad_norm": 18.5,
+      "learning_rate": 3.0747710398366616e-07,
+      "loss": 0.5943,
+      "step": 18139
+    },
+    {
+      "epoch": 1.545803152961227,
+      "grad_norm": 17.375,
+      "learning_rate": 3.0736673777887983e-07,
+      "loss": 0.3744,
+      "step": 18140
+    },
+    {
+      "epoch": 1.545888368129527,
+      "grad_norm": 13.375,
+      "learning_rate": 3.072563886084241e-07,
+      "loss": 0.6065,
+      "step": 18141
+    },
+    {
+      "epoch": 1.545973583297827,
+      "grad_norm": 10.875,
+      "learning_rate": 3.071460564742937e-07,
+      "loss": 0.3043,
+      "step": 18142
+    },
+    {
+      "epoch": 1.546058798466127,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.070357413784819e-07,
+      "loss": 0.6737,
+      "step": 18143
+    },
+    {
+      "epoch": 1.5461440136344269,
+      "grad_norm": 19.375,
+      "learning_rate": 3.0692544332298234e-07,
+      "loss": 0.8263,
+      "step": 18144
+    },
+    {
+      "epoch": 1.5462292288027268,
+      "grad_norm": 13.875,
+      "learning_rate": 3.0681516230978855e-07,
+      "loss": 0.4584,
+      "step": 18145
+    },
+    {
+      "epoch": 1.5463144439710268,
+      "grad_norm": 20.75,
+      "learning_rate": 3.067048983408935e-07,
+      "loss": 0.5865,
+      "step": 18146
+    },
+    {
+      "epoch": 1.5463996591393268,
+      "grad_norm": 11.125,
+      "learning_rate": 3.065946514182891e-07,
+      "loss": 0.3218,
+      "step": 18147
+    },
+    {
+      "epoch": 1.5464848743076267,
+      "grad_norm": 12.0,
+      "learning_rate": 3.064844215439683e-07,
+      "loss": 0.4027,
+      "step": 18148
+    },
+    {
+      "epoch": 1.5465700894759267,
+      "grad_norm": 17.0,
+      "learning_rate": 3.0637420871992255e-07,
+      "loss": 0.7887,
+      "step": 18149
+    },
+    {
+      "epoch": 1.5466553046442266,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.062640129481437e-07,
+      "loss": 0.4705,
+      "step": 18150
+    },
+    {
+      "epoch": 1.5467405198125266,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.0615383423062347e-07,
+      "loss": 0.603,
+      "step": 18151
+    },
+    {
+      "epoch": 1.5468257349808265,
+      "grad_norm": 15.0,
+      "learning_rate": 3.060436725693522e-07,
+      "loss": 0.6974,
+      "step": 18152
+    },
+    {
+      "epoch": 1.5469109501491265,
+      "grad_norm": 9.75,
+      "learning_rate": 3.05933527966321e-07,
+      "loss": 0.2105,
+      "step": 18153
+    },
+    {
+      "epoch": 1.5469961653174265,
+      "grad_norm": 17.125,
+      "learning_rate": 3.058234004235207e-07,
+      "loss": 0.4721,
+      "step": 18154
+    },
+    {
+      "epoch": 1.5470813804857264,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.057132899429406e-07,
+      "loss": 0.4539,
+      "step": 18155
+    },
+    {
+      "epoch": 1.5471665956540264,
+      "grad_norm": 14.375,
+      "learning_rate": 3.056031965265713e-07,
+      "loss": 0.6603,
+      "step": 18156
+    },
+    {
+      "epoch": 1.5472518108223263,
+      "grad_norm": 41.5,
+      "learning_rate": 3.054931201764015e-07,
+      "loss": 0.8076,
+      "step": 18157
+    },
+    {
+      "epoch": 1.5473370259906263,
+      "grad_norm": 14.75,
+      "learning_rate": 3.053830608944208e-07,
+      "loss": 0.5673,
+      "step": 18158
+    },
+    {
+      "epoch": 1.5474222411589262,
+      "grad_norm": 17.625,
+      "learning_rate": 3.052730186826185e-07,
+      "loss": 0.9168,
+      "step": 18159
+    },
+    {
+      "epoch": 1.5475074563272262,
+      "grad_norm": 16.875,
+      "learning_rate": 3.051629935429823e-07,
+      "loss": 0.2476,
+      "step": 18160
+    },
+    {
+      "epoch": 1.5475926714955262,
+      "grad_norm": 17.625,
+      "learning_rate": 3.050529854775008e-07,
+      "loss": 1.1721,
+      "step": 18161
+    },
+    {
+      "epoch": 1.547677886663826,
+      "grad_norm": 20.125,
+      "learning_rate": 3.049429944881623e-07,
+      "loss": 0.996,
+      "step": 18162
+    },
+    {
+      "epoch": 1.547763101832126,
+      "grad_norm": 14.875,
+      "learning_rate": 3.0483302057695454e-07,
+      "loss": 0.6625,
+      "step": 18163
+    },
+    {
+      "epoch": 1.547848317000426,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.047230637458644e-07,
+      "loss": 0.7355,
+      "step": 18164
+    },
+    {
+      "epoch": 1.547933532168726,
+      "grad_norm": 19.0,
+      "learning_rate": 3.046131239968787e-07,
+      "loss": 0.4716,
+      "step": 18165
+    },
+    {
+      "epoch": 1.548018747337026,
+      "grad_norm": 13.75,
+      "learning_rate": 3.0450320133198474e-07,
+      "loss": 0.6945,
+      "step": 18166
+    },
+    {
+      "epoch": 1.548103962505326,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.043932957531688e-07,
+      "loss": 0.5598,
+      "step": 18167
+    },
+    {
+      "epoch": 1.5481891776736258,
+      "grad_norm": 16.0,
+      "learning_rate": 3.042834072624166e-07,
+      "loss": 0.5223,
+      "step": 18168
+    },
+    {
+      "epoch": 1.5482743928419258,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.0417353586171427e-07,
+      "loss": 0.8531,
+      "step": 18169
+    },
+    {
+      "epoch": 1.5483596080102258,
+      "grad_norm": 13.6875,
+      "learning_rate": 3.040636815530472e-07,
+      "loss": 0.45,
+      "step": 18170
+    },
+    {
+      "epoch": 1.5484448231785257,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.0395384433840103e-07,
+      "loss": 0.4013,
+      "step": 18171
+    },
+    {
+      "epoch": 1.5485300383468257,
+      "grad_norm": 16.0,
+      "learning_rate": 3.038440242197602e-07,
+      "loss": 0.6198,
+      "step": 18172
+    },
+    {
+      "epoch": 1.5486152535151256,
+      "grad_norm": 13.6875,
+      "learning_rate": 3.0373422119910885e-07,
+      "loss": 0.4915,
+      "step": 18173
+    },
+    {
+      "epoch": 1.5487004686834256,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.0362443527843176e-07,
+      "loss": 0.6626,
+      "step": 18174
+    },
+    {
+      "epoch": 1.5487856838517255,
+      "grad_norm": 18.25,
+      "learning_rate": 3.0351466645971287e-07,
+      "loss": 0.7268,
+      "step": 18175
+    },
+    {
+      "epoch": 1.5488708990200255,
+      "grad_norm": 16.875,
+      "learning_rate": 3.0340491474493585e-07,
+      "loss": 0.612,
+      "step": 18176
+    },
+    {
+      "epoch": 1.5489561141883255,
+      "grad_norm": 14.0,
+      "learning_rate": 3.032951801360837e-07,
+      "loss": 0.3543,
+      "step": 18177
+    },
+    {
+      "epoch": 1.5490413293566254,
+      "grad_norm": 29.5,
+      "learning_rate": 3.031854626351395e-07,
+      "loss": 0.7592,
+      "step": 18178
+    },
+    {
+      "epoch": 1.5491265445249254,
+      "grad_norm": 16.0,
+      "learning_rate": 3.030757622440862e-07,
+      "loss": 0.673,
+      "step": 18179
+    },
+    {
+      "epoch": 1.5492117596932253,
+      "grad_norm": 11.625,
+      "learning_rate": 3.0296607896490635e-07,
+      "loss": 0.4766,
+      "step": 18180
+    },
+    {
+      "epoch": 1.5492969748615253,
+      "grad_norm": 19.375,
+      "learning_rate": 3.0285641279958185e-07,
+      "loss": 0.5371,
+      "step": 18181
+    },
+    {
+      "epoch": 1.5493821900298252,
+      "grad_norm": 14.625,
+      "learning_rate": 3.02746763750094e-07,
+      "loss": 0.6792,
+      "step": 18182
+    },
+    {
+      "epoch": 1.5494674051981252,
+      "grad_norm": 14.0,
+      "learning_rate": 3.026371318184247e-07,
+      "loss": 0.5179,
+      "step": 18183
+    },
+    {
+      "epoch": 1.5495526203664252,
+      "grad_norm": 16.0,
+      "learning_rate": 3.0252751700655544e-07,
+      "loss": 0.4493,
+      "step": 18184
+    },
+    {
+      "epoch": 1.5496378355347251,
+      "grad_norm": 17.125,
+      "learning_rate": 3.0241791931646644e-07,
+      "loss": 0.6506,
+      "step": 18185
+    },
+    {
+      "epoch": 1.549723050703025,
+      "grad_norm": 16.625,
+      "learning_rate": 3.0230833875013827e-07,
+      "loss": 0.7194,
+      "step": 18186
+    },
+    {
+      "epoch": 1.549808265871325,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.0219877530955147e-07,
+      "loss": 0.592,
+      "step": 18187
+    },
+    {
+      "epoch": 1.549893481039625,
+      "grad_norm": 15.75,
+      "learning_rate": 3.0208922899668634e-07,
+      "loss": 0.672,
+      "step": 18188
+    },
+    {
+      "epoch": 1.549978696207925,
+      "grad_norm": 13.75,
+      "learning_rate": 3.01979699813522e-07,
+      "loss": 0.4713,
+      "step": 18189
+    },
+    {
+      "epoch": 1.550063911376225,
+      "grad_norm": 17.75,
+      "learning_rate": 3.018701877620374e-07,
+      "loss": 0.8693,
+      "step": 18190
+    },
+    {
+      "epoch": 1.5501491265445249,
+      "grad_norm": 17.375,
+      "learning_rate": 3.0176069284421195e-07,
+      "loss": 0.6229,
+      "step": 18191
+    },
+    {
+      "epoch": 1.5502343417128248,
+      "grad_norm": 17.5,
+      "learning_rate": 3.016512150620243e-07,
+      "loss": 0.839,
+      "step": 18192
+    },
+    {
+      "epoch": 1.5503195568811248,
+      "grad_norm": 12.125,
+      "learning_rate": 3.015417544174533e-07,
+      "loss": 0.3162,
+      "step": 18193
+    },
+    {
+      "epoch": 1.5504047720494247,
+      "grad_norm": 14.5,
+      "learning_rate": 3.01432310912476e-07,
+      "loss": 0.4831,
+      "step": 18194
+    },
+    {
+      "epoch": 1.5504899872177247,
+      "grad_norm": 17.625,
+      "learning_rate": 3.013228845490708e-07,
+      "loss": 0.5869,
+      "step": 18195
+    },
+    {
+      "epoch": 1.5505752023860246,
+      "grad_norm": 18.125,
+      "learning_rate": 3.0121347532921533e-07,
+      "loss": 0.518,
+      "step": 18196
+    },
+    {
+      "epoch": 1.5506604175543246,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.0110408325488614e-07,
+      "loss": 0.638,
+      "step": 18197
+    },
+    {
+      "epoch": 1.5507456327226246,
+      "grad_norm": 16.625,
+      "learning_rate": 3.009947083280608e-07,
+      "loss": 0.5814,
+      "step": 18198
+    },
+    {
+      "epoch": 1.5508308478909245,
+      "grad_norm": 12.375,
+      "learning_rate": 3.008853505507149e-07,
+      "loss": 0.5837,
+      "step": 18199
+    },
+    {
+      "epoch": 1.5509160630592245,
+      "grad_norm": 18.25,
+      "learning_rate": 3.007760099248251e-07,
+      "loss": 0.77,
+      "step": 18200
+    },
+    {
+      "epoch": 1.5510012782275244,
+      "grad_norm": 16.875,
+      "learning_rate": 3.0066668645236774e-07,
+      "loss": 0.9167,
+      "step": 18201
+    },
+    {
+      "epoch": 1.5510864933958244,
+      "grad_norm": 12.5,
+      "learning_rate": 3.0055738013531776e-07,
+      "loss": 0.5106,
+      "step": 18202
+    },
+    {
+      "epoch": 1.5511717085641243,
+      "grad_norm": 15.5,
+      "learning_rate": 3.004480909756505e-07,
+      "loss": 0.8513,
+      "step": 18203
+    },
+    {
+      "epoch": 1.5512569237324243,
+      "grad_norm": 14.25,
+      "learning_rate": 3.00338818975341e-07,
+      "loss": 0.375,
+      "step": 18204
+    },
+    {
+      "epoch": 1.5513421389007243,
+      "grad_norm": 18.625,
+      "learning_rate": 3.002295641363645e-07,
+      "loss": 0.731,
+      "step": 18205
+    },
+    {
+      "epoch": 1.5514273540690242,
+      "grad_norm": 11.375,
+      "learning_rate": 3.0012032646069485e-07,
+      "loss": 0.2764,
+      "step": 18206
+    },
+    {
+      "epoch": 1.5515125692373242,
+      "grad_norm": 14.0,
+      "learning_rate": 3.000111059503057e-07,
+      "loss": 0.5755,
+      "step": 18207
+    },
+    {
+      "epoch": 1.5515977844056241,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.9990190260717105e-07,
+      "loss": 0.389,
+      "step": 18208
+    },
+    {
+      "epoch": 1.551682999573924,
+      "grad_norm": 26.375,
+      "learning_rate": 2.997927164332645e-07,
+      "loss": 0.7195,
+      "step": 18209
+    },
+    {
+      "epoch": 1.551768214742224,
+      "grad_norm": 9.0625,
+      "learning_rate": 2.996835474305594e-07,
+      "loss": 0.7298,
+      "step": 18210
+    },
+    {
+      "epoch": 1.551853429910524,
+      "grad_norm": 53.25,
+      "learning_rate": 2.995743956010279e-07,
+      "loss": 0.8251,
+      "step": 18211
+    },
+    {
+      "epoch": 1.551938645078824,
+      "grad_norm": 16.75,
+      "learning_rate": 2.994652609466428e-07,
+      "loss": 0.6133,
+      "step": 18212
+    },
+    {
+      "epoch": 1.552023860247124,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.993561434693766e-07,
+      "loss": 0.6612,
+      "step": 18213
+    },
+    {
+      "epoch": 1.5521090754154239,
+      "grad_norm": 11.625,
+      "learning_rate": 2.992470431712009e-07,
+      "loss": 0.4034,
+      "step": 18214
+    },
+    {
+      "epoch": 1.5521942905837238,
+      "grad_norm": 18.375,
+      "learning_rate": 2.991379600540867e-07,
+      "loss": 0.4562,
+      "step": 18215
+    },
+    {
+      "epoch": 1.5522795057520238,
+      "grad_norm": 18.875,
+      "learning_rate": 2.990288941200058e-07,
+      "loss": 0.5251,
+      "step": 18216
+    },
+    {
+      "epoch": 1.5523647209203237,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.989198453709291e-07,
+      "loss": 0.4322,
+      "step": 18217
+    },
+    {
+      "epoch": 1.5524499360886237,
+      "grad_norm": 12.0,
+      "learning_rate": 2.9881081380882747e-07,
+      "loss": 0.3745,
+      "step": 18218
+    },
+    {
+      "epoch": 1.5525351512569237,
+      "grad_norm": 16.875,
+      "learning_rate": 2.9870179943567056e-07,
+      "loss": 0.7131,
+      "step": 18219
+    },
+    {
+      "epoch": 1.5526203664252236,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.985928022534289e-07,
+      "loss": 0.4414,
+      "step": 18220
+    },
+    {
+      "epoch": 1.5527055815935236,
+      "grad_norm": 15.625,
+      "learning_rate": 2.984838222640722e-07,
+      "loss": 0.5665,
+      "step": 18221
+    },
+    {
+      "epoch": 1.5527907967618235,
+      "grad_norm": 30.75,
+      "learning_rate": 2.9837485946956927e-07,
+      "loss": 0.994,
+      "step": 18222
+    },
+    {
+      "epoch": 1.5528760119301235,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.9826591387189006e-07,
+      "loss": 0.3535,
+      "step": 18223
+    },
+    {
+      "epoch": 1.5529612270984234,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.9815698547300235e-07,
+      "loss": 0.6345,
+      "step": 18224
+    },
+    {
+      "epoch": 1.5530464422667234,
+      "grad_norm": 9.5,
+      "learning_rate": 2.980480742748752e-07,
+      "loss": 0.2725,
+      "step": 18225
+    },
+    {
+      "epoch": 1.5531316574350233,
+      "grad_norm": 17.125,
+      "learning_rate": 2.979391802794765e-07,
+      "loss": 0.6158,
+      "step": 18226
+    },
+    {
+      "epoch": 1.5532168726033233,
+      "grad_norm": 18.25,
+      "learning_rate": 2.978303034887746e-07,
+      "loss": 0.8655,
+      "step": 18227
+    },
+    {
+      "epoch": 1.5533020877716233,
+      "grad_norm": 15.375,
+      "learning_rate": 2.9772144390473616e-07,
+      "loss": 0.4809,
+      "step": 18228
+    },
+    {
+      "epoch": 1.5533873029399232,
+      "grad_norm": 22.125,
+      "learning_rate": 2.976126015293293e-07,
+      "loss": 0.7585,
+      "step": 18229
+    },
+    {
+      "epoch": 1.5534725181082232,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.9750377636452e-07,
+      "loss": 0.5745,
+      "step": 18230
+    },
+    {
+      "epoch": 1.5535577332765231,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.9739496841227567e-07,
+      "loss": 0.466,
+      "step": 18231
+    },
+    {
+      "epoch": 1.553642948444823,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.9728617767456167e-07,
+      "loss": 0.6781,
+      "step": 18232
+    },
+    {
+      "epoch": 1.553728163613123,
+      "grad_norm": 15.0,
+      "learning_rate": 2.9717740415334454e-07,
+      "loss": 0.5491,
+      "step": 18233
+    },
+    {
+      "epoch": 1.553813378781423,
+      "grad_norm": 26.5,
+      "learning_rate": 2.970686478505899e-07,
+      "loss": 0.5826,
+      "step": 18234
+    },
+    {
+      "epoch": 1.553898593949723,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.969599087682634e-07,
+      "loss": 0.3537,
+      "step": 18235
+    },
+    {
+      "epoch": 1.553983809118023,
+      "grad_norm": 17.125,
+      "learning_rate": 2.9685118690832947e-07,
+      "loss": 0.395,
+      "step": 18236
+    },
+    {
+      "epoch": 1.5540690242863229,
+      "grad_norm": 17.125,
+      "learning_rate": 2.9674248227275324e-07,
+      "loss": 0.6078,
+      "step": 18237
+    },
+    {
+      "epoch": 1.5541542394546228,
+      "grad_norm": 16.75,
+      "learning_rate": 2.9663379486349855e-07,
+      "loss": 0.4337,
+      "step": 18238
+    },
+    {
+      "epoch": 1.5542394546229228,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.9652512468253e-07,
+      "loss": 0.6212,
+      "step": 18239
+    },
+    {
+      "epoch": 1.5543246697912227,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.9641647173181144e-07,
+      "loss": 0.4846,
+      "step": 18240
+    },
+    {
+      "epoch": 1.5544098849595227,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.963078360133059e-07,
+      "loss": 0.4001,
+      "step": 18241
+    },
+    {
+      "epoch": 1.5544951001278227,
+      "grad_norm": 11.375,
+      "learning_rate": 2.961992175289768e-07,
+      "loss": 0.364,
+      "step": 18242
+    },
+    {
+      "epoch": 1.5545803152961226,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.9609061628078693e-07,
+      "loss": 0.39,
+      "step": 18243
+    },
+    {
+      "epoch": 1.5546655304644226,
+      "grad_norm": 18.25,
+      "learning_rate": 2.959820322706991e-07,
+      "loss": 0.8767,
+      "step": 18244
+    },
+    {
+      "epoch": 1.5547507456327225,
+      "grad_norm": 16.0,
+      "learning_rate": 2.958734655006755e-07,
+      "loss": 0.3333,
+      "step": 18245
+    },
+    {
+      "epoch": 1.5548359608010225,
+      "grad_norm": 17.625,
+      "learning_rate": 2.9576491597267735e-07,
+      "loss": 0.6168,
+      "step": 18246
+    },
+    {
+      "epoch": 1.5549211759693224,
+      "grad_norm": 16.5,
+      "learning_rate": 2.956563836886667e-07,
+      "loss": 0.3021,
+      "step": 18247
+    },
+    {
+      "epoch": 1.5550063911376224,
+      "grad_norm": 11.5,
+      "learning_rate": 2.9554786865060523e-07,
+      "loss": 0.3279,
+      "step": 18248
+    },
+    {
+      "epoch": 1.5550916063059224,
+      "grad_norm": 22.75,
+      "learning_rate": 2.9543937086045317e-07,
+      "loss": 0.7473,
+      "step": 18249
+    },
+    {
+      "epoch": 1.5551768214742223,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.9533089032017166e-07,
+      "loss": 0.3894,
+      "step": 18250
+    },
+    {
+      "epoch": 1.5552620366425223,
+      "grad_norm": 20.75,
+      "learning_rate": 2.9522242703172084e-07,
+      "loss": 1.0367,
+      "step": 18251
+    },
+    {
+      "epoch": 1.5553472518108222,
+      "grad_norm": 38.5,
+      "learning_rate": 2.951139809970613e-07,
+      "loss": 1.6008,
+      "step": 18252
+    },
+    {
+      "epoch": 1.5554324669791222,
+      "grad_norm": 24.125,
+      "learning_rate": 2.9500555221815233e-07,
+      "loss": 0.8166,
+      "step": 18253
+    },
+    {
+      "epoch": 1.5555176821474221,
+      "grad_norm": 11.625,
+      "learning_rate": 2.948971406969529e-07,
+      "loss": 0.2904,
+      "step": 18254
+    },
+    {
+      "epoch": 1.555602897315722,
+      "grad_norm": 11.25,
+      "learning_rate": 2.9478874643542265e-07,
+      "loss": 0.3079,
+      "step": 18255
+    },
+    {
+      "epoch": 1.555688112484022,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.946803694355203e-07,
+      "loss": 0.4667,
+      "step": 18256
+    },
+    {
+      "epoch": 1.555773327652322,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.945720096992047e-07,
+      "loss": 0.317,
+      "step": 18257
+    },
+    {
+      "epoch": 1.555858542820622,
+      "grad_norm": 19.25,
+      "learning_rate": 2.944636672284333e-07,
+      "loss": 0.9343,
+      "step": 18258
+    },
+    {
+      "epoch": 1.555943757988922,
+      "grad_norm": 12.0,
+      "learning_rate": 2.943553420251642e-07,
+      "loss": 0.4059,
+      "step": 18259
+    },
+    {
+      "epoch": 1.5560289731572219,
+      "grad_norm": 36.75,
+      "learning_rate": 2.9424703409135517e-07,
+      "loss": 0.6904,
+      "step": 18260
+    },
+    {
+      "epoch": 1.5561141883255218,
+      "grad_norm": 16.0,
+      "learning_rate": 2.9413874342896385e-07,
+      "loss": 0.5821,
+      "step": 18261
+    },
+    {
+      "epoch": 1.5561994034938218,
+      "grad_norm": 15.875,
+      "learning_rate": 2.94030470039946e-07,
+      "loss": 0.4557,
+      "step": 18262
+    },
+    {
+      "epoch": 1.5562846186621218,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.939222139262589e-07,
+      "loss": 0.3218,
+      "step": 18263
+    },
+    {
+      "epoch": 1.5563698338304217,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.938139750898589e-07,
+      "loss": 0.36,
+      "step": 18264
+    },
+    {
+      "epoch": 1.5564550489987217,
+      "grad_norm": 29.75,
+      "learning_rate": 2.9370575353270204e-07,
+      "loss": 1.1815,
+      "step": 18265
+    },
+    {
+      "epoch": 1.5565402641670216,
+      "grad_norm": 14.125,
+      "learning_rate": 2.935975492567436e-07,
+      "loss": 0.5413,
+      "step": 18266
+    },
+    {
+      "epoch": 1.5566254793353216,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.934893622639391e-07,
+      "loss": 0.5462,
+      "step": 18267
+    },
+    {
+      "epoch": 1.5567106945036215,
+      "grad_norm": 11.5,
+      "learning_rate": 2.933811925562438e-07,
+      "loss": 0.2213,
+      "step": 18268
+    },
+    {
+      "epoch": 1.5567959096719215,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.932730401356126e-07,
+      "loss": 0.3461,
+      "step": 18269
+    },
+    {
+      "epoch": 1.5568811248402215,
+      "grad_norm": 16.625,
+      "learning_rate": 2.931649050039996e-07,
+      "loss": 0.4414,
+      "step": 18270
+    },
+    {
+      "epoch": 1.5569663400085214,
+      "grad_norm": 21.5,
+      "learning_rate": 2.9305678716335856e-07,
+      "loss": 0.4036,
+      "step": 18271
+    },
+    {
+      "epoch": 1.5570515551768214,
+      "grad_norm": 27.75,
+      "learning_rate": 2.9294868661564375e-07,
+      "loss": 0.8809,
+      "step": 18272
+    },
+    {
+      "epoch": 1.5571367703451213,
+      "grad_norm": 22.25,
+      "learning_rate": 2.9284060336280855e-07,
+      "loss": 0.7155,
+      "step": 18273
+    },
+    {
+      "epoch": 1.5572219855134213,
+      "grad_norm": 26.375,
+      "learning_rate": 2.927325374068064e-07,
+      "loss": 0.3937,
+      "step": 18274
+    },
+    {
+      "epoch": 1.5573072006817212,
+      "grad_norm": 18.75,
+      "learning_rate": 2.9262448874958965e-07,
+      "loss": 1.0006,
+      "step": 18275
+    },
+    {
+      "epoch": 1.5573924158500212,
+      "grad_norm": 18.25,
+      "learning_rate": 2.925164573931111e-07,
+      "loss": 0.574,
+      "step": 18276
+    },
+    {
+      "epoch": 1.5574776310183212,
+      "grad_norm": 15.0,
+      "learning_rate": 2.9240844333932333e-07,
+      "loss": 0.6314,
+      "step": 18277
+    },
+    {
+      "epoch": 1.557562846186621,
+      "grad_norm": 19.375,
+      "learning_rate": 2.9230044659017785e-07,
+      "loss": 0.3731,
+      "step": 18278
+    },
+    {
+      "epoch": 1.557648061354921,
+      "grad_norm": 18.0,
+      "learning_rate": 2.92192467147626e-07,
+      "loss": 0.7989,
+      "step": 18279
+    },
+    {
+      "epoch": 1.557733276523221,
+      "grad_norm": 13.625,
+      "learning_rate": 2.9208450501361954e-07,
+      "loss": 0.3797,
+      "step": 18280
+    },
+    {
+      "epoch": 1.557818491691521,
+      "grad_norm": 11.875,
+      "learning_rate": 2.9197656019010925e-07,
+      "loss": 0.419,
+      "step": 18281
+    },
+    {
+      "epoch": 1.557903706859821,
+      "grad_norm": 15.875,
+      "learning_rate": 2.918686326790461e-07,
+      "loss": 0.6322,
+      "step": 18282
+    },
+    {
+      "epoch": 1.557988922028121,
+      "grad_norm": 12.375,
+      "learning_rate": 2.917607224823801e-07,
+      "loss": 0.3976,
+      "step": 18283
+    },
+    {
+      "epoch": 1.5580741371964208,
+      "grad_norm": 19.25,
+      "learning_rate": 2.9165282960206126e-07,
+      "loss": 0.5949,
+      "step": 18284
+    },
+    {
+      "epoch": 1.5581593523647208,
+      "grad_norm": 14.0,
+      "learning_rate": 2.915449540400396e-07,
+      "loss": 0.3596,
+      "step": 18285
+    },
+    {
+      "epoch": 1.5582445675330208,
+      "grad_norm": 19.875,
+      "learning_rate": 2.914370957982647e-07,
+      "loss": 0.4221,
+      "step": 18286
+    },
+    {
+      "epoch": 1.5583297827013207,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.9132925487868556e-07,
+      "loss": 0.7852,
+      "step": 18287
+    },
+    {
+      "epoch": 1.5584149978696207,
+      "grad_norm": 16.875,
+      "learning_rate": 2.9122143128325024e-07,
+      "loss": 0.403,
+      "step": 18288
+    },
+    {
+      "epoch": 1.5585002130379206,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.9111362501390795e-07,
+      "loss": 0.4639,
+      "step": 18289
+    },
+    {
+      "epoch": 1.5585854282062206,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.910058360726066e-07,
+      "loss": 0.495,
+      "step": 18290
+    },
+    {
+      "epoch": 1.5586706433745205,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.9089806446129436e-07,
+      "loss": 0.8543,
+      "step": 18291
+    },
+    {
+      "epoch": 1.5587558585428205,
+      "grad_norm": 17.625,
+      "learning_rate": 2.9079031018191847e-07,
+      "loss": 0.5971,
+      "step": 18292
+    },
+    {
+      "epoch": 1.5588410737111205,
+      "grad_norm": 16.125,
+      "learning_rate": 2.9068257323642603e-07,
+      "loss": 0.7796,
+      "step": 18293
+    },
+    {
+      "epoch": 1.5589262888794204,
+      "grad_norm": 10.75,
+      "learning_rate": 2.9057485362676453e-07,
+      "loss": 0.9313,
+      "step": 18294
+    },
+    {
+      "epoch": 1.5590115040477204,
+      "grad_norm": 15.625,
+      "learning_rate": 2.9046715135488025e-07,
+      "loss": 0.3727,
+      "step": 18295
+    },
+    {
+      "epoch": 1.5590967192160203,
+      "grad_norm": 15.875,
+      "learning_rate": 2.9035946642271913e-07,
+      "loss": 0.6305,
+      "step": 18296
+    },
+    {
+      "epoch": 1.5591819343843203,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.902517988322273e-07,
+      "loss": 0.5754,
+      "step": 18297
+    },
+    {
+      "epoch": 1.5592671495526202,
+      "grad_norm": 16.75,
+      "learning_rate": 2.901441485853507e-07,
+      "loss": 0.5426,
+      "step": 18298
+    },
+    {
+      "epoch": 1.5593523647209202,
+      "grad_norm": 15.75,
+      "learning_rate": 2.9003651568403483e-07,
+      "loss": 0.7613,
+      "step": 18299
+    },
+    {
+      "epoch": 1.5594375798892202,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.899289001302241e-07,
+      "loss": 0.4962,
+      "step": 18300
+    },
+    {
+      "epoch": 1.5595227950575201,
+      "grad_norm": 18.25,
+      "learning_rate": 2.898213019258636e-07,
+      "loss": 0.8255,
+      "step": 18301
+    },
+    {
+      "epoch": 1.55960801022582,
+      "grad_norm": 20.875,
+      "learning_rate": 2.8971372107289816e-07,
+      "loss": 0.9768,
+      "step": 18302
+    },
+    {
+      "epoch": 1.55969322539412,
+      "grad_norm": 16.125,
+      "learning_rate": 2.896061575732709e-07,
+      "loss": 0.6189,
+      "step": 18303
+    },
+    {
+      "epoch": 1.55977844056242,
+      "grad_norm": 12.25,
+      "learning_rate": 2.894986114289265e-07,
+      "loss": 0.3068,
+      "step": 18304
+    },
+    {
+      "epoch": 1.55986365573072,
+      "grad_norm": 14.75,
+      "learning_rate": 2.893910826418077e-07,
+      "loss": 0.346,
+      "step": 18305
+    },
+    {
+      "epoch": 1.55994887089902,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.892835712138581e-07,
+      "loss": 0.7032,
+      "step": 18306
+    },
+    {
+      "epoch": 1.5600340860673199,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.8917607714702063e-07,
+      "loss": 0.5629,
+      "step": 18307
+    },
+    {
+      "epoch": 1.5601193012356198,
+      "grad_norm": 13.375,
+      "learning_rate": 2.8906860044323734e-07,
+      "loss": 0.5289,
+      "step": 18308
+    },
+    {
+      "epoch": 1.5602045164039198,
+      "grad_norm": 17.5,
+      "learning_rate": 2.889611411044509e-07,
+      "loss": 0.5603,
+      "step": 18309
+    },
+    {
+      "epoch": 1.5602897315722197,
+      "grad_norm": 14.375,
+      "learning_rate": 2.8885369913260324e-07,
+      "loss": 0.4301,
+      "step": 18310
+    },
+    {
+      "epoch": 1.5603749467405197,
+      "grad_norm": 15.5,
+      "learning_rate": 2.8874627452963533e-07,
+      "loss": 0.7183,
+      "step": 18311
+    },
+    {
+      "epoch": 1.5604601619088196,
+      "grad_norm": 14.25,
+      "learning_rate": 2.8863886729748934e-07,
+      "loss": 0.4862,
+      "step": 18312
+    },
+    {
+      "epoch": 1.5605453770771196,
+      "grad_norm": 14.125,
+      "learning_rate": 2.8853147743810525e-07,
+      "loss": 0.4342,
+      "step": 18313
+    },
+    {
+      "epoch": 1.5606305922454196,
+      "grad_norm": 16.25,
+      "learning_rate": 2.8842410495342436e-07,
+      "loss": 0.5707,
+      "step": 18314
+    },
+    {
+      "epoch": 1.5607158074137195,
+      "grad_norm": 17.75,
+      "learning_rate": 2.883167498453868e-07,
+      "loss": 0.356,
+      "step": 18315
+    },
+    {
+      "epoch": 1.5608010225820195,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.8820941211593276e-07,
+      "loss": 0.5358,
+      "step": 18316
+    },
+    {
+      "epoch": 1.5608862377503194,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.881020917670016e-07,
+      "loss": 0.5579,
+      "step": 18317
+    },
+    {
+      "epoch": 1.5609714529186194,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.8799478880053325e-07,
+      "loss": 0.5988,
+      "step": 18318
+    },
+    {
+      "epoch": 1.5610566680869193,
+      "grad_norm": 15.625,
+      "learning_rate": 2.878875032184662e-07,
+      "loss": 0.5068,
+      "step": 18319
+    },
+    {
+      "epoch": 1.5611418832552193,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.8778023502273935e-07,
+      "loss": 0.4186,
+      "step": 18320
+    },
+    {
+      "epoch": 1.5612270984235193,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.876729842152916e-07,
+      "loss": 0.7328,
+      "step": 18321
+    },
+    {
+      "epoch": 1.5613123135918192,
+      "grad_norm": 21.125,
+      "learning_rate": 2.875657507980603e-07,
+      "loss": 0.9197,
+      "step": 18322
+    },
+    {
+      "epoch": 1.5613975287601192,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.874585347729837e-07,
+      "loss": 0.3865,
+      "step": 18323
+    },
+    {
+      "epoch": 1.5614827439284191,
+      "grad_norm": 19.25,
+      "learning_rate": 2.873513361419997e-07,
+      "loss": 0.6067,
+      "step": 18324
+    },
+    {
+      "epoch": 1.561567959096719,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.872441549070447e-07,
+      "loss": 0.4283,
+      "step": 18325
+    },
+    {
+      "epoch": 1.561653174265019,
+      "grad_norm": 16.75,
+      "learning_rate": 2.8713699107005625e-07,
+      "loss": 0.7615,
+      "step": 18326
+    },
+    {
+      "epoch": 1.561738389433319,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.870298446329704e-07,
+      "loss": 0.5689,
+      "step": 18327
+    },
+    {
+      "epoch": 1.561823604601619,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.869227155977235e-07,
+      "loss": 0.6617,
+      "step": 18328
+    },
+    {
+      "epoch": 1.561908819769919,
+      "grad_norm": 21.875,
+      "learning_rate": 2.868156039662519e-07,
+      "loss": 0.5673,
+      "step": 18329
+    },
+    {
+      "epoch": 1.5619940349382189,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.867085097404905e-07,
+      "loss": 0.3517,
+      "step": 18330
+    },
+    {
+      "epoch": 1.5620792501065188,
+      "grad_norm": 14.5,
+      "learning_rate": 2.8660143292237514e-07,
+      "loss": 0.403,
+      "step": 18331
+    },
+    {
+      "epoch": 1.5621644652748188,
+      "grad_norm": 22.0,
+      "learning_rate": 2.8649437351384063e-07,
+      "loss": 0.6531,
+      "step": 18332
+    },
+    {
+      "epoch": 1.5622496804431187,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.86387331516822e-07,
+      "loss": 0.3981,
+      "step": 18333
+    },
+    {
+      "epoch": 1.5623348956114187,
+      "grad_norm": 26.875,
+      "learning_rate": 2.8628030693325316e-07,
+      "loss": 1.0107,
+      "step": 18334
+    },
+    {
+      "epoch": 1.5624201107797187,
+      "grad_norm": 14.875,
+      "learning_rate": 2.8617329976506803e-07,
+      "loss": 0.649,
+      "step": 18335
+    },
+    {
+      "epoch": 1.5625053259480186,
+      "grad_norm": 13.75,
+      "learning_rate": 2.8606631001420054e-07,
+      "loss": 0.5941,
+      "step": 18336
+    },
+    {
+      "epoch": 1.5625905411163186,
+      "grad_norm": 12.5,
+      "learning_rate": 2.859593376825845e-07,
+      "loss": 0.3596,
+      "step": 18337
+    },
+    {
+      "epoch": 1.5626757562846185,
+      "grad_norm": 17.75,
+      "learning_rate": 2.8585238277215226e-07,
+      "loss": 0.4282,
+      "step": 18338
+    },
+    {
+      "epoch": 1.5627609714529185,
+      "grad_norm": 17.625,
+      "learning_rate": 2.857454452848371e-07,
+      "loss": 0.8426,
+      "step": 18339
+    },
+    {
+      "epoch": 1.5628461866212184,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.856385252225713e-07,
+      "loss": 0.4469,
+      "step": 18340
+    },
+    {
+      "epoch": 1.5629314017895184,
+      "grad_norm": 50.5,
+      "learning_rate": 2.8553162258728743e-07,
+      "loss": 0.6292,
+      "step": 18341
+    },
+    {
+      "epoch": 1.5630166169578183,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.854247373809166e-07,
+      "loss": 0.5563,
+      "step": 18342
+    },
+    {
+      "epoch": 1.5631018321261183,
+      "grad_norm": 16.375,
+      "learning_rate": 2.8531786960539114e-07,
+      "loss": 0.895,
+      "step": 18343
+    },
+    {
+      "epoch": 1.5631870472944183,
+      "grad_norm": 18.625,
+      "learning_rate": 2.8521101926264145e-07,
+      "loss": 0.8199,
+      "step": 18344
+    },
+    {
+      "epoch": 1.5632722624627182,
+      "grad_norm": 21.875,
+      "learning_rate": 2.8510418635459893e-07,
+      "loss": 0.7596,
+      "step": 18345
+    },
+    {
+      "epoch": 1.5633574776310182,
+      "grad_norm": 23.25,
+      "learning_rate": 2.849973708831942e-07,
+      "loss": 1.0762,
+      "step": 18346
+    },
+    {
+      "epoch": 1.5634426927993181,
+      "grad_norm": 13.5,
+      "learning_rate": 2.848905728503572e-07,
+      "loss": 0.5918,
+      "step": 18347
+    },
+    {
+      "epoch": 1.563527907967618,
+      "grad_norm": 17.375,
+      "learning_rate": 2.8478379225801797e-07,
+      "loss": 0.6026,
+      "step": 18348
+    },
+    {
+      "epoch": 1.563613123135918,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.846770291081061e-07,
+      "loss": 0.5549,
+      "step": 18349
+    },
+    {
+      "epoch": 1.563698338304218,
+      "grad_norm": 19.5,
+      "learning_rate": 2.8457028340255154e-07,
+      "loss": 0.8989,
+      "step": 18350
+    },
+    {
+      "epoch": 1.563783553472518,
+      "grad_norm": 18.125,
+      "learning_rate": 2.844635551432827e-07,
+      "loss": 0.5535,
+      "step": 18351
+    },
+    {
+      "epoch": 1.563868768640818,
+      "grad_norm": 20.75,
+      "learning_rate": 2.8435684433222795e-07,
+      "loss": 0.7023,
+      "step": 18352
+    },
+    {
+      "epoch": 1.5639539838091179,
+      "grad_norm": 15.375,
+      "learning_rate": 2.8425015097131606e-07,
+      "loss": 0.6022,
+      "step": 18353
+    },
+    {
+      "epoch": 1.5640391989774178,
+      "grad_norm": 13.625,
+      "learning_rate": 2.841434750624755e-07,
+      "loss": 0.4727,
+      "step": 18354
+    },
+    {
+      "epoch": 1.5641244141457178,
+      "grad_norm": 17.625,
+      "learning_rate": 2.840368166076332e-07,
+      "loss": 0.6407,
+      "step": 18355
+    },
+    {
+      "epoch": 1.5642096293140177,
+      "grad_norm": 17.5,
+      "learning_rate": 2.839301756087169e-07,
+      "loss": 0.8453,
+      "step": 18356
+    },
+    {
+      "epoch": 1.5642948444823177,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.83823552067654e-07,
+      "loss": 0.2862,
+      "step": 18357
+    },
+    {
+      "epoch": 1.5643800596506177,
+      "grad_norm": 13.25,
+      "learning_rate": 2.837169459863712e-07,
+      "loss": 0.6984,
+      "step": 18358
+    },
+    {
+      "epoch": 1.5644652748189176,
+      "grad_norm": 10.6875,
+      "learning_rate": 2.836103573667949e-07,
+      "loss": 0.2799,
+      "step": 18359
+    },
+    {
+      "epoch": 1.5645504899872176,
+      "grad_norm": 21.375,
+      "learning_rate": 2.83503786210851e-07,
+      "loss": 0.7326,
+      "step": 18360
+    },
+    {
+      "epoch": 1.5646357051555175,
+      "grad_norm": 23.375,
+      "learning_rate": 2.833972325204655e-07,
+      "loss": 0.7949,
+      "step": 18361
+    },
+    {
+      "epoch": 1.5647209203238175,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.83290696297564e-07,
+      "loss": 0.6501,
+      "step": 18362
+    },
+    {
+      "epoch": 1.5648061354921174,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.8318417754407206e-07,
+      "loss": 0.471,
+      "step": 18363
+    },
+    {
+      "epoch": 1.5648913506604174,
+      "grad_norm": 26.5,
+      "learning_rate": 2.830776762619139e-07,
+      "loss": 1.012,
+      "step": 18364
+    },
+    {
+      "epoch": 1.5649765658287174,
+      "grad_norm": 18.0,
+      "learning_rate": 2.8297119245301443e-07,
+      "loss": 0.7449,
+      "step": 18365
+    },
+    {
+      "epoch": 1.5650617809970173,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.8286472611929794e-07,
+      "loss": 0.6536,
+      "step": 18366
+    },
+    {
+      "epoch": 1.5651469961653173,
+      "grad_norm": 13.625,
+      "learning_rate": 2.8275827726268897e-07,
+      "loss": 0.5513,
+      "step": 18367
+    },
+    {
+      "epoch": 1.5652322113336172,
+      "grad_norm": 16.25,
+      "learning_rate": 2.8265184588511007e-07,
+      "loss": 0.6335,
+      "step": 18368
+    },
+    {
+      "epoch": 1.5653174265019172,
+      "grad_norm": 20.375,
+      "learning_rate": 2.825454319884849e-07,
+      "loss": 0.8774,
+      "step": 18369
+    },
+    {
+      "epoch": 1.5654026416702171,
+      "grad_norm": 16.5,
+      "learning_rate": 2.8243903557473666e-07,
+      "loss": 0.5222,
+      "step": 18370
+    },
+    {
+      "epoch": 1.565487856838517,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.8233265664578847e-07,
+      "loss": 0.4482,
+      "step": 18371
+    },
+    {
+      "epoch": 1.565573072006817,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.822262952035618e-07,
+      "loss": 0.5475,
+      "step": 18372
+    },
+    {
+      "epoch": 1.565658287175117,
+      "grad_norm": 22.625,
+      "learning_rate": 2.821199512499792e-07,
+      "loss": 0.7336,
+      "step": 18373
+    },
+    {
+      "epoch": 1.565743502343417,
+      "grad_norm": 14.625,
+      "learning_rate": 2.820136247869623e-07,
+      "loss": 0.7298,
+      "step": 18374
+    },
+    {
+      "epoch": 1.565828717511717,
+      "grad_norm": 16.5,
+      "learning_rate": 2.819073158164329e-07,
+      "loss": 0.674,
+      "step": 18375
+    },
+    {
+      "epoch": 1.5659139326800169,
+      "grad_norm": 22.5,
+      "learning_rate": 2.8180102434031185e-07,
+      "loss": 0.6195,
+      "step": 18376
+    },
+    {
+      "epoch": 1.5659991478483168,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.8169475036051956e-07,
+      "loss": 0.5279,
+      "step": 18377
+    },
+    {
+      "epoch": 1.5660843630166168,
+      "grad_norm": 20.0,
+      "learning_rate": 2.8158849387897674e-07,
+      "loss": 0.4234,
+      "step": 18378
+    },
+    {
+      "epoch": 1.5661695781849168,
+      "grad_norm": 16.5,
+      "learning_rate": 2.814822548976037e-07,
+      "loss": 0.5858,
+      "step": 18379
+    },
+    {
+      "epoch": 1.5662547933532167,
+      "grad_norm": 29.0,
+      "learning_rate": 2.813760334183206e-07,
+      "loss": 1.2449,
+      "step": 18380
+    },
+    {
+      "epoch": 1.5663400085215167,
+      "grad_norm": 13.875,
+      "learning_rate": 2.812698294430463e-07,
+      "loss": 0.3739,
+      "step": 18381
+    },
+    {
+      "epoch": 1.5664252236898166,
+      "grad_norm": 17.75,
+      "learning_rate": 2.8116364297370025e-07,
+      "loss": 0.9608,
+      "step": 18382
+    },
+    {
+      "epoch": 1.5665104388581166,
+      "grad_norm": 11.375,
+      "learning_rate": 2.810574740122018e-07,
+      "loss": 0.5017,
+      "step": 18383
+    },
+    {
+      "epoch": 1.5665956540264165,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.8095132256046917e-07,
+      "loss": 0.6418,
+      "step": 18384
+    },
+    {
+      "epoch": 1.5666808691947165,
+      "grad_norm": 14.125,
+      "learning_rate": 2.808451886204203e-07,
+      "loss": 0.5668,
+      "step": 18385
+    },
+    {
+      "epoch": 1.5667660843630165,
+      "grad_norm": 16.25,
+      "learning_rate": 2.8073907219397317e-07,
+      "loss": 0.7968,
+      "step": 18386
+    },
+    {
+      "epoch": 1.5668512995313164,
+      "grad_norm": 13.75,
+      "learning_rate": 2.8063297328304594e-07,
+      "loss": 0.6196,
+      "step": 18387
+    },
+    {
+      "epoch": 1.5669365146996164,
+      "grad_norm": 16.25,
+      "learning_rate": 2.8052689188955584e-07,
+      "loss": 0.4646,
+      "step": 18388
+    },
+    {
+      "epoch": 1.5670217298679163,
+      "grad_norm": 13.75,
+      "learning_rate": 2.8042082801541945e-07,
+      "loss": 0.4478,
+      "step": 18389
+    },
+    {
+      "epoch": 1.5671069450362163,
+      "grad_norm": 24.75,
+      "learning_rate": 2.803147816625536e-07,
+      "loss": 0.9445,
+      "step": 18390
+    },
+    {
+      "epoch": 1.5671921602045162,
+      "grad_norm": 13.625,
+      "learning_rate": 2.802087528328751e-07,
+      "loss": 0.4044,
+      "step": 18391
+    },
+    {
+      "epoch": 1.5672773753728162,
+      "grad_norm": 16.875,
+      "learning_rate": 2.8010274152829934e-07,
+      "loss": 0.8602,
+      "step": 18392
+    },
+    {
+      "epoch": 1.5673625905411164,
+      "grad_norm": 19.25,
+      "learning_rate": 2.799967477507426e-07,
+      "loss": 0.4261,
+      "step": 18393
+    },
+    {
+      "epoch": 1.5674478057094163,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.7989077150211985e-07,
+      "loss": 0.4724,
+      "step": 18394
+    },
+    {
+      "epoch": 1.5675330208777163,
+      "grad_norm": 15.25,
+      "learning_rate": 2.7978481278434623e-07,
+      "loss": 0.7191,
+      "step": 18395
+    },
+    {
+      "epoch": 1.5676182360460162,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.7967887159933675e-07,
+      "loss": 0.5552,
+      "step": 18396
+    },
+    {
+      "epoch": 1.5677034512143162,
+      "grad_norm": 13.75,
+      "learning_rate": 2.7957294794900614e-07,
+      "loss": 0.3924,
+      "step": 18397
+    },
+    {
+      "epoch": 1.5677886663826162,
+      "grad_norm": 16.625,
+      "learning_rate": 2.7946704183526783e-07,
+      "loss": 0.7877,
+      "step": 18398
+    },
+    {
+      "epoch": 1.5678738815509161,
+      "grad_norm": 16.25,
+      "learning_rate": 2.793611532600363e-07,
+      "loss": 0.7158,
+      "step": 18399
+    },
+    {
+      "epoch": 1.567959096719216,
+      "grad_norm": 20.5,
+      "learning_rate": 2.7925528222522454e-07,
+      "loss": 0.848,
+      "step": 18400
+    },
+    {
+      "epoch": 1.568044311887516,
+      "grad_norm": 21.5,
+      "learning_rate": 2.7914942873274634e-07,
+      "loss": 0.8966,
+      "step": 18401
+    },
+    {
+      "epoch": 1.568129527055816,
+      "grad_norm": 15.625,
+      "learning_rate": 2.790435927845138e-07,
+      "loss": 0.6607,
+      "step": 18402
+    },
+    {
+      "epoch": 1.568214742224116,
+      "grad_norm": 15.125,
+      "learning_rate": 2.7893777438243994e-07,
+      "loss": 0.5567,
+      "step": 18403
+    },
+    {
+      "epoch": 1.568299957392416,
+      "grad_norm": 16.125,
+      "learning_rate": 2.78831973528437e-07,
+      "loss": 0.4404,
+      "step": 18404
+    },
+    {
+      "epoch": 1.5683851725607159,
+      "grad_norm": 14.25,
+      "learning_rate": 2.787261902244172e-07,
+      "loss": 0.4489,
+      "step": 18405
+    },
+    {
+      "epoch": 1.5684703877290158,
+      "grad_norm": 13.875,
+      "learning_rate": 2.786204244722915e-07,
+      "loss": 0.7203,
+      "step": 18406
+    },
+    {
+      "epoch": 1.5685556028973158,
+      "grad_norm": 16.125,
+      "learning_rate": 2.7851467627397153e-07,
+      "loss": 0.5947,
+      "step": 18407
+    },
+    {
+      "epoch": 1.5686408180656157,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.784089456313686e-07,
+      "loss": 0.3658,
+      "step": 18408
+    },
+    {
+      "epoch": 1.5687260332339157,
+      "grad_norm": 18.5,
+      "learning_rate": 2.7830323254639276e-07,
+      "loss": 0.6681,
+      "step": 18409
+    },
+    {
+      "epoch": 1.5688112484022156,
+      "grad_norm": 13.375,
+      "learning_rate": 2.7819753702095486e-07,
+      "loss": 0.5404,
+      "step": 18410
+    },
+    {
+      "epoch": 1.5688964635705156,
+      "grad_norm": 17.875,
+      "learning_rate": 2.780918590569644e-07,
+      "loss": 0.964,
+      "step": 18411
+    },
+    {
+      "epoch": 1.5689816787388156,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.7798619865633136e-07,
+      "loss": 0.6041,
+      "step": 18412
+    },
+    {
+      "epoch": 1.5690668939071155,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.778805558209652e-07,
+      "loss": 0.5975,
+      "step": 18413
+    },
+    {
+      "epoch": 1.5691521090754155,
+      "grad_norm": 17.375,
+      "learning_rate": 2.777749305527752e-07,
+      "loss": 0.4763,
+      "step": 18414
+    },
+    {
+      "epoch": 1.5692373242437154,
+      "grad_norm": 11.125,
+      "learning_rate": 2.776693228536696e-07,
+      "loss": 0.4782,
+      "step": 18415
+    },
+    {
+      "epoch": 1.5693225394120154,
+      "grad_norm": 16.5,
+      "learning_rate": 2.775637327255573e-07,
+      "loss": 0.6821,
+      "step": 18416
+    },
+    {
+      "epoch": 1.5694077545803153,
+      "grad_norm": 19.0,
+      "learning_rate": 2.7745816017034605e-07,
+      "loss": 0.7034,
+      "step": 18417
+    },
+    {
+      "epoch": 1.5694929697486153,
+      "grad_norm": 16.25,
+      "learning_rate": 2.77352605189944e-07,
+      "loss": 0.5299,
+      "step": 18418
+    },
+    {
+      "epoch": 1.5695781849169153,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.772470677862582e-07,
+      "loss": 0.3171,
+      "step": 18419
+    },
+    {
+      "epoch": 1.5696634000852152,
+      "grad_norm": 16.75,
+      "learning_rate": 2.7714154796119595e-07,
+      "loss": 0.6154,
+      "step": 18420
+    },
+    {
+      "epoch": 1.5697486152535152,
+      "grad_norm": 16.625,
+      "learning_rate": 2.770360457166643e-07,
+      "loss": 0.5214,
+      "step": 18421
+    },
+    {
+      "epoch": 1.5698338304218151,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.7693056105457e-07,
+      "loss": 0.4467,
+      "step": 18422
+    },
+    {
+      "epoch": 1.569919045590115,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.768250939768187e-07,
+      "loss": 0.4451,
+      "step": 18423
+    },
+    {
+      "epoch": 1.570004260758415,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.767196444853168e-07,
+      "loss": 0.4872,
+      "step": 18424
+    },
+    {
+      "epoch": 1.570089475926715,
+      "grad_norm": 15.625,
+      "learning_rate": 2.766142125819693e-07,
+      "loss": 0.5646,
+      "step": 18425
+    },
+    {
+      "epoch": 1.570174691095015,
+      "grad_norm": 14.875,
+      "learning_rate": 2.765087982686819e-07,
+      "loss": 0.6746,
+      "step": 18426
+    },
+    {
+      "epoch": 1.570259906263315,
+      "grad_norm": 22.0,
+      "learning_rate": 2.7640340154735964e-07,
+      "loss": 0.5188,
+      "step": 18427
+    },
+    {
+      "epoch": 1.5703451214316149,
+      "grad_norm": 16.5,
+      "learning_rate": 2.762980224199066e-07,
+      "loss": 0.5773,
+      "step": 18428
+    },
+    {
+      "epoch": 1.5704303365999148,
+      "grad_norm": 16.5,
+      "learning_rate": 2.761926608882276e-07,
+      "loss": 0.6369,
+      "step": 18429
+    },
+    {
+      "epoch": 1.5705155517682148,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.7608731695422664e-07,
+      "loss": 0.6914,
+      "step": 18430
+    },
+    {
+      "epoch": 1.5706007669365147,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.759819906198069e-07,
+      "loss": 0.3968,
+      "step": 18431
+    },
+    {
+      "epoch": 1.5706859821048147,
+      "grad_norm": 13.125,
+      "learning_rate": 2.7587668188687244e-07,
+      "loss": 0.3522,
+      "step": 18432
+    },
+    {
+      "epoch": 1.5707711972731146,
+      "grad_norm": 16.5,
+      "learning_rate": 2.757713907573255e-07,
+      "loss": 0.4076,
+      "step": 18433
+    },
+    {
+      "epoch": 1.5708564124414146,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.756661172330692e-07,
+      "loss": 0.421,
+      "step": 18434
+    },
+    {
+      "epoch": 1.5709416276097146,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.755608613160063e-07,
+      "loss": 0.3806,
+      "step": 18435
+    },
+    {
+      "epoch": 1.5710268427780145,
+      "grad_norm": 16.375,
+      "learning_rate": 2.754556230080381e-07,
+      "loss": 0.7098,
+      "step": 18436
+    },
+    {
+      "epoch": 1.5711120579463145,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.753504023110667e-07,
+      "loss": 0.3752,
+      "step": 18437
+    },
+    {
+      "epoch": 1.5711972731146144,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.7524519922699375e-07,
+      "loss": 0.5133,
+      "step": 18438
+    },
+    {
+      "epoch": 1.5712824882829144,
+      "grad_norm": 16.75,
+      "learning_rate": 2.7514001375772064e-07,
+      "loss": 0.6404,
+      "step": 18439
+    },
+    {
+      "epoch": 1.5713677034512143,
+      "grad_norm": 18.75,
+      "learning_rate": 2.750348459051476e-07,
+      "loss": 0.773,
+      "step": 18440
+    },
+    {
+      "epoch": 1.5714529186195143,
+      "grad_norm": 12.625,
+      "learning_rate": 2.7492969567117514e-07,
+      "loss": 0.5183,
+      "step": 18441
+    },
+    {
+      "epoch": 1.5715381337878143,
+      "grad_norm": 14.25,
+      "learning_rate": 2.7482456305770344e-07,
+      "loss": 0.6035,
+      "step": 18442
+    },
+    {
+      "epoch": 1.5716233489561142,
+      "grad_norm": 18.0,
+      "learning_rate": 2.7471944806663266e-07,
+      "loss": 0.9852,
+      "step": 18443
+    },
+    {
+      "epoch": 1.5717085641244142,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.7461435069986235e-07,
+      "loss": 0.684,
+      "step": 18444
+    },
+    {
+      "epoch": 1.5717937792927141,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.7450927095929127e-07,
+      "loss": 0.3859,
+      "step": 18445
+    },
+    {
+      "epoch": 1.571878994461014,
+      "grad_norm": 25.25,
+      "learning_rate": 2.7440420884681863e-07,
+      "loss": 0.4881,
+      "step": 18446
+    },
+    {
+      "epoch": 1.571964209629314,
+      "grad_norm": 20.125,
+      "learning_rate": 2.742991643643432e-07,
+      "loss": 0.7466,
+      "step": 18447
+    },
+    {
+      "epoch": 1.572049424797614,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.7419413751376314e-07,
+      "loss": 0.3455,
+      "step": 18448
+    },
+    {
+      "epoch": 1.572134639965914,
+      "grad_norm": 17.5,
+      "learning_rate": 2.7408912829697584e-07,
+      "loss": 0.6424,
+      "step": 18449
+    },
+    {
+      "epoch": 1.572219855134214,
+      "grad_norm": 18.25,
+      "learning_rate": 2.739841367158795e-07,
+      "loss": 0.893,
+      "step": 18450
+    },
+    {
+      "epoch": 1.5723050703025139,
+      "grad_norm": 11.0,
+      "learning_rate": 2.7387916277237106e-07,
+      "loss": 0.3577,
+      "step": 18451
+    },
+    {
+      "epoch": 1.5723902854708138,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.7377420646834824e-07,
+      "loss": 0.5559,
+      "step": 18452
+    },
+    {
+      "epoch": 1.5724755006391138,
+      "grad_norm": 17.625,
+      "learning_rate": 2.7366926780570666e-07,
+      "loss": 0.9137,
+      "step": 18453
+    },
+    {
+      "epoch": 1.5725607158074137,
+      "grad_norm": 15.5,
+      "learning_rate": 2.735643467863433e-07,
+      "loss": 0.6162,
+      "step": 18454
+    },
+    {
+      "epoch": 1.5726459309757137,
+      "grad_norm": 19.5,
+      "learning_rate": 2.73459443412154e-07,
+      "loss": 0.788,
+      "step": 18455
+    },
+    {
+      "epoch": 1.5727311461440137,
+      "grad_norm": 20.5,
+      "learning_rate": 2.7335455768503494e-07,
+      "loss": 0.7222,
+      "step": 18456
+    },
+    {
+      "epoch": 1.5728163613123136,
+      "grad_norm": 29.75,
+      "learning_rate": 2.732496896068812e-07,
+      "loss": 0.6256,
+      "step": 18457
+    },
+    {
+      "epoch": 1.5729015764806136,
+      "grad_norm": 38.75,
+      "learning_rate": 2.7314483917958724e-07,
+      "loss": 1.3189,
+      "step": 18458
+    },
+    {
+      "epoch": 1.5729867916489135,
+      "grad_norm": 16.25,
+      "learning_rate": 2.7304000640504855e-07,
+      "loss": 0.6024,
+      "step": 18459
+    },
+    {
+      "epoch": 1.5730720068172135,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.729351912851595e-07,
+      "loss": 0.5381,
+      "step": 18460
+    },
+    {
+      "epoch": 1.5731572219855134,
+      "grad_norm": 23.75,
+      "learning_rate": 2.728303938218138e-07,
+      "loss": 0.7995,
+      "step": 18461
+    },
+    {
+      "epoch": 1.5732424371538134,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.7272561401690557e-07,
+      "loss": 0.3877,
+      "step": 18462
+    },
+    {
+      "epoch": 1.5733276523221134,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.726208518723282e-07,
+      "loss": 0.3317,
+      "step": 18463
+    },
+    {
+      "epoch": 1.5734128674904133,
+      "grad_norm": 11.0,
+      "learning_rate": 2.7251610738997515e-07,
+      "loss": 0.4387,
+      "step": 18464
+    },
+    {
+      "epoch": 1.5734980826587133,
+      "grad_norm": 18.25,
+      "learning_rate": 2.724113805717389e-07,
+      "loss": 0.9723,
+      "step": 18465
+    },
+    {
+      "epoch": 1.5735832978270132,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.723066714195119e-07,
+      "loss": 0.3894,
+      "step": 18466
+    },
+    {
+      "epoch": 1.5736685129953132,
+      "grad_norm": 18.125,
+      "learning_rate": 2.722019799351863e-07,
+      "loss": 0.4327,
+      "step": 18467
+    },
+    {
+      "epoch": 1.5737537281636131,
+      "grad_norm": 10.0,
+      "learning_rate": 2.720973061206543e-07,
+      "loss": 0.3784,
+      "step": 18468
+    },
+    {
+      "epoch": 1.573838943331913,
+      "grad_norm": 12.25,
+      "learning_rate": 2.7199264997780765e-07,
+      "loss": 0.3998,
+      "step": 18469
+    },
+    {
+      "epoch": 1.573924158500213,
+      "grad_norm": 19.875,
+      "learning_rate": 2.718880115085369e-07,
+      "loss": 0.7107,
+      "step": 18470
+    },
+    {
+      "epoch": 1.574009373668513,
+      "grad_norm": 16.125,
+      "learning_rate": 2.7178339071473337e-07,
+      "loss": 0.8626,
+      "step": 18471
+    },
+    {
+      "epoch": 1.574094588836813,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.7167878759828764e-07,
+      "loss": 0.7088,
+      "step": 18472
+    },
+    {
+      "epoch": 1.574179804005113,
+      "grad_norm": 20.875,
+      "learning_rate": 2.7157420216109035e-07,
+      "loss": 0.9969,
+      "step": 18473
+    },
+    {
+      "epoch": 1.5742650191734129,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.714696344050312e-07,
+      "loss": 0.3811,
+      "step": 18474
+    },
+    {
+      "epoch": 1.5743502343417128,
+      "grad_norm": 16.125,
+      "learning_rate": 2.7136508433199926e-07,
+      "loss": 0.7363,
+      "step": 18475
+    },
+    {
+      "epoch": 1.5744354495100128,
+      "grad_norm": 12.0,
+      "learning_rate": 2.7126055194388433e-07,
+      "loss": 0.3281,
+      "step": 18476
+    },
+    {
+      "epoch": 1.5745206646783128,
+      "grad_norm": 14.5,
+      "learning_rate": 2.7115603724257586e-07,
+      "loss": 0.3907,
+      "step": 18477
+    },
+    {
+      "epoch": 1.5746058798466127,
+      "grad_norm": 16.375,
+      "learning_rate": 2.7105154022996167e-07,
+      "loss": 0.6436,
+      "step": 18478
+    },
+    {
+      "epoch": 1.5746910950149127,
+      "grad_norm": 13.625,
+      "learning_rate": 2.7094706090793065e-07,
+      "loss": 0.6307,
+      "step": 18479
+    },
+    {
+      "epoch": 1.5747763101832126,
+      "grad_norm": 21.75,
+      "learning_rate": 2.708425992783707e-07,
+      "loss": 0.7852,
+      "step": 18480
+    },
+    {
+      "epoch": 1.5748615253515126,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.7073815534317e-07,
+      "loss": 0.4404,
+      "step": 18481
+    },
+    {
+      "epoch": 1.5749467405198125,
+      "grad_norm": 18.125,
+      "learning_rate": 2.706337291042155e-07,
+      "loss": 0.6337,
+      "step": 18482
+    },
+    {
+      "epoch": 1.5750319556881125,
+      "grad_norm": 18.75,
+      "learning_rate": 2.7052932056339396e-07,
+      "loss": 0.9609,
+      "step": 18483
+    },
+    {
+      "epoch": 1.5751171708564125,
+      "grad_norm": 13.625,
+      "learning_rate": 2.7042492972259263e-07,
+      "loss": 0.589,
+      "step": 18484
+    },
+    {
+      "epoch": 1.5752023860247124,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.7032055658369777e-07,
+      "loss": 0.5916,
+      "step": 18485
+    },
+    {
+      "epoch": 1.5752876011930124,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.702162011485961e-07,
+      "loss": 0.2893,
+      "step": 18486
+    },
+    {
+      "epoch": 1.5753728163613123,
+      "grad_norm": 13.125,
+      "learning_rate": 2.7011186341917237e-07,
+      "loss": 0.4326,
+      "step": 18487
+    },
+    {
+      "epoch": 1.5754580315296123,
+      "grad_norm": 19.125,
+      "learning_rate": 2.7000754339731277e-07,
+      "loss": 0.6215,
+      "step": 18488
+    },
+    {
+      "epoch": 1.5755432466979122,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.6990324108490256e-07,
+      "loss": 0.5567,
+      "step": 18489
+    },
+    {
+      "epoch": 1.5756284618662122,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.6979895648382603e-07,
+      "loss": 0.5002,
+      "step": 18490
+    },
+    {
+      "epoch": 1.5757136770345122,
+      "grad_norm": 14.875,
+      "learning_rate": 2.6969468959596845e-07,
+      "loss": 0.4885,
+      "step": 18491
+    },
+    {
+      "epoch": 1.575798892202812,
+      "grad_norm": 11.125,
+      "learning_rate": 2.6959044042321316e-07,
+      "loss": 0.4522,
+      "step": 18492
+    },
+    {
+      "epoch": 1.575884107371112,
+      "grad_norm": 22.0,
+      "learning_rate": 2.694862089674445e-07,
+      "loss": 0.6917,
+      "step": 18493
+    },
+    {
+      "epoch": 1.575969322539412,
+      "grad_norm": 11.75,
+      "learning_rate": 2.693819952305464e-07,
+      "loss": 0.3125,
+      "step": 18494
+    },
+    {
+      "epoch": 1.576054537707712,
+      "grad_norm": 12.375,
+      "learning_rate": 2.6927779921440145e-07,
+      "loss": 0.5023,
+      "step": 18495
+    },
+    {
+      "epoch": 1.576139752876012,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.691736209208928e-07,
+      "loss": 0.3659,
+      "step": 18496
+    },
+    {
+      "epoch": 1.576224968044312,
+      "grad_norm": 14.625,
+      "learning_rate": 2.690694603519034e-07,
+      "loss": 0.5715,
+      "step": 18497
+    },
+    {
+      "epoch": 1.5763101832126118,
+      "grad_norm": 17.0,
+      "learning_rate": 2.68965317509315e-07,
+      "loss": 0.7785,
+      "step": 18498
+    },
+    {
+      "epoch": 1.5763953983809118,
+      "grad_norm": 11.625,
+      "learning_rate": 2.688611923950102e-07,
+      "loss": 0.4601,
+      "step": 18499
+    },
+    {
+      "epoch": 1.5764806135492118,
+      "grad_norm": 19.75,
+      "learning_rate": 2.6875708501086976e-07,
+      "loss": 0.8054,
+      "step": 18500
+    },
+    {
+      "epoch": 1.5765658287175117,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.686529953587756e-07,
+      "loss": 0.4458,
+      "step": 18501
+    },
+    {
+      "epoch": 1.5766510438858117,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.685489234406087e-07,
+      "loss": 0.3875,
+      "step": 18502
+    },
+    {
+      "epoch": 1.5767362590541116,
+      "grad_norm": 12.875,
+      "learning_rate": 2.684448692582499e-07,
+      "loss": 0.589,
+      "step": 18503
+    },
+    {
+      "epoch": 1.5768214742224116,
+      "grad_norm": 16.375,
+      "learning_rate": 2.6834083281357904e-07,
+      "loss": 0.518,
+      "step": 18504
+    },
+    {
+      "epoch": 1.5769066893907115,
+      "grad_norm": 18.875,
+      "learning_rate": 2.682368141084769e-07,
+      "loss": 0.691,
+      "step": 18505
+    },
+    {
+      "epoch": 1.5769919045590115,
+      "grad_norm": 31.0,
+      "learning_rate": 2.681328131448223e-07,
+      "loss": 0.6377,
+      "step": 18506
+    },
+    {
+      "epoch": 1.5770771197273115,
+      "grad_norm": 17.5,
+      "learning_rate": 2.680288299244956e-07,
+      "loss": 0.7858,
+      "step": 18507
+    },
+    {
+      "epoch": 1.5771623348956114,
+      "grad_norm": 18.0,
+      "learning_rate": 2.679248644493748e-07,
+      "loss": 0.732,
+      "step": 18508
+    },
+    {
+      "epoch": 1.5772475500639114,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.6782091672133957e-07,
+      "loss": 0.4285,
+      "step": 18509
+    },
+    {
+      "epoch": 1.5773327652322113,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.677169867422678e-07,
+      "loss": 0.4973,
+      "step": 18510
+    },
+    {
+      "epoch": 1.5774179804005113,
+      "grad_norm": 18.375,
+      "learning_rate": 2.6761307451403833e-07,
+      "loss": 0.6311,
+      "step": 18511
+    },
+    {
+      "epoch": 1.5775031955688112,
+      "grad_norm": 16.875,
+      "learning_rate": 2.6750918003852796e-07,
+      "loss": 0.7236,
+      "step": 18512
+    },
+    {
+      "epoch": 1.5775884107371112,
+      "grad_norm": 18.25,
+      "learning_rate": 2.6740530331761506e-07,
+      "loss": 0.7436,
+      "step": 18513
+    },
+    {
+      "epoch": 1.5776736259054112,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.6730144435317617e-07,
+      "loss": 0.2203,
+      "step": 18514
+    },
+    {
+      "epoch": 1.5777588410737111,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.671976031470883e-07,
+      "loss": 0.6666,
+      "step": 18515
+    },
+    {
+      "epoch": 1.577844056242011,
+      "grad_norm": 15.5,
+      "learning_rate": 2.670937797012284e-07,
+      "loss": 0.6546,
+      "step": 18516
+    },
+    {
+      "epoch": 1.577929271410311,
+      "grad_norm": 18.25,
+      "learning_rate": 2.6698997401747196e-07,
+      "loss": 0.6795,
+      "step": 18517
+    },
+    {
+      "epoch": 1.578014486578611,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.668861860976951e-07,
+      "loss": 0.5499,
+      "step": 18518
+    },
+    {
+      "epoch": 1.578099701746911,
+      "grad_norm": 16.875,
+      "learning_rate": 2.6678241594377346e-07,
+      "loss": 0.6227,
+      "step": 18519
+    },
+    {
+      "epoch": 1.578184916915211,
+      "grad_norm": 17.0,
+      "learning_rate": 2.666786635575827e-07,
+      "loss": 0.8236,
+      "step": 18520
+    },
+    {
+      "epoch": 1.5782701320835109,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.665749289409972e-07,
+      "loss": 0.49,
+      "step": 18521
+    },
+    {
+      "epoch": 1.5783553472518108,
+      "grad_norm": 15.5,
+      "learning_rate": 2.6647121209589136e-07,
+      "loss": 0.6541,
+      "step": 18522
+    },
+    {
+      "epoch": 1.5784405624201108,
+      "grad_norm": 19.25,
+      "learning_rate": 2.663675130241397e-07,
+      "loss": 0.8065,
+      "step": 18523
+    },
+    {
+      "epoch": 1.5785257775884107,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.662638317276166e-07,
+      "loss": 0.3863,
+      "step": 18524
+    },
+    {
+      "epoch": 1.5786109927567107,
+      "grad_norm": 18.0,
+      "learning_rate": 2.661601682081949e-07,
+      "loss": 0.5858,
+      "step": 18525
+    },
+    {
+      "epoch": 1.5786962079250106,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.660565224677482e-07,
+      "loss": 0.2623,
+      "step": 18526
+    },
+    {
+      "epoch": 1.5787814230933106,
+      "grad_norm": 12.0,
+      "learning_rate": 2.6595289450814966e-07,
+      "loss": 0.3062,
+      "step": 18527
+    },
+    {
+      "epoch": 1.5788666382616106,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.6584928433127227e-07,
+      "loss": 0.7284,
+      "step": 18528
+    },
+    {
+      "epoch": 1.5789518534299105,
+      "grad_norm": 18.625,
+      "learning_rate": 2.657456919389878e-07,
+      "loss": 0.6329,
+      "step": 18529
+    },
+    {
+      "epoch": 1.5790370685982105,
+      "grad_norm": 10.75,
+      "learning_rate": 2.6564211733316817e-07,
+      "loss": 0.247,
+      "step": 18530
+    },
+    {
+      "epoch": 1.5791222837665104,
+      "grad_norm": 14.0,
+      "learning_rate": 2.655385605156853e-07,
+      "loss": 0.4894,
+      "step": 18531
+    },
+    {
+      "epoch": 1.5792074989348104,
+      "grad_norm": 17.0,
+      "learning_rate": 2.6543502148841065e-07,
+      "loss": 0.8102,
+      "step": 18532
+    },
+    {
+      "epoch": 1.5792927141031103,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.6533150025321547e-07,
+      "loss": 0.7557,
+      "step": 18533
+    },
+    {
+      "epoch": 1.5793779292714103,
+      "grad_norm": 14.0,
+      "learning_rate": 2.652279968119699e-07,
+      "loss": 0.5465,
+      "step": 18534
+    },
+    {
+      "epoch": 1.5794631444397103,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.651245111665449e-07,
+      "loss": 0.5156,
+      "step": 18535
+    },
+    {
+      "epoch": 1.5795483596080102,
+      "grad_norm": 16.75,
+      "learning_rate": 2.6502104331881007e-07,
+      "loss": 0.5967,
+      "step": 18536
+    },
+    {
+      "epoch": 1.5796335747763102,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.649175932706363e-07,
+      "loss": 0.4047,
+      "step": 18537
+    },
+    {
+      "epoch": 1.5797187899446101,
+      "grad_norm": 16.75,
+      "learning_rate": 2.6481416102389147e-07,
+      "loss": 0.4963,
+      "step": 18538
+    },
+    {
+      "epoch": 1.57980400511291,
+      "grad_norm": 14.75,
+      "learning_rate": 2.6471074658044537e-07,
+      "loss": 0.49,
+      "step": 18539
+    },
+    {
+      "epoch": 1.57988922028121,
+      "grad_norm": 18.75,
+      "learning_rate": 2.64607349942167e-07,
+      "loss": 0.5962,
+      "step": 18540
+    },
+    {
+      "epoch": 1.57997443544951,
+      "grad_norm": 22.125,
+      "learning_rate": 2.6450397111092497e-07,
+      "loss": 0.5567,
+      "step": 18541
+    },
+    {
+      "epoch": 1.58005965061781,
+      "grad_norm": 19.875,
+      "learning_rate": 2.644006100885868e-07,
+      "loss": 0.7394,
+      "step": 18542
+    },
+    {
+      "epoch": 1.58014486578611,
+      "grad_norm": 13.25,
+      "learning_rate": 2.642972668770208e-07,
+      "loss": 0.5319,
+      "step": 18543
+    },
+    {
+      "epoch": 1.5802300809544099,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.6419394147809434e-07,
+      "loss": 1.1369,
+      "step": 18544
+    },
+    {
+      "epoch": 1.5803152961227098,
+      "grad_norm": 13.75,
+      "learning_rate": 2.6409063389367497e-07,
+      "loss": 0.5735,
+      "step": 18545
+    },
+    {
+      "epoch": 1.5804005112910098,
+      "grad_norm": 19.375,
+      "learning_rate": 2.6398734412562925e-07,
+      "loss": 0.6798,
+      "step": 18546
+    },
+    {
+      "epoch": 1.5804857264593097,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.6388407217582345e-07,
+      "loss": 0.3749,
+      "step": 18547
+    },
+    {
+      "epoch": 1.5805709416276097,
+      "grad_norm": 17.5,
+      "learning_rate": 2.6378081804612407e-07,
+      "loss": 0.557,
+      "step": 18548
+    },
+    {
+      "epoch": 1.5806561567959097,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.63677581738397e-07,
+      "loss": 0.3307,
+      "step": 18549
+    },
+    {
+      "epoch": 1.5807413719642096,
+      "grad_norm": 35.0,
+      "learning_rate": 2.6357436325450817e-07,
+      "loss": 0.8709,
+      "step": 18550
+    },
+    {
+      "epoch": 1.5808265871325096,
+      "grad_norm": 16.25,
+      "learning_rate": 2.6347116259632226e-07,
+      "loss": 0.6508,
+      "step": 18551
+    },
+    {
+      "epoch": 1.5809118023008095,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.633679797657043e-07,
+      "loss": 0.3805,
+      "step": 18552
+    },
+    {
+      "epoch": 1.5809970174691095,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.632648147645193e-07,
+      "loss": 0.502,
+      "step": 18553
+    },
+    {
+      "epoch": 1.5810822326374094,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.631616675946319e-07,
+      "loss": 0.291,
+      "step": 18554
+    },
+    {
+      "epoch": 1.5811674478057094,
+      "grad_norm": 17.25,
+      "learning_rate": 2.630585382579048e-07,
+      "loss": 0.7447,
+      "step": 18555
+    },
+    {
+      "epoch": 1.5812526629740093,
+      "grad_norm": 15.0,
+      "learning_rate": 2.629554267562025e-07,
+      "loss": 0.6566,
+      "step": 18556
+    },
+    {
+      "epoch": 1.5813378781423093,
+      "grad_norm": 16.5,
+      "learning_rate": 2.6285233309138807e-07,
+      "loss": 0.7805,
+      "step": 18557
+    },
+    {
+      "epoch": 1.5814230933106093,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.627492572653251e-07,
+      "loss": 0.7582,
+      "step": 18558
+    },
+    {
+      "epoch": 1.5815083084789092,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.6264619927987527e-07,
+      "loss": 0.5567,
+      "step": 18559
+    },
+    {
+      "epoch": 1.5815935236472092,
+      "grad_norm": 16.5,
+      "learning_rate": 2.625431591369017e-07,
+      "loss": 0.8404,
+      "step": 18560
+    },
+    {
+      "epoch": 1.5816787388155091,
+      "grad_norm": 16.75,
+      "learning_rate": 2.6244013683826603e-07,
+      "loss": 0.7119,
+      "step": 18561
+    },
+    {
+      "epoch": 1.581763953983809,
+      "grad_norm": 15.75,
+      "learning_rate": 2.623371323858305e-07,
+      "loss": 0.4319,
+      "step": 18562
+    },
+    {
+      "epoch": 1.581849169152109,
+      "grad_norm": 13.5,
+      "learning_rate": 2.6223414578145633e-07,
+      "loss": 0.4016,
+      "step": 18563
+    },
+    {
+      "epoch": 1.581934384320409,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.6213117702700395e-07,
+      "loss": 0.3826,
+      "step": 18564
+    },
+    {
+      "epoch": 1.582019599488709,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.6202822612433453e-07,
+      "loss": 0.3079,
+      "step": 18565
+    },
+    {
+      "epoch": 1.582104814657009,
+      "grad_norm": 17.0,
+      "learning_rate": 2.619252930753087e-07,
+      "loss": 0.6956,
+      "step": 18566
+    },
+    {
+      "epoch": 1.5821900298253089,
+      "grad_norm": 17.75,
+      "learning_rate": 2.618223778817866e-07,
+      "loss": 0.6405,
+      "step": 18567
+    },
+    {
+      "epoch": 1.5822752449936088,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.617194805456276e-07,
+      "loss": 0.5696,
+      "step": 18568
+    },
+    {
+      "epoch": 1.5823604601619088,
+      "grad_norm": 15.375,
+      "learning_rate": 2.616166010686913e-07,
+      "loss": 0.6368,
+      "step": 18569
+    },
+    {
+      "epoch": 1.5824456753302087,
+      "grad_norm": 14.5,
+      "learning_rate": 2.6151373945283734e-07,
+      "loss": 0.476,
+      "step": 18570
+    },
+    {
+      "epoch": 1.5825308904985087,
+      "grad_norm": 15.625,
+      "learning_rate": 2.614108956999241e-07,
+      "loss": 0.5464,
+      "step": 18571
+    },
+    {
+      "epoch": 1.5826161056668087,
+      "grad_norm": 15.75,
+      "learning_rate": 2.6130806981180965e-07,
+      "loss": 0.5787,
+      "step": 18572
+    },
+    {
+      "epoch": 1.5827013208351086,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.6120526179035263e-07,
+      "loss": 0.3786,
+      "step": 18573
+    },
+    {
+      "epoch": 1.5827865360034086,
+      "grad_norm": 15.75,
+      "learning_rate": 2.611024716374108e-07,
+      "loss": 0.587,
+      "step": 18574
+    },
+    {
+      "epoch": 1.5828717511717085,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.6099969935484205e-07,
+      "loss": 0.5273,
+      "step": 18575
+    },
+    {
+      "epoch": 1.5829569663400085,
+      "grad_norm": 19.0,
+      "learning_rate": 2.60896944944503e-07,
+      "loss": 0.5657,
+      "step": 18576
+    },
+    {
+      "epoch": 1.5830421815083084,
+      "grad_norm": 16.625,
+      "learning_rate": 2.6079420840825073e-07,
+      "loss": 0.655,
+      "step": 18577
+    },
+    {
+      "epoch": 1.5831273966766084,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.606914897479422e-07,
+      "loss": 0.691,
+      "step": 18578
+    },
+    {
+      "epoch": 1.5832126118449084,
+      "grad_norm": 26.875,
+      "learning_rate": 2.6058878896543294e-07,
+      "loss": 0.8032,
+      "step": 18579
+    },
+    {
+      "epoch": 1.5832978270132083,
+      "grad_norm": 12.375,
+      "learning_rate": 2.6048610606257946e-07,
+      "loss": 0.3089,
+      "step": 18580
+    },
+    {
+      "epoch": 1.5833830421815083,
+      "grad_norm": 13.875,
+      "learning_rate": 2.6038344104123673e-07,
+      "loss": 0.4432,
+      "step": 18581
+    },
+    {
+      "epoch": 1.5834682573498082,
+      "grad_norm": 25.5,
+      "learning_rate": 2.602807939032605e-07,
+      "loss": 0.8797,
+      "step": 18582
+    },
+    {
+      "epoch": 1.5835534725181082,
+      "grad_norm": 22.0,
+      "learning_rate": 2.601781646505054e-07,
+      "loss": 0.7887,
+      "step": 18583
+    },
+    {
+      "epoch": 1.5836386876864081,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.6007555328482663e-07,
+      "loss": 0.2908,
+      "step": 18584
+    },
+    {
+      "epoch": 1.583723902854708,
+      "grad_norm": 19.5,
+      "learning_rate": 2.5997295980807775e-07,
+      "loss": 0.7696,
+      "step": 18585
+    },
+    {
+      "epoch": 1.583809118023008,
+      "grad_norm": 19.75,
+      "learning_rate": 2.598703842221133e-07,
+      "loss": 0.6926,
+      "step": 18586
+    },
+    {
+      "epoch": 1.583894333191308,
+      "grad_norm": 15.0,
+      "learning_rate": 2.597678265287865e-07,
+      "loss": 0.5404,
+      "step": 18587
+    },
+    {
+      "epoch": 1.583979548359608,
+      "grad_norm": 22.0,
+      "learning_rate": 2.5966528672995097e-07,
+      "loss": 0.3412,
+      "step": 18588
+    },
+    {
+      "epoch": 1.584064763527908,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.595627648274593e-07,
+      "loss": 0.7037,
+      "step": 18589
+    },
+    {
+      "epoch": 1.5841499786962079,
+      "grad_norm": 15.0,
+      "learning_rate": 2.594602608231643e-07,
+      "loss": 0.6279,
+      "step": 18590
+    },
+    {
+      "epoch": 1.5842351938645078,
+      "grad_norm": 16.875,
+      "learning_rate": 2.5935777471891867e-07,
+      "loss": 0.947,
+      "step": 18591
+    },
+    {
+      "epoch": 1.5843204090328078,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.592553065165744e-07,
+      "loss": 0.3969,
+      "step": 18592
+    },
+    {
+      "epoch": 1.5844056242011078,
+      "grad_norm": 16.0,
+      "learning_rate": 2.591528562179828e-07,
+      "loss": 0.564,
+      "step": 18593
+    },
+    {
+      "epoch": 1.5844908393694077,
+      "grad_norm": 22.25,
+      "learning_rate": 2.5905042382499567e-07,
+      "loss": 0.9315,
+      "step": 18594
+    },
+    {
+      "epoch": 1.5845760545377077,
+      "grad_norm": 18.75,
+      "learning_rate": 2.5894800933946365e-07,
+      "loss": 0.5244,
+      "step": 18595
+    },
+    {
+      "epoch": 1.5846612697060076,
+      "grad_norm": 21.125,
+      "learning_rate": 2.588456127632376e-07,
+      "loss": 0.4891,
+      "step": 18596
+    },
+    {
+      "epoch": 1.5847464848743076,
+      "grad_norm": 18.5,
+      "learning_rate": 2.5874323409816824e-07,
+      "loss": 0.6796,
+      "step": 18597
+    },
+    {
+      "epoch": 1.5848317000426075,
+      "grad_norm": 25.75,
+      "learning_rate": 2.586408733461053e-07,
+      "loss": 0.7024,
+      "step": 18598
+    },
+    {
+      "epoch": 1.5849169152109075,
+      "grad_norm": 14.375,
+      "learning_rate": 2.5853853050889845e-07,
+      "loss": 0.58,
+      "step": 18599
+    },
+    {
+      "epoch": 1.5850021303792075,
+      "grad_norm": 13.5,
+      "learning_rate": 2.584362055883979e-07,
+      "loss": 0.5231,
+      "step": 18600
+    },
+    {
+      "epoch": 1.5850873455475074,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.583338985864517e-07,
+      "loss": 0.6363,
+      "step": 18601
+    },
+    {
+      "epoch": 1.5851725607158074,
+      "grad_norm": 17.5,
+      "learning_rate": 2.5823160950490963e-07,
+      "loss": 0.7232,
+      "step": 18602
+    },
+    {
+      "epoch": 1.5852577758841073,
+      "grad_norm": 14.75,
+      "learning_rate": 2.5812933834561933e-07,
+      "loss": 0.5057,
+      "step": 18603
+    },
+    {
+      "epoch": 1.5853429910524073,
+      "grad_norm": 25.125,
+      "learning_rate": 2.580270851104291e-07,
+      "loss": 1.0878,
+      "step": 18604
+    },
+    {
+      "epoch": 1.5854282062207072,
+      "grad_norm": 18.375,
+      "learning_rate": 2.5792484980118743e-07,
+      "loss": 0.8084,
+      "step": 18605
+    },
+    {
+      "epoch": 1.5855134213890072,
+      "grad_norm": 12.375,
+      "learning_rate": 2.5782263241974107e-07,
+      "loss": 0.4712,
+      "step": 18606
+    },
+    {
+      "epoch": 1.5855986365573072,
+      "grad_norm": 12.5,
+      "learning_rate": 2.577204329679374e-07,
+      "loss": 0.2318,
+      "step": 18607
+    },
+    {
+      "epoch": 1.585683851725607,
+      "grad_norm": 16.5,
+      "learning_rate": 2.576182514476232e-07,
+      "loss": 0.6264,
+      "step": 18608
+    },
+    {
+      "epoch": 1.585769066893907,
+      "grad_norm": 23.25,
+      "learning_rate": 2.575160878606456e-07,
+      "loss": 0.8851,
+      "step": 18609
+    },
+    {
+      "epoch": 1.585854282062207,
+      "grad_norm": 11.875,
+      "learning_rate": 2.5741394220884993e-07,
+      "loss": 0.3612,
+      "step": 18610
+    },
+    {
+      "epoch": 1.585939497230507,
+      "grad_norm": 16.375,
+      "learning_rate": 2.5731181449408266e-07,
+      "loss": 0.789,
+      "step": 18611
+    },
+    {
+      "epoch": 1.586024712398807,
+      "grad_norm": 19.125,
+      "learning_rate": 2.57209704718189e-07,
+      "loss": 0.4576,
+      "step": 18612
+    },
+    {
+      "epoch": 1.586109927567107,
+      "grad_norm": 14.625,
+      "learning_rate": 2.5710761288301414e-07,
+      "loss": 0.7679,
+      "step": 18613
+    },
+    {
+      "epoch": 1.5861951427354068,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.570055389904036e-07,
+      "loss": 0.6355,
+      "step": 18614
+    },
+    {
+      "epoch": 1.5862803579037068,
+      "grad_norm": 17.125,
+      "learning_rate": 2.569034830422011e-07,
+      "loss": 0.3793,
+      "step": 18615
+    },
+    {
+      "epoch": 1.5863655730720068,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.568014450402513e-07,
+      "loss": 0.5668,
+      "step": 18616
+    },
+    {
+      "epoch": 1.5864507882403067,
+      "grad_norm": 16.875,
+      "learning_rate": 2.566994249863984e-07,
+      "loss": 0.7687,
+      "step": 18617
+    },
+    {
+      "epoch": 1.5865360034086067,
+      "grad_norm": 16.625,
+      "learning_rate": 2.5659742288248554e-07,
+      "loss": 0.4177,
+      "step": 18618
+    },
+    {
+      "epoch": 1.5866212185769069,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.564954387303564e-07,
+      "loss": 0.4472,
+      "step": 18619
+    },
+    {
+      "epoch": 1.5867064337452068,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.5639347253185345e-07,
+      "loss": 0.5648,
+      "step": 18620
+    },
+    {
+      "epoch": 1.5867916489135068,
+      "grad_norm": 22.375,
+      "learning_rate": 2.562915242888196e-07,
+      "loss": 0.832,
+      "step": 18621
+    },
+    {
+      "epoch": 1.5868768640818067,
+      "grad_norm": 17.875,
+      "learning_rate": 2.561895940030976e-07,
+      "loss": 1.0256,
+      "step": 18622
+    },
+    {
+      "epoch": 1.5869620792501067,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.560876816765284e-07,
+      "loss": 0.4695,
+      "step": 18623
+    },
+    {
+      "epoch": 1.5870472944184066,
+      "grad_norm": 12.75,
+      "learning_rate": 2.5598578731095445e-07,
+      "loss": 0.4743,
+      "step": 18624
+    },
+    {
+      "epoch": 1.5871325095867066,
+      "grad_norm": 16.125,
+      "learning_rate": 2.5588391090821687e-07,
+      "loss": 0.6687,
+      "step": 18625
+    },
+    {
+      "epoch": 1.5872177247550066,
+      "grad_norm": 13.125,
+      "learning_rate": 2.557820524701569e-07,
+      "loss": 0.4836,
+      "step": 18626
+    },
+    {
+      "epoch": 1.5873029399233065,
+      "grad_norm": 10.125,
+      "learning_rate": 2.5568021199861515e-07,
+      "loss": 0.2403,
+      "step": 18627
+    },
+    {
+      "epoch": 1.5873881550916065,
+      "grad_norm": 17.375,
+      "learning_rate": 2.555783894954314e-07,
+      "loss": 0.6275,
+      "step": 18628
+    },
+    {
+      "epoch": 1.5874733702599064,
+      "grad_norm": 14.0,
+      "learning_rate": 2.5547658496244606e-07,
+      "loss": 0.4196,
+      "step": 18629
+    },
+    {
+      "epoch": 1.5875585854282064,
+      "grad_norm": 13.25,
+      "learning_rate": 2.553747984014994e-07,
+      "loss": 0.4438,
+      "step": 18630
+    },
+    {
+      "epoch": 1.5876438005965063,
+      "grad_norm": 14.375,
+      "learning_rate": 2.5527302981442987e-07,
+      "loss": 0.3848,
+      "step": 18631
+    },
+    {
+      "epoch": 1.5877290157648063,
+      "grad_norm": 19.375,
+      "learning_rate": 2.5517127920307695e-07,
+      "loss": 0.7541,
+      "step": 18632
+    },
+    {
+      "epoch": 1.5878142309331063,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.550695465692794e-07,
+      "loss": 0.6279,
+      "step": 18633
+    },
+    {
+      "epoch": 1.5878994461014062,
+      "grad_norm": 12.0,
+      "learning_rate": 2.5496783191487603e-07,
+      "loss": 0.48,
+      "step": 18634
+    },
+    {
+      "epoch": 1.5879846612697062,
+      "grad_norm": 12.375,
+      "learning_rate": 2.548661352417045e-07,
+      "loss": 0.3846,
+      "step": 18635
+    },
+    {
+      "epoch": 1.5880698764380061,
+      "grad_norm": 16.25,
+      "learning_rate": 2.547644565516022e-07,
+      "loss": 0.622,
+      "step": 18636
+    },
+    {
+      "epoch": 1.588155091606306,
+      "grad_norm": 17.875,
+      "learning_rate": 2.5466279584640705e-07,
+      "loss": 0.5017,
+      "step": 18637
+    },
+    {
+      "epoch": 1.588240306774606,
+      "grad_norm": 20.625,
+      "learning_rate": 2.54561153127956e-07,
+      "loss": 0.8189,
+      "step": 18638
+    },
+    {
+      "epoch": 1.588325521942906,
+      "grad_norm": 15.5,
+      "learning_rate": 2.544595283980862e-07,
+      "loss": 0.4061,
+      "step": 18639
+    },
+    {
+      "epoch": 1.588410737111206,
+      "grad_norm": 15.125,
+      "learning_rate": 2.5435792165863353e-07,
+      "loss": 0.6542,
+      "step": 18640
+    },
+    {
+      "epoch": 1.588495952279506,
+      "grad_norm": 18.375,
+      "learning_rate": 2.542563329114345e-07,
+      "loss": 0.5429,
+      "step": 18641
+    },
+    {
+      "epoch": 1.5885811674478059,
+      "grad_norm": 16.625,
+      "learning_rate": 2.541547621583247e-07,
+      "loss": 0.4337,
+      "step": 18642
+    },
+    {
+      "epoch": 1.5886663826161058,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.540532094011401e-07,
+      "loss": 0.3385,
+      "step": 18643
+    },
+    {
+      "epoch": 1.5887515977844058,
+      "grad_norm": 12.125,
+      "learning_rate": 2.5395167464171545e-07,
+      "loss": 0.2758,
+      "step": 18644
+    },
+    {
+      "epoch": 1.5888368129527057,
+      "grad_norm": 14.625,
+      "learning_rate": 2.538501578818854e-07,
+      "loss": 0.7563,
+      "step": 18645
+    },
+    {
+      "epoch": 1.5889220281210057,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.537486591234847e-07,
+      "loss": 0.5784,
+      "step": 18646
+    },
+    {
+      "epoch": 1.5890072432893056,
+      "grad_norm": 19.125,
+      "learning_rate": 2.536471783683479e-07,
+      "loss": 1.0476,
+      "step": 18647
+    },
+    {
+      "epoch": 1.5890924584576056,
+      "grad_norm": 14.125,
+      "learning_rate": 2.535457156183081e-07,
+      "loss": 0.5884,
+      "step": 18648
+    },
+    {
+      "epoch": 1.5891776736259056,
+      "grad_norm": 13.875,
+      "learning_rate": 2.534442708751993e-07,
+      "loss": 0.4893,
+      "step": 18649
+    },
+    {
+      "epoch": 1.5892628887942055,
+      "grad_norm": 14.875,
+      "learning_rate": 2.533428441408546e-07,
+      "loss": 0.7088,
+      "step": 18650
+    },
+    {
+      "epoch": 1.5893481039625055,
+      "grad_norm": 36.0,
+      "learning_rate": 2.5324143541710727e-07,
+      "loss": 0.6284,
+      "step": 18651
+    },
+    {
+      "epoch": 1.5894333191308054,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.531400447057895e-07,
+      "loss": 0.7716,
+      "step": 18652
+    },
+    {
+      "epoch": 1.5895185342991054,
+      "grad_norm": 19.0,
+      "learning_rate": 2.530386720087333e-07,
+      "loss": 0.5941,
+      "step": 18653
+    },
+    {
+      "epoch": 1.5896037494674053,
+      "grad_norm": 16.75,
+      "learning_rate": 2.529373173277706e-07,
+      "loss": 0.3928,
+      "step": 18654
+    },
+    {
+      "epoch": 1.5896889646357053,
+      "grad_norm": 17.25,
+      "learning_rate": 2.528359806647333e-07,
+      "loss": 0.7314,
+      "step": 18655
+    },
+    {
+      "epoch": 1.5897741798040053,
+      "grad_norm": 27.875,
+      "learning_rate": 2.5273466202145295e-07,
+      "loss": 0.7509,
+      "step": 18656
+    },
+    {
+      "epoch": 1.5898593949723052,
+      "grad_norm": 19.875,
+      "learning_rate": 2.5263336139975976e-07,
+      "loss": 0.6531,
+      "step": 18657
+    },
+    {
+      "epoch": 1.5899446101406052,
+      "grad_norm": 17.25,
+      "learning_rate": 2.525320788014847e-07,
+      "loss": 0.6442,
+      "step": 18658
+    },
+    {
+      "epoch": 1.5900298253089051,
+      "grad_norm": 16.5,
+      "learning_rate": 2.5243081422845816e-07,
+      "loss": 0.6333,
+      "step": 18659
+    },
+    {
+      "epoch": 1.590115040477205,
+      "grad_norm": 13.375,
+      "learning_rate": 2.523295676825101e-07,
+      "loss": 0.4523,
+      "step": 18660
+    },
+    {
+      "epoch": 1.590200255645505,
+      "grad_norm": 14.5,
+      "learning_rate": 2.5222833916546967e-07,
+      "loss": 0.8157,
+      "step": 18661
+    },
+    {
+      "epoch": 1.590285470813805,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.521271286791663e-07,
+      "loss": 0.2267,
+      "step": 18662
+    },
+    {
+      "epoch": 1.590370685982105,
+      "grad_norm": 16.375,
+      "learning_rate": 2.5202593622542934e-07,
+      "loss": 0.7226,
+      "step": 18663
+    },
+    {
+      "epoch": 1.590455901150405,
+      "grad_norm": 23.625,
+      "learning_rate": 2.5192476180608744e-07,
+      "loss": 0.8938,
+      "step": 18664
+    },
+    {
+      "epoch": 1.5905411163187049,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.518236054229685e-07,
+      "loss": 0.4867,
+      "step": 18665
+    },
+    {
+      "epoch": 1.5906263314870048,
+      "grad_norm": 20.0,
+      "learning_rate": 2.5172246707790076e-07,
+      "loss": 0.8384,
+      "step": 18666
+    },
+    {
+      "epoch": 1.5907115466553048,
+      "grad_norm": 16.625,
+      "learning_rate": 2.5162134677271214e-07,
+      "loss": 0.7599,
+      "step": 18667
+    },
+    {
+      "epoch": 1.5907967618236047,
+      "grad_norm": 13.0,
+      "learning_rate": 2.515202445092295e-07,
+      "loss": 0.4949,
+      "step": 18668
+    },
+    {
+      "epoch": 1.5908819769919047,
+      "grad_norm": 16.25,
+      "learning_rate": 2.514191602892804e-07,
+      "loss": 0.6829,
+      "step": 18669
+    },
+    {
+      "epoch": 1.5909671921602047,
+      "grad_norm": 11.25,
+      "learning_rate": 2.513180941146909e-07,
+      "loss": 0.3216,
+      "step": 18670
+    },
+    {
+      "epoch": 1.5910524073285046,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.512170459872876e-07,
+      "loss": 0.6503,
+      "step": 18671
+    },
+    {
+      "epoch": 1.5911376224968046,
+      "grad_norm": 13.25,
+      "learning_rate": 2.511160159088967e-07,
+      "loss": 0.3412,
+      "step": 18672
+    },
+    {
+      "epoch": 1.5912228376651045,
+      "grad_norm": 17.875,
+      "learning_rate": 2.5101500388134415e-07,
+      "loss": 0.7667,
+      "step": 18673
+    },
+    {
+      "epoch": 1.5913080528334045,
+      "grad_norm": 20.75,
+      "learning_rate": 2.509140099064547e-07,
+      "loss": 0.9123,
+      "step": 18674
+    },
+    {
+      "epoch": 1.5913932680017044,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.5081303398605387e-07,
+      "loss": 0.5971,
+      "step": 18675
+    },
+    {
+      "epoch": 1.5914784831700044,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.507120761219664e-07,
+      "loss": 0.2272,
+      "step": 18676
+    },
+    {
+      "epoch": 1.5915636983383044,
+      "grad_norm": 14.0,
+      "learning_rate": 2.506111363160167e-07,
+      "loss": 0.4967,
+      "step": 18677
+    },
+    {
+      "epoch": 1.5916489135066043,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.5051021457002847e-07,
+      "loss": 0.5229,
+      "step": 18678
+    },
+    {
+      "epoch": 1.5917341286749043,
+      "grad_norm": 12.875,
+      "learning_rate": 2.5040931088582564e-07,
+      "loss": 0.4295,
+      "step": 18679
+    },
+    {
+      "epoch": 1.5918193438432042,
+      "grad_norm": 14.875,
+      "learning_rate": 2.503084252652317e-07,
+      "loss": 0.57,
+      "step": 18680
+    },
+    {
+      "epoch": 1.5919045590115042,
+      "grad_norm": 19.375,
+      "learning_rate": 2.5020755771007015e-07,
+      "loss": 0.7784,
+      "step": 18681
+    },
+    {
+      "epoch": 1.5919897741798041,
+      "grad_norm": 16.125,
+      "learning_rate": 2.501067082221631e-07,
+      "loss": 0.6915,
+      "step": 18682
+    },
+    {
+      "epoch": 1.592074989348104,
+      "grad_norm": 17.5,
+      "learning_rate": 2.500058768033332e-07,
+      "loss": 0.7322,
+      "step": 18683
+    },
+    {
+      "epoch": 1.592160204516404,
+      "grad_norm": 21.0,
+      "learning_rate": 2.4990506345540316e-07,
+      "loss": 1.0505,
+      "step": 18684
+    },
+    {
+      "epoch": 1.592245419684704,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4980426818019376e-07,
+      "loss": 0.5769,
+      "step": 18685
+    },
+    {
+      "epoch": 1.592330634853004,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4970349097952746e-07,
+      "loss": 0.7815,
+      "step": 18686
+    },
+    {
+      "epoch": 1.592415850021304,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.496027318552247e-07,
+      "loss": 0.3663,
+      "step": 18687
+    },
+    {
+      "epoch": 1.5925010651896039,
+      "grad_norm": 13.0,
+      "learning_rate": 2.495019908091065e-07,
+      "loss": 0.6091,
+      "step": 18688
+    },
+    {
+      "epoch": 1.5925862803579038,
+      "grad_norm": 21.625,
+      "learning_rate": 2.4940126784299343e-07,
+      "loss": 0.7644,
+      "step": 18689
+    },
+    {
+      "epoch": 1.5926714955262038,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4930056295870584e-07,
+      "loss": 0.3038,
+      "step": 18690
+    },
+    {
+      "epoch": 1.5927567106945038,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.4919987615806313e-07,
+      "loss": 0.6516,
+      "step": 18691
+    },
+    {
+      "epoch": 1.5928419258628037,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.4909920744288537e-07,
+      "loss": 0.5797,
+      "step": 18692
+    },
+    {
+      "epoch": 1.5929271410311037,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.489985568149911e-07,
+      "loss": 0.5171,
+      "step": 18693
+    },
+    {
+      "epoch": 1.5930123561994036,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4889792427619966e-07,
+      "loss": 0.8088,
+      "step": 18694
+    },
+    {
+      "epoch": 1.5930975713677036,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4879730982832933e-07,
+      "loss": 0.366,
+      "step": 18695
+    },
+    {
+      "epoch": 1.5931827865360035,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.486967134731982e-07,
+      "loss": 0.5143,
+      "step": 18696
+    },
+    {
+      "epoch": 1.5932680017043035,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4859613521262443e-07,
+      "loss": 0.6527,
+      "step": 18697
+    },
+    {
+      "epoch": 1.5933532168726035,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.484955750484258e-07,
+      "loss": 0.6022,
+      "step": 18698
+    },
+    {
+      "epoch": 1.5934384320409034,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.483950329824189e-07,
+      "loss": 0.7407,
+      "step": 18699
+    },
+    {
+      "epoch": 1.5935236472092034,
+      "grad_norm": 18.5,
+      "learning_rate": 2.482945090164213e-07,
+      "loss": 0.7191,
+      "step": 18700
+    },
+    {
+      "epoch": 1.5936088623775033,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.481940031522488e-07,
+      "loss": 0.2934,
+      "step": 18701
+    },
+    {
+      "epoch": 1.5936940775458033,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4809351539171815e-07,
+      "loss": 0.5601,
+      "step": 18702
+    },
+    {
+      "epoch": 1.5937792927141032,
+      "grad_norm": 16.375,
+      "learning_rate": 2.479930457366454e-07,
+      "loss": 0.4799,
+      "step": 18703
+    },
+    {
+      "epoch": 1.5938645078824032,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4789259418884564e-07,
+      "loss": 0.3071,
+      "step": 18704
+    },
+    {
+      "epoch": 1.5939497230507031,
+      "grad_norm": 24.25,
+      "learning_rate": 2.4779216075013447e-07,
+      "loss": 0.8387,
+      "step": 18705
+    },
+    {
+      "epoch": 1.594034938219003,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.476917454223267e-07,
+      "loss": 0.4521,
+      "step": 18706
+    },
+    {
+      "epoch": 1.594120153387303,
+      "grad_norm": 19.25,
+      "learning_rate": 2.475913482072373e-07,
+      "loss": 0.648,
+      "step": 18707
+    },
+    {
+      "epoch": 1.594205368555603,
+      "grad_norm": 12.75,
+      "learning_rate": 2.474909691066803e-07,
+      "loss": 0.3155,
+      "step": 18708
+    },
+    {
+      "epoch": 1.594290583723903,
+      "grad_norm": 17.625,
+      "learning_rate": 2.473906081224692e-07,
+      "loss": 0.9932,
+      "step": 18709
+    },
+    {
+      "epoch": 1.594375798892203,
+      "grad_norm": 17.125,
+      "learning_rate": 2.47290265256418e-07,
+      "loss": 0.5818,
+      "step": 18710
+    },
+    {
+      "epoch": 1.594461014060503,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.4718994051034034e-07,
+      "loss": 0.7625,
+      "step": 18711
+    },
+    {
+      "epoch": 1.5945462292288028,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4708963388604847e-07,
+      "loss": 0.5002,
+      "step": 18712
+    },
+    {
+      "epoch": 1.5946314443971028,
+      "grad_norm": 16.625,
+      "learning_rate": 2.469893453853556e-07,
+      "loss": 1.0032,
+      "step": 18713
+    },
+    {
+      "epoch": 1.5947166595654028,
+      "grad_norm": 16.625,
+      "learning_rate": 2.4688907501007375e-07,
+      "loss": 0.5558,
+      "step": 18714
+    },
+    {
+      "epoch": 1.5948018747337027,
+      "grad_norm": 19.625,
+      "learning_rate": 2.4678882276201537e-07,
+      "loss": 0.5802,
+      "step": 18715
+    },
+    {
+      "epoch": 1.5948870899020027,
+      "grad_norm": 20.75,
+      "learning_rate": 2.4668858864299166e-07,
+      "loss": 0.7337,
+      "step": 18716
+    },
+    {
+      "epoch": 1.5949723050703026,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4658837265481376e-07,
+      "loss": 0.7707,
+      "step": 18717
+    },
+    {
+      "epoch": 1.5950575202386026,
+      "grad_norm": 16.875,
+      "learning_rate": 2.464881747992931e-07,
+      "loss": 0.573,
+      "step": 18718
+    },
+    {
+      "epoch": 1.5951427354069025,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.463879950782402e-07,
+      "loss": 0.4791,
+      "step": 18719
+    },
+    {
+      "epoch": 1.5952279505752025,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4628783349346566e-07,
+      "loss": 0.8522,
+      "step": 18720
+    },
+    {
+      "epoch": 1.5953131657435025,
+      "grad_norm": 19.0,
+      "learning_rate": 2.4618769004677914e-07,
+      "loss": 0.5529,
+      "step": 18721
+    },
+    {
+      "epoch": 1.5953983809118024,
+      "grad_norm": 17.375,
+      "learning_rate": 2.4608756473999033e-07,
+      "loss": 0.7735,
+      "step": 18722
+    },
+    {
+      "epoch": 1.5954835960801024,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.459874575749091e-07,
+      "loss": 0.4652,
+      "step": 18723
+    },
+    {
+      "epoch": 1.5955688112484023,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4588736855334415e-07,
+      "loss": 0.4393,
+      "step": 18724
+    },
+    {
+      "epoch": 1.5956540264167023,
+      "grad_norm": 12.0,
+      "learning_rate": 2.4578729767710376e-07,
+      "loss": 0.4578,
+      "step": 18725
+    },
+    {
+      "epoch": 1.5957392415850022,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.4568724494799677e-07,
+      "loss": 0.5,
+      "step": 18726
+    },
+    {
+      "epoch": 1.5958244567533022,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.4558721036783127e-07,
+      "loss": 0.3416,
+      "step": 18727
+    },
+    {
+      "epoch": 1.5959096719216022,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.45487193938415e-07,
+      "loss": 0.6808,
+      "step": 18728
+    },
+    {
+      "epoch": 1.5959948870899021,
+      "grad_norm": 24.5,
+      "learning_rate": 2.453871956615551e-07,
+      "loss": 0.9701,
+      "step": 18729
+    },
+    {
+      "epoch": 1.596080102258202,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.452872155390587e-07,
+      "loss": 0.4936,
+      "step": 18730
+    },
+    {
+      "epoch": 1.596165317426502,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.4518725357273274e-07,
+      "loss": 0.3895,
+      "step": 18731
+    },
+    {
+      "epoch": 1.596250532594802,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.450873097643838e-07,
+      "loss": 0.5221,
+      "step": 18732
+    },
+    {
+      "epoch": 1.596335747763102,
+      "grad_norm": 15.0,
+      "learning_rate": 2.4498738411581763e-07,
+      "loss": 0.7583,
+      "step": 18733
+    },
+    {
+      "epoch": 1.596420962931402,
+      "grad_norm": 14.375,
+      "learning_rate": 2.4488747662883977e-07,
+      "loss": 0.5887,
+      "step": 18734
+    },
+    {
+      "epoch": 1.5965061780997019,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.447875873052559e-07,
+      "loss": 0.428,
+      "step": 18735
+    },
+    {
+      "epoch": 1.5965913932680018,
+      "grad_norm": 26.875,
+      "learning_rate": 2.4468771614687125e-07,
+      "loss": 0.7837,
+      "step": 18736
+    },
+    {
+      "epoch": 1.5966766084363018,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.445878631554907e-07,
+      "loss": 0.49,
+      "step": 18737
+    },
+    {
+      "epoch": 1.5967618236046017,
+      "grad_norm": 17.25,
+      "learning_rate": 2.444880283329182e-07,
+      "loss": 1.0393,
+      "step": 18738
+    },
+    {
+      "epoch": 1.5968470387729017,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.4438821168095816e-07,
+      "loss": 0.306,
+      "step": 18739
+    },
+    {
+      "epoch": 1.5969322539412016,
+      "grad_norm": 19.25,
+      "learning_rate": 2.4428841320141463e-07,
+      "loss": 1.0694,
+      "step": 18740
+    },
+    {
+      "epoch": 1.5970174691095016,
+      "grad_norm": 16.75,
+      "learning_rate": 2.4418863289609075e-07,
+      "loss": 0.5005,
+      "step": 18741
+    },
+    {
+      "epoch": 1.5971026842778016,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.440888707667895e-07,
+      "loss": 0.6171,
+      "step": 18742
+    },
+    {
+      "epoch": 1.5971878994461015,
+      "grad_norm": 29.25,
+      "learning_rate": 2.4398912681531385e-07,
+      "loss": 0.5317,
+      "step": 18743
+    },
+    {
+      "epoch": 1.5972731146144015,
+      "grad_norm": 23.5,
+      "learning_rate": 2.438894010434663e-07,
+      "loss": 0.937,
+      "step": 18744
+    },
+    {
+      "epoch": 1.5973583297827014,
+      "grad_norm": 15.125,
+      "learning_rate": 2.4378969345304927e-07,
+      "loss": 0.7253,
+      "step": 18745
+    },
+    {
+      "epoch": 1.5974435449510014,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.4369000404586397e-07,
+      "loss": 0.3986,
+      "step": 18746
+    },
+    {
+      "epoch": 1.5975287601193013,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.435903328237123e-07,
+      "loss": 0.2375,
+      "step": 18747
+    },
+    {
+      "epoch": 1.5976139752876013,
+      "grad_norm": 14.5,
+      "learning_rate": 2.434906797883953e-07,
+      "loss": 0.5873,
+      "step": 18748
+    },
+    {
+      "epoch": 1.5976991904559013,
+      "grad_norm": 16.125,
+      "learning_rate": 2.43391044941714e-07,
+      "loss": 0.8699,
+      "step": 18749
+    },
+    {
+      "epoch": 1.5977844056242012,
+      "grad_norm": 23.875,
+      "learning_rate": 2.43291428285469e-07,
+      "loss": 0.8254,
+      "step": 18750
+    },
+    {
+      "epoch": 1.5978696207925012,
+      "grad_norm": 28.125,
+      "learning_rate": 2.431918298214597e-07,
+      "loss": 0.8934,
+      "step": 18751
+    },
+    {
+      "epoch": 1.5979548359608011,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.4309224955148663e-07,
+      "loss": 0.526,
+      "step": 18752
+    },
+    {
+      "epoch": 1.598040051129101,
+      "grad_norm": 15.25,
+      "learning_rate": 2.42992687477349e-07,
+      "loss": 0.658,
+      "step": 18753
+    },
+    {
+      "epoch": 1.598125266297401,
+      "grad_norm": 19.75,
+      "learning_rate": 2.428931436008465e-07,
+      "loss": 0.7171,
+      "step": 18754
+    },
+    {
+      "epoch": 1.598210481465701,
+      "grad_norm": 16.75,
+      "learning_rate": 2.427936179237773e-07,
+      "loss": 0.6928,
+      "step": 18755
+    },
+    {
+      "epoch": 1.598295696634001,
+      "grad_norm": 17.0,
+      "learning_rate": 2.4269411044794024e-07,
+      "loss": 0.4457,
+      "step": 18756
+    },
+    {
+      "epoch": 1.598380911802301,
+      "grad_norm": 18.0,
+      "learning_rate": 2.4259462117513384e-07,
+      "loss": 0.5815,
+      "step": 18757
+    },
+    {
+      "epoch": 1.5984661269706009,
+      "grad_norm": 20.0,
+      "learning_rate": 2.424951501071557e-07,
+      "loss": 0.7767,
+      "step": 18758
+    },
+    {
+      "epoch": 1.5985513421389008,
+      "grad_norm": 18.0,
+      "learning_rate": 2.4239569724580304e-07,
+      "loss": 0.9372,
+      "step": 18759
+    },
+    {
+      "epoch": 1.5986365573072008,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.422962625928732e-07,
+      "loss": 0.6742,
+      "step": 18760
+    },
+    {
+      "epoch": 1.5987217724755007,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.4219684615016336e-07,
+      "loss": 0.654,
+      "step": 18761
+    },
+    {
+      "epoch": 1.5988069876438007,
+      "grad_norm": 14.875,
+      "learning_rate": 2.4209744791947013e-07,
+      "loss": 0.5966,
+      "step": 18762
+    },
+    {
+      "epoch": 1.5988922028121006,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.4199806790258936e-07,
+      "loss": 0.3408,
+      "step": 18763
+    },
+    {
+      "epoch": 1.5989774179804006,
+      "grad_norm": 17.125,
+      "learning_rate": 2.4189870610131714e-07,
+      "loss": 0.7999,
+      "step": 18764
+    },
+    {
+      "epoch": 1.5990626331487006,
+      "grad_norm": 13.125,
+      "learning_rate": 2.4179936251744917e-07,
+      "loss": 0.5337,
+      "step": 18765
+    },
+    {
+      "epoch": 1.5991478483170005,
+      "grad_norm": 14.375,
+      "learning_rate": 2.417000371527804e-07,
+      "loss": 0.7124,
+      "step": 18766
+    },
+    {
+      "epoch": 1.5992330634853005,
+      "grad_norm": 11.875,
+      "learning_rate": 2.4160073000910613e-07,
+      "loss": 0.3955,
+      "step": 18767
+    },
+    {
+      "epoch": 1.5993182786536004,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4150144108822036e-07,
+      "loss": 0.4551,
+      "step": 18768
+    },
+    {
+      "epoch": 1.5994034938219004,
+      "grad_norm": 21.5,
+      "learning_rate": 2.4140217039191766e-07,
+      "loss": 0.5333,
+      "step": 18769
+    },
+    {
+      "epoch": 1.5994887089902003,
+      "grad_norm": 11.625,
+      "learning_rate": 2.4130291792199236e-07,
+      "loss": 0.2712,
+      "step": 18770
+    },
+    {
+      "epoch": 1.5995739241585003,
+      "grad_norm": 22.875,
+      "learning_rate": 2.4120368368023727e-07,
+      "loss": 0.9137,
+      "step": 18771
+    },
+    {
+      "epoch": 1.5996591393268003,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.4110446766844596e-07,
+      "loss": 0.3845,
+      "step": 18772
+    },
+    {
+      "epoch": 1.5997443544951002,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.410052698884119e-07,
+      "loss": 0.4656,
+      "step": 18773
+    },
+    {
+      "epoch": 1.5998295696634002,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.4090609034192693e-07,
+      "loss": 0.5611,
+      "step": 18774
+    },
+    {
+      "epoch": 1.5999147848317001,
+      "grad_norm": 18.5,
+      "learning_rate": 2.408069290307838e-07,
+      "loss": 0.6448,
+      "step": 18775
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 21.125,
+      "learning_rate": 2.407077859567741e-07,
+      "loss": 0.5192,
+      "step": 18776
+    },
+    {
+      "epoch": 1.6000852151683,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4060866112168957e-07,
+      "loss": 0.8147,
+      "step": 18777
+    },
+    {
+      "epoch": 1.6001704303366,
+      "grad_norm": 20.0,
+      "learning_rate": 2.405095545273216e-07,
+      "loss": 0.6353,
+      "step": 18778
+    },
+    {
+      "epoch": 1.6002556455049,
+      "grad_norm": 24.375,
+      "learning_rate": 2.4041046617546136e-07,
+      "loss": 0.7615,
+      "step": 18779
+    },
+    {
+      "epoch": 1.6003408606732,
+      "grad_norm": 14.5,
+      "learning_rate": 2.403113960678989e-07,
+      "loss": 0.4928,
+      "step": 18780
+    },
+    {
+      "epoch": 1.6004260758414999,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.4021234420642525e-07,
+      "loss": 0.1688,
+      "step": 18781
+    },
+    {
+      "epoch": 1.6005112910097998,
+      "grad_norm": 11.0,
+      "learning_rate": 2.401133105928295e-07,
+      "loss": 0.4235,
+      "step": 18782
+    },
+    {
+      "epoch": 1.6005965061780998,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.400142952289018e-07,
+      "loss": 0.5192,
+      "step": 18783
+    },
+    {
+      "epoch": 1.6006817213463997,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.399152981164318e-07,
+      "loss": 0.5804,
+      "step": 18784
+    },
+    {
+      "epoch": 1.6007669365146997,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3981631925720767e-07,
+      "loss": 1.0167,
+      "step": 18785
+    },
+    {
+      "epoch": 1.6008521516829997,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3971735865301855e-07,
+      "loss": 0.594,
+      "step": 18786
+    },
+    {
+      "epoch": 1.6009373668512996,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.3961841630565295e-07,
+      "loss": 0.4664,
+      "step": 18787
+    },
+    {
+      "epoch": 1.6010225820195996,
+      "grad_norm": 15.5,
+      "learning_rate": 2.395194922168982e-07,
+      "loss": 0.898,
+      "step": 18788
+    },
+    {
+      "epoch": 1.6011077971878995,
+      "grad_norm": 30.25,
+      "learning_rate": 2.394205863885428e-07,
+      "loss": 0.8917,
+      "step": 18789
+    },
+    {
+      "epoch": 1.6011930123561995,
+      "grad_norm": 13.125,
+      "learning_rate": 2.3932169882237314e-07,
+      "loss": 0.5951,
+      "step": 18790
+    },
+    {
+      "epoch": 1.6012782275244994,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.392228295201769e-07,
+      "loss": 0.7615,
+      "step": 18791
+    },
+    {
+      "epoch": 1.6013634426927994,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.391239784837407e-07,
+      "loss": 0.4873,
+      "step": 18792
+    },
+    {
+      "epoch": 1.6014486578610994,
+      "grad_norm": 17.75,
+      "learning_rate": 2.390251457148504e-07,
+      "loss": 0.7804,
+      "step": 18793
+    },
+    {
+      "epoch": 1.6015338730293993,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3892633121529244e-07,
+      "loss": 0.3813,
+      "step": 18794
+    },
+    {
+      "epoch": 1.6016190881976993,
+      "grad_norm": 22.125,
+      "learning_rate": 2.388275349868524e-07,
+      "loss": 0.7056,
+      "step": 18795
+    },
+    {
+      "epoch": 1.6017043033659992,
+      "grad_norm": 17.25,
+      "learning_rate": 2.387287570313158e-07,
+      "loss": 0.9108,
+      "step": 18796
+    },
+    {
+      "epoch": 1.6017895185342992,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3862999735046756e-07,
+      "loss": 0.6095,
+      "step": 18797
+    },
+    {
+      "epoch": 1.6018747337025991,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.38531255946092e-07,
+      "loss": 0.5042,
+      "step": 18798
+    },
+    {
+      "epoch": 1.601959948870899,
+      "grad_norm": 19.125,
+      "learning_rate": 2.3843253281997366e-07,
+      "loss": 0.6324,
+      "step": 18799
+    },
+    {
+      "epoch": 1.602045164039199,
+      "grad_norm": 15.25,
+      "learning_rate": 2.38333827973897e-07,
+      "loss": 0.6849,
+      "step": 18800
+    },
+    {
+      "epoch": 1.602130379207499,
+      "grad_norm": 19.5,
+      "learning_rate": 2.382351414096451e-07,
+      "loss": 0.985,
+      "step": 18801
+    },
+    {
+      "epoch": 1.602215594375799,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.381364731290016e-07,
+      "loss": 0.2165,
+      "step": 18802
+    },
+    {
+      "epoch": 1.602300809544099,
+      "grad_norm": 14.5,
+      "learning_rate": 2.380378231337495e-07,
+      "loss": 0.7024,
+      "step": 18803
+    },
+    {
+      "epoch": 1.6023860247123989,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3793919142567178e-07,
+      "loss": 0.7469,
+      "step": 18804
+    },
+    {
+      "epoch": 1.6024712398806988,
+      "grad_norm": 14.875,
+      "learning_rate": 2.3784057800655068e-07,
+      "loss": 0.6323,
+      "step": 18805
+    },
+    {
+      "epoch": 1.6025564550489988,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3774198287816781e-07,
+      "loss": 0.6103,
+      "step": 18806
+    },
+    {
+      "epoch": 1.6026416702172988,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3764340604230524e-07,
+      "loss": 0.6129,
+      "step": 18807
+    },
+    {
+      "epoch": 1.6027268853855987,
+      "grad_norm": 19.5,
+      "learning_rate": 2.3754484750074428e-07,
+      "loss": 0.6792,
+      "step": 18808
+    },
+    {
+      "epoch": 1.6028121005538987,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3744630725526642e-07,
+      "loss": 0.8101,
+      "step": 18809
+    },
+    {
+      "epoch": 1.6028973157221986,
+      "grad_norm": 13.625,
+      "learning_rate": 2.373477853076517e-07,
+      "loss": 0.5101,
+      "step": 18810
+    },
+    {
+      "epoch": 1.6029825308904986,
+      "grad_norm": 11.5,
+      "learning_rate": 2.3724928165968066e-07,
+      "loss": 0.3176,
+      "step": 18811
+    },
+    {
+      "epoch": 1.6030677460587985,
+      "grad_norm": 14.625,
+      "learning_rate": 2.371507963131335e-07,
+      "loss": 0.4864,
+      "step": 18812
+    },
+    {
+      "epoch": 1.6031529612270985,
+      "grad_norm": 21.625,
+      "learning_rate": 2.3705232926979045e-07,
+      "loss": 1.0411,
+      "step": 18813
+    },
+    {
+      "epoch": 1.6032381763953985,
+      "grad_norm": 32.25,
+      "learning_rate": 2.3695388053143017e-07,
+      "loss": 0.9117,
+      "step": 18814
+    },
+    {
+      "epoch": 1.6033233915636984,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.3685545009983186e-07,
+      "loss": 0.6072,
+      "step": 18815
+    },
+    {
+      "epoch": 1.6034086067319984,
+      "grad_norm": 19.75,
+      "learning_rate": 2.3675703797677418e-07,
+      "loss": 0.7835,
+      "step": 18816
+    },
+    {
+      "epoch": 1.6034938219002983,
+      "grad_norm": 30.125,
+      "learning_rate": 2.3665864416403617e-07,
+      "loss": 0.8277,
+      "step": 18817
+    },
+    {
+      "epoch": 1.6035790370685983,
+      "grad_norm": 15.75,
+      "learning_rate": 2.3656026866339513e-07,
+      "loss": 0.9781,
+      "step": 18818
+    },
+    {
+      "epoch": 1.6036642522368982,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3646191147662908e-07,
+      "loss": 0.5267,
+      "step": 18819
+    },
+    {
+      "epoch": 1.6037494674051982,
+      "grad_norm": 16.125,
+      "learning_rate": 2.3636357260551553e-07,
+      "loss": 0.5873,
+      "step": 18820
+    },
+    {
+      "epoch": 1.6038346825734981,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3626525205183174e-07,
+      "loss": 0.4482,
+      "step": 18821
+    },
+    {
+      "epoch": 1.603919897741798,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3616694981735424e-07,
+      "loss": 0.4091,
+      "step": 18822
+    },
+    {
+      "epoch": 1.604005112910098,
+      "grad_norm": 18.0,
+      "learning_rate": 2.3606866590385914e-07,
+      "loss": 0.9149,
+      "step": 18823
+    },
+    {
+      "epoch": 1.604090328078398,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3597040031312286e-07,
+      "loss": 0.5366,
+      "step": 18824
+    },
+    {
+      "epoch": 1.604175543246698,
+      "grad_norm": 13.375,
+      "learning_rate": 2.3587215304692116e-07,
+      "loss": 0.5246,
+      "step": 18825
+    },
+    {
+      "epoch": 1.604260758414998,
+      "grad_norm": 19.75,
+      "learning_rate": 2.3577392410702975e-07,
+      "loss": 0.6386,
+      "step": 18826
+    },
+    {
+      "epoch": 1.604345973583298,
+      "grad_norm": 30.75,
+      "learning_rate": 2.3567571349522312e-07,
+      "loss": 0.3958,
+      "step": 18827
+    },
+    {
+      "epoch": 1.6044311887515978,
+      "grad_norm": 10.625,
+      "learning_rate": 2.355775212132763e-07,
+      "loss": 0.8644,
+      "step": 18828
+    },
+    {
+      "epoch": 1.6045164039198978,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3547934726296366e-07,
+      "loss": 0.5665,
+      "step": 18829
+    },
+    {
+      "epoch": 1.6046016190881978,
+      "grad_norm": 13.25,
+      "learning_rate": 2.353811916460602e-07,
+      "loss": 0.4578,
+      "step": 18830
+    },
+    {
+      "epoch": 1.6046868342564977,
+      "grad_norm": 14.625,
+      "learning_rate": 2.3528305436433808e-07,
+      "loss": 0.4603,
+      "step": 18831
+    },
+    {
+      "epoch": 1.6047720494247977,
+      "grad_norm": 14.5,
+      "learning_rate": 2.3518493541957163e-07,
+      "loss": 0.6296,
+      "step": 18832
+    },
+    {
+      "epoch": 1.6048572645930976,
+      "grad_norm": 16.25,
+      "learning_rate": 2.3508683481353393e-07,
+      "loss": 0.5881,
+      "step": 18833
+    },
+    {
+      "epoch": 1.6049424797613976,
+      "grad_norm": 23.125,
+      "learning_rate": 2.3498875254799796e-07,
+      "loss": 0.5631,
+      "step": 18834
+    },
+    {
+      "epoch": 1.6050276949296975,
+      "grad_norm": 17.375,
+      "learning_rate": 2.3489068862473554e-07,
+      "loss": 0.8261,
+      "step": 18835
+    },
+    {
+      "epoch": 1.6051129100979975,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3479264304551913e-07,
+      "loss": 0.6818,
+      "step": 18836
+    },
+    {
+      "epoch": 1.6051981252662975,
+      "grad_norm": 19.375,
+      "learning_rate": 2.3469461581212064e-07,
+      "loss": 0.7859,
+      "step": 18837
+    },
+    {
+      "epoch": 1.6052833404345974,
+      "grad_norm": 14.0,
+      "learning_rate": 2.345966069263117e-07,
+      "loss": 0.5161,
+      "step": 18838
+    },
+    {
+      "epoch": 1.6053685556028974,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3449861638986314e-07,
+      "loss": 0.6446,
+      "step": 18839
+    },
+    {
+      "epoch": 1.6054537707711973,
+      "grad_norm": 20.75,
+      "learning_rate": 2.3440064420454545e-07,
+      "loss": 0.5604,
+      "step": 18840
+    },
+    {
+      "epoch": 1.6055389859394973,
+      "grad_norm": 13.5,
+      "learning_rate": 2.3430269037212933e-07,
+      "loss": 0.6236,
+      "step": 18841
+    },
+    {
+      "epoch": 1.6056242011077972,
+      "grad_norm": 16.875,
+      "learning_rate": 2.3420475489438487e-07,
+      "loss": 0.5447,
+      "step": 18842
+    },
+    {
+      "epoch": 1.6057094162760972,
+      "grad_norm": 15.375,
+      "learning_rate": 2.3410683777308235e-07,
+      "loss": 0.5969,
+      "step": 18843
+    },
+    {
+      "epoch": 1.6057946314443972,
+      "grad_norm": 13.625,
+      "learning_rate": 2.340089390099906e-07,
+      "loss": 0.4832,
+      "step": 18844
+    },
+    {
+      "epoch": 1.6058798466126971,
+      "grad_norm": 16.625,
+      "learning_rate": 2.3391105860687894e-07,
+      "loss": 0.6368,
+      "step": 18845
+    },
+    {
+      "epoch": 1.605965061780997,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3381319656551647e-07,
+      "loss": 0.8314,
+      "step": 18846
+    },
+    {
+      "epoch": 1.606050276949297,
+      "grad_norm": 15.25,
+      "learning_rate": 2.3371535288767137e-07,
+      "loss": 0.5549,
+      "step": 18847
+    },
+    {
+      "epoch": 1.606135492117597,
+      "grad_norm": 23.25,
+      "learning_rate": 2.3361752757511152e-07,
+      "loss": 0.6812,
+      "step": 18848
+    },
+    {
+      "epoch": 1.606220707285897,
+      "grad_norm": 15.0,
+      "learning_rate": 2.33519720629605e-07,
+      "loss": 0.6904,
+      "step": 18849
+    },
+    {
+      "epoch": 1.606305922454197,
+      "grad_norm": 11.625,
+      "learning_rate": 2.3342193205291936e-07,
+      "loss": 0.2608,
+      "step": 18850
+    },
+    {
+      "epoch": 1.6063911376224969,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3332416184682183e-07,
+      "loss": 0.6077,
+      "step": 18851
+    },
+    {
+      "epoch": 1.6064763527907968,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3322641001307878e-07,
+      "loss": 0.5989,
+      "step": 18852
+    },
+    {
+      "epoch": 1.6065615679590968,
+      "grad_norm": 16.5,
+      "learning_rate": 2.3312867655345698e-07,
+      "loss": 0.635,
+      "step": 18853
+    },
+    {
+      "epoch": 1.6066467831273967,
+      "grad_norm": 24.25,
+      "learning_rate": 2.330309614697228e-07,
+      "loss": 0.6872,
+      "step": 18854
+    },
+    {
+      "epoch": 1.6067319982956967,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.3293326476364162e-07,
+      "loss": 0.2417,
+      "step": 18855
+    },
+    {
+      "epoch": 1.6068172134639966,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3283558643697926e-07,
+      "loss": 0.6158,
+      "step": 18856
+    },
+    {
+      "epoch": 1.6069024286322966,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.327379264915003e-07,
+      "loss": 0.2301,
+      "step": 18857
+    },
+    {
+      "epoch": 1.6069876438005966,
+      "grad_norm": 18.625,
+      "learning_rate": 2.326402849289701e-07,
+      "loss": 0.5618,
+      "step": 18858
+    },
+    {
+      "epoch": 1.6070728589688965,
+      "grad_norm": 21.0,
+      "learning_rate": 2.3254266175115295e-07,
+      "loss": 0.5473,
+      "step": 18859
+    },
+    {
+      "epoch": 1.6071580741371965,
+      "grad_norm": 17.625,
+      "learning_rate": 2.324450569598133e-07,
+      "loss": 0.4874,
+      "step": 18860
+    },
+    {
+      "epoch": 1.6072432893054964,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.323474705567144e-07,
+      "loss": 0.385,
+      "step": 18861
+    },
+    {
+      "epoch": 1.6073285044737964,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3224990254362034e-07,
+      "loss": 0.8734,
+      "step": 18862
+    },
+    {
+      "epoch": 1.6074137196420963,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3215235292229368e-07,
+      "loss": 0.4393,
+      "step": 18863
+    },
+    {
+      "epoch": 1.6074989348103963,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3205482169449789e-07,
+      "loss": 0.3763,
+      "step": 18864
+    },
+    {
+      "epoch": 1.6075841499786963,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3195730886199475e-07,
+      "loss": 0.6207,
+      "step": 18865
+    },
+    {
+      "epoch": 1.6076693651469962,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3185981442654673e-07,
+      "loss": 0.6964,
+      "step": 18866
+    },
+    {
+      "epoch": 1.6077545803152962,
+      "grad_norm": 20.125,
+      "learning_rate": 2.3176233838991562e-07,
+      "loss": 0.4676,
+      "step": 18867
+    },
+    {
+      "epoch": 1.6078397954835961,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.3166488075386336e-07,
+      "loss": 0.3362,
+      "step": 18868
+    },
+    {
+      "epoch": 1.607925010651896,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3156744152015048e-07,
+      "loss": 0.6524,
+      "step": 18869
+    },
+    {
+      "epoch": 1.608010225820196,
+      "grad_norm": 17.125,
+      "learning_rate": 2.3147002069053833e-07,
+      "loss": 0.5693,
+      "step": 18870
+    },
+    {
+      "epoch": 1.608095440988496,
+      "grad_norm": 18.625,
+      "learning_rate": 2.3137261826678677e-07,
+      "loss": 0.4716,
+      "step": 18871
+    },
+    {
+      "epoch": 1.608180656156796,
+      "grad_norm": 18.0,
+      "learning_rate": 2.312752342506562e-07,
+      "loss": 0.7448,
+      "step": 18872
+    },
+    {
+      "epoch": 1.608265871325096,
+      "grad_norm": 18.875,
+      "learning_rate": 2.3117786864390701e-07,
+      "loss": 0.8164,
+      "step": 18873
+    },
+    {
+      "epoch": 1.6083510864933959,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.3108052144829792e-07,
+      "loss": 0.3929,
+      "step": 18874
+    },
+    {
+      "epoch": 1.6084363016616958,
+      "grad_norm": 20.0,
+      "learning_rate": 2.3098319266558838e-07,
+      "loss": 0.7361,
+      "step": 18875
+    },
+    {
+      "epoch": 1.6085215168299958,
+      "grad_norm": 22.75,
+      "learning_rate": 2.3088588229753712e-07,
+      "loss": 0.7829,
+      "step": 18876
+    },
+    {
+      "epoch": 1.6086067319982957,
+      "grad_norm": 35.25,
+      "learning_rate": 2.3078859034590325e-07,
+      "loss": 0.7101,
+      "step": 18877
+    },
+    {
+      "epoch": 1.6086919471665957,
+      "grad_norm": 21.5,
+      "learning_rate": 2.3069131681244403e-07,
+      "loss": 0.5062,
+      "step": 18878
+    },
+    {
+      "epoch": 1.6087771623348956,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.3059406169891816e-07,
+      "loss": 0.393,
+      "step": 18879
+    },
+    {
+      "epoch": 1.6088623775031956,
+      "grad_norm": 20.625,
+      "learning_rate": 2.3049682500708228e-07,
+      "loss": 0.5155,
+      "step": 18880
+    },
+    {
+      "epoch": 1.6089475926714956,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3039960673869433e-07,
+      "loss": 0.6625,
+      "step": 18881
+    },
+    {
+      "epoch": 1.6090328078397955,
+      "grad_norm": 26.875,
+      "learning_rate": 2.3030240689551037e-07,
+      "loss": 0.6586,
+      "step": 18882
+    },
+    {
+      "epoch": 1.6091180230080955,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.3020522547928733e-07,
+      "loss": 0.4726,
+      "step": 18883
+    },
+    {
+      "epoch": 1.6092032381763954,
+      "grad_norm": 19.125,
+      "learning_rate": 2.3010806249178132e-07,
+      "loss": 0.983,
+      "step": 18884
+    },
+    {
+      "epoch": 1.6092884533446954,
+      "grad_norm": 13.25,
+      "learning_rate": 2.3001091793474858e-07,
+      "loss": 0.5155,
+      "step": 18885
+    },
+    {
+      "epoch": 1.6093736685129953,
+      "grad_norm": 14.75,
+      "learning_rate": 2.2991379180994395e-07,
+      "loss": 0.4991,
+      "step": 18886
+    },
+    {
+      "epoch": 1.6094588836812953,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2981668411912308e-07,
+      "loss": 0.5437,
+      "step": 18887
+    },
+    {
+      "epoch": 1.6095440988495953,
+      "grad_norm": 17.75,
+      "learning_rate": 2.2971959486404045e-07,
+      "loss": 0.5197,
+      "step": 18888
+    },
+    {
+      "epoch": 1.6096293140178952,
+      "grad_norm": 18.5,
+      "learning_rate": 2.2962252404645058e-07,
+      "loss": 0.9474,
+      "step": 18889
+    },
+    {
+      "epoch": 1.6097145291861952,
+      "grad_norm": 16.375,
+      "learning_rate": 2.2952547166810806e-07,
+      "loss": 0.4562,
+      "step": 18890
+    },
+    {
+      "epoch": 1.6097997443544951,
+      "grad_norm": 15.75,
+      "learning_rate": 2.2942843773076622e-07,
+      "loss": 0.657,
+      "step": 18891
+    },
+    {
+      "epoch": 1.609884959522795,
+      "grad_norm": 13.375,
+      "learning_rate": 2.2933142223617865e-07,
+      "loss": 0.3246,
+      "step": 18892
+    },
+    {
+      "epoch": 1.609970174691095,
+      "grad_norm": 16.375,
+      "learning_rate": 2.292344251860991e-07,
+      "loss": 0.5714,
+      "step": 18893
+    },
+    {
+      "epoch": 1.610055389859395,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.291374465822796e-07,
+      "loss": 0.736,
+      "step": 18894
+    },
+    {
+      "epoch": 1.610140605027695,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2904048642647323e-07,
+      "loss": 0.5963,
+      "step": 18895
+    },
+    {
+      "epoch": 1.610225820195995,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2894354472043162e-07,
+      "loss": 0.5651,
+      "step": 18896
+    },
+    {
+      "epoch": 1.6103110353642949,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.28846621465907e-07,
+      "loss": 0.5897,
+      "step": 18897
+    },
+    {
+      "epoch": 1.6103962505325948,
+      "grad_norm": 14.25,
+      "learning_rate": 2.28749716664651e-07,
+      "loss": 0.5992,
+      "step": 18898
+    },
+    {
+      "epoch": 1.6104814657008948,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2865283031841434e-07,
+      "loss": 0.6712,
+      "step": 18899
+    },
+    {
+      "epoch": 1.6105666808691947,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.2855596242894796e-07,
+      "loss": 0.494,
+      "step": 18900
+    },
+    {
+      "epoch": 1.6106518960374947,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2845911299800252e-07,
+      "loss": 0.4914,
+      "step": 18901
+    },
+    {
+      "epoch": 1.6107371112057947,
+      "grad_norm": 17.0,
+      "learning_rate": 2.283622820273286e-07,
+      "loss": 0.4247,
+      "step": 18902
+    },
+    {
+      "epoch": 1.6108223263740946,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2826546951867548e-07,
+      "loss": 0.4652,
+      "step": 18903
+    },
+    {
+      "epoch": 1.6109075415423946,
+      "grad_norm": 16.5,
+      "learning_rate": 2.2816867547379232e-07,
+      "loss": 0.7342,
+      "step": 18904
+    },
+    {
+      "epoch": 1.6109927567106945,
+      "grad_norm": 14.0,
+      "learning_rate": 2.2807189989442895e-07,
+      "loss": 0.4671,
+      "step": 18905
+    },
+    {
+      "epoch": 1.6110779718789945,
+      "grad_norm": 20.0,
+      "learning_rate": 2.2797514278233388e-07,
+      "loss": 0.8333,
+      "step": 18906
+    },
+    {
+      "epoch": 1.6111631870472944,
+      "grad_norm": 18.0,
+      "learning_rate": 2.2787840413925593e-07,
+      "loss": 0.7056,
+      "step": 18907
+    },
+    {
+      "epoch": 1.6112484022155944,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.2778168396694278e-07,
+      "loss": 0.367,
+      "step": 18908
+    },
+    {
+      "epoch": 1.6113336173838944,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2768498226714258e-07,
+      "loss": 0.3515,
+      "step": 18909
+    },
+    {
+      "epoch": 1.6114188325521943,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.27588299041603e-07,
+      "loss": 0.3745,
+      "step": 18910
+    },
+    {
+      "epoch": 1.6115040477204943,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2749163429207105e-07,
+      "loss": 0.5195,
+      "step": 18911
+    },
+    {
+      "epoch": 1.6115892628887942,
+      "grad_norm": 18.25,
+      "learning_rate": 2.2739498802029318e-07,
+      "loss": 0.516,
+      "step": 18912
+    },
+    {
+      "epoch": 1.6116744780570942,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.2729836022801612e-07,
+      "loss": 0.4664,
+      "step": 18913
+    },
+    {
+      "epoch": 1.6117596932253941,
+      "grad_norm": 10.875,
+      "learning_rate": 2.2720175091698603e-07,
+      "loss": 0.3204,
+      "step": 18914
+    },
+    {
+      "epoch": 1.611844908393694,
+      "grad_norm": 17.625,
+      "learning_rate": 2.2710516008894928e-07,
+      "loss": 0.6475,
+      "step": 18915
+    },
+    {
+      "epoch": 1.611930123561994,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2700858774565044e-07,
+      "loss": 0.7274,
+      "step": 18916
+    },
+    {
+      "epoch": 1.612015338730294,
+      "grad_norm": 15.75,
+      "learning_rate": 2.269120338888352e-07,
+      "loss": 0.7989,
+      "step": 18917
+    },
+    {
+      "epoch": 1.612100553898594,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2681549852024823e-07,
+      "loss": 0.4669,
+      "step": 18918
+    },
+    {
+      "epoch": 1.612185769066894,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.267189816416346e-07,
+      "loss": 0.3724,
+      "step": 18919
+    },
+    {
+      "epoch": 1.6122709842351939,
+      "grad_norm": 17.75,
+      "learning_rate": 2.266224832547377e-07,
+      "loss": 0.7062,
+      "step": 18920
+    },
+    {
+      "epoch": 1.6123561994034938,
+      "grad_norm": 11.6875,
+      "learning_rate": 2.265260033613015e-07,
+      "loss": 0.2924,
+      "step": 18921
+    },
+    {
+      "epoch": 1.6124414145717938,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2642954196306953e-07,
+      "loss": 0.8064,
+      "step": 18922
+    },
+    {
+      "epoch": 1.6125266297400938,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.2633309906178526e-07,
+      "loss": 0.4289,
+      "step": 18923
+    },
+    {
+      "epoch": 1.6126118449083937,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2623667465919098e-07,
+      "loss": 0.4171,
+      "step": 18924
+    },
+    {
+      "epoch": 1.6126970600766937,
+      "grad_norm": 9.875,
+      "learning_rate": 2.261402687570295e-07,
+      "loss": 0.2209,
+      "step": 18925
+    },
+    {
+      "epoch": 1.6127822752449936,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2604388135704286e-07,
+      "loss": 0.4362,
+      "step": 18926
+    },
+    {
+      "epoch": 1.6128674904132936,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.2594751246097334e-07,
+      "loss": 0.46,
+      "step": 18927
+    },
+    {
+      "epoch": 1.6129527055815935,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.258511620705621e-07,
+      "loss": 0.5378,
+      "step": 18928
+    },
+    {
+      "epoch": 1.6130379207498935,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.257548301875498e-07,
+      "loss": 0.3998,
+      "step": 18929
+    },
+    {
+      "epoch": 1.6131231359181935,
+      "grad_norm": 19.875,
+      "learning_rate": 2.2565851681367768e-07,
+      "loss": 0.8357,
+      "step": 18930
+    },
+    {
+      "epoch": 1.6132083510864934,
+      "grad_norm": 13.5,
+      "learning_rate": 2.255622219506863e-07,
+      "loss": 0.3133,
+      "step": 18931
+    },
+    {
+      "epoch": 1.6132935662547934,
+      "grad_norm": 12.0,
+      "learning_rate": 2.2546594560031597e-07,
+      "loss": 0.3148,
+      "step": 18932
+    },
+    {
+      "epoch": 1.6133787814230933,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2536968776430593e-07,
+      "loss": 0.7453,
+      "step": 18933
+    },
+    {
+      "epoch": 1.6134639965913933,
+      "grad_norm": 23.125,
+      "learning_rate": 2.252734484443961e-07,
+      "loss": 0.8936,
+      "step": 18934
+    },
+    {
+      "epoch": 1.6135492117596932,
+      "grad_norm": 15.25,
+      "learning_rate": 2.2517722764232558e-07,
+      "loss": 0.5723,
+      "step": 18935
+    },
+    {
+      "epoch": 1.6136344269279932,
+      "grad_norm": 14.125,
+      "learning_rate": 2.250810253598329e-07,
+      "loss": 0.6366,
+      "step": 18936
+    },
+    {
+      "epoch": 1.6137196420962931,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2498484159865706e-07,
+      "loss": 0.6842,
+      "step": 18937
+    },
+    {
+      "epoch": 1.613804857264593,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2488867636053554e-07,
+      "loss": 0.409,
+      "step": 18938
+    },
+    {
+      "epoch": 1.613890072432893,
+      "grad_norm": 23.5,
+      "learning_rate": 2.2479252964720658e-07,
+      "loss": 1.0129,
+      "step": 18939
+    },
+    {
+      "epoch": 1.613975287601193,
+      "grad_norm": 11.5,
+      "learning_rate": 2.2469640146040763e-07,
+      "loss": 0.3593,
+      "step": 18940
+    },
+    {
+      "epoch": 1.614060502769493,
+      "grad_norm": 14.25,
+      "learning_rate": 2.2460029180187554e-07,
+      "loss": 0.6708,
+      "step": 18941
+    },
+    {
+      "epoch": 1.614145717937793,
+      "grad_norm": 13.375,
+      "learning_rate": 2.2450420067334735e-07,
+      "loss": 0.4351,
+      "step": 18942
+    },
+    {
+      "epoch": 1.614230933106093,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.2440812807655962e-07,
+      "loss": 0.6171,
+      "step": 18943
+    },
+    {
+      "epoch": 1.6143161482743928,
+      "grad_norm": 18.875,
+      "learning_rate": 2.2431207401324816e-07,
+      "loss": 0.7788,
+      "step": 18944
+    },
+    {
+      "epoch": 1.6144013634426928,
+      "grad_norm": 20.25,
+      "learning_rate": 2.2421603848514924e-07,
+      "loss": 0.4486,
+      "step": 18945
+    },
+    {
+      "epoch": 1.6144865786109928,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2412002149399758e-07,
+      "loss": 0.7137,
+      "step": 18946
+    },
+    {
+      "epoch": 1.6145717937792927,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.2402402304152886e-07,
+      "loss": 0.2267,
+      "step": 18947
+    },
+    {
+      "epoch": 1.6146570089475927,
+      "grad_norm": 19.75,
+      "learning_rate": 2.2392804312947768e-07,
+      "loss": 0.9764,
+      "step": 18948
+    },
+    {
+      "epoch": 1.6147422241158926,
+      "grad_norm": 13.875,
+      "learning_rate": 2.238320817595789e-07,
+      "loss": 0.4837,
+      "step": 18949
+    },
+    {
+      "epoch": 1.6148274392841926,
+      "grad_norm": 22.75,
+      "learning_rate": 2.2373613893356584e-07,
+      "loss": 1.129,
+      "step": 18950
+    },
+    {
+      "epoch": 1.6149126544524925,
+      "grad_norm": 18.0,
+      "learning_rate": 2.2364021465317285e-07,
+      "loss": 0.8148,
+      "step": 18951
+    },
+    {
+      "epoch": 1.6149978696207925,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.2354430892013362e-07,
+      "loss": 0.7075,
+      "step": 18952
+    },
+    {
+      "epoch": 1.6150830847890925,
+      "grad_norm": 16.125,
+      "learning_rate": 2.2344842173618084e-07,
+      "loss": 0.6647,
+      "step": 18953
+    },
+    {
+      "epoch": 1.6151682999573924,
+      "grad_norm": 18.625,
+      "learning_rate": 2.2335255310304714e-07,
+      "loss": 0.9228,
+      "step": 18954
+    },
+    {
+      "epoch": 1.6152535151256924,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.2325670302246512e-07,
+      "loss": 0.2184,
+      "step": 18955
+    },
+    {
+      "epoch": 1.6153387302939923,
+      "grad_norm": 15.0,
+      "learning_rate": 2.2316087149616694e-07,
+      "loss": 0.5058,
+      "step": 18956
+    },
+    {
+      "epoch": 1.6154239454622923,
+      "grad_norm": 11.75,
+      "learning_rate": 2.2306505852588463e-07,
+      "loss": 0.3358,
+      "step": 18957
+    },
+    {
+      "epoch": 1.6155091606305922,
+      "grad_norm": 20.625,
+      "learning_rate": 2.229692641133492e-07,
+      "loss": 0.6196,
+      "step": 18958
+    },
+    {
+      "epoch": 1.6155943757988922,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2287348826029175e-07,
+      "loss": 0.3448,
+      "step": 18959
+    },
+    {
+      "epoch": 1.6156795909671922,
+      "grad_norm": 10.25,
+      "learning_rate": 2.227777309684437e-07,
+      "loss": 0.4847,
+      "step": 18960
+    },
+    {
+      "epoch": 1.6157648061354921,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2268199223953457e-07,
+      "loss": 0.6079,
+      "step": 18961
+    },
+    {
+      "epoch": 1.615850021303792,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.2258627207529526e-07,
+      "loss": 0.4541,
+      "step": 18962
+    },
+    {
+      "epoch": 1.615935236472092,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.2249057047745493e-07,
+      "loss": 0.3674,
+      "step": 18963
+    },
+    {
+      "epoch": 1.616020451640392,
+      "grad_norm": 32.5,
+      "learning_rate": 2.2239488744774314e-07,
+      "loss": 1.1958,
+      "step": 18964
+    },
+    {
+      "epoch": 1.616105666808692,
+      "grad_norm": 14.125,
+      "learning_rate": 2.2229922298788898e-07,
+      "loss": 0.5229,
+      "step": 18965
+    },
+    {
+      "epoch": 1.616190881976992,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.222035770996217e-07,
+      "loss": 0.4778,
+      "step": 18966
+    },
+    {
+      "epoch": 1.6162760971452919,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.2210794978466913e-07,
+      "loss": 0.4959,
+      "step": 18967
+    },
+    {
+      "epoch": 1.6163613123135918,
+      "grad_norm": 21.875,
+      "learning_rate": 2.2201234104475959e-07,
+      "loss": 0.8219,
+      "step": 18968
+    },
+    {
+      "epoch": 1.6164465274818918,
+      "grad_norm": 14.5,
+      "learning_rate": 2.2191675088162057e-07,
+      "loss": 0.6508,
+      "step": 18969
+    },
+    {
+      "epoch": 1.6165317426501917,
+      "grad_norm": 21.125,
+      "learning_rate": 2.2182117929698002e-07,
+      "loss": 0.6952,
+      "step": 18970
+    },
+    {
+      "epoch": 1.6166169578184917,
+      "grad_norm": 22.375,
+      "learning_rate": 2.2172562629256434e-07,
+      "loss": 1.0118,
+      "step": 18971
+    },
+    {
+      "epoch": 1.6167021729867916,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2163009187010059e-07,
+      "loss": 0.5616,
+      "step": 18972
+    },
+    {
+      "epoch": 1.6167873881550916,
+      "grad_norm": 20.5,
+      "learning_rate": 2.215345760313152e-07,
+      "loss": 0.7652,
+      "step": 18973
+    },
+    {
+      "epoch": 1.6168726033233916,
+      "grad_norm": 28.25,
+      "learning_rate": 2.2143907877793456e-07,
+      "loss": 0.7624,
+      "step": 18974
+    },
+    {
+      "epoch": 1.6169578184916915,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.213436001116838e-07,
+      "loss": 0.647,
+      "step": 18975
+    },
+    {
+      "epoch": 1.6170430336599915,
+      "grad_norm": 13.5,
+      "learning_rate": 2.2124814003428877e-07,
+      "loss": 0.3172,
+      "step": 18976
+    },
+    {
+      "epoch": 1.6171282488282914,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.2115269854747407e-07,
+      "loss": 0.4153,
+      "step": 18977
+    },
+    {
+      "epoch": 1.6172134639965914,
+      "grad_norm": 17.875,
+      "learning_rate": 2.2105727565296469e-07,
+      "loss": 0.7889,
+      "step": 18978
+    },
+    {
+      "epoch": 1.6172986791648913,
+      "grad_norm": 17.875,
+      "learning_rate": 2.209618713524854e-07,
+      "loss": 0.7738,
+      "step": 18979
+    },
+    {
+      "epoch": 1.6173838943331913,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.208664856477595e-07,
+      "loss": 0.6177,
+      "step": 18980
+    },
+    {
+      "epoch": 1.6174691095014913,
+      "grad_norm": 15.375,
+      "learning_rate": 2.2077111854051106e-07,
+      "loss": 0.6108,
+      "step": 18981
+    },
+    {
+      "epoch": 1.6175543246697912,
+      "grad_norm": 19.25,
+      "learning_rate": 2.2067577003246354e-07,
+      "loss": 0.7637,
+      "step": 18982
+    },
+    {
+      "epoch": 1.6176395398380912,
+      "grad_norm": 18.375,
+      "learning_rate": 2.2058044012534017e-07,
+      "loss": 0.8011,
+      "step": 18983
+    },
+    {
+      "epoch": 1.6177247550063911,
+      "grad_norm": 42.5,
+      "learning_rate": 2.2048512882086343e-07,
+      "loss": 1.0336,
+      "step": 18984
+    },
+    {
+      "epoch": 1.617809970174691,
+      "grad_norm": 18.0,
+      "learning_rate": 2.2038983612075542e-07,
+      "loss": 0.8031,
+      "step": 18985
+    },
+    {
+      "epoch": 1.617895185342991,
+      "grad_norm": 13.125,
+      "learning_rate": 2.2029456202673826e-07,
+      "loss": 0.5233,
+      "step": 18986
+    },
+    {
+      "epoch": 1.617980400511291,
+      "grad_norm": 16.25,
+      "learning_rate": 2.201993065405343e-07,
+      "loss": 0.624,
+      "step": 18987
+    },
+    {
+      "epoch": 1.618065615679591,
+      "grad_norm": 9.375,
+      "learning_rate": 2.20104069663864e-07,
+      "loss": 0.2774,
+      "step": 18988
+    },
+    {
+      "epoch": 1.618150830847891,
+      "grad_norm": 18.25,
+      "learning_rate": 2.2000885139844886e-07,
+      "loss": 0.6238,
+      "step": 18989
+    },
+    {
+      "epoch": 1.6182360460161909,
+      "grad_norm": 19.0,
+      "learning_rate": 2.1991365174600944e-07,
+      "loss": 0.7261,
+      "step": 18990
+    },
+    {
+      "epoch": 1.6183212611844908,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.1981847070826664e-07,
+      "loss": 0.469,
+      "step": 18991
+    },
+    {
+      "epoch": 1.6184064763527908,
+      "grad_norm": 22.75,
+      "learning_rate": 2.197233082869399e-07,
+      "loss": 0.8964,
+      "step": 18992
+    },
+    {
+      "epoch": 1.6184916915210907,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.1962816448374864e-07,
+      "loss": 0.2543,
+      "step": 18993
+    },
+    {
+      "epoch": 1.6185769066893907,
+      "grad_norm": 16.5,
+      "learning_rate": 2.1953303930041262e-07,
+      "loss": 0.6343,
+      "step": 18994
+    },
+    {
+      "epoch": 1.6186621218576907,
+      "grad_norm": 13.75,
+      "learning_rate": 2.1943793273865074e-07,
+      "loss": 0.5399,
+      "step": 18995
+    },
+    {
+      "epoch": 1.6187473370259906,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1934284480018207e-07,
+      "loss": 0.4913,
+      "step": 18996
+    },
+    {
+      "epoch": 1.6188325521942906,
+      "grad_norm": 20.0,
+      "learning_rate": 2.1924777548672436e-07,
+      "loss": 0.701,
+      "step": 18997
+    },
+    {
+      "epoch": 1.6189177673625905,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1915272479999568e-07,
+      "loss": 0.3874,
+      "step": 18998
+    },
+    {
+      "epoch": 1.6190029825308905,
+      "grad_norm": 13.5,
+      "learning_rate": 2.19057692741714e-07,
+      "loss": 0.5942,
+      "step": 18999
+    },
+    {
+      "epoch": 1.6190881976991904,
+      "grad_norm": 19.625,
+      "learning_rate": 2.1896267931359711e-07,
+      "loss": 0.5702,
+      "step": 19000
+    },
+    {
+      "epoch": 1.6191734128674904,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.1886768451736067e-07,
+      "loss": 0.358,
+      "step": 19001
+    },
+    {
+      "epoch": 1.6192586280357903,
+      "grad_norm": 22.25,
+      "learning_rate": 2.1877270835472225e-07,
+      "loss": 0.79,
+      "step": 19002
+    },
+    {
+      "epoch": 1.6193438432040903,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.186777508273978e-07,
+      "loss": 0.4998,
+      "step": 19003
+    },
+    {
+      "epoch": 1.6194290583723903,
+      "grad_norm": 17.375,
+      "learning_rate": 2.185828119371039e-07,
+      "loss": 0.8208,
+      "step": 19004
+    },
+    {
+      "epoch": 1.6195142735406902,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1848789168555527e-07,
+      "loss": 0.4181,
+      "step": 19005
+    },
+    {
+      "epoch": 1.6195994887089902,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.1839299007446789e-07,
+      "loss": 0.8593,
+      "step": 19006
+    },
+    {
+      "epoch": 1.6196847038772901,
+      "grad_norm": 11.5,
+      "learning_rate": 2.1829810710555648e-07,
+      "loss": 0.3258,
+      "step": 19007
+    },
+    {
+      "epoch": 1.61976991904559,
+      "grad_norm": 26.75,
+      "learning_rate": 2.1820324278053598e-07,
+      "loss": 0.609,
+      "step": 19008
+    },
+    {
+      "epoch": 1.61985513421389,
+      "grad_norm": 30.625,
+      "learning_rate": 2.1810839710112053e-07,
+      "loss": 1.1832,
+      "step": 19009
+    },
+    {
+      "epoch": 1.61994034938219,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1801357006902364e-07,
+      "loss": 0.3867,
+      "step": 19010
+    },
+    {
+      "epoch": 1.62002556455049,
+      "grad_norm": 16.875,
+      "learning_rate": 2.179187616859592e-07,
+      "loss": 0.6775,
+      "step": 19011
+    },
+    {
+      "epoch": 1.62011077971879,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.1782397195364068e-07,
+      "loss": 0.3081,
+      "step": 19012
+    },
+    {
+      "epoch": 1.6201959948870899,
+      "grad_norm": 11.75,
+      "learning_rate": 2.177292008737812e-07,
+      "loss": 0.5052,
+      "step": 19013
+    },
+    {
+      "epoch": 1.6202812100553898,
+      "grad_norm": 16.625,
+      "learning_rate": 2.1763444844809272e-07,
+      "loss": 0.5751,
+      "step": 19014
+    },
+    {
+      "epoch": 1.6203664252236898,
+      "grad_norm": 17.0,
+      "learning_rate": 2.1753971467828802e-07,
+      "loss": 0.6328,
+      "step": 19015
+    },
+    {
+      "epoch": 1.6204516403919897,
+      "grad_norm": 18.625,
+      "learning_rate": 2.1744499956607905e-07,
+      "loss": 0.5177,
+      "step": 19016
+    },
+    {
+      "epoch": 1.6205368555602897,
+      "grad_norm": 16.25,
+      "learning_rate": 2.173503031131774e-07,
+      "loss": 0.5115,
+      "step": 19017
+    },
+    {
+      "epoch": 1.6206220707285897,
+      "grad_norm": 12.375,
+      "learning_rate": 2.1725562532129374e-07,
+      "loss": 0.3917,
+      "step": 19018
+    },
+    {
+      "epoch": 1.6207072858968896,
+      "grad_norm": 22.375,
+      "learning_rate": 2.171609661921395e-07,
+      "loss": 0.7338,
+      "step": 19019
+    },
+    {
+      "epoch": 1.6207925010651896,
+      "grad_norm": 23.375,
+      "learning_rate": 2.1706632572742513e-07,
+      "loss": 0.8863,
+      "step": 19020
+    },
+    {
+      "epoch": 1.6208777162334895,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.169717039288613e-07,
+      "loss": 0.4285,
+      "step": 19021
+    },
+    {
+      "epoch": 1.6209629314017895,
+      "grad_norm": 24.0,
+      "learning_rate": 2.1687710079815738e-07,
+      "loss": 1.2616,
+      "step": 19022
+    },
+    {
+      "epoch": 1.6210481465700894,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1678251633702296e-07,
+      "loss": 0.6176,
+      "step": 19023
+    },
+    {
+      "epoch": 1.6211333617383894,
+      "grad_norm": 16.25,
+      "learning_rate": 2.1668795054716753e-07,
+      "loss": 0.7763,
+      "step": 19024
+    },
+    {
+      "epoch": 1.6212185769066894,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.1659340343030022e-07,
+      "loss": 0.599,
+      "step": 19025
+    },
+    {
+      "epoch": 1.6213037920749893,
+      "grad_norm": 18.625,
+      "learning_rate": 2.164988749881293e-07,
+      "loss": 0.6235,
+      "step": 19026
+    },
+    {
+      "epoch": 1.6213890072432893,
+      "grad_norm": 19.25,
+      "learning_rate": 2.1640436522236268e-07,
+      "loss": 0.8484,
+      "step": 19027
+    },
+    {
+      "epoch": 1.6214742224115892,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1630987413470847e-07,
+      "loss": 0.5622,
+      "step": 19028
+    },
+    {
+      "epoch": 1.6215594375798892,
+      "grad_norm": 22.125,
+      "learning_rate": 2.162154017268743e-07,
+      "loss": 0.7588,
+      "step": 19029
+    },
+    {
+      "epoch": 1.6216446527481891,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.161209480005677e-07,
+      "loss": 0.7635,
+      "step": 19030
+    },
+    {
+      "epoch": 1.621729867916489,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.1602651295749497e-07,
+      "loss": 0.2991,
+      "step": 19031
+    },
+    {
+      "epoch": 1.621815083084789,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1593209659936277e-07,
+      "loss": 0.6824,
+      "step": 19032
+    },
+    {
+      "epoch": 1.621900298253089,
+      "grad_norm": 14.25,
+      "learning_rate": 2.158376989278778e-07,
+      "loss": 0.605,
+      "step": 19033
+    },
+    {
+      "epoch": 1.621985513421389,
+      "grad_norm": 22.75,
+      "learning_rate": 2.1574331994474547e-07,
+      "loss": 1.0675,
+      "step": 19034
+    },
+    {
+      "epoch": 1.622070728589689,
+      "grad_norm": 18.5,
+      "learning_rate": 2.1564895965167116e-07,
+      "loss": 0.5317,
+      "step": 19035
+    },
+    {
+      "epoch": 1.6221559437579889,
+      "grad_norm": 39.5,
+      "learning_rate": 2.1555461805036023e-07,
+      "loss": 0.7862,
+      "step": 19036
+    },
+    {
+      "epoch": 1.6222411589262888,
+      "grad_norm": 12.5,
+      "learning_rate": 2.1546029514251762e-07,
+      "loss": 0.3959,
+      "step": 19037
+    },
+    {
+      "epoch": 1.6223263740945888,
+      "grad_norm": 12.25,
+      "learning_rate": 2.1536599092984806e-07,
+      "loss": 0.387,
+      "step": 19038
+    },
+    {
+      "epoch": 1.6224115892628888,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.1527170541405518e-07,
+      "loss": 0.4703,
+      "step": 19039
+    },
+    {
+      "epoch": 1.6224968044311887,
+      "grad_norm": 17.75,
+      "learning_rate": 2.151774385968429e-07,
+      "loss": 0.7273,
+      "step": 19040
+    },
+    {
+      "epoch": 1.6225820195994887,
+      "grad_norm": 16.0,
+      "learning_rate": 2.1508319047991542e-07,
+      "loss": 0.8319,
+      "step": 19041
+    },
+    {
+      "epoch": 1.6226672347677886,
+      "grad_norm": 14.5,
+      "learning_rate": 2.1498896106497496e-07,
+      "loss": 0.6021,
+      "step": 19042
+    },
+    {
+      "epoch": 1.6227524499360886,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.1489475035372507e-07,
+      "loss": 0.3779,
+      "step": 19043
+    },
+    {
+      "epoch": 1.6228376651043885,
+      "grad_norm": 19.375,
+      "learning_rate": 2.1480055834786767e-07,
+      "loss": 0.5258,
+      "step": 19044
+    },
+    {
+      "epoch": 1.6229228802726885,
+      "grad_norm": 18.25,
+      "learning_rate": 2.1470638504910505e-07,
+      "loss": 0.6699,
+      "step": 19045
+    },
+    {
+      "epoch": 1.6230080954409885,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.146122304591393e-07,
+      "loss": 0.5442,
+      "step": 19046
+    },
+    {
+      "epoch": 1.6230933106092884,
+      "grad_norm": 19.625,
+      "learning_rate": 2.1451809457967183e-07,
+      "loss": 0.5053,
+      "step": 19047
+    },
+    {
+      "epoch": 1.6231785257775884,
+      "grad_norm": 12.625,
+      "learning_rate": 2.1442397741240352e-07,
+      "loss": 0.4152,
+      "step": 19048
+    },
+    {
+      "epoch": 1.6232637409458883,
+      "grad_norm": 40.0,
+      "learning_rate": 2.1432987895903547e-07,
+      "loss": 0.4815,
+      "step": 19049
+    },
+    {
+      "epoch": 1.6233489561141883,
+      "grad_norm": 16.125,
+      "learning_rate": 2.1423579922126764e-07,
+      "loss": 0.488,
+      "step": 19050
+    },
+    {
+      "epoch": 1.6234341712824882,
+      "grad_norm": 16.625,
+      "learning_rate": 2.1414173820080085e-07,
+      "loss": 0.9597,
+      "step": 19051
+    },
+    {
+      "epoch": 1.6235193864507882,
+      "grad_norm": 19.375,
+      "learning_rate": 2.14047695899334e-07,
+      "loss": 0.7146,
+      "step": 19052
+    },
+    {
+      "epoch": 1.6236046016190882,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.1395367231856715e-07,
+      "loss": 0.3526,
+      "step": 19053
+    },
+    {
+      "epoch": 1.623689816787388,
+      "grad_norm": 17.75,
+      "learning_rate": 2.1385966746019935e-07,
+      "loss": 0.5707,
+      "step": 19054
+    },
+    {
+      "epoch": 1.623775031955688,
+      "grad_norm": 10.125,
+      "learning_rate": 2.137656813259295e-07,
+      "loss": 0.2695,
+      "step": 19055
+    },
+    {
+      "epoch": 1.623860247123988,
+      "grad_norm": 20.0,
+      "learning_rate": 2.1367171391745545e-07,
+      "loss": 0.8899,
+      "step": 19056
+    },
+    {
+      "epoch": 1.623945462292288,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.1357776523647593e-07,
+      "loss": 0.806,
+      "step": 19057
+    },
+    {
+      "epoch": 1.624030677460588,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.1348383528468824e-07,
+      "loss": 0.6306,
+      "step": 19058
+    },
+    {
+      "epoch": 1.624115892628888,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.1338992406378976e-07,
+      "loss": 0.725,
+      "step": 19059
+    },
+    {
+      "epoch": 1.6242011077971878,
+      "grad_norm": 16.75,
+      "learning_rate": 2.1329603157547811e-07,
+      "loss": 0.5382,
+      "step": 19060
+    },
+    {
+      "epoch": 1.6242863229654878,
+      "grad_norm": 16.375,
+      "learning_rate": 2.132021578214495e-07,
+      "loss": 0.4606,
+      "step": 19061
+    },
+    {
+      "epoch": 1.6243715381337878,
+      "grad_norm": 17.625,
+      "learning_rate": 2.1310830280340034e-07,
+      "loss": 0.8391,
+      "step": 19062
+    },
+    {
+      "epoch": 1.6244567533020877,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.1301446652302715e-07,
+      "loss": 0.7079,
+      "step": 19063
+    },
+    {
+      "epoch": 1.6245419684703877,
+      "grad_norm": 23.25,
+      "learning_rate": 2.12920648982025e-07,
+      "loss": 0.8438,
+      "step": 19064
+    },
+    {
+      "epoch": 1.6246271836386876,
+      "grad_norm": 17.125,
+      "learning_rate": 2.1282685018208986e-07,
+      "loss": 0.7646,
+      "step": 19065
+    },
+    {
+      "epoch": 1.6247123988069876,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1273307012491626e-07,
+      "loss": 0.517,
+      "step": 19066
+    },
+    {
+      "epoch": 1.6247976139752875,
+      "grad_norm": 29.75,
+      "learning_rate": 2.1263930881219904e-07,
+      "loss": 0.6678,
+      "step": 19067
+    },
+    {
+      "epoch": 1.6248828291435875,
+      "grad_norm": 15.625,
+      "learning_rate": 2.1254556624563314e-07,
+      "loss": 0.5106,
+      "step": 19068
+    },
+    {
+      "epoch": 1.6249680443118875,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.124518424269116e-07,
+      "loss": 0.4258,
+      "step": 19069
+    },
+    {
+      "epoch": 1.6250532594801874,
+      "grad_norm": 20.5,
+      "learning_rate": 2.1235813735772866e-07,
+      "loss": 0.7577,
+      "step": 19070
+    },
+    {
+      "epoch": 1.6251384746484874,
+      "grad_norm": 15.625,
+      "learning_rate": 2.122644510397777e-07,
+      "loss": 0.8039,
+      "step": 19071
+    },
+    {
+      "epoch": 1.6252236898167873,
+      "grad_norm": 15.375,
+      "learning_rate": 2.121707834747519e-07,
+      "loss": 0.646,
+      "step": 19072
+    },
+    {
+      "epoch": 1.6253089049850873,
+      "grad_norm": 13.875,
+      "learning_rate": 2.1207713466434354e-07,
+      "loss": 0.7292,
+      "step": 19073
+    },
+    {
+      "epoch": 1.6253941201533872,
+      "grad_norm": 21.0,
+      "learning_rate": 2.119835046102449e-07,
+      "loss": 1.0844,
+      "step": 19074
+    },
+    {
+      "epoch": 1.6254793353216872,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.1188989331414806e-07,
+      "loss": 0.5018,
+      "step": 19075
+    },
+    {
+      "epoch": 1.6255645504899872,
+      "grad_norm": 17.625,
+      "learning_rate": 2.117963007777449e-07,
+      "loss": 0.7229,
+      "step": 19076
+    },
+    {
+      "epoch": 1.6256497656582871,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.1170272700272664e-07,
+      "loss": 0.6906,
+      "step": 19077
+    },
+    {
+      "epoch": 1.625734980826587,
+      "grad_norm": 15.375,
+      "learning_rate": 2.1160917199078408e-07,
+      "loss": 0.6723,
+      "step": 19078
+    },
+    {
+      "epoch": 1.625820195994887,
+      "grad_norm": 14.125,
+      "learning_rate": 2.1151563574360795e-07,
+      "loss": 0.5626,
+      "step": 19079
+    },
+    {
+      "epoch": 1.625905411163187,
+      "grad_norm": 16.625,
+      "learning_rate": 2.114221182628888e-07,
+      "loss": 0.4812,
+      "step": 19080
+    },
+    {
+      "epoch": 1.625990626331487,
+      "grad_norm": 17.0,
+      "learning_rate": 2.11328619550316e-07,
+      "loss": 0.5573,
+      "step": 19081
+    },
+    {
+      "epoch": 1.626075841499787,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.1123513960757986e-07,
+      "loss": 0.1548,
+      "step": 19082
+    },
+    {
+      "epoch": 1.6261610566680869,
+      "grad_norm": 12.0,
+      "learning_rate": 2.11141678436369e-07,
+      "loss": 0.3418,
+      "step": 19083
+    },
+    {
+      "epoch": 1.6262462718363868,
+      "grad_norm": 20.375,
+      "learning_rate": 2.110482360383727e-07,
+      "loss": 0.8664,
+      "step": 19084
+    },
+    {
+      "epoch": 1.6263314870046868,
+      "grad_norm": 12.625,
+      "learning_rate": 2.109548124152798e-07,
+      "loss": 0.419,
+      "step": 19085
+    },
+    {
+      "epoch": 1.6264167021729867,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.10861407568778e-07,
+      "loss": 0.6381,
+      "step": 19086
+    },
+    {
+      "epoch": 1.6265019173412867,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.1076802150055544e-07,
+      "loss": 0.2664,
+      "step": 19087
+    },
+    {
+      "epoch": 1.6265871325095866,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.1067465421229977e-07,
+      "loss": 0.6256,
+      "step": 19088
+    },
+    {
+      "epoch": 1.6266723476778866,
+      "grad_norm": 21.375,
+      "learning_rate": 2.1058130570569866e-07,
+      "loss": 0.6922,
+      "step": 19089
+    },
+    {
+      "epoch": 1.6267575628461866,
+      "grad_norm": 20.625,
+      "learning_rate": 2.104879759824384e-07,
+      "loss": 1.0169,
+      "step": 19090
+    },
+    {
+      "epoch": 1.6268427780144865,
+      "grad_norm": 17.125,
+      "learning_rate": 2.103946650442054e-07,
+      "loss": 0.6213,
+      "step": 19091
+    },
+    {
+      "epoch": 1.6269279931827865,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.103013728926863e-07,
+      "loss": 0.6833,
+      "step": 19092
+    },
+    {
+      "epoch": 1.6270132083510864,
+      "grad_norm": 18.875,
+      "learning_rate": 2.1020809952956732e-07,
+      "loss": 0.3505,
+      "step": 19093
+    },
+    {
+      "epoch": 1.6270984235193864,
+      "grad_norm": 18.125,
+      "learning_rate": 2.101148449565331e-07,
+      "loss": 0.5466,
+      "step": 19094
+    },
+    {
+      "epoch": 1.6271836386876863,
+      "grad_norm": 17.875,
+      "learning_rate": 2.1002160917526923e-07,
+      "loss": 0.5418,
+      "step": 19095
+    },
+    {
+      "epoch": 1.6272688538559863,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0992839218746077e-07,
+      "loss": 0.6152,
+      "step": 19096
+    },
+    {
+      "epoch": 1.6273540690242863,
+      "grad_norm": 15.625,
+      "learning_rate": 2.0983519399479248e-07,
+      "loss": 0.7784,
+      "step": 19097
+    },
+    {
+      "epoch": 1.6274392841925862,
+      "grad_norm": 18.875,
+      "learning_rate": 2.0974201459894816e-07,
+      "loss": 0.707,
+      "step": 19098
+    },
+    {
+      "epoch": 1.6275244993608862,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.0964885400161145e-07,
+      "loss": 0.6867,
+      "step": 19099
+    },
+    {
+      "epoch": 1.6276097145291861,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0955571220446599e-07,
+      "loss": 0.7038,
+      "step": 19100
+    },
+    {
+      "epoch": 1.627694929697486,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.094625892091952e-07,
+      "loss": 0.4516,
+      "step": 19101
+    },
+    {
+      "epoch": 1.627780144865786,
+      "grad_norm": 14.125,
+      "learning_rate": 2.0936948501748198e-07,
+      "loss": 0.5599,
+      "step": 19102
+    },
+    {
+      "epoch": 1.627865360034086,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.0927639963100836e-07,
+      "loss": 0.4817,
+      "step": 19103
+    },
+    {
+      "epoch": 1.627950575202386,
+      "grad_norm": 16.75,
+      "learning_rate": 2.0918333305145658e-07,
+      "loss": 0.6433,
+      "step": 19104
+    },
+    {
+      "epoch": 1.628035790370686,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.0909028528050878e-07,
+      "loss": 0.2253,
+      "step": 19105
+    },
+    {
+      "epoch": 1.6281210055389859,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0899725631984642e-07,
+      "loss": 0.6834,
+      "step": 19106
+    },
+    {
+      "epoch": 1.6282062207072858,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0890424617115047e-07,
+      "loss": 0.5394,
+      "step": 19107
+    },
+    {
+      "epoch": 1.6282914358755858,
+      "grad_norm": 13.8125,
+      "learning_rate": 2.088112548361014e-07,
+      "loss": 0.5533,
+      "step": 19108
+    },
+    {
+      "epoch": 1.6283766510438857,
+      "grad_norm": 13.0,
+      "learning_rate": 2.0871828231637998e-07,
+      "loss": 0.5744,
+      "step": 19109
+    },
+    {
+      "epoch": 1.6284618662121857,
+      "grad_norm": 14.375,
+      "learning_rate": 2.086253286136665e-07,
+      "loss": 0.713,
+      "step": 19110
+    },
+    {
+      "epoch": 1.6285470813804857,
+      "grad_norm": 18.625,
+      "learning_rate": 2.0853239372964032e-07,
+      "loss": 0.7767,
+      "step": 19111
+    },
+    {
+      "epoch": 1.6286322965487856,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.0843947766598107e-07,
+      "loss": 0.672,
+      "step": 19112
+    },
+    {
+      "epoch": 1.6287175117170856,
+      "grad_norm": 18.75,
+      "learning_rate": 2.083465804243677e-07,
+      "loss": 0.5927,
+      "step": 19113
+    },
+    {
+      "epoch": 1.6288027268853855,
+      "grad_norm": 16.0,
+      "learning_rate": 2.0825370200647955e-07,
+      "loss": 0.7323,
+      "step": 19114
+    },
+    {
+      "epoch": 1.6288879420536855,
+      "grad_norm": 23.375,
+      "learning_rate": 2.0816084241399444e-07,
+      "loss": 0.7287,
+      "step": 19115
+    },
+    {
+      "epoch": 1.6289731572219854,
+      "grad_norm": 13.25,
+      "learning_rate": 2.0806800164859032e-07,
+      "loss": 0.4175,
+      "step": 19116
+    },
+    {
+      "epoch": 1.6290583723902854,
+      "grad_norm": 24.625,
+      "learning_rate": 2.0797517971194503e-07,
+      "loss": 0.8377,
+      "step": 19117
+    },
+    {
+      "epoch": 1.6291435875585853,
+      "grad_norm": 13.375,
+      "learning_rate": 2.0788237660573626e-07,
+      "loss": 0.6084,
+      "step": 19118
+    },
+    {
+      "epoch": 1.6292288027268853,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.07789592331641e-07,
+      "loss": 0.4501,
+      "step": 19119
+    },
+    {
+      "epoch": 1.6293140178951853,
+      "grad_norm": 16.0,
+      "learning_rate": 2.076968268913357e-07,
+      "loss": 0.5525,
+      "step": 19120
+    },
+    {
+      "epoch": 1.6293992330634852,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.0760408028649675e-07,
+      "loss": 0.5471,
+      "step": 19121
+    },
+    {
+      "epoch": 1.6294844482317852,
+      "grad_norm": 12.75,
+      "learning_rate": 2.0751135251880063e-07,
+      "loss": 0.2358,
+      "step": 19122
+    },
+    {
+      "epoch": 1.6295696634000851,
+      "grad_norm": 21.25,
+      "learning_rate": 2.0741864358992268e-07,
+      "loss": 0.8726,
+      "step": 19123
+    },
+    {
+      "epoch": 1.629654878568385,
+      "grad_norm": 15.875,
+      "learning_rate": 2.073259535015379e-07,
+      "loss": 0.4473,
+      "step": 19124
+    },
+    {
+      "epoch": 1.629740093736685,
+      "grad_norm": 15.875,
+      "learning_rate": 2.0723328225532162e-07,
+      "loss": 0.6868,
+      "step": 19125
+    },
+    {
+      "epoch": 1.629825308904985,
+      "grad_norm": 12.75,
+      "learning_rate": 2.0714062985294854e-07,
+      "loss": 0.549,
+      "step": 19126
+    },
+    {
+      "epoch": 1.629910524073285,
+      "grad_norm": 11.375,
+      "learning_rate": 2.070479962960932e-07,
+      "loss": 0.74,
+      "step": 19127
+    },
+    {
+      "epoch": 1.629995739241585,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0695538158642904e-07,
+      "loss": 0.6589,
+      "step": 19128
+    },
+    {
+      "epoch": 1.6300809544098849,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0686278572563002e-07,
+      "loss": 0.7531,
+      "step": 19129
+    },
+    {
+      "epoch": 1.6301661695781848,
+      "grad_norm": 19.25,
+      "learning_rate": 2.0677020871536973e-07,
+      "loss": 0.9152,
+      "step": 19130
+    },
+    {
+      "epoch": 1.6302513847464848,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0667765055732052e-07,
+      "loss": 0.8161,
+      "step": 19131
+    },
+    {
+      "epoch": 1.6303365999147847,
+      "grad_norm": 22.0,
+      "learning_rate": 2.065851112531557e-07,
+      "loss": 1.0395,
+      "step": 19132
+    },
+    {
+      "epoch": 1.6304218150830847,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0649259080454674e-07,
+      "loss": 0.8496,
+      "step": 19133
+    },
+    {
+      "epoch": 1.6305070302513847,
+      "grad_norm": 19.125,
+      "learning_rate": 2.0640008921316595e-07,
+      "loss": 0.5147,
+      "step": 19134
+    },
+    {
+      "epoch": 1.6305922454196846,
+      "grad_norm": 22.875,
+      "learning_rate": 2.06307606480685e-07,
+      "loss": 0.5134,
+      "step": 19135
+    },
+    {
+      "epoch": 1.6306774605879846,
+      "grad_norm": 11.75,
+      "learning_rate": 2.062151426087755e-07,
+      "loss": 0.4841,
+      "step": 19136
+    },
+    {
+      "epoch": 1.6307626757562845,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.0612269759910763e-07,
+      "loss": 0.5701,
+      "step": 19137
+    },
+    {
+      "epoch": 1.6308478909245845,
+      "grad_norm": 21.0,
+      "learning_rate": 2.0603027145335268e-07,
+      "loss": 1.0739,
+      "step": 19138
+    },
+    {
+      "epoch": 1.6309331060928844,
+      "grad_norm": 13.375,
+      "learning_rate": 2.0593786417318005e-07,
+      "loss": 0.4565,
+      "step": 19139
+    },
+    {
+      "epoch": 1.6310183212611844,
+      "grad_norm": 18.125,
+      "learning_rate": 2.0584547576026057e-07,
+      "loss": 0.6731,
+      "step": 19140
+    },
+    {
+      "epoch": 1.6311035364294844,
+      "grad_norm": 26.75,
+      "learning_rate": 2.0575310621626293e-07,
+      "loss": 0.8974,
+      "step": 19141
+    },
+    {
+      "epoch": 1.6311887515977843,
+      "grad_norm": 12.875,
+      "learning_rate": 2.056607555428568e-07,
+      "loss": 0.5142,
+      "step": 19142
+    },
+    {
+      "epoch": 1.6312739667660843,
+      "grad_norm": 11.0625,
+      "learning_rate": 2.0556842374171092e-07,
+      "loss": 0.3983,
+      "step": 19143
+    },
+    {
+      "epoch": 1.6313591819343842,
+      "grad_norm": 15.625,
+      "learning_rate": 2.054761108144941e-07,
+      "loss": 0.8841,
+      "step": 19144
+    },
+    {
+      "epoch": 1.6314443971026842,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.053838167628741e-07,
+      "loss": 0.2462,
+      "step": 19145
+    },
+    {
+      "epoch": 1.6315296122709841,
+      "grad_norm": 14.875,
+      "learning_rate": 2.052915415885194e-07,
+      "loss": 0.6933,
+      "step": 19146
+    },
+    {
+      "epoch": 1.631614827439284,
+      "grad_norm": 15.625,
+      "learning_rate": 2.051992852930966e-07,
+      "loss": 0.5042,
+      "step": 19147
+    },
+    {
+      "epoch": 1.631700042607584,
+      "grad_norm": 17.0,
+      "learning_rate": 2.0510704787827337e-07,
+      "loss": 0.8469,
+      "step": 19148
+    },
+    {
+      "epoch": 1.631785257775884,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.0501482934571687e-07,
+      "loss": 0.5831,
+      "step": 19149
+    },
+    {
+      "epoch": 1.631870472944184,
+      "grad_norm": 17.5,
+      "learning_rate": 2.0492262969709283e-07,
+      "loss": 0.5598,
+      "step": 19150
+    },
+    {
+      "epoch": 1.631955688112484,
+      "grad_norm": 29.25,
+      "learning_rate": 2.0483044893406786e-07,
+      "loss": 1.0469,
+      "step": 19151
+    },
+    {
+      "epoch": 1.6320409032807839,
+      "grad_norm": 17.375,
+      "learning_rate": 2.047382870583077e-07,
+      "loss": 0.8309,
+      "step": 19152
+    },
+    {
+      "epoch": 1.6321261184490838,
+      "grad_norm": 14.0,
+      "learning_rate": 2.0464614407147795e-07,
+      "loss": 0.5365,
+      "step": 19153
+    },
+    {
+      "epoch": 1.6322113336173838,
+      "grad_norm": 19.875,
+      "learning_rate": 2.045540199752434e-07,
+      "loss": 0.6484,
+      "step": 19154
+    },
+    {
+      "epoch": 1.6322965487856838,
+      "grad_norm": 23.0,
+      "learning_rate": 2.0446191477126927e-07,
+      "loss": 0.6926,
+      "step": 19155
+    },
+    {
+      "epoch": 1.6323817639539837,
+      "grad_norm": 20.875,
+      "learning_rate": 2.043698284612193e-07,
+      "loss": 0.8586,
+      "step": 19156
+    },
+    {
+      "epoch": 1.6324669791222837,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.0427776104675834e-07,
+      "loss": 0.6592,
+      "step": 19157
+    },
+    {
+      "epoch": 1.6325521942905836,
+      "grad_norm": 17.0,
+      "learning_rate": 2.041857125295496e-07,
+      "loss": 0.4598,
+      "step": 19158
+    },
+    {
+      "epoch": 1.6326374094588836,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.040936829112565e-07,
+      "loss": 0.545,
+      "step": 19159
+    },
+    {
+      "epoch": 1.6327226246271835,
+      "grad_norm": 18.75,
+      "learning_rate": 2.0400167219354243e-07,
+      "loss": 0.7947,
+      "step": 19160
+    },
+    {
+      "epoch": 1.6328078397954835,
+      "grad_norm": 16.125,
+      "learning_rate": 2.0390968037807017e-07,
+      "loss": 0.7596,
+      "step": 19161
+    },
+    {
+      "epoch": 1.6328930549637835,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0381770746650168e-07,
+      "loss": 0.5368,
+      "step": 19162
+    },
+    {
+      "epoch": 1.6329782701320834,
+      "grad_norm": 12.75,
+      "learning_rate": 2.0372575346049958e-07,
+      "loss": 0.3962,
+      "step": 19163
+    },
+    {
+      "epoch": 1.6330634853003834,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0363381836172493e-07,
+      "loss": 0.4097,
+      "step": 19164
+    },
+    {
+      "epoch": 1.6331487004686833,
+      "grad_norm": 14.25,
+      "learning_rate": 2.0354190217183935e-07,
+      "loss": 0.6555,
+      "step": 19165
+    },
+    {
+      "epoch": 1.6332339156369833,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.0345000489250408e-07,
+      "loss": 0.6969,
+      "step": 19166
+    },
+    {
+      "epoch": 1.6333191308052832,
+      "grad_norm": 21.0,
+      "learning_rate": 2.0335812652537938e-07,
+      "loss": 0.9809,
+      "step": 19167
+    },
+    {
+      "epoch": 1.6334043459735832,
+      "grad_norm": 24.5,
+      "learning_rate": 2.032662670721258e-07,
+      "loss": 0.7171,
+      "step": 19168
+    },
+    {
+      "epoch": 1.6334895611418832,
+      "grad_norm": 18.75,
+      "learning_rate": 2.031744265344035e-07,
+      "loss": 0.874,
+      "step": 19169
+    },
+    {
+      "epoch": 1.633574776310183,
+      "grad_norm": 12.25,
+      "learning_rate": 2.0308260491387204e-07,
+      "loss": 0.388,
+      "step": 19170
+    },
+    {
+      "epoch": 1.633659991478483,
+      "grad_norm": 14.5,
+      "learning_rate": 2.0299080221219082e-07,
+      "loss": 0.4038,
+      "step": 19171
+    },
+    {
+      "epoch": 1.633745206646783,
+      "grad_norm": 22.5,
+      "learning_rate": 2.028990184310184e-07,
+      "loss": 0.8247,
+      "step": 19172
+    },
+    {
+      "epoch": 1.633830421815083,
+      "grad_norm": 11.5,
+      "learning_rate": 2.0280725357201354e-07,
+      "loss": 0.2662,
+      "step": 19173
+    },
+    {
+      "epoch": 1.633915636983383,
+      "grad_norm": 15.375,
+      "learning_rate": 2.0271550763683493e-07,
+      "loss": 0.7404,
+      "step": 19174
+    },
+    {
+      "epoch": 1.634000852151683,
+      "grad_norm": 15.875,
+      "learning_rate": 2.0262378062714e-07,
+      "loss": 0.7432,
+      "step": 19175
+    },
+    {
+      "epoch": 1.6340860673199828,
+      "grad_norm": 25.625,
+      "learning_rate": 2.0253207254458647e-07,
+      "loss": 0.6961,
+      "step": 19176
+    },
+    {
+      "epoch": 1.6341712824882828,
+      "grad_norm": 17.25,
+      "learning_rate": 2.0244038339083185e-07,
+      "loss": 0.7056,
+      "step": 19177
+    },
+    {
+      "epoch": 1.6342564976565828,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.0234871316753308e-07,
+      "loss": 0.6173,
+      "step": 19178
+    },
+    {
+      "epoch": 1.6343417128248827,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.0225706187634655e-07,
+      "loss": 0.454,
+      "step": 19179
+    },
+    {
+      "epoch": 1.6344269279931827,
+      "grad_norm": 34.75,
+      "learning_rate": 2.0216542951892833e-07,
+      "loss": 0.7762,
+      "step": 19180
+    },
+    {
+      "epoch": 1.6345121431614826,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.0207381609693433e-07,
+      "loss": 0.4802,
+      "step": 19181
+    },
+    {
+      "epoch": 1.6345973583297826,
+      "grad_norm": 15.75,
+      "learning_rate": 2.0198222161202042e-07,
+      "loss": 0.6747,
+      "step": 19182
+    },
+    {
+      "epoch": 1.6346825734980825,
+      "grad_norm": 26.5,
+      "learning_rate": 2.018906460658418e-07,
+      "loss": 1.0639,
+      "step": 19183
+    },
+    {
+      "epoch": 1.6347677886663825,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.0179908946005287e-07,
+      "loss": 0.3698,
+      "step": 19184
+    },
+    {
+      "epoch": 1.6348530038346825,
+      "grad_norm": 17.5,
+      "learning_rate": 2.017075517963084e-07,
+      "loss": 0.512,
+      "step": 19185
+    },
+    {
+      "epoch": 1.6349382190029824,
+      "grad_norm": 18.0,
+      "learning_rate": 2.0161603307626302e-07,
+      "loss": 0.7948,
+      "step": 19186
+    },
+    {
+      "epoch": 1.6350234341712824,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.0152453330157001e-07,
+      "loss": 0.3664,
+      "step": 19187
+    },
+    {
+      "epoch": 1.6351086493395823,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0143305247388278e-07,
+      "loss": 0.5645,
+      "step": 19188
+    },
+    {
+      "epoch": 1.6351938645078823,
+      "grad_norm": 17.625,
+      "learning_rate": 2.0134159059485472e-07,
+      "loss": 0.7854,
+      "step": 19189
+    },
+    {
+      "epoch": 1.6352790796761822,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.0125014766613867e-07,
+      "loss": 0.4956,
+      "step": 19190
+    },
+    {
+      "epoch": 1.6353642948444822,
+      "grad_norm": 10.375,
+      "learning_rate": 2.0115872368938734e-07,
+      "loss": 0.2675,
+      "step": 19191
+    },
+    {
+      "epoch": 1.6354495100127822,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.010673186662522e-07,
+      "loss": 0.4401,
+      "step": 19192
+    },
+    {
+      "epoch": 1.6355347251810821,
+      "grad_norm": 14.75,
+      "learning_rate": 2.0097593259838537e-07,
+      "loss": 0.5151,
+      "step": 19193
+    },
+    {
+      "epoch": 1.635619940349382,
+      "grad_norm": 14.625,
+      "learning_rate": 2.0088456548743848e-07,
+      "loss": 0.6012,
+      "step": 19194
+    },
+    {
+      "epoch": 1.635705155517682,
+      "grad_norm": 17.125,
+      "learning_rate": 2.0079321733506255e-07,
+      "loss": 0.7165,
+      "step": 19195
+    },
+    {
+      "epoch": 1.635790370685982,
+      "grad_norm": 22.0,
+      "learning_rate": 2.0070188814290834e-07,
+      "loss": 0.6291,
+      "step": 19196
+    },
+    {
+      "epoch": 1.635875585854282,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0061057791262593e-07,
+      "loss": 0.7295,
+      "step": 19197
+    },
+    {
+      "epoch": 1.635960801022582,
+      "grad_norm": 19.125,
+      "learning_rate": 2.005192866458655e-07,
+      "loss": 0.7184,
+      "step": 19198
+    },
+    {
+      "epoch": 1.6360460161908819,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.0042801434427702e-07,
+      "loss": 0.4732,
+      "step": 19199
+    },
+    {
+      "epoch": 1.6361312313591818,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.0033676100951e-07,
+      "loss": 0.5279,
+      "step": 19200
+    },
+    {
+      "epoch": 1.6362164465274818,
+      "grad_norm": 14.625,
+      "learning_rate": 2.0024552664321282e-07,
+      "loss": 0.5978,
+      "step": 19201
+    },
+    {
+      "epoch": 1.6363016616957817,
+      "grad_norm": 16.875,
+      "learning_rate": 2.0015431124703472e-07,
+      "loss": 0.5472,
+      "step": 19202
+    },
+    {
+      "epoch": 1.6363868768640817,
+      "grad_norm": 18.375,
+      "learning_rate": 2.000631148226241e-07,
+      "loss": 0.977,
+      "step": 19203
+    },
+    {
+      "epoch": 1.6364720920323816,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9997193737162885e-07,
+      "loss": 0.6837,
+      "step": 19204
+    },
+    {
+      "epoch": 1.6365573072006816,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9988077889569634e-07,
+      "loss": 0.8697,
+      "step": 19205
+    },
+    {
+      "epoch": 1.6366425223689816,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.9978963939647405e-07,
+      "loss": 0.4756,
+      "step": 19206
+    },
+    {
+      "epoch": 1.6367277375372815,
+      "grad_norm": 20.875,
+      "learning_rate": 1.9969851887560897e-07,
+      "loss": 1.0981,
+      "step": 19207
+    },
+    {
+      "epoch": 1.6368129527055815,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.996074173347483e-07,
+      "loss": 0.6102,
+      "step": 19208
+    },
+    {
+      "epoch": 1.6368981678738814,
+      "grad_norm": 25.5,
+      "learning_rate": 1.9951633477553746e-07,
+      "loss": 0.5971,
+      "step": 19209
+    },
+    {
+      "epoch": 1.6369833830421814,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.994252711996228e-07,
+      "loss": 0.2685,
+      "step": 19210
+    },
+    {
+      "epoch": 1.6370685982104813,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9933422660865024e-07,
+      "loss": 1.0529,
+      "step": 19211
+    },
+    {
+      "epoch": 1.6371538133787813,
+      "grad_norm": 18.75,
+      "learning_rate": 1.9924320100426428e-07,
+      "loss": 0.6649,
+      "step": 19212
+    },
+    {
+      "epoch": 1.6372390285470813,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9915219438811084e-07,
+      "loss": 0.5465,
+      "step": 19213
+    },
+    {
+      "epoch": 1.6373242437153812,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9906120676183347e-07,
+      "loss": 0.3934,
+      "step": 19214
+    },
+    {
+      "epoch": 1.6374094588836812,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9897023812707694e-07,
+      "loss": 0.67,
+      "step": 19215
+    },
+    {
+      "epoch": 1.6374946740519811,
+      "grad_norm": 17.5,
+      "learning_rate": 1.988792884854855e-07,
+      "loss": 0.4604,
+      "step": 19216
+    },
+    {
+      "epoch": 1.637579889220281,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.987883578387019e-07,
+      "loss": 0.353,
+      "step": 19217
+    },
+    {
+      "epoch": 1.637665104388581,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9869744618836978e-07,
+      "loss": 0.4494,
+      "step": 19218
+    },
+    {
+      "epoch": 1.637750319556881,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9860655353613188e-07,
+      "loss": 0.5565,
+      "step": 19219
+    },
+    {
+      "epoch": 1.637835534725181,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9851567988363128e-07,
+      "loss": 0.9302,
+      "step": 19220
+    },
+    {
+      "epoch": 1.637920749893481,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9842482523250952e-07,
+      "loss": 0.3044,
+      "step": 19221
+    },
+    {
+      "epoch": 1.6380059650617809,
+      "grad_norm": 21.25,
+      "learning_rate": 1.9833398958440844e-07,
+      "loss": 0.9634,
+      "step": 19222
+    },
+    {
+      "epoch": 1.6380911802300808,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9824317294096964e-07,
+      "loss": 0.6796,
+      "step": 19223
+    },
+    {
+      "epoch": 1.6381763953983808,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9815237530383431e-07,
+      "loss": 0.4766,
+      "step": 19224
+    },
+    {
+      "epoch": 1.6382616105666807,
+      "grad_norm": 19.75,
+      "learning_rate": 1.9806159667464367e-07,
+      "loss": 0.8462,
+      "step": 19225
+    },
+    {
+      "epoch": 1.6383468257349807,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9797083705503749e-07,
+      "loss": 0.7259,
+      "step": 19226
+    },
+    {
+      "epoch": 1.6384320409032807,
+      "grad_norm": 16.75,
+      "learning_rate": 1.978800964466561e-07,
+      "loss": 0.7671,
+      "step": 19227
+    },
+    {
+      "epoch": 1.6385172560715806,
+      "grad_norm": 16.625,
+      "learning_rate": 1.977893748511396e-07,
+      "loss": 0.5474,
+      "step": 19228
+    },
+    {
+      "epoch": 1.6386024712398806,
+      "grad_norm": 12.625,
+      "learning_rate": 1.97698672270127e-07,
+      "loss": 0.5465,
+      "step": 19229
+    },
+    {
+      "epoch": 1.6386876864081805,
+      "grad_norm": 20.75,
+      "learning_rate": 1.9760798870525788e-07,
+      "loss": 0.5657,
+      "step": 19230
+    },
+    {
+      "epoch": 1.6387729015764805,
+      "grad_norm": 20.75,
+      "learning_rate": 1.975173241581703e-07,
+      "loss": 0.3315,
+      "step": 19231
+    },
+    {
+      "epoch": 1.6388581167447804,
+      "grad_norm": 20.375,
+      "learning_rate": 1.974266786305029e-07,
+      "loss": 1.0918,
+      "step": 19232
+    },
+    {
+      "epoch": 1.6389433319130804,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9733605212389428e-07,
+      "loss": 0.4472,
+      "step": 19233
+    },
+    {
+      "epoch": 1.6390285470813803,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9724544463998158e-07,
+      "loss": 0.9391,
+      "step": 19234
+    },
+    {
+      "epoch": 1.6391137622496803,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9715485618040208e-07,
+      "loss": 0.7093,
+      "step": 19235
+    },
+    {
+      "epoch": 1.6391989774179803,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9706428674679354e-07,
+      "loss": 0.8956,
+      "step": 19236
+    },
+    {
+      "epoch": 1.6392841925862802,
+      "grad_norm": 16.25,
+      "learning_rate": 1.969737363407917e-07,
+      "loss": 0.5911,
+      "step": 19237
+    },
+    {
+      "epoch": 1.6393694077545802,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9688320496403374e-07,
+      "loss": 0.4869,
+      "step": 19238
+    },
+    {
+      "epoch": 1.6394546229228801,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9679269261815488e-07,
+      "loss": 0.4933,
+      "step": 19239
+    },
+    {
+      "epoch": 1.63953983809118,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9670219930479114e-07,
+      "loss": 0.9891,
+      "step": 19240
+    },
+    {
+      "epoch": 1.63962505325948,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9661172502557793e-07,
+      "loss": 0.5033,
+      "step": 19241
+    },
+    {
+      "epoch": 1.63971026842778,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9652126978215032e-07,
+      "loss": 0.881,
+      "step": 19242
+    },
+    {
+      "epoch": 1.63979548359608,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9643083357614252e-07,
+      "loss": 0.641,
+      "step": 19243
+    },
+    {
+      "epoch": 1.63988069876438,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9634041640918923e-07,
+      "loss": 0.4672,
+      "step": 19244
+    },
+    {
+      "epoch": 1.6399659139326799,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9625001828292384e-07,
+      "loss": 0.4367,
+      "step": 19245
+    },
+    {
+      "epoch": 1.6400511291009798,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9615963919898047e-07,
+      "loss": 0.2846,
+      "step": 19246
+    },
+    {
+      "epoch": 1.6401363442692798,
+      "grad_norm": 20.375,
+      "learning_rate": 1.9606927915899201e-07,
+      "loss": 0.6163,
+      "step": 19247
+    },
+    {
+      "epoch": 1.6402215594375797,
+      "grad_norm": 18.625,
+      "learning_rate": 1.9597893816459144e-07,
+      "loss": 0.6651,
+      "step": 19248
+    },
+    {
+      "epoch": 1.6403067746058797,
+      "grad_norm": 12.0,
+      "learning_rate": 1.958886162174113e-07,
+      "loss": 0.2826,
+      "step": 19249
+    },
+    {
+      "epoch": 1.6403919897741797,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9579831331908426e-07,
+      "loss": 0.6941,
+      "step": 19250
+    },
+    {
+      "epoch": 1.6404772049424796,
+      "grad_norm": 17.25,
+      "learning_rate": 1.957080294712413e-07,
+      "loss": 0.4887,
+      "step": 19251
+    },
+    {
+      "epoch": 1.6405624201107796,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9561776467551488e-07,
+      "loss": 0.3961,
+      "step": 19252
+    },
+    {
+      "epoch": 1.6406476352790795,
+      "grad_norm": 12.75,
+      "learning_rate": 1.955275189335354e-07,
+      "loss": 0.4397,
+      "step": 19253
+    },
+    {
+      "epoch": 1.6407328504473795,
+      "grad_norm": 32.5,
+      "learning_rate": 1.9543729224693385e-07,
+      "loss": 0.5277,
+      "step": 19254
+    },
+    {
+      "epoch": 1.6408180656156794,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9534708461734138e-07,
+      "loss": 0.6993,
+      "step": 19255
+    },
+    {
+      "epoch": 1.6409032807839794,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9525689604638716e-07,
+      "loss": 0.504,
+      "step": 19256
+    },
+    {
+      "epoch": 1.6409884959522794,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9516672653570157e-07,
+      "loss": 0.4653,
+      "step": 19257
+    },
+    {
+      "epoch": 1.6410737111205793,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9507657608691388e-07,
+      "loss": 0.8008,
+      "step": 19258
+    },
+    {
+      "epoch": 1.6411589262888793,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.9498644470165363e-07,
+      "loss": 0.7395,
+      "step": 19259
+    },
+    {
+      "epoch": 1.6412441414571792,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.948963323815492e-07,
+      "loss": 0.2985,
+      "step": 19260
+    },
+    {
+      "epoch": 1.6413293566254792,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9480623912822877e-07,
+      "loss": 0.6626,
+      "step": 19261
+    },
+    {
+      "epoch": 1.6414145717937791,
+      "grad_norm": 19.25,
+      "learning_rate": 1.947161649433206e-07,
+      "loss": 0.6038,
+      "step": 19262
+    },
+    {
+      "epoch": 1.641499786962079,
+      "grad_norm": 24.0,
+      "learning_rate": 1.9462610982845284e-07,
+      "loss": 0.9866,
+      "step": 19263
+    },
+    {
+      "epoch": 1.641585002130379,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9453607378525223e-07,
+      "loss": 0.594,
+      "step": 19264
+    },
+    {
+      "epoch": 1.641670217298679,
+      "grad_norm": 20.125,
+      "learning_rate": 1.9444605681534612e-07,
+      "loss": 0.9843,
+      "step": 19265
+    },
+    {
+      "epoch": 1.641755432466979,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.9435605892036122e-07,
+      "loss": 0.4618,
+      "step": 19266
+    },
+    {
+      "epoch": 1.641840647635279,
+      "grad_norm": 21.0,
+      "learning_rate": 1.9426608010192416e-07,
+      "loss": 0.5147,
+      "step": 19267
+    },
+    {
+      "epoch": 1.6419258628035789,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9417612036166059e-07,
+      "loss": 0.5629,
+      "step": 19268
+    },
+    {
+      "epoch": 1.6420110779718788,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.94086179701196e-07,
+      "loss": 0.6224,
+      "step": 19269
+    },
+    {
+      "epoch": 1.6420962931401788,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.939962581221559e-07,
+      "loss": 0.4071,
+      "step": 19270
+    },
+    {
+      "epoch": 1.6421815083084788,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.939063556261654e-07,
+      "loss": 0.6087,
+      "step": 19271
+    },
+    {
+      "epoch": 1.6422667234767787,
+      "grad_norm": 39.75,
+      "learning_rate": 1.938164722148493e-07,
+      "loss": 0.5465,
+      "step": 19272
+    },
+    {
+      "epoch": 1.6423519386450787,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.937266078898313e-07,
+      "loss": 0.8606,
+      "step": 19273
+    },
+    {
+      "epoch": 1.6424371538133786,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9363676265273565e-07,
+      "loss": 0.4762,
+      "step": 19274
+    },
+    {
+      "epoch": 1.6425223689816786,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9354693650518592e-07,
+      "loss": 0.4164,
+      "step": 19275
+    },
+    {
+      "epoch": 1.6426075841499785,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.9345712944880607e-07,
+      "loss": 0.6559,
+      "step": 19276
+    },
+    {
+      "epoch": 1.6426927993182785,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9336734148521775e-07,
+      "loss": 0.3196,
+      "step": 19277
+    },
+    {
+      "epoch": 1.6427780144865785,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9327757261604394e-07,
+      "loss": 0.6215,
+      "step": 19278
+    },
+    {
+      "epoch": 1.6428632296548784,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9318782284290711e-07,
+      "loss": 0.5282,
+      "step": 19279
+    },
+    {
+      "epoch": 1.6429484448231784,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9309809216742942e-07,
+      "loss": 0.4708,
+      "step": 19280
+    },
+    {
+      "epoch": 1.6430336599914783,
+      "grad_norm": 21.125,
+      "learning_rate": 1.9300838059123154e-07,
+      "loss": 0.7098,
+      "step": 19281
+    },
+    {
+      "epoch": 1.6431188751597783,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.929186881159352e-07,
+      "loss": 0.7243,
+      "step": 19282
+    },
+    {
+      "epoch": 1.6432040903280782,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.928290147431612e-07,
+      "loss": 0.3831,
+      "step": 19283
+    },
+    {
+      "epoch": 1.6432893054963782,
+      "grad_norm": 31.25,
+      "learning_rate": 1.927393604745302e-07,
+      "loss": 1.1392,
+      "step": 19284
+    },
+    {
+      "epoch": 1.6433745206646782,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9264972531166202e-07,
+      "loss": 0.5971,
+      "step": 19285
+    },
+    {
+      "epoch": 1.643459735832978,
+      "grad_norm": 21.25,
+      "learning_rate": 1.9256010925617643e-07,
+      "loss": 0.7387,
+      "step": 19286
+    },
+    {
+      "epoch": 1.643544951001278,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9247051230969288e-07,
+      "loss": 0.7771,
+      "step": 19287
+    },
+    {
+      "epoch": 1.643630166169578,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.923809344738306e-07,
+      "loss": 0.4712,
+      "step": 19288
+    },
+    {
+      "epoch": 1.643715381337878,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9229137575020874e-07,
+      "loss": 0.4438,
+      "step": 19289
+    },
+    {
+      "epoch": 1.643800596506178,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.92201836140445e-07,
+      "loss": 0.4888,
+      "step": 19290
+    },
+    {
+      "epoch": 1.643885811674478,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.9211231564615768e-07,
+      "loss": 0.4234,
+      "step": 19291
+    },
+    {
+      "epoch": 1.6439710268427778,
+      "grad_norm": 20.0,
+      "learning_rate": 1.920228142689648e-07,
+      "loss": 0.5276,
+      "step": 19292
+    },
+    {
+      "epoch": 1.6440562420110778,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9193333201048395e-07,
+      "loss": 0.805,
+      "step": 19293
+    },
+    {
+      "epoch": 1.6441414571793778,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.918438688723312e-07,
+      "loss": 0.5112,
+      "step": 19294
+    },
+    {
+      "epoch": 1.6442266723476777,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9175442485612385e-07,
+      "loss": 0.8751,
+      "step": 19295
+    },
+    {
+      "epoch": 1.644311887515978,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.9166499996347828e-07,
+      "loss": 0.302,
+      "step": 19296
+    },
+    {
+      "epoch": 1.6443971026842779,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9157559419601055e-07,
+      "loss": 0.486,
+      "step": 19297
+    },
+    {
+      "epoch": 1.6444823178525778,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9148620755533588e-07,
+      "loss": 0.6681,
+      "step": 19298
+    },
+    {
+      "epoch": 1.6445675330208778,
+      "grad_norm": 16.75,
+      "learning_rate": 1.913968400430699e-07,
+      "loss": 0.5504,
+      "step": 19299
+    },
+    {
+      "epoch": 1.6446527481891777,
+      "grad_norm": 20.125,
+      "learning_rate": 1.9130749166082749e-07,
+      "loss": 0.9259,
+      "step": 19300
+    },
+    {
+      "epoch": 1.6447379633574777,
+      "grad_norm": 23.875,
+      "learning_rate": 1.912181624102237e-07,
+      "loss": 0.8031,
+      "step": 19301
+    },
+    {
+      "epoch": 1.6448231785257776,
+      "grad_norm": 14.0,
+      "learning_rate": 1.9112885229287227e-07,
+      "loss": 1.0276,
+      "step": 19302
+    },
+    {
+      "epoch": 1.6449083936940776,
+      "grad_norm": 14.5,
+      "learning_rate": 1.91039561310387e-07,
+      "loss": 0.5001,
+      "step": 19303
+    },
+    {
+      "epoch": 1.6449936088623776,
+      "grad_norm": 16.0,
+      "learning_rate": 1.909502894643818e-07,
+      "loss": 0.7078,
+      "step": 19304
+    },
+    {
+      "epoch": 1.6450788240306775,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9086103675646978e-07,
+      "loss": 0.5766,
+      "step": 19305
+    },
+    {
+      "epoch": 1.6451640391989775,
+      "grad_norm": 17.125,
+      "learning_rate": 1.907718031882641e-07,
+      "loss": 0.5363,
+      "step": 19306
+    },
+    {
+      "epoch": 1.6452492543672774,
+      "grad_norm": 10.625,
+      "learning_rate": 1.9068258876137682e-07,
+      "loss": 0.3351,
+      "step": 19307
+    },
+    {
+      "epoch": 1.6453344695355774,
+      "grad_norm": 12.0,
+      "learning_rate": 1.905933934774204e-07,
+      "loss": 0.2664,
+      "step": 19308
+    },
+    {
+      "epoch": 1.6454196847038773,
+      "grad_norm": 16.125,
+      "learning_rate": 1.90504217338007e-07,
+      "loss": 0.8447,
+      "step": 19309
+    },
+    {
+      "epoch": 1.6455048998721773,
+      "grad_norm": 14.625,
+      "learning_rate": 1.904150603447477e-07,
+      "loss": 0.8133,
+      "step": 19310
+    },
+    {
+      "epoch": 1.6455901150404773,
+      "grad_norm": 20.375,
+      "learning_rate": 1.903259224992536e-07,
+      "loss": 0.1961,
+      "step": 19311
+    },
+    {
+      "epoch": 1.6456753302087772,
+      "grad_norm": 11.5,
+      "learning_rate": 1.902368038031356e-07,
+      "loss": 0.3076,
+      "step": 19312
+    },
+    {
+      "epoch": 1.6457605453770772,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9014770425800425e-07,
+      "loss": 0.6216,
+      "step": 19313
+    },
+    {
+      "epoch": 1.6458457605453771,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.900586238654699e-07,
+      "loss": 0.654,
+      "step": 19314
+    },
+    {
+      "epoch": 1.645930975713677,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8996956262714166e-07,
+      "loss": 0.4777,
+      "step": 19315
+    },
+    {
+      "epoch": 1.646016190881977,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8988052054462954e-07,
+      "loss": 0.5721,
+      "step": 19316
+    },
+    {
+      "epoch": 1.646101406050277,
+      "grad_norm": 18.375,
+      "learning_rate": 1.897914976195428e-07,
+      "loss": 0.8751,
+      "step": 19317
+    },
+    {
+      "epoch": 1.646186621218577,
+      "grad_norm": 16.75,
+      "learning_rate": 1.897024938534893e-07,
+      "loss": 0.5409,
+      "step": 19318
+    },
+    {
+      "epoch": 1.646271836386877,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8961350924807844e-07,
+      "loss": 0.6097,
+      "step": 19319
+    },
+    {
+      "epoch": 1.6463570515551769,
+      "grad_norm": 17.75,
+      "learning_rate": 1.8952454380491744e-07,
+      "loss": 0.6218,
+      "step": 19320
+    },
+    {
+      "epoch": 1.6464422667234768,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8943559752561417e-07,
+      "loss": 0.4256,
+      "step": 19321
+    },
+    {
+      "epoch": 1.6465274818917768,
+      "grad_norm": 18.625,
+      "learning_rate": 1.8934667041177633e-07,
+      "loss": 0.9468,
+      "step": 19322
+    },
+    {
+      "epoch": 1.6466126970600767,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.892577624650109e-07,
+      "loss": 0.5616,
+      "step": 19323
+    },
+    {
+      "epoch": 1.6466979122283767,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.8916887368692406e-07,
+      "loss": 0.2948,
+      "step": 19324
+    },
+    {
+      "epoch": 1.6467831273966766,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8908000407912284e-07,
+      "loss": 0.3557,
+      "step": 19325
+    },
+    {
+      "epoch": 1.6468683425649766,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8899115364321236e-07,
+      "loss": 0.7896,
+      "step": 19326
+    },
+    {
+      "epoch": 1.6469535577332766,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8890232238079897e-07,
+      "loss": 0.2526,
+      "step": 19327
+    },
+    {
+      "epoch": 1.6470387729015765,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8881351029348737e-07,
+      "loss": 0.9509,
+      "step": 19328
+    },
+    {
+      "epoch": 1.6471239880698765,
+      "grad_norm": 13.375,
+      "learning_rate": 1.8872471738288282e-07,
+      "loss": 0.3303,
+      "step": 19329
+    },
+    {
+      "epoch": 1.6472092032381764,
+      "grad_norm": 16.125,
+      "learning_rate": 1.8863594365058968e-07,
+      "loss": 0.714,
+      "step": 19330
+    },
+    {
+      "epoch": 1.6472944184064764,
+      "grad_norm": 17.375,
+      "learning_rate": 1.8854718909821267e-07,
+      "loss": 0.5752,
+      "step": 19331
+    },
+    {
+      "epoch": 1.6473796335747763,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8845845372735511e-07,
+      "loss": 0.6939,
+      "step": 19332
+    },
+    {
+      "epoch": 1.6474648487430763,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8836973753962096e-07,
+      "loss": 0.6664,
+      "step": 19333
+    },
+    {
+      "epoch": 1.6475500639113763,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.8828104053661298e-07,
+      "loss": 0.5217,
+      "step": 19334
+    },
+    {
+      "epoch": 1.6476352790796762,
+      "grad_norm": 18.5,
+      "learning_rate": 1.8819236271993418e-07,
+      "loss": 0.9461,
+      "step": 19335
+    },
+    {
+      "epoch": 1.6477204942479762,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8810370409118734e-07,
+      "loss": 0.7008,
+      "step": 19336
+    },
+    {
+      "epoch": 1.6478057094162761,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.880150646519742e-07,
+      "loss": 0.2939,
+      "step": 19337
+    },
+    {
+      "epoch": 1.647890924584576,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8792644440389668e-07,
+      "loss": 0.5609,
+      "step": 19338
+    },
+    {
+      "epoch": 1.647976139752876,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8783784334855642e-07,
+      "loss": 0.4804,
+      "step": 19339
+    },
+    {
+      "epoch": 1.648061354921176,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8774926148755464e-07,
+      "loss": 0.4724,
+      "step": 19340
+    },
+    {
+      "epoch": 1.648146570089476,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8766069882249188e-07,
+      "loss": 0.4475,
+      "step": 19341
+    },
+    {
+      "epoch": 1.648231785257776,
+      "grad_norm": 21.5,
+      "learning_rate": 1.8757215535496823e-07,
+      "loss": 0.8342,
+      "step": 19342
+    },
+    {
+      "epoch": 1.6483170004260759,
+      "grad_norm": 22.25,
+      "learning_rate": 1.8748363108658408e-07,
+      "loss": 0.9996,
+      "step": 19343
+    },
+    {
+      "epoch": 1.6484022155943758,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8739512601893952e-07,
+      "loss": 0.696,
+      "step": 19344
+    },
+    {
+      "epoch": 1.6484874307626758,
+      "grad_norm": 27.25,
+      "learning_rate": 1.8730664015363332e-07,
+      "loss": 0.7625,
+      "step": 19345
+    },
+    {
+      "epoch": 1.6485726459309757,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.8721817349226457e-07,
+      "loss": 0.5164,
+      "step": 19346
+    },
+    {
+      "epoch": 1.6486578610992757,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8712972603643228e-07,
+      "loss": 0.4156,
+      "step": 19347
+    },
+    {
+      "epoch": 1.6487430762675757,
+      "grad_norm": 22.25,
+      "learning_rate": 1.870412977877349e-07,
+      "loss": 1.2572,
+      "step": 19348
+    },
+    {
+      "epoch": 1.6488282914358756,
+      "grad_norm": 16.25,
+      "learning_rate": 1.869528887477702e-07,
+      "loss": 0.7531,
+      "step": 19349
+    },
+    {
+      "epoch": 1.6489135066041756,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8686449891813536e-07,
+      "loss": 0.5164,
+      "step": 19350
+    },
+    {
+      "epoch": 1.6489987217724755,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8677612830042823e-07,
+      "loss": 0.3172,
+      "step": 19351
+    },
+    {
+      "epoch": 1.6490839369407755,
+      "grad_norm": 11.875,
+      "learning_rate": 1.866877768962455e-07,
+      "loss": 0.4066,
+      "step": 19352
+    },
+    {
+      "epoch": 1.6491691521090754,
+      "grad_norm": 18.5,
+      "learning_rate": 1.8659944470718422e-07,
+      "loss": 0.7412,
+      "step": 19353
+    },
+    {
+      "epoch": 1.6492543672773754,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.8651113173484004e-07,
+      "loss": 0.6349,
+      "step": 19354
+    },
+    {
+      "epoch": 1.6493395824456754,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8642283798080917e-07,
+      "loss": 0.6759,
+      "step": 19355
+    },
+    {
+      "epoch": 1.6494247976139753,
+      "grad_norm": 17.25,
+      "learning_rate": 1.8633456344668733e-07,
+      "loss": 0.718,
+      "step": 19356
+    },
+    {
+      "epoch": 1.6495100127822753,
+      "grad_norm": 12.25,
+      "learning_rate": 1.862463081340693e-07,
+      "loss": 0.2772,
+      "step": 19357
+    },
+    {
+      "epoch": 1.6495952279505752,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8615807204455035e-07,
+      "loss": 0.8108,
+      "step": 19358
+    },
+    {
+      "epoch": 1.6496804431188752,
+      "grad_norm": 19.75,
+      "learning_rate": 1.8606985517972464e-07,
+      "loss": 0.7674,
+      "step": 19359
+    },
+    {
+      "epoch": 1.6497656582871751,
+      "grad_norm": 20.75,
+      "learning_rate": 1.8598165754118643e-07,
+      "loss": 0.8205,
+      "step": 19360
+    },
+    {
+      "epoch": 1.649850873455475,
+      "grad_norm": 21.0,
+      "learning_rate": 1.8589347913053003e-07,
+      "loss": 0.8357,
+      "step": 19361
+    },
+    {
+      "epoch": 1.649936088623775,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8580531994934813e-07,
+      "loss": 0.5318,
+      "step": 19362
+    },
+    {
+      "epoch": 1.650021303792075,
+      "grad_norm": 24.75,
+      "learning_rate": 1.8571717999923424e-07,
+      "loss": 0.4585,
+      "step": 19363
+    },
+    {
+      "epoch": 1.650106518960375,
+      "grad_norm": 16.625,
+      "learning_rate": 1.856290592817811e-07,
+      "loss": 0.4659,
+      "step": 19364
+    },
+    {
+      "epoch": 1.650191734128675,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8554095779858145e-07,
+      "loss": 0.7044,
+      "step": 19365
+    },
+    {
+      "epoch": 1.6502769492969749,
+      "grad_norm": 23.5,
+      "learning_rate": 1.8545287555122706e-07,
+      "loss": 0.7706,
+      "step": 19366
+    },
+    {
+      "epoch": 1.6503621644652748,
+      "grad_norm": 14.5,
+      "learning_rate": 1.8536481254130948e-07,
+      "loss": 0.5212,
+      "step": 19367
+    },
+    {
+      "epoch": 1.6504473796335748,
+      "grad_norm": 17.75,
+      "learning_rate": 1.852767687704203e-07,
+      "loss": 0.7551,
+      "step": 19368
+    },
+    {
+      "epoch": 1.6505325948018748,
+      "grad_norm": 10.5,
+      "learning_rate": 1.8518874424015065e-07,
+      "loss": 0.1772,
+      "step": 19369
+    },
+    {
+      "epoch": 1.6506178099701747,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8510073895209133e-07,
+      "loss": 0.5679,
+      "step": 19370
+    },
+    {
+      "epoch": 1.6507030251384747,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8501275290783217e-07,
+      "loss": 0.4533,
+      "step": 19371
+    },
+    {
+      "epoch": 1.6507882403067746,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8492478610896355e-07,
+      "loss": 0.519,
+      "step": 19372
+    },
+    {
+      "epoch": 1.6508734554750746,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.8483683855707536e-07,
+      "loss": 0.4941,
+      "step": 19373
+    },
+    {
+      "epoch": 1.6509586706433745,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8474891025375654e-07,
+      "loss": 0.6649,
+      "step": 19374
+    },
+    {
+      "epoch": 1.6510438858116745,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8466100120059587e-07,
+      "loss": 0.6033,
+      "step": 19375
+    },
+    {
+      "epoch": 1.6511291009799745,
+      "grad_norm": 20.75,
+      "learning_rate": 1.8457311139918222e-07,
+      "loss": 0.7452,
+      "step": 19376
+    },
+    {
+      "epoch": 1.6512143161482744,
+      "grad_norm": 14.75,
+      "learning_rate": 1.8448524085110373e-07,
+      "loss": 0.8775,
+      "step": 19377
+    },
+    {
+      "epoch": 1.6512995313165744,
+      "grad_norm": 16.5,
+      "learning_rate": 1.8439738955794872e-07,
+      "loss": 0.5007,
+      "step": 19378
+    },
+    {
+      "epoch": 1.6513847464848743,
+      "grad_norm": 11.25,
+      "learning_rate": 1.8430955752130415e-07,
+      "loss": 0.3371,
+      "step": 19379
+    },
+    {
+      "epoch": 1.6514699616531743,
+      "grad_norm": 12.625,
+      "learning_rate": 1.842217447427576e-07,
+      "loss": 0.6475,
+      "step": 19380
+    },
+    {
+      "epoch": 1.6515551768214742,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8413395122389572e-07,
+      "loss": 0.6553,
+      "step": 19381
+    },
+    {
+      "epoch": 1.6516403919897742,
+      "grad_norm": 11.0,
+      "learning_rate": 1.8404617696630544e-07,
+      "loss": 0.2985,
+      "step": 19382
+    },
+    {
+      "epoch": 1.6517256071580741,
+      "grad_norm": 20.625,
+      "learning_rate": 1.8395842197157273e-07,
+      "loss": 0.7027,
+      "step": 19383
+    },
+    {
+      "epoch": 1.651810822326374,
+      "grad_norm": 18.5,
+      "learning_rate": 1.83870686241283e-07,
+      "loss": 0.5273,
+      "step": 19384
+    },
+    {
+      "epoch": 1.651896037494674,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8378296977702201e-07,
+      "loss": 0.2964,
+      "step": 19385
+    },
+    {
+      "epoch": 1.651981252662974,
+      "grad_norm": 19.0,
+      "learning_rate": 1.836952725803752e-07,
+      "loss": 0.7612,
+      "step": 19386
+    },
+    {
+      "epoch": 1.652066467831274,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8360759465292685e-07,
+      "loss": 0.3732,
+      "step": 19387
+    },
+    {
+      "epoch": 1.652151682999574,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8351993599626154e-07,
+      "loss": 0.5553,
+      "step": 19388
+    },
+    {
+      "epoch": 1.652236898167874,
+      "grad_norm": 18.375,
+      "learning_rate": 1.8343229661196337e-07,
+      "loss": 0.6769,
+      "step": 19389
+    },
+    {
+      "epoch": 1.6523221133361738,
+      "grad_norm": 19.0,
+      "learning_rate": 1.8334467650161656e-07,
+      "loss": 0.8116,
+      "step": 19390
+    },
+    {
+      "epoch": 1.6524073285044738,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.8325707566680394e-07,
+      "loss": 0.6021,
+      "step": 19391
+    },
+    {
+      "epoch": 1.6524925436727738,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8316949410910833e-07,
+      "loss": 0.8711,
+      "step": 19392
+    },
+    {
+      "epoch": 1.6525777588410737,
+      "grad_norm": 18.875,
+      "learning_rate": 1.830819318301129e-07,
+      "loss": 0.7389,
+      "step": 19393
+    },
+    {
+      "epoch": 1.6526629740093737,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8299438883139967e-07,
+      "loss": 0.8028,
+      "step": 19394
+    },
+    {
+      "epoch": 1.6527481891776736,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8290686511455117e-07,
+      "loss": 0.4817,
+      "step": 19395
+    },
+    {
+      "epoch": 1.6528334043459736,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8281936068114832e-07,
+      "loss": 0.5762,
+      "step": 19396
+    },
+    {
+      "epoch": 1.6529186195142735,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8273187553277284e-07,
+      "loss": 0.7924,
+      "step": 19397
+    },
+    {
+      "epoch": 1.6530038346825735,
+      "grad_norm": 30.0,
+      "learning_rate": 1.8264440967100574e-07,
+      "loss": 0.8572,
+      "step": 19398
+    },
+    {
+      "epoch": 1.6530890498508735,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.8255696309742722e-07,
+      "loss": 0.4286,
+      "step": 19399
+    },
+    {
+      "epoch": 1.6531742650191734,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.8246953581361819e-07,
+      "loss": 0.5601,
+      "step": 19400
+    },
+    {
+      "epoch": 1.6532594801874734,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.8238212782115766e-07,
+      "loss": 0.6186,
+      "step": 19401
+    },
+    {
+      "epoch": 1.6533446953557733,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.822947391216258e-07,
+      "loss": 0.635,
+      "step": 19402
+    },
+    {
+      "epoch": 1.6534299105240733,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8220736971660186e-07,
+      "loss": 0.7179,
+      "step": 19403
+    },
+    {
+      "epoch": 1.6535151256923732,
+      "grad_norm": 20.125,
+      "learning_rate": 1.8212001960766433e-07,
+      "loss": 0.7307,
+      "step": 19404
+    },
+    {
+      "epoch": 1.6536003408606732,
+      "grad_norm": 19.25,
+      "learning_rate": 1.8203268879639176e-07,
+      "loss": 1.0239,
+      "step": 19405
+    },
+    {
+      "epoch": 1.6536855560289732,
+      "grad_norm": 16.0,
+      "learning_rate": 1.8194537728436267e-07,
+      "loss": 0.6888,
+      "step": 19406
+    },
+    {
+      "epoch": 1.6537707711972731,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.8185808507315453e-07,
+      "loss": 0.65,
+      "step": 19407
+    },
+    {
+      "epoch": 1.653855986365573,
+      "grad_norm": 16.875,
+      "learning_rate": 1.817708121643451e-07,
+      "loss": 0.5776,
+      "step": 19408
+    },
+    {
+      "epoch": 1.653941201533873,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8168355855951085e-07,
+      "loss": 0.8526,
+      "step": 19409
+    },
+    {
+      "epoch": 1.654026416702173,
+      "grad_norm": 22.75,
+      "learning_rate": 1.8159632426022908e-07,
+      "loss": 0.6855,
+      "step": 19410
+    },
+    {
+      "epoch": 1.654111631870473,
+      "grad_norm": 13.25,
+      "learning_rate": 1.815091092680761e-07,
+      "loss": 0.413,
+      "step": 19411
+    },
+    {
+      "epoch": 1.654196847038773,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.8142191358462833e-07,
+      "loss": 0.5441,
+      "step": 19412
+    },
+    {
+      "epoch": 1.6542820622070729,
+      "grad_norm": 14.0,
+      "learning_rate": 1.8133473721146072e-07,
+      "loss": 0.6318,
+      "step": 19413
+    },
+    {
+      "epoch": 1.6543672773753728,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.812475801501494e-07,
+      "loss": 0.2501,
+      "step": 19414
+    },
+    {
+      "epoch": 1.6544524925436728,
+      "grad_norm": 15.375,
+      "learning_rate": 1.8116044240226866e-07,
+      "loss": 0.7859,
+      "step": 19415
+    },
+    {
+      "epoch": 1.6545377077119727,
+      "grad_norm": 17.0,
+      "learning_rate": 1.810733239693939e-07,
+      "loss": 0.8148,
+      "step": 19416
+    },
+    {
+      "epoch": 1.6546229228802727,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8098622485309873e-07,
+      "loss": 0.5387,
+      "step": 19417
+    },
+    {
+      "epoch": 1.6547081380485726,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8089914505495744e-07,
+      "loss": 0.3067,
+      "step": 19418
+    },
+    {
+      "epoch": 1.6547933532168726,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.8081208457654362e-07,
+      "loss": 0.4237,
+      "step": 19419
+    },
+    {
+      "epoch": 1.6548785683851726,
+      "grad_norm": 17.5,
+      "learning_rate": 1.8072504341943089e-07,
+      "loss": 0.6383,
+      "step": 19420
+    },
+    {
+      "epoch": 1.6549637835534725,
+      "grad_norm": 25.625,
+      "learning_rate": 1.8063802158519161e-07,
+      "loss": 0.6936,
+      "step": 19421
+    },
+    {
+      "epoch": 1.6550489987217725,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.805510190753987e-07,
+      "loss": 0.3968,
+      "step": 19422
+    },
+    {
+      "epoch": 1.6551342138900724,
+      "grad_norm": 13.5,
+      "learning_rate": 1.8046403589162446e-07,
+      "loss": 0.5926,
+      "step": 19423
+    },
+    {
+      "epoch": 1.6552194290583724,
+      "grad_norm": 12.5,
+      "learning_rate": 1.803770720354403e-07,
+      "loss": 0.397,
+      "step": 19424
+    },
+    {
+      "epoch": 1.6553046442266723,
+      "grad_norm": 18.125,
+      "learning_rate": 1.802901275084183e-07,
+      "loss": 0.7631,
+      "step": 19425
+    },
+    {
+      "epoch": 1.6553898593949723,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.8020320231212919e-07,
+      "loss": 0.6995,
+      "step": 19426
+    },
+    {
+      "epoch": 1.6554750745632723,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.80116296448144e-07,
+      "loss": 0.6697,
+      "step": 19427
+    },
+    {
+      "epoch": 1.6555602897315722,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.8002940991803303e-07,
+      "loss": 0.3863,
+      "step": 19428
+    },
+    {
+      "epoch": 1.6556455048998722,
+      "grad_norm": 12.5,
+      "learning_rate": 1.7994254272336698e-07,
+      "loss": 0.398,
+      "step": 19429
+    },
+    {
+      "epoch": 1.6557307200681721,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.7985569486571486e-07,
+      "loss": 0.6979,
+      "step": 19430
+    },
+    {
+      "epoch": 1.655815935236472,
+      "grad_norm": 10.75,
+      "learning_rate": 1.7976886634664682e-07,
+      "loss": 0.3375,
+      "step": 19431
+    },
+    {
+      "epoch": 1.655901150404772,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7968205716773116e-07,
+      "loss": 0.5716,
+      "step": 19432
+    },
+    {
+      "epoch": 1.655986365573072,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7959526733053733e-07,
+      "loss": 0.5077,
+      "step": 19433
+    },
+    {
+      "epoch": 1.656071580741372,
+      "grad_norm": 25.375,
+      "learning_rate": 1.7950849683663312e-07,
+      "loss": 0.6731,
+      "step": 19434
+    },
+    {
+      "epoch": 1.656156795909672,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7942174568758686e-07,
+      "loss": 0.4194,
+      "step": 19435
+    },
+    {
+      "epoch": 1.6562420110779719,
+      "grad_norm": 24.625,
+      "learning_rate": 1.7933501388496616e-07,
+      "loss": 1.1108,
+      "step": 19436
+    },
+    {
+      "epoch": 1.6563272262462718,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.7924830143033866e-07,
+      "loss": 0.7367,
+      "step": 19437
+    },
+    {
+      "epoch": 1.6564124414145718,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.791616083252709e-07,
+      "loss": 0.6487,
+      "step": 19438
+    },
+    {
+      "epoch": 1.6564976565828717,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7907493457132995e-07,
+      "loss": 0.696,
+      "step": 19439
+    },
+    {
+      "epoch": 1.6565828717511717,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7898828017008151e-07,
+      "loss": 0.7518,
+      "step": 19440
+    },
+    {
+      "epoch": 1.6566680869194716,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7890164512309184e-07,
+      "loss": 0.3174,
+      "step": 19441
+    },
+    {
+      "epoch": 1.6567533020877716,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.7881502943192672e-07,
+      "loss": 0.5122,
+      "step": 19442
+    },
+    {
+      "epoch": 1.6568385172560716,
+      "grad_norm": 13.25,
+      "learning_rate": 1.7872843309815092e-07,
+      "loss": 0.3762,
+      "step": 19443
+    },
+    {
+      "epoch": 1.6569237324243715,
+      "grad_norm": 18.5,
+      "learning_rate": 1.786418561233297e-07,
+      "loss": 0.6564,
+      "step": 19444
+    },
+    {
+      "epoch": 1.6570089475926715,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7855529850902733e-07,
+      "loss": 0.382,
+      "step": 19445
+    },
+    {
+      "epoch": 1.6570941627609714,
+      "grad_norm": 11.5,
+      "learning_rate": 1.784687602568086e-07,
+      "loss": 0.3701,
+      "step": 19446
+    },
+    {
+      "epoch": 1.6571793779292714,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7838224136823666e-07,
+      "loss": 0.4403,
+      "step": 19447
+    },
+    {
+      "epoch": 1.6572645930975713,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7829574184487511e-07,
+      "loss": 0.6807,
+      "step": 19448
+    },
+    {
+      "epoch": 1.6573498082658713,
+      "grad_norm": 12.5,
+      "learning_rate": 1.7820926168828703e-07,
+      "loss": 0.4887,
+      "step": 19449
+    },
+    {
+      "epoch": 1.6574350234341713,
+      "grad_norm": 22.625,
+      "learning_rate": 1.7812280090003574e-07,
+      "loss": 0.7665,
+      "step": 19450
+    },
+    {
+      "epoch": 1.6575202386024712,
+      "grad_norm": 12.0,
+      "learning_rate": 1.7803635948168307e-07,
+      "loss": 0.4414,
+      "step": 19451
+    },
+    {
+      "epoch": 1.6576054537707712,
+      "grad_norm": 19.875,
+      "learning_rate": 1.779499374347912e-07,
+      "loss": 1.1401,
+      "step": 19452
+    },
+    {
+      "epoch": 1.6576906689390711,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7786353476092214e-07,
+      "loss": 0.6111,
+      "step": 19453
+    },
+    {
+      "epoch": 1.657775884107371,
+      "grad_norm": 13.0,
+      "learning_rate": 1.777771514616372e-07,
+      "loss": 0.6027,
+      "step": 19454
+    },
+    {
+      "epoch": 1.657861099275671,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7769078753849756e-07,
+      "loss": 0.6748,
+      "step": 19455
+    },
+    {
+      "epoch": 1.657946314443971,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.7760444299306317e-07,
+      "loss": 0.3051,
+      "step": 19456
+    },
+    {
+      "epoch": 1.658031529612271,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7751811782689505e-07,
+      "loss": 1.0536,
+      "step": 19457
+    },
+    {
+      "epoch": 1.658116744780571,
+      "grad_norm": 17.25,
+      "learning_rate": 1.7743181204155287e-07,
+      "loss": 0.5162,
+      "step": 19458
+    },
+    {
+      "epoch": 1.6582019599488709,
+      "grad_norm": 29.625,
+      "learning_rate": 1.7734552563859683e-07,
+      "loss": 0.791,
+      "step": 19459
+    },
+    {
+      "epoch": 1.6582871751171708,
+      "grad_norm": 15.375,
+      "learning_rate": 1.7725925861958538e-07,
+      "loss": 0.6744,
+      "step": 19460
+    },
+    {
+      "epoch": 1.6583723902854708,
+      "grad_norm": 24.875,
+      "learning_rate": 1.7717301098607794e-07,
+      "loss": 0.2824,
+      "step": 19461
+    },
+    {
+      "epoch": 1.6584576054537707,
+      "grad_norm": 30.375,
+      "learning_rate": 1.770867827396329e-07,
+      "loss": 0.9232,
+      "step": 19462
+    },
+    {
+      "epoch": 1.6585428206220707,
+      "grad_norm": 10.875,
+      "learning_rate": 1.7700057388180926e-07,
+      "loss": 0.2941,
+      "step": 19463
+    },
+    {
+      "epoch": 1.6586280357903707,
+      "grad_norm": 13.625,
+      "learning_rate": 1.7691438441416368e-07,
+      "loss": 0.3183,
+      "step": 19464
+    },
+    {
+      "epoch": 1.6587132509586706,
+      "grad_norm": 16.375,
+      "learning_rate": 1.768282143382541e-07,
+      "loss": 0.622,
+      "step": 19465
+    },
+    {
+      "epoch": 1.6587984661269706,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.7674206365563802e-07,
+      "loss": 0.3902,
+      "step": 19466
+    },
+    {
+      "epoch": 1.6588836812952705,
+      "grad_norm": 13.625,
+      "learning_rate": 1.7665593236787224e-07,
+      "loss": 0.483,
+      "step": 19467
+    },
+    {
+      "epoch": 1.6589688964635705,
+      "grad_norm": 15.375,
+      "learning_rate": 1.765698204765129e-07,
+      "loss": 0.7258,
+      "step": 19468
+    },
+    {
+      "epoch": 1.6590541116318704,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7648372798311622e-07,
+      "loss": 0.4711,
+      "step": 19469
+    },
+    {
+      "epoch": 1.6591393268001704,
+      "grad_norm": 20.125,
+      "learning_rate": 1.7639765488923806e-07,
+      "loss": 0.931,
+      "step": 19470
+    },
+    {
+      "epoch": 1.6592245419684704,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7631160119643428e-07,
+      "loss": 0.4046,
+      "step": 19471
+    },
+    {
+      "epoch": 1.6593097571367703,
+      "grad_norm": 18.625,
+      "learning_rate": 1.7622556690625944e-07,
+      "loss": 0.8349,
+      "step": 19472
+    },
+    {
+      "epoch": 1.6593949723050703,
+      "grad_norm": 21.5,
+      "learning_rate": 1.7613955202026815e-07,
+      "loss": 0.5973,
+      "step": 19473
+    },
+    {
+      "epoch": 1.6594801874733702,
+      "grad_norm": 11.0,
+      "learning_rate": 1.760535565400151e-07,
+      "loss": 0.295,
+      "step": 19474
+    },
+    {
+      "epoch": 1.6595654026416702,
+      "grad_norm": 15.375,
+      "learning_rate": 1.759675804670541e-07,
+      "loss": 0.7016,
+      "step": 19475
+    },
+    {
+      "epoch": 1.6596506178099701,
+      "grad_norm": 16.375,
+      "learning_rate": 1.7588162380293913e-07,
+      "loss": 0.5547,
+      "step": 19476
+    },
+    {
+      "epoch": 1.65973583297827,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7579568654922317e-07,
+      "loss": 0.8985,
+      "step": 19477
+    },
+    {
+      "epoch": 1.65982104814657,
+      "grad_norm": 20.875,
+      "learning_rate": 1.7570976870745937e-07,
+      "loss": 0.7554,
+      "step": 19478
+    },
+    {
+      "epoch": 1.65990626331487,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7562387027920055e-07,
+      "loss": 0.5377,
+      "step": 19479
+    },
+    {
+      "epoch": 1.65999147848317,
+      "grad_norm": 24.625,
+      "learning_rate": 1.755379912659988e-07,
+      "loss": 0.5473,
+      "step": 19480
+    },
+    {
+      "epoch": 1.66007669365147,
+      "grad_norm": 14.5,
+      "learning_rate": 1.7545213166940577e-07,
+      "loss": 0.3896,
+      "step": 19481
+    },
+    {
+      "epoch": 1.6601619088197699,
+      "grad_norm": 18.125,
+      "learning_rate": 1.7536629149097318e-07,
+      "loss": 0.8575,
+      "step": 19482
+    },
+    {
+      "epoch": 1.6602471239880698,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.752804707322524e-07,
+      "loss": 0.5296,
+      "step": 19483
+    },
+    {
+      "epoch": 1.6603323391563698,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.7519466939479437e-07,
+      "loss": 0.4234,
+      "step": 19484
+    },
+    {
+      "epoch": 1.6604175543246698,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.751088874801493e-07,
+      "loss": 0.3034,
+      "step": 19485
+    },
+    {
+      "epoch": 1.6605027694929697,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.7502312498986747e-07,
+      "loss": 0.4663,
+      "step": 19486
+    },
+    {
+      "epoch": 1.6605879846612697,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7493738192549888e-07,
+      "loss": 0.5164,
+      "step": 19487
+    },
+    {
+      "epoch": 1.6606731998295696,
+      "grad_norm": 12.125,
+      "learning_rate": 1.7485165828859298e-07,
+      "loss": 0.3834,
+      "step": 19488
+    },
+    {
+      "epoch": 1.6607584149978696,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.7476595408069882e-07,
+      "loss": 0.6588,
+      "step": 19489
+    },
+    {
+      "epoch": 1.6608436301661695,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7468026930336474e-07,
+      "loss": 0.6609,
+      "step": 19490
+    },
+    {
+      "epoch": 1.6609288453344695,
+      "grad_norm": 18.5,
+      "learning_rate": 1.7459460395813963e-07,
+      "loss": 0.7481,
+      "step": 19491
+    },
+    {
+      "epoch": 1.6610140605027695,
+      "grad_norm": 14.25,
+      "learning_rate": 1.7450895804657143e-07,
+      "loss": 0.5069,
+      "step": 19492
+    },
+    {
+      "epoch": 1.6610992756710694,
+      "grad_norm": 17.5,
+      "learning_rate": 1.7442333157020806e-07,
+      "loss": 0.4555,
+      "step": 19493
+    },
+    {
+      "epoch": 1.6611844908393694,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7433772453059633e-07,
+      "loss": 0.4616,
+      "step": 19494
+    },
+    {
+      "epoch": 1.6612697060076693,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.7425213692928377e-07,
+      "loss": 0.5307,
+      "step": 19495
+    },
+    {
+      "epoch": 1.6613549211759693,
+      "grad_norm": 18.375,
+      "learning_rate": 1.7416656876781694e-07,
+      "loss": 0.5097,
+      "step": 19496
+    },
+    {
+      "epoch": 1.6614401363442692,
+      "grad_norm": 17.5,
+      "learning_rate": 1.740810200477422e-07,
+      "loss": 0.528,
+      "step": 19497
+    },
+    {
+      "epoch": 1.6615253515125692,
+      "grad_norm": 22.0,
+      "learning_rate": 1.7399549077060484e-07,
+      "loss": 0.9126,
+      "step": 19498
+    },
+    {
+      "epoch": 1.6616105666808691,
+      "grad_norm": 13.875,
+      "learning_rate": 1.7390998093795117e-07,
+      "loss": 0.4675,
+      "step": 19499
+    },
+    {
+      "epoch": 1.661695781849169,
+      "grad_norm": 30.5,
+      "learning_rate": 1.7382449055132616e-07,
+      "loss": 0.8561,
+      "step": 19500
+    },
+    {
+      "epoch": 1.661780997017469,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7373901961227513e-07,
+      "loss": 0.4115,
+      "step": 19501
+    },
+    {
+      "epoch": 1.661866212185769,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7365356812234197e-07,
+      "loss": 0.7084,
+      "step": 19502
+    },
+    {
+      "epoch": 1.661951427354069,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.7356813608307114e-07,
+      "loss": 0.5722,
+      "step": 19503
+    },
+    {
+      "epoch": 1.662036642522369,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.7348272349600682e-07,
+      "loss": 0.5985,
+      "step": 19504
+    },
+    {
+      "epoch": 1.662121857690669,
+      "grad_norm": 17.125,
+      "learning_rate": 1.7339733036269194e-07,
+      "loss": 0.5964,
+      "step": 19505
+    },
+    {
+      "epoch": 1.6622070728589688,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7331195668467013e-07,
+      "loss": 0.8222,
+      "step": 19506
+    },
+    {
+      "epoch": 1.6622922880272688,
+      "grad_norm": 13.625,
+      "learning_rate": 1.7322660246348378e-07,
+      "loss": 0.3949,
+      "step": 19507
+    },
+    {
+      "epoch": 1.6623775031955688,
+      "grad_norm": 14.125,
+      "learning_rate": 1.7314126770067552e-07,
+      "loss": 0.4898,
+      "step": 19508
+    },
+    {
+      "epoch": 1.6624627183638687,
+      "grad_norm": 26.375,
+      "learning_rate": 1.730559523977876e-07,
+      "loss": 1.0845,
+      "step": 19509
+    },
+    {
+      "epoch": 1.6625479335321687,
+      "grad_norm": 14.25,
+      "learning_rate": 1.729706565563613e-07,
+      "loss": 0.6164,
+      "step": 19510
+    },
+    {
+      "epoch": 1.6626331487004686,
+      "grad_norm": 15.625,
+      "learning_rate": 1.728853801779383e-07,
+      "loss": 0.4716,
+      "step": 19511
+    },
+    {
+      "epoch": 1.6627183638687686,
+      "grad_norm": 20.625,
+      "learning_rate": 1.7280012326405986e-07,
+      "loss": 0.646,
+      "step": 19512
+    },
+    {
+      "epoch": 1.6628035790370685,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.7271488581626625e-07,
+      "loss": 0.4666,
+      "step": 19513
+    },
+    {
+      "epoch": 1.6628887942053685,
+      "grad_norm": 14.875,
+      "learning_rate": 1.7262966783609808e-07,
+      "loss": 0.4652,
+      "step": 19514
+    },
+    {
+      "epoch": 1.6629740093736685,
+      "grad_norm": 20.375,
+      "learning_rate": 1.7254446932509494e-07,
+      "loss": 0.7882,
+      "step": 19515
+    },
+    {
+      "epoch": 1.6630592245419684,
+      "grad_norm": 17.375,
+      "learning_rate": 1.724592902847967e-07,
+      "loss": 0.5199,
+      "step": 19516
+    },
+    {
+      "epoch": 1.6631444397102684,
+      "grad_norm": 22.5,
+      "learning_rate": 1.723741307167427e-07,
+      "loss": 0.6446,
+      "step": 19517
+    },
+    {
+      "epoch": 1.6632296548785683,
+      "grad_norm": 25.875,
+      "learning_rate": 1.72288990622472e-07,
+      "loss": 1.0557,
+      "step": 19518
+    },
+    {
+      "epoch": 1.6633148700468683,
+      "grad_norm": 15.625,
+      "learning_rate": 1.7220387000352265e-07,
+      "loss": 0.5639,
+      "step": 19519
+    },
+    {
+      "epoch": 1.6634000852151682,
+      "grad_norm": 30.5,
+      "learning_rate": 1.7211876886143355e-07,
+      "loss": 0.9343,
+      "step": 19520
+    },
+    {
+      "epoch": 1.6634853003834684,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7203368719774182e-07,
+      "loss": 0.4428,
+      "step": 19521
+    },
+    {
+      "epoch": 1.6635705155517684,
+      "grad_norm": 14.75,
+      "learning_rate": 1.719486250139854e-07,
+      "loss": 0.6097,
+      "step": 19522
+    },
+    {
+      "epoch": 1.6636557307200683,
+      "grad_norm": 12.375,
+      "learning_rate": 1.7186358231170167e-07,
+      "loss": 0.5344,
+      "step": 19523
+    },
+    {
+      "epoch": 1.6637409458883683,
+      "grad_norm": 20.625,
+      "learning_rate": 1.717785590924269e-07,
+      "loss": 0.6887,
+      "step": 19524
+    },
+    {
+      "epoch": 1.6638261610566683,
+      "grad_norm": 12.75,
+      "learning_rate": 1.7169355535769775e-07,
+      "loss": 0.4937,
+      "step": 19525
+    },
+    {
+      "epoch": 1.6639113762249682,
+      "grad_norm": 16.0,
+      "learning_rate": 1.716085711090508e-07,
+      "loss": 0.4135,
+      "step": 19526
+    },
+    {
+      "epoch": 1.6639965913932682,
+      "grad_norm": 17.75,
+      "learning_rate": 1.7152360634802092e-07,
+      "loss": 1.01,
+      "step": 19527
+    },
+    {
+      "epoch": 1.6640818065615681,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.7143866107614426e-07,
+      "loss": 0.6504,
+      "step": 19528
+    },
+    {
+      "epoch": 1.664167021729868,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.713537352949554e-07,
+      "loss": 0.4485,
+      "step": 19529
+    },
+    {
+      "epoch": 1.664252236898168,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7126882900598912e-07,
+      "loss": 0.688,
+      "step": 19530
+    },
+    {
+      "epoch": 1.664337452066468,
+      "grad_norm": 23.0,
+      "learning_rate": 1.7118394221078013e-07,
+      "loss": 0.9478,
+      "step": 19531
+    },
+    {
+      "epoch": 1.664422667234768,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.7109907491086195e-07,
+      "loss": 0.6269,
+      "step": 19532
+    },
+    {
+      "epoch": 1.664507882403068,
+      "grad_norm": 16.875,
+      "learning_rate": 1.7101422710776832e-07,
+      "loss": 0.499,
+      "step": 19533
+    },
+    {
+      "epoch": 1.6645930975713679,
+      "grad_norm": 13.125,
+      "learning_rate": 1.7092939880303263e-07,
+      "loss": 0.6598,
+      "step": 19534
+    },
+    {
+      "epoch": 1.6646783127396678,
+      "grad_norm": 29.5,
+      "learning_rate": 1.7084458999818795e-07,
+      "loss": 0.5009,
+      "step": 19535
+    },
+    {
+      "epoch": 1.6647635279079678,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7075980069476693e-07,
+      "loss": 0.5432,
+      "step": 19536
+    },
+    {
+      "epoch": 1.6648487430762677,
+      "grad_norm": 17.875,
+      "learning_rate": 1.7067503089430116e-07,
+      "loss": 0.6029,
+      "step": 19537
+    },
+    {
+      "epoch": 1.6649339582445677,
+      "grad_norm": 22.625,
+      "learning_rate": 1.7059028059832298e-07,
+      "loss": 0.4113,
+      "step": 19538
+    },
+    {
+      "epoch": 1.6650191734128676,
+      "grad_norm": 16.25,
+      "learning_rate": 1.705055498083641e-07,
+      "loss": 0.8155,
+      "step": 19539
+    },
+    {
+      "epoch": 1.6651043885811676,
+      "grad_norm": 9.25,
+      "learning_rate": 1.7042083852595509e-07,
+      "loss": 0.1954,
+      "step": 19540
+    },
+    {
+      "epoch": 1.6651896037494676,
+      "grad_norm": 21.375,
+      "learning_rate": 1.7033614675262726e-07,
+      "loss": 0.8048,
+      "step": 19541
+    },
+    {
+      "epoch": 1.6652748189177675,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.7025147448991087e-07,
+      "loss": 0.294,
+      "step": 19542
+    },
+    {
+      "epoch": 1.6653600340860675,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7016682173933652e-07,
+      "loss": 0.5275,
+      "step": 19543
+    },
+    {
+      "epoch": 1.6654452492543674,
+      "grad_norm": 18.75,
+      "learning_rate": 1.7008218850243342e-07,
+      "loss": 1.06,
+      "step": 19544
+    },
+    {
+      "epoch": 1.6655304644226674,
+      "grad_norm": 13.875,
+      "learning_rate": 1.6999757478073086e-07,
+      "loss": 0.5648,
+      "step": 19545
+    },
+    {
+      "epoch": 1.6656156795909673,
+      "grad_norm": 24.5,
+      "learning_rate": 1.6991298057575807e-07,
+      "loss": 0.497,
+      "step": 19546
+    },
+    {
+      "epoch": 1.6657008947592673,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.6982840588904397e-07,
+      "loss": 0.4411,
+      "step": 19547
+    },
+    {
+      "epoch": 1.6657861099275673,
+      "grad_norm": 16.375,
+      "learning_rate": 1.6974385072211705e-07,
+      "loss": 0.7338,
+      "step": 19548
+    },
+    {
+      "epoch": 1.6658713250958672,
+      "grad_norm": 12.875,
+      "learning_rate": 1.6965931507650468e-07,
+      "loss": 0.7609,
+      "step": 19549
+    },
+    {
+      "epoch": 1.6659565402641672,
+      "grad_norm": 15.25,
+      "learning_rate": 1.6957479895373482e-07,
+      "loss": 0.7155,
+      "step": 19550
+    },
+    {
+      "epoch": 1.6660417554324671,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.6949030235533488e-07,
+      "loss": 0.2808,
+      "step": 19551
+    },
+    {
+      "epoch": 1.666126970600767,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.6940582528283206e-07,
+      "loss": 0.669,
+      "step": 19552
+    },
+    {
+      "epoch": 1.666212185769067,
+      "grad_norm": 13.375,
+      "learning_rate": 1.693213677377524e-07,
+      "loss": 0.6048,
+      "step": 19553
+    },
+    {
+      "epoch": 1.666297400937367,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.6923692972162218e-07,
+      "loss": 0.5538,
+      "step": 19554
+    },
+    {
+      "epoch": 1.666382616105667,
+      "grad_norm": 20.375,
+      "learning_rate": 1.6915251123596737e-07,
+      "loss": 0.4744,
+      "step": 19555
+    },
+    {
+      "epoch": 1.666467831273967,
+      "grad_norm": 19.0,
+      "learning_rate": 1.6906811228231388e-07,
+      "loss": 0.815,
+      "step": 19556
+    },
+    {
+      "epoch": 1.6665530464422669,
+      "grad_norm": 13.625,
+      "learning_rate": 1.6898373286218627e-07,
+      "loss": 0.3131,
+      "step": 19557
+    },
+    {
+      "epoch": 1.6666382616105668,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6889937297710977e-07,
+      "loss": 0.3792,
+      "step": 19558
+    },
+    {
+      "epoch": 1.6667234767788668,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.688150326286085e-07,
+      "loss": 0.5057,
+      "step": 19559
+    },
+    {
+      "epoch": 1.6668086919471667,
+      "grad_norm": 13.5,
+      "learning_rate": 1.687307118182073e-07,
+      "loss": 0.4515,
+      "step": 19560
+    },
+    {
+      "epoch": 1.6668939071154667,
+      "grad_norm": 15.375,
+      "learning_rate": 1.6864641054742934e-07,
+      "loss": 0.4161,
+      "step": 19561
+    },
+    {
+      "epoch": 1.6669791222837667,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.6856212881779786e-07,
+      "loss": 0.307,
+      "step": 19562
+    },
+    {
+      "epoch": 1.6670643374520666,
+      "grad_norm": 15.875,
+      "learning_rate": 1.6847786663083607e-07,
+      "loss": 0.7221,
+      "step": 19563
+    },
+    {
+      "epoch": 1.6671495526203666,
+      "grad_norm": 17.625,
+      "learning_rate": 1.6839362398806695e-07,
+      "loss": 0.7561,
+      "step": 19564
+    },
+    {
+      "epoch": 1.6672347677886665,
+      "grad_norm": 41.25,
+      "learning_rate": 1.6830940089101285e-07,
+      "loss": 1.2363,
+      "step": 19565
+    },
+    {
+      "epoch": 1.6673199829569665,
+      "grad_norm": 23.125,
+      "learning_rate": 1.682251973411954e-07,
+      "loss": 0.7129,
+      "step": 19566
+    },
+    {
+      "epoch": 1.6674051981252664,
+      "grad_norm": 15.125,
+      "learning_rate": 1.681410133401365e-07,
+      "loss": 0.6204,
+      "step": 19567
+    },
+    {
+      "epoch": 1.6674904132935664,
+      "grad_norm": 18.25,
+      "learning_rate": 1.6805684888935718e-07,
+      "loss": 0.9542,
+      "step": 19568
+    },
+    {
+      "epoch": 1.6675756284618664,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.6797270399037932e-07,
+      "loss": 0.5049,
+      "step": 19569
+    },
+    {
+      "epoch": 1.6676608436301663,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6788857864472207e-07,
+      "loss": 0.6156,
+      "step": 19570
+    },
+    {
+      "epoch": 1.6677460587984663,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.6780447285390646e-07,
+      "loss": 0.4022,
+      "step": 19571
+    },
+    {
+      "epoch": 1.6678312739667662,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6772038661945226e-07,
+      "loss": 0.6107,
+      "step": 19572
+    },
+    {
+      "epoch": 1.6679164891350662,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.6763631994287917e-07,
+      "loss": 0.5795,
+      "step": 19573
+    },
+    {
+      "epoch": 1.6680017043033661,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.675522728257059e-07,
+      "loss": 0.6579,
+      "step": 19574
+    },
+    {
+      "epoch": 1.668086919471666,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6746824526945163e-07,
+      "loss": 0.4819,
+      "step": 19575
+    },
+    {
+      "epoch": 1.668172134639966,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.673842372756347e-07,
+      "loss": 0.3505,
+      "step": 19576
+    },
+    {
+      "epoch": 1.668257349808266,
+      "grad_norm": 14.25,
+      "learning_rate": 1.6730024884577354e-07,
+      "loss": 0.7329,
+      "step": 19577
+    },
+    {
+      "epoch": 1.668342564976566,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6721627998138562e-07,
+      "loss": 0.453,
+      "step": 19578
+    },
+    {
+      "epoch": 1.668427780144866,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.671323306839881e-07,
+      "loss": 0.4497,
+      "step": 19579
+    },
+    {
+      "epoch": 1.6685129953131659,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.6704840095509823e-07,
+      "loss": 0.425,
+      "step": 19580
+    },
+    {
+      "epoch": 1.6685982104814658,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.6696449079623282e-07,
+      "loss": 0.5173,
+      "step": 19581
+    },
+    {
+      "epoch": 1.6686834256497658,
+      "grad_norm": 19.25,
+      "learning_rate": 1.6688060020890837e-07,
+      "loss": 0.8994,
+      "step": 19582
+    },
+    {
+      "epoch": 1.6687686408180658,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6679672919464026e-07,
+      "loss": 0.5687,
+      "step": 19583
+    },
+    {
+      "epoch": 1.6688538559863657,
+      "grad_norm": 20.0,
+      "learning_rate": 1.6671287775494469e-07,
+      "loss": 0.4264,
+      "step": 19584
+    },
+    {
+      "epoch": 1.6689390711546657,
+      "grad_norm": 12.875,
+      "learning_rate": 1.666290458913368e-07,
+      "loss": 0.4121,
+      "step": 19585
+    },
+    {
+      "epoch": 1.6690242863229656,
+      "grad_norm": 19.125,
+      "learning_rate": 1.6654523360533166e-07,
+      "loss": 0.645,
+      "step": 19586
+    },
+    {
+      "epoch": 1.6691095014912656,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.664614408984433e-07,
+      "loss": 0.3593,
+      "step": 19587
+    },
+    {
+      "epoch": 1.6691947166595655,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.6637766777218622e-07,
+      "loss": 0.3309,
+      "step": 19588
+    },
+    {
+      "epoch": 1.6692799318278655,
+      "grad_norm": 20.875,
+      "learning_rate": 1.662939142280745e-07,
+      "loss": 0.6345,
+      "step": 19589
+    },
+    {
+      "epoch": 1.6693651469961654,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6621018026762164e-07,
+      "loss": 0.5042,
+      "step": 19590
+    },
+    {
+      "epoch": 1.6694503621644654,
+      "grad_norm": 12.25,
+      "learning_rate": 1.6612646589234046e-07,
+      "loss": 0.6791,
+      "step": 19591
+    },
+    {
+      "epoch": 1.6695355773327654,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6604277110374392e-07,
+      "loss": 0.6099,
+      "step": 19592
+    },
+    {
+      "epoch": 1.6696207925010653,
+      "grad_norm": 26.5,
+      "learning_rate": 1.659590959033447e-07,
+      "loss": 0.9235,
+      "step": 19593
+    },
+    {
+      "epoch": 1.6697060076693653,
+      "grad_norm": 16.5,
+      "learning_rate": 1.6587544029265462e-07,
+      "loss": 0.8553,
+      "step": 19594
+    },
+    {
+      "epoch": 1.6697912228376652,
+      "grad_norm": 21.5,
+      "learning_rate": 1.6579180427318555e-07,
+      "loss": 0.877,
+      "step": 19595
+    },
+    {
+      "epoch": 1.6698764380059652,
+      "grad_norm": 16.125,
+      "learning_rate": 1.657081878464488e-07,
+      "loss": 0.8051,
+      "step": 19596
+    },
+    {
+      "epoch": 1.6699616531742651,
+      "grad_norm": 20.0,
+      "learning_rate": 1.6562459101395535e-07,
+      "loss": 0.9766,
+      "step": 19597
+    },
+    {
+      "epoch": 1.670046868342565,
+      "grad_norm": 18.625,
+      "learning_rate": 1.6554101377721592e-07,
+      "loss": 0.8149,
+      "step": 19598
+    },
+    {
+      "epoch": 1.670132083510865,
+      "grad_norm": 21.625,
+      "learning_rate": 1.6545745613774117e-07,
+      "loss": 0.8536,
+      "step": 19599
+    },
+    {
+      "epoch": 1.670217298679165,
+      "grad_norm": 13.125,
+      "learning_rate": 1.6537391809704039e-07,
+      "loss": 0.4464,
+      "step": 19600
+    },
+    {
+      "epoch": 1.670302513847465,
+      "grad_norm": 32.25,
+      "learning_rate": 1.6529039965662408e-07,
+      "loss": 0.4111,
+      "step": 19601
+    },
+    {
+      "epoch": 1.670387729015765,
+      "grad_norm": 21.75,
+      "learning_rate": 1.652069008180006e-07,
+      "loss": 0.7956,
+      "step": 19602
+    },
+    {
+      "epoch": 1.670472944184065,
+      "grad_norm": 19.125,
+      "learning_rate": 1.6512342158267956e-07,
+      "loss": 0.5453,
+      "step": 19603
+    },
+    {
+      "epoch": 1.6705581593523648,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6503996195216892e-07,
+      "loss": 0.4224,
+      "step": 19604
+    },
+    {
+      "epoch": 1.6706433745206648,
+      "grad_norm": 11.875,
+      "learning_rate": 1.6495652192797724e-07,
+      "loss": 0.3406,
+      "step": 19605
+    },
+    {
+      "epoch": 1.6707285896889648,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.6487310151161229e-07,
+      "loss": 0.5211,
+      "step": 19606
+    },
+    {
+      "epoch": 1.6708138048572647,
+      "grad_norm": 19.875,
+      "learning_rate": 1.6478970070458178e-07,
+      "loss": 0.6165,
+      "step": 19607
+    },
+    {
+      "epoch": 1.6708990200255647,
+      "grad_norm": 16.125,
+      "learning_rate": 1.6470631950839242e-07,
+      "loss": 0.8308,
+      "step": 19608
+    },
+    {
+      "epoch": 1.6709842351938646,
+      "grad_norm": 29.375,
+      "learning_rate": 1.646229579245516e-07,
+      "loss": 0.5105,
+      "step": 19609
+    },
+    {
+      "epoch": 1.6710694503621646,
+      "grad_norm": 17.875,
+      "learning_rate": 1.6453961595456494e-07,
+      "loss": 0.4058,
+      "step": 19610
+    },
+    {
+      "epoch": 1.6711546655304645,
+      "grad_norm": 20.875,
+      "learning_rate": 1.6445629359993901e-07,
+      "loss": 1.0022,
+      "step": 19611
+    },
+    {
+      "epoch": 1.6712398806987645,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.6437299086217967e-07,
+      "loss": 0.5625,
+      "step": 19612
+    },
+    {
+      "epoch": 1.6713250958670645,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.6428970774279185e-07,
+      "loss": 0.4387,
+      "step": 19613
+    },
+    {
+      "epoch": 1.6714103110353644,
+      "grad_norm": 22.25,
+      "learning_rate": 1.6420644424328073e-07,
+      "loss": 0.8694,
+      "step": 19614
+    },
+    {
+      "epoch": 1.6714955262036644,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.6412320036515093e-07,
+      "loss": 0.4558,
+      "step": 19615
+    },
+    {
+      "epoch": 1.6715807413719643,
+      "grad_norm": 12.875,
+      "learning_rate": 1.640399761099072e-07,
+      "loss": 0.4908,
+      "step": 19616
+    },
+    {
+      "epoch": 1.6716659565402643,
+      "grad_norm": 20.875,
+      "learning_rate": 1.6395677147905312e-07,
+      "loss": 1.0584,
+      "step": 19617
+    },
+    {
+      "epoch": 1.6717511717085642,
+      "grad_norm": 20.875,
+      "learning_rate": 1.6387358647409203e-07,
+      "loss": 0.857,
+      "step": 19618
+    },
+    {
+      "epoch": 1.6718363868768642,
+      "grad_norm": 12.25,
+      "learning_rate": 1.6379042109652743e-07,
+      "loss": 0.4299,
+      "step": 19619
+    },
+    {
+      "epoch": 1.6719216020451642,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.6370727534786247e-07,
+      "loss": 0.4542,
+      "step": 19620
+    },
+    {
+      "epoch": 1.6720068172134641,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.6362414922959897e-07,
+      "loss": 0.3407,
+      "step": 19621
+    },
+    {
+      "epoch": 1.672092032381764,
+      "grad_norm": 41.75,
+      "learning_rate": 1.6354104274323977e-07,
+      "loss": 0.7355,
+      "step": 19622
+    },
+    {
+      "epoch": 1.672177247550064,
+      "grad_norm": 20.75,
+      "learning_rate": 1.634579558902863e-07,
+      "loss": 0.8174,
+      "step": 19623
+    },
+    {
+      "epoch": 1.672262462718364,
+      "grad_norm": 16.5,
+      "learning_rate": 1.633748886722404e-07,
+      "loss": 0.83,
+      "step": 19624
+    },
+    {
+      "epoch": 1.672347677886664,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.6329184109060273e-07,
+      "loss": 0.3509,
+      "step": 19625
+    },
+    {
+      "epoch": 1.672432893054964,
+      "grad_norm": 17.5,
+      "learning_rate": 1.6320881314687453e-07,
+      "loss": 0.7026,
+      "step": 19626
+    },
+    {
+      "epoch": 1.6725181082232639,
+      "grad_norm": 19.875,
+      "learning_rate": 1.6312580484255562e-07,
+      "loss": 0.9518,
+      "step": 19627
+    },
+    {
+      "epoch": 1.6726033233915638,
+      "grad_norm": 14.0,
+      "learning_rate": 1.6304281617914642e-07,
+      "loss": 0.3117,
+      "step": 19628
+    },
+    {
+      "epoch": 1.6726885385598638,
+      "grad_norm": 13.75,
+      "learning_rate": 1.6295984715814687e-07,
+      "loss": 0.4702,
+      "step": 19629
+    },
+    {
+      "epoch": 1.6727737537281637,
+      "grad_norm": 16.375,
+      "learning_rate": 1.6287689778105563e-07,
+      "loss": 0.484,
+      "step": 19630
+    },
+    {
+      "epoch": 1.6728589688964637,
+      "grad_norm": 18.125,
+      "learning_rate": 1.6279396804937203e-07,
+      "loss": 0.8856,
+      "step": 19631
+    },
+    {
+      "epoch": 1.6729441840647636,
+      "grad_norm": 11.875,
+      "learning_rate": 1.6271105796459464e-07,
+      "loss": 0.3434,
+      "step": 19632
+    },
+    {
+      "epoch": 1.6730293992330636,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.6262816752822223e-07,
+      "loss": 0.5004,
+      "step": 19633
+    },
+    {
+      "epoch": 1.6731146144013636,
+      "grad_norm": 21.125,
+      "learning_rate": 1.6254529674175206e-07,
+      "loss": 1.0232,
+      "step": 19634
+    },
+    {
+      "epoch": 1.6731998295696635,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6246244560668186e-07,
+      "loss": 0.5048,
+      "step": 19635
+    },
+    {
+      "epoch": 1.6732850447379635,
+      "grad_norm": 18.25,
+      "learning_rate": 1.6237961412450877e-07,
+      "loss": 0.9212,
+      "step": 19636
+    },
+    {
+      "epoch": 1.6733702599062634,
+      "grad_norm": 14.75,
+      "learning_rate": 1.6229680229672987e-07,
+      "loss": 0.6671,
+      "step": 19637
+    },
+    {
+      "epoch": 1.6734554750745634,
+      "grad_norm": 13.25,
+      "learning_rate": 1.6221401012484136e-07,
+      "loss": 0.5043,
+      "step": 19638
+    },
+    {
+      "epoch": 1.6735406902428633,
+      "grad_norm": 12.875,
+      "learning_rate": 1.6213123761033957e-07,
+      "loss": 0.6146,
+      "step": 19639
+    },
+    {
+      "epoch": 1.6736259054111633,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.620484847547203e-07,
+      "loss": 0.4952,
+      "step": 19640
+    },
+    {
+      "epoch": 1.6737111205794633,
+      "grad_norm": 26.125,
+      "learning_rate": 1.6196575155947914e-07,
+      "loss": 0.9596,
+      "step": 19641
+    },
+    {
+      "epoch": 1.6737963357477632,
+      "grad_norm": 19.25,
+      "learning_rate": 1.618830380261109e-07,
+      "loss": 0.6887,
+      "step": 19642
+    },
+    {
+      "epoch": 1.6738815509160632,
+      "grad_norm": 26.25,
+      "learning_rate": 1.6180034415611014e-07,
+      "loss": 0.6722,
+      "step": 19643
+    },
+    {
+      "epoch": 1.6739667660843631,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.617176699509715e-07,
+      "loss": 0.4559,
+      "step": 19644
+    },
+    {
+      "epoch": 1.674051981252663,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.6163501541218874e-07,
+      "loss": 0.4614,
+      "step": 19645
+    },
+    {
+      "epoch": 1.674137196420963,
+      "grad_norm": 15.25,
+      "learning_rate": 1.615523805412561e-07,
+      "loss": 0.64,
+      "step": 19646
+    },
+    {
+      "epoch": 1.674222411589263,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.6146976533966616e-07,
+      "loss": 0.5161,
+      "step": 19647
+    },
+    {
+      "epoch": 1.674307626757563,
+      "grad_norm": 23.375,
+      "learning_rate": 1.6138716980891214e-07,
+      "loss": 0.9481,
+      "step": 19648
+    },
+    {
+      "epoch": 1.674392841925863,
+      "grad_norm": 27.5,
+      "learning_rate": 1.6130459395048688e-07,
+      "loss": 0.8486,
+      "step": 19649
+    },
+    {
+      "epoch": 1.6744780570941629,
+      "grad_norm": 19.0,
+      "learning_rate": 1.6122203776588241e-07,
+      "loss": 0.6844,
+      "step": 19650
+    },
+    {
+      "epoch": 1.6745632722624628,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6113950125659028e-07,
+      "loss": 0.5532,
+      "step": 19651
+    },
+    {
+      "epoch": 1.6746484874307628,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.6105698442410236e-07,
+      "loss": 0.3066,
+      "step": 19652
+    },
+    {
+      "epoch": 1.6747337025990627,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.6097448726990968e-07,
+      "loss": 0.7653,
+      "step": 19653
+    },
+    {
+      "epoch": 1.6748189177673627,
+      "grad_norm": 15.25,
+      "learning_rate": 1.6089200979550342e-07,
+      "loss": 0.7557,
+      "step": 19654
+    },
+    {
+      "epoch": 1.6749041329356626,
+      "grad_norm": 10.5,
+      "learning_rate": 1.608095520023735e-07,
+      "loss": 0.1761,
+      "step": 19655
+    },
+    {
+      "epoch": 1.6749893481039626,
+      "grad_norm": 21.375,
+      "learning_rate": 1.6072711389201023e-07,
+      "loss": 0.8089,
+      "step": 19656
+    },
+    {
+      "epoch": 1.6750745632722626,
+      "grad_norm": 11.0,
+      "learning_rate": 1.6064469546590327e-07,
+      "loss": 0.3148,
+      "step": 19657
+    },
+    {
+      "epoch": 1.6751597784405625,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6056229672554242e-07,
+      "loss": 0.6035,
+      "step": 19658
+    },
+    {
+      "epoch": 1.6752449936088625,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.6047991767241645e-07,
+      "loss": 0.4517,
+      "step": 19659
+    },
+    {
+      "epoch": 1.6753302087771624,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6039755830801362e-07,
+      "loss": 0.4029,
+      "step": 19660
+    },
+    {
+      "epoch": 1.6754154239454624,
+      "grad_norm": 14.875,
+      "learning_rate": 1.6031521863382249e-07,
+      "loss": 0.5574,
+      "step": 19661
+    },
+    {
+      "epoch": 1.6755006391137623,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.6023289865133128e-07,
+      "loss": 0.412,
+      "step": 19662
+    },
+    {
+      "epoch": 1.6755858542820623,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6015059836202759e-07,
+      "loss": 0.3952,
+      "step": 19663
+    },
+    {
+      "epoch": 1.6756710694503623,
+      "grad_norm": 15.125,
+      "learning_rate": 1.600683177673984e-07,
+      "loss": 0.5004,
+      "step": 19664
+    },
+    {
+      "epoch": 1.6757562846186622,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5998605686893055e-07,
+      "loss": 0.7931,
+      "step": 19665
+    },
+    {
+      "epoch": 1.6758414997869622,
+      "grad_norm": 18.75,
+      "learning_rate": 1.5990381566811112e-07,
+      "loss": 0.6097,
+      "step": 19666
+    },
+    {
+      "epoch": 1.6759267149552621,
+      "grad_norm": 19.0,
+      "learning_rate": 1.5982159416642593e-07,
+      "loss": 0.7413,
+      "step": 19667
+    },
+    {
+      "epoch": 1.676011930123562,
+      "grad_norm": 23.125,
+      "learning_rate": 1.5973939236536048e-07,
+      "loss": 0.9017,
+      "step": 19668
+    },
+    {
+      "epoch": 1.676097145291862,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.5965721026640055e-07,
+      "loss": 0.4956,
+      "step": 19669
+    },
+    {
+      "epoch": 1.676182360460162,
+      "grad_norm": 20.625,
+      "learning_rate": 1.595750478710313e-07,
+      "loss": 0.7898,
+      "step": 19670
+    },
+    {
+      "epoch": 1.676267575628462,
+      "grad_norm": 20.25,
+      "learning_rate": 1.594929051807377e-07,
+      "loss": 1.1608,
+      "step": 19671
+    },
+    {
+      "epoch": 1.676352790796762,
+      "grad_norm": 17.5,
+      "learning_rate": 1.5941078219700353e-07,
+      "loss": 0.6206,
+      "step": 19672
+    },
+    {
+      "epoch": 1.6764380059650619,
+      "grad_norm": 15.75,
+      "learning_rate": 1.5932867892131332e-07,
+      "loss": 0.51,
+      "step": 19673
+    },
+    {
+      "epoch": 1.6765232211333618,
+      "grad_norm": 25.375,
+      "learning_rate": 1.5924659535515087e-07,
+      "loss": 0.9831,
+      "step": 19674
+    },
+    {
+      "epoch": 1.6766084363016618,
+      "grad_norm": 19.0,
+      "learning_rate": 1.59164531499999e-07,
+      "loss": 0.7045,
+      "step": 19675
+    },
+    {
+      "epoch": 1.6766936514699617,
+      "grad_norm": 28.125,
+      "learning_rate": 1.5908248735734114e-07,
+      "loss": 0.5495,
+      "step": 19676
+    },
+    {
+      "epoch": 1.6767788666382617,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5900046292865957e-07,
+      "loss": 0.4374,
+      "step": 19677
+    },
+    {
+      "epoch": 1.6768640818065617,
+      "grad_norm": 17.125,
+      "learning_rate": 1.589184582154367e-07,
+      "loss": 0.8149,
+      "step": 19678
+    },
+    {
+      "epoch": 1.6769492969748616,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.588364732191547e-07,
+      "loss": 0.4686,
+      "step": 19679
+    },
+    {
+      "epoch": 1.6770345121431616,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5875450794129472e-07,
+      "loss": 0.677,
+      "step": 19680
+    },
+    {
+      "epoch": 1.6771197273114615,
+      "grad_norm": 11.125,
+      "learning_rate": 1.5867256238333795e-07,
+      "loss": 0.2074,
+      "step": 19681
+    },
+    {
+      "epoch": 1.6772049424797615,
+      "grad_norm": 24.25,
+      "learning_rate": 1.5859063654676585e-07,
+      "loss": 0.997,
+      "step": 19682
+    },
+    {
+      "epoch": 1.6772901576480614,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.5850873043305807e-07,
+      "loss": 0.2566,
+      "step": 19683
+    },
+    {
+      "epoch": 1.6773753728163614,
+      "grad_norm": 14.875,
+      "learning_rate": 1.5842684404369534e-07,
+      "loss": 0.4794,
+      "step": 19684
+    },
+    {
+      "epoch": 1.6774605879846614,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5834497738015694e-07,
+      "loss": 0.3119,
+      "step": 19685
+    },
+    {
+      "epoch": 1.6775458031529613,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.5826313044392262e-07,
+      "loss": 0.5822,
+      "step": 19686
+    },
+    {
+      "epoch": 1.6776310183212613,
+      "grad_norm": 22.875,
+      "learning_rate": 1.5818130323647124e-07,
+      "loss": 0.8634,
+      "step": 19687
+    },
+    {
+      "epoch": 1.6777162334895612,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.58099495759282e-07,
+      "loss": 0.6015,
+      "step": 19688
+    },
+    {
+      "epoch": 1.6778014486578612,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.5801770801383258e-07,
+      "loss": 0.6848,
+      "step": 19689
+    },
+    {
+      "epoch": 1.6778866638261611,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.5793594000160117e-07,
+      "loss": 0.4451,
+      "step": 19690
+    },
+    {
+      "epoch": 1.677971878994461,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.5785419172406585e-07,
+      "loss": 0.7328,
+      "step": 19691
+    },
+    {
+      "epoch": 1.678057094162761,
+      "grad_norm": 20.0,
+      "learning_rate": 1.5777246318270324e-07,
+      "loss": 0.8237,
+      "step": 19692
+    },
+    {
+      "epoch": 1.678142309331061,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.5769075437899072e-07,
+      "loss": 0.4048,
+      "step": 19693
+    },
+    {
+      "epoch": 1.678227524499361,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.576090653144044e-07,
+      "loss": 0.5528,
+      "step": 19694
+    },
+    {
+      "epoch": 1.678312739667661,
+      "grad_norm": 23.0,
+      "learning_rate": 1.5752739599042084e-07,
+      "loss": 1.0978,
+      "step": 19695
+    },
+    {
+      "epoch": 1.6783979548359609,
+      "grad_norm": 32.25,
+      "learning_rate": 1.5744574640851607e-07,
+      "loss": 0.7942,
+      "step": 19696
+    },
+    {
+      "epoch": 1.6784831700042608,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.5736411657016504e-07,
+      "loss": 0.4339,
+      "step": 19697
+    },
+    {
+      "epoch": 1.6785683851725608,
+      "grad_norm": 19.25,
+      "learning_rate": 1.572825064768431e-07,
+      "loss": 0.9457,
+      "step": 19698
+    },
+    {
+      "epoch": 1.6786536003408608,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.572009161300253e-07,
+      "loss": 0.714,
+      "step": 19699
+    },
+    {
+      "epoch": 1.6787388155091607,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5711934553118577e-07,
+      "loss": 0.595,
+      "step": 19700
+    },
+    {
+      "epoch": 1.6788240306774607,
+      "grad_norm": 22.75,
+      "learning_rate": 1.570377946817987e-07,
+      "loss": 0.7886,
+      "step": 19701
+    },
+    {
+      "epoch": 1.6789092458457606,
+      "grad_norm": 20.125,
+      "learning_rate": 1.5695626358333758e-07,
+      "loss": 0.7049,
+      "step": 19702
+    },
+    {
+      "epoch": 1.6789944610140606,
+      "grad_norm": 11.75,
+      "learning_rate": 1.5687475223727585e-07,
+      "loss": 0.5709,
+      "step": 19703
+    },
+    {
+      "epoch": 1.6790796761823605,
+      "grad_norm": 13.5,
+      "learning_rate": 1.567932606450867e-07,
+      "loss": 0.5358,
+      "step": 19704
+    },
+    {
+      "epoch": 1.6791648913506605,
+      "grad_norm": 12.875,
+      "learning_rate": 1.5671178880824284e-07,
+      "loss": 0.2842,
+      "step": 19705
+    },
+    {
+      "epoch": 1.6792501065189605,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.5663033672821598e-07,
+      "loss": 0.5186,
+      "step": 19706
+    },
+    {
+      "epoch": 1.6793353216872604,
+      "grad_norm": 17.375,
+      "learning_rate": 1.565489044064787e-07,
+      "loss": 0.9706,
+      "step": 19707
+    },
+    {
+      "epoch": 1.6794205368555604,
+      "grad_norm": 14.375,
+      "learning_rate": 1.5646749184450207e-07,
+      "loss": 0.6231,
+      "step": 19708
+    },
+    {
+      "epoch": 1.6795057520238603,
+      "grad_norm": 18.875,
+      "learning_rate": 1.5638609904375782e-07,
+      "loss": 0.7224,
+      "step": 19709
+    },
+    {
+      "epoch": 1.6795909671921603,
+      "grad_norm": 18.25,
+      "learning_rate": 1.56304726005716e-07,
+      "loss": 0.5201,
+      "step": 19710
+    },
+    {
+      "epoch": 1.6796761823604602,
+      "grad_norm": 17.625,
+      "learning_rate": 1.562233727318478e-07,
+      "loss": 0.5264,
+      "step": 19711
+    },
+    {
+      "epoch": 1.6797613975287602,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5614203922362292e-07,
+      "loss": 0.4242,
+      "step": 19712
+    },
+    {
+      "epoch": 1.6798466126970601,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.560607254825118e-07,
+      "loss": 0.5372,
+      "step": 19713
+    },
+    {
+      "epoch": 1.67993182786536,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.5597943150998302e-07,
+      "loss": 0.3765,
+      "step": 19714
+    },
+    {
+      "epoch": 1.68001704303366,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5589815730750634e-07,
+      "loss": 0.4476,
+      "step": 19715
+    },
+    {
+      "epoch": 1.68010225820196,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5581690287654994e-07,
+      "loss": 0.5117,
+      "step": 19716
+    },
+    {
+      "epoch": 1.68018747337026,
+      "grad_norm": 16.875,
+      "learning_rate": 1.5573566821858233e-07,
+      "loss": 0.7862,
+      "step": 19717
+    },
+    {
+      "epoch": 1.68027268853856,
+      "grad_norm": 18.875,
+      "learning_rate": 1.5565445333507191e-07,
+      "loss": 0.4951,
+      "step": 19718
+    },
+    {
+      "epoch": 1.68035790370686,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.5557325822748562e-07,
+      "loss": 0.3699,
+      "step": 19719
+    },
+    {
+      "epoch": 1.6804431188751598,
+      "grad_norm": 14.25,
+      "learning_rate": 1.5549208289729112e-07,
+      "loss": 0.9187,
+      "step": 19720
+    },
+    {
+      "epoch": 1.6805283340434598,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5541092734595532e-07,
+      "loss": 0.6602,
+      "step": 19721
+    },
+    {
+      "epoch": 1.6806135492117598,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5532979157494493e-07,
+      "loss": 0.5393,
+      "step": 19722
+    },
+    {
+      "epoch": 1.6806987643800597,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.5524867558572617e-07,
+      "loss": 0.4877,
+      "step": 19723
+    },
+    {
+      "epoch": 1.6807839795483597,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5516757937976427e-07,
+      "loss": 0.5432,
+      "step": 19724
+    },
+    {
+      "epoch": 1.6808691947166596,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5508650295852523e-07,
+      "loss": 0.6153,
+      "step": 19725
+    },
+    {
+      "epoch": 1.6809544098849596,
+      "grad_norm": 30.125,
+      "learning_rate": 1.5500544632347432e-07,
+      "loss": 1.0579,
+      "step": 19726
+    },
+    {
+      "epoch": 1.6810396250532595,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.5492440947607573e-07,
+      "loss": 0.5595,
+      "step": 19727
+    },
+    {
+      "epoch": 1.6811248402215595,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.5484339241779445e-07,
+      "loss": 0.6792,
+      "step": 19728
+    },
+    {
+      "epoch": 1.6812100553898595,
+      "grad_norm": 19.75,
+      "learning_rate": 1.5476239515009413e-07,
+      "loss": 0.8399,
+      "step": 19729
+    },
+    {
+      "epoch": 1.6812952705581594,
+      "grad_norm": 14.375,
+      "learning_rate": 1.546814176744392e-07,
+      "loss": 0.6664,
+      "step": 19730
+    },
+    {
+      "epoch": 1.6813804857264594,
+      "grad_norm": 18.25,
+      "learning_rate": 1.546004599922922e-07,
+      "loss": 0.9092,
+      "step": 19731
+    },
+    {
+      "epoch": 1.6814657008947593,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5451952210511628e-07,
+      "loss": 0.8535,
+      "step": 19732
+    },
+    {
+      "epoch": 1.6815509160630593,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5443860401437417e-07,
+      "loss": 0.4278,
+      "step": 19733
+    },
+    {
+      "epoch": 1.6816361312313592,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.5435770572152813e-07,
+      "loss": 0.6655,
+      "step": 19734
+    },
+    {
+      "epoch": 1.6817213463996592,
+      "grad_norm": 20.375,
+      "learning_rate": 1.5427682722804052e-07,
+      "loss": 0.5804,
+      "step": 19735
+    },
+    {
+      "epoch": 1.6818065615679592,
+      "grad_norm": 11.625,
+      "learning_rate": 1.5419596853537207e-07,
+      "loss": 0.2571,
+      "step": 19736
+    },
+    {
+      "epoch": 1.6818917767362591,
+      "grad_norm": 13.375,
+      "learning_rate": 1.541151296449843e-07,
+      "loss": 0.7082,
+      "step": 19737
+    },
+    {
+      "epoch": 1.681976991904559,
+      "grad_norm": 19.0,
+      "learning_rate": 1.5403431055833834e-07,
+      "loss": 0.7951,
+      "step": 19738
+    },
+    {
+      "epoch": 1.682062207072859,
+      "grad_norm": 15.75,
+      "learning_rate": 1.5395351127689486e-07,
+      "loss": 0.6159,
+      "step": 19739
+    },
+    {
+      "epoch": 1.682147422241159,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5387273180211312e-07,
+      "loss": 0.3712,
+      "step": 19740
+    },
+    {
+      "epoch": 1.682232637409459,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5379197213545334e-07,
+      "loss": 0.2122,
+      "step": 19741
+    },
+    {
+      "epoch": 1.682317852577759,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5371123227837488e-07,
+      "loss": 0.9449,
+      "step": 19742
+    },
+    {
+      "epoch": 1.6824030677460589,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5363051223233712e-07,
+      "loss": 0.8562,
+      "step": 19743
+    },
+    {
+      "epoch": 1.6824882829143588,
+      "grad_norm": 11.125,
+      "learning_rate": 1.535498119987984e-07,
+      "loss": 0.3278,
+      "step": 19744
+    },
+    {
+      "epoch": 1.6825734980826588,
+      "grad_norm": 16.5,
+      "learning_rate": 1.5346913157921694e-07,
+      "loss": 0.6628,
+      "step": 19745
+    },
+    {
+      "epoch": 1.6826587132509587,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.5338847097505105e-07,
+      "loss": 0.4233,
+      "step": 19746
+    },
+    {
+      "epoch": 1.6827439284192587,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.5330783018775844e-07,
+      "loss": 0.2306,
+      "step": 19747
+    },
+    {
+      "epoch": 1.6828291435875586,
+      "grad_norm": 12.375,
+      "learning_rate": 1.5322720921879614e-07,
+      "loss": 0.2267,
+      "step": 19748
+    },
+    {
+      "epoch": 1.6829143587558586,
+      "grad_norm": 21.75,
+      "learning_rate": 1.5314660806962076e-07,
+      "loss": 1.1723,
+      "step": 19749
+    },
+    {
+      "epoch": 1.6829995739241586,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.530660267416892e-07,
+      "loss": 0.3432,
+      "step": 19750
+    },
+    {
+      "epoch": 1.6830847890924585,
+      "grad_norm": 12.75,
+      "learning_rate": 1.5298546523645762e-07,
+      "loss": 0.4834,
+      "step": 19751
+    },
+    {
+      "epoch": 1.6831700042607585,
+      "grad_norm": 16.625,
+      "learning_rate": 1.52904923555382e-07,
+      "loss": 0.4959,
+      "step": 19752
+    },
+    {
+      "epoch": 1.6832552194290584,
+      "grad_norm": 21.75,
+      "learning_rate": 1.528244016999174e-07,
+      "loss": 0.6294,
+      "step": 19753
+    },
+    {
+      "epoch": 1.6833404345973584,
+      "grad_norm": 17.25,
+      "learning_rate": 1.5274389967151916e-07,
+      "loss": 0.6183,
+      "step": 19754
+    },
+    {
+      "epoch": 1.6834256497656583,
+      "grad_norm": 17.625,
+      "learning_rate": 1.5266341747164224e-07,
+      "loss": 0.6248,
+      "step": 19755
+    },
+    {
+      "epoch": 1.6835108649339583,
+      "grad_norm": 18.375,
+      "learning_rate": 1.525829551017409e-07,
+      "loss": 0.6868,
+      "step": 19756
+    },
+    {
+      "epoch": 1.6835960801022583,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5250251256326884e-07,
+      "loss": 0.5363,
+      "step": 19757
+    },
+    {
+      "epoch": 1.6836812952705582,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.524220898576799e-07,
+      "loss": 0.333,
+      "step": 19758
+    },
+    {
+      "epoch": 1.6837665104388582,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.523416869864275e-07,
+      "loss": 0.7298,
+      "step": 19759
+    },
+    {
+      "epoch": 1.6838517256071581,
+      "grad_norm": 15.5,
+      "learning_rate": 1.5226130395096503e-07,
+      "loss": 0.5665,
+      "step": 19760
+    },
+    {
+      "epoch": 1.683936940775458,
+      "grad_norm": 16.875,
+      "learning_rate": 1.5218094075274428e-07,
+      "loss": 0.7613,
+      "step": 19761
+    },
+    {
+      "epoch": 1.684022155943758,
+      "grad_norm": 22.625,
+      "learning_rate": 1.5210059739321783e-07,
+      "loss": 0.7901,
+      "step": 19762
+    },
+    {
+      "epoch": 1.684107371112058,
+      "grad_norm": 18.25,
+      "learning_rate": 1.520202738738376e-07,
+      "loss": 0.6076,
+      "step": 19763
+    },
+    {
+      "epoch": 1.684192586280358,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.519399701960554e-07,
+      "loss": 0.4777,
+      "step": 19764
+    },
+    {
+      "epoch": 1.684277801448658,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5185968636132227e-07,
+      "loss": 0.4876,
+      "step": 19765
+    },
+    {
+      "epoch": 1.6843630166169579,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5177942237108846e-07,
+      "loss": 0.5104,
+      "step": 19766
+    },
+    {
+      "epoch": 1.6844482317852578,
+      "grad_norm": 14.625,
+      "learning_rate": 1.5169917822680485e-07,
+      "loss": 0.5821,
+      "step": 19767
+    },
+    {
+      "epoch": 1.6845334469535578,
+      "grad_norm": 29.25,
+      "learning_rate": 1.5161895392992162e-07,
+      "loss": 0.5703,
+      "step": 19768
+    },
+    {
+      "epoch": 1.6846186621218577,
+      "grad_norm": 10.5,
+      "learning_rate": 1.515387494818886e-07,
+      "loss": 0.3133,
+      "step": 19769
+    },
+    {
+      "epoch": 1.6847038772901577,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.514585648841546e-07,
+      "loss": 0.5234,
+      "step": 19770
+    },
+    {
+      "epoch": 1.6847890924584576,
+      "grad_norm": 17.375,
+      "learning_rate": 1.513784001381692e-07,
+      "loss": 0.847,
+      "step": 19771
+    },
+    {
+      "epoch": 1.6848743076267576,
+      "grad_norm": 19.625,
+      "learning_rate": 1.5129825524538095e-07,
+      "loss": 1.0173,
+      "step": 19772
+    },
+    {
+      "epoch": 1.6849595227950576,
+      "grad_norm": 20.5,
+      "learning_rate": 1.5121813020723812e-07,
+      "loss": 0.675,
+      "step": 19773
+    },
+    {
+      "epoch": 1.6850447379633575,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.511380250251883e-07,
+      "loss": 0.456,
+      "step": 19774
+    },
+    {
+      "epoch": 1.6851299531316575,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5105793970067936e-07,
+      "loss": 0.4614,
+      "step": 19775
+    },
+    {
+      "epoch": 1.6852151682999574,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.5097787423515864e-07,
+      "loss": 0.4237,
+      "step": 19776
+    },
+    {
+      "epoch": 1.6853003834682574,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.5089782863007302e-07,
+      "loss": 0.4394,
+      "step": 19777
+    },
+    {
+      "epoch": 1.6853855986365573,
+      "grad_norm": 14.875,
+      "learning_rate": 1.508178028868687e-07,
+      "loss": 0.5848,
+      "step": 19778
+    },
+    {
+      "epoch": 1.6854708138048573,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.507377970069919e-07,
+      "loss": 0.405,
+      "step": 19779
+    },
+    {
+      "epoch": 1.6855560289731573,
+      "grad_norm": 17.75,
+      "learning_rate": 1.5065781099188881e-07,
+      "loss": 0.4306,
+      "step": 19780
+    },
+    {
+      "epoch": 1.6856412441414572,
+      "grad_norm": 24.0,
+      "learning_rate": 1.5057784484300425e-07,
+      "loss": 0.7947,
+      "step": 19781
+    },
+    {
+      "epoch": 1.6857264593097572,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5049789856178388e-07,
+      "loss": 0.7904,
+      "step": 19782
+    },
+    {
+      "epoch": 1.6858116744780571,
+      "grad_norm": 17.375,
+      "learning_rate": 1.504179721496718e-07,
+      "loss": 0.7298,
+      "step": 19783
+    },
+    {
+      "epoch": 1.685896889646357,
+      "grad_norm": 23.5,
+      "learning_rate": 1.503380656081127e-07,
+      "loss": 0.94,
+      "step": 19784
+    },
+    {
+      "epoch": 1.685982104814657,
+      "grad_norm": 14.0,
+      "learning_rate": 1.502581789385507e-07,
+      "loss": 0.3489,
+      "step": 19785
+    },
+    {
+      "epoch": 1.686067319982957,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.501783121424294e-07,
+      "loss": 0.6618,
+      "step": 19786
+    },
+    {
+      "epoch": 1.686152535151257,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5009846522119177e-07,
+      "loss": 0.3954,
+      "step": 19787
+    },
+    {
+      "epoch": 1.686237750319557,
+      "grad_norm": 26.25,
+      "learning_rate": 1.500186381762811e-07,
+      "loss": 0.6644,
+      "step": 19788
+    },
+    {
+      "epoch": 1.6863229654878569,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4993883100913946e-07,
+      "loss": 0.5749,
+      "step": 19789
+    },
+    {
+      "epoch": 1.6864081806561568,
+      "grad_norm": 18.0,
+      "learning_rate": 1.4985904372120969e-07,
+      "loss": 0.8556,
+      "step": 19790
+    },
+    {
+      "epoch": 1.6864933958244568,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.4977927631393302e-07,
+      "loss": 0.5212,
+      "step": 19791
+    },
+    {
+      "epoch": 1.6865786109927567,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4969952878875108e-07,
+      "loss": 0.843,
+      "step": 19792
+    },
+    {
+      "epoch": 1.6866638261610567,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.496198011471052e-07,
+      "loss": 0.3448,
+      "step": 19793
+    },
+    {
+      "epoch": 1.6867490413293567,
+      "grad_norm": 21.875,
+      "learning_rate": 1.4954009339043618e-07,
+      "loss": 0.7323,
+      "step": 19794
+    },
+    {
+      "epoch": 1.6868342564976566,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.494604055201841e-07,
+      "loss": 0.5456,
+      "step": 19795
+    },
+    {
+      "epoch": 1.6869194716659566,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.493807375377894e-07,
+      "loss": 0.6925,
+      "step": 19796
+    },
+    {
+      "epoch": 1.6870046868342565,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4930108944469126e-07,
+      "loss": 0.5471,
+      "step": 19797
+    },
+    {
+      "epoch": 1.6870899020025565,
+      "grad_norm": 25.0,
+      "learning_rate": 1.4922146124232929e-07,
+      "loss": 0.8355,
+      "step": 19798
+    },
+    {
+      "epoch": 1.6871751171708564,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.491418529321427e-07,
+      "loss": 0.4994,
+      "step": 19799
+    },
+    {
+      "epoch": 1.6872603323391564,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4906226451556968e-07,
+      "loss": 0.3923,
+      "step": 19800
+    },
+    {
+      "epoch": 1.6873455475074564,
+      "grad_norm": 17.625,
+      "learning_rate": 1.489826959940484e-07,
+      "loss": 1.0337,
+      "step": 19801
+    },
+    {
+      "epoch": 1.6874307626757563,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4890314736901737e-07,
+      "loss": 0.7032,
+      "step": 19802
+    },
+    {
+      "epoch": 1.6875159778440563,
+      "grad_norm": 16.875,
+      "learning_rate": 1.488236186419134e-07,
+      "loss": 0.7413,
+      "step": 19803
+    },
+    {
+      "epoch": 1.6876011930123562,
+      "grad_norm": 19.5,
+      "learning_rate": 1.4874410981417426e-07,
+      "loss": 0.7288,
+      "step": 19804
+    },
+    {
+      "epoch": 1.6876864081806562,
+      "grad_norm": 16.0,
+      "learning_rate": 1.486646208872361e-07,
+      "loss": 0.6496,
+      "step": 19805
+    },
+    {
+      "epoch": 1.6877716233489561,
+      "grad_norm": 17.25,
+      "learning_rate": 1.4858515186253587e-07,
+      "loss": 0.5152,
+      "step": 19806
+    },
+    {
+      "epoch": 1.687856838517256,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.485057027415096e-07,
+      "loss": 0.3354,
+      "step": 19807
+    },
+    {
+      "epoch": 1.687942053685556,
+      "grad_norm": 19.5,
+      "learning_rate": 1.4842627352559272e-07,
+      "loss": 0.8337,
+      "step": 19808
+    },
+    {
+      "epoch": 1.688027268853856,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4834686421622086e-07,
+      "loss": 0.5229,
+      "step": 19809
+    },
+    {
+      "epoch": 1.688112484022156,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4826747481482885e-07,
+      "loss": 0.6045,
+      "step": 19810
+    },
+    {
+      "epoch": 1.688197699190456,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.4818810532285178e-07,
+      "loss": 0.3666,
+      "step": 19811
+    },
+    {
+      "epoch": 1.6882829143587559,
+      "grad_norm": 14.875,
+      "learning_rate": 1.4810875574172352e-07,
+      "loss": 0.4309,
+      "step": 19812
+    },
+    {
+      "epoch": 1.6883681295270558,
+      "grad_norm": 13.875,
+      "learning_rate": 1.480294260728779e-07,
+      "loss": 0.4722,
+      "step": 19813
+    },
+    {
+      "epoch": 1.6884533446953558,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4795011631774866e-07,
+      "loss": 0.8184,
+      "step": 19814
+    },
+    {
+      "epoch": 1.6885385598636558,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4787082647776908e-07,
+      "loss": 0.9364,
+      "step": 19815
+    },
+    {
+      "epoch": 1.6886237750319557,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.477915565543721e-07,
+      "loss": 0.607,
+      "step": 19816
+    },
+    {
+      "epoch": 1.6887089902002557,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.477123065489898e-07,
+      "loss": 0.4657,
+      "step": 19817
+    },
+    {
+      "epoch": 1.6887942053685556,
+      "grad_norm": 15.75,
+      "learning_rate": 1.476330764630546e-07,
+      "loss": 0.5009,
+      "step": 19818
+    },
+    {
+      "epoch": 1.6888794205368556,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.475538662979985e-07,
+      "loss": 0.6768,
+      "step": 19819
+    },
+    {
+      "epoch": 1.6889646357051555,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4747467605525257e-07,
+      "loss": 0.3975,
+      "step": 19820
+    },
+    {
+      "epoch": 1.6890498508734555,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4739550573624764e-07,
+      "loss": 0.888,
+      "step": 19821
+    },
+    {
+      "epoch": 1.6891350660417555,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.4731635534241477e-07,
+      "loss": 0.2654,
+      "step": 19822
+    },
+    {
+      "epoch": 1.6892202812100554,
+      "grad_norm": 11.375,
+      "learning_rate": 1.4723722487518412e-07,
+      "loss": 0.3611,
+      "step": 19823
+    },
+    {
+      "epoch": 1.6893054963783554,
+      "grad_norm": 18.75,
+      "learning_rate": 1.4715811433598598e-07,
+      "loss": 0.7624,
+      "step": 19824
+    },
+    {
+      "epoch": 1.6893907115466553,
+      "grad_norm": 27.5,
+      "learning_rate": 1.4707902372624954e-07,
+      "loss": 0.9125,
+      "step": 19825
+    },
+    {
+      "epoch": 1.6894759267149553,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.469999530474041e-07,
+      "loss": 0.7287,
+      "step": 19826
+    },
+    {
+      "epoch": 1.6895611418832552,
+      "grad_norm": 14.25,
+      "learning_rate": 1.4692090230087871e-07,
+      "loss": 0.7356,
+      "step": 19827
+    },
+    {
+      "epoch": 1.6896463570515552,
+      "grad_norm": 15.125,
+      "learning_rate": 1.4684187148810217e-07,
+      "loss": 0.6066,
+      "step": 19828
+    },
+    {
+      "epoch": 1.6897315722198551,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.467628606105022e-07,
+      "loss": 0.549,
+      "step": 19829
+    },
+    {
+      "epoch": 1.689816787388155,
+      "grad_norm": 16.625,
+      "learning_rate": 1.466838696695065e-07,
+      "loss": 0.6604,
+      "step": 19830
+    },
+    {
+      "epoch": 1.689902002556455,
+      "grad_norm": 14.5,
+      "learning_rate": 1.466048986665429e-07,
+      "loss": 0.4092,
+      "step": 19831
+    },
+    {
+      "epoch": 1.689987217724755,
+      "grad_norm": 20.625,
+      "learning_rate": 1.465259476030384e-07,
+      "loss": 0.7878,
+      "step": 19832
+    },
+    {
+      "epoch": 1.690072432893055,
+      "grad_norm": 21.375,
+      "learning_rate": 1.464470164804195e-07,
+      "loss": 0.7124,
+      "step": 19833
+    },
+    {
+      "epoch": 1.690157648061355,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4636810530011258e-07,
+      "loss": 0.7486,
+      "step": 19834
+    },
+    {
+      "epoch": 1.690242863229655,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.46289214063544e-07,
+      "loss": 0.8039,
+      "step": 19835
+    },
+    {
+      "epoch": 1.6903280783979548,
+      "grad_norm": 15.0,
+      "learning_rate": 1.4621034277213925e-07,
+      "loss": 0.7368,
+      "step": 19836
+    },
+    {
+      "epoch": 1.6904132935662548,
+      "grad_norm": 15.625,
+      "learning_rate": 1.4613149142732374e-07,
+      "loss": 0.7952,
+      "step": 19837
+    },
+    {
+      "epoch": 1.6904985087345548,
+      "grad_norm": 19.625,
+      "learning_rate": 1.4605266003052178e-07,
+      "loss": 0.5878,
+      "step": 19838
+    },
+    {
+      "epoch": 1.6905837239028547,
+      "grad_norm": 16.25,
+      "learning_rate": 1.459738485831584e-07,
+      "loss": 0.7525,
+      "step": 19839
+    },
+    {
+      "epoch": 1.6906689390711547,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4589505708665782e-07,
+      "loss": 0.6464,
+      "step": 19840
+    },
+    {
+      "epoch": 1.6907541542394546,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4581628554244402e-07,
+      "loss": 0.6383,
+      "step": 19841
+    },
+    {
+      "epoch": 1.6908393694077546,
+      "grad_norm": 15.5,
+      "learning_rate": 1.4573753395194013e-07,
+      "loss": 0.7119,
+      "step": 19842
+    },
+    {
+      "epoch": 1.6909245845760545,
+      "grad_norm": 20.75,
+      "learning_rate": 1.4565880231656934e-07,
+      "loss": 0.4111,
+      "step": 19843
+    },
+    {
+      "epoch": 1.6910097997443545,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.4558009063775442e-07,
+      "loss": 0.5719,
+      "step": 19844
+    },
+    {
+      "epoch": 1.6910950149126545,
+      "grad_norm": 17.625,
+      "learning_rate": 1.4550139891691824e-07,
+      "loss": 0.6986,
+      "step": 19845
+    },
+    {
+      "epoch": 1.6911802300809544,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.454227271554824e-07,
+      "loss": 0.4287,
+      "step": 19846
+    },
+    {
+      "epoch": 1.6912654452492544,
+      "grad_norm": 29.125,
+      "learning_rate": 1.4534407535486836e-07,
+      "loss": 0.7954,
+      "step": 19847
+    },
+    {
+      "epoch": 1.6913506604175543,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.4526544351649768e-07,
+      "loss": 0.3777,
+      "step": 19848
+    },
+    {
+      "epoch": 1.6914358755858543,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4518683164179147e-07,
+      "loss": 0.3116,
+      "step": 19849
+    },
+    {
+      "epoch": 1.6915210907541542,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.4510823973217014e-07,
+      "loss": 0.5906,
+      "step": 19850
+    },
+    {
+      "epoch": 1.6916063059224542,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4502966778905379e-07,
+      "loss": 0.4817,
+      "step": 19851
+    },
+    {
+      "epoch": 1.6916915210907542,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4495111581386264e-07,
+      "loss": 0.4834,
+      "step": 19852
+    },
+    {
+      "epoch": 1.6917767362590541,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.4487258380801617e-07,
+      "loss": 0.3179,
+      "step": 19853
+    },
+    {
+      "epoch": 1.691861951427354,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4479407177293348e-07,
+      "loss": 0.5432,
+      "step": 19854
+    },
+    {
+      "epoch": 1.691947166595654,
+      "grad_norm": 10.25,
+      "learning_rate": 1.4471557971003304e-07,
+      "loss": 0.2705,
+      "step": 19855
+    },
+    {
+      "epoch": 1.692032381763954,
+      "grad_norm": 17.875,
+      "learning_rate": 1.446371076207334e-07,
+      "loss": 0.7952,
+      "step": 19856
+    },
+    {
+      "epoch": 1.692117596932254,
+      "grad_norm": 10.375,
+      "learning_rate": 1.4455865550645292e-07,
+      "loss": 0.2833,
+      "step": 19857
+    },
+    {
+      "epoch": 1.692202812100554,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4448022336860933e-07,
+      "loss": 0.821,
+      "step": 19858
+    },
+    {
+      "epoch": 1.6922880272688539,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4440181120861942e-07,
+      "loss": 0.391,
+      "step": 19859
+    },
+    {
+      "epoch": 1.6923732424371538,
+      "grad_norm": 21.625,
+      "learning_rate": 1.4432341902790068e-07,
+      "loss": 0.9421,
+      "step": 19860
+    },
+    {
+      "epoch": 1.6924584576054538,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4424504682786985e-07,
+      "loss": 0.6095,
+      "step": 19861
+    },
+    {
+      "epoch": 1.6925436727737537,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.4416669460994265e-07,
+      "loss": 0.4578,
+      "step": 19862
+    },
+    {
+      "epoch": 1.6926288879420537,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4408836237553555e-07,
+      "loss": 0.4686,
+      "step": 19863
+    },
+    {
+      "epoch": 1.6927141031103536,
+      "grad_norm": 15.125,
+      "learning_rate": 1.4401005012606357e-07,
+      "loss": 0.6618,
+      "step": 19864
+    },
+    {
+      "epoch": 1.6927993182786536,
+      "grad_norm": 16.625,
+      "learning_rate": 1.439317578629422e-07,
+      "loss": 0.5734,
+      "step": 19865
+    },
+    {
+      "epoch": 1.6928845334469536,
+      "grad_norm": 14.375,
+      "learning_rate": 1.438534855875863e-07,
+      "loss": 0.5537,
+      "step": 19866
+    },
+    {
+      "epoch": 1.6929697486152535,
+      "grad_norm": 10.375,
+      "learning_rate": 1.4377523330141016e-07,
+      "loss": 0.2787,
+      "step": 19867
+    },
+    {
+      "epoch": 1.6930549637835535,
+      "grad_norm": 20.0,
+      "learning_rate": 1.436970010058278e-07,
+      "loss": 0.6065,
+      "step": 19868
+    },
+    {
+      "epoch": 1.6931401789518534,
+      "grad_norm": 11.125,
+      "learning_rate": 1.4361878870225345e-07,
+      "loss": 0.6043,
+      "step": 19869
+    },
+    {
+      "epoch": 1.6932253941201534,
+      "grad_norm": 15.375,
+      "learning_rate": 1.435405963920998e-07,
+      "loss": 0.5849,
+      "step": 19870
+    },
+    {
+      "epoch": 1.6933106092884533,
+      "grad_norm": 14.125,
+      "learning_rate": 1.4346242407678054e-07,
+      "loss": 0.5293,
+      "step": 19871
+    },
+    {
+      "epoch": 1.6933958244567533,
+      "grad_norm": 17.875,
+      "learning_rate": 1.4338427175770775e-07,
+      "loss": 0.5661,
+      "step": 19872
+    },
+    {
+      "epoch": 1.6934810396250533,
+      "grad_norm": 12.375,
+      "learning_rate": 1.4330613943629391e-07,
+      "loss": 0.407,
+      "step": 19873
+    },
+    {
+      "epoch": 1.6935662547933532,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.4322802711395106e-07,
+      "loss": 0.6975,
+      "step": 19874
+    },
+    {
+      "epoch": 1.6936514699616532,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4314993479209088e-07,
+      "loss": 0.3579,
+      "step": 19875
+    },
+    {
+      "epoch": 1.6937366851299531,
+      "grad_norm": 13.875,
+      "learning_rate": 1.4307186247212418e-07,
+      "loss": 0.4404,
+      "step": 19876
+    },
+    {
+      "epoch": 1.693821900298253,
+      "grad_norm": 18.0,
+      "learning_rate": 1.429938101554622e-07,
+      "loss": 0.7064,
+      "step": 19877
+    },
+    {
+      "epoch": 1.693907115466553,
+      "grad_norm": 13.5,
+      "learning_rate": 1.4291577784351516e-07,
+      "loss": 0.5175,
+      "step": 19878
+    },
+    {
+      "epoch": 1.693992330634853,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4283776553769336e-07,
+      "loss": 0.5695,
+      "step": 19879
+    },
+    {
+      "epoch": 1.694077545803153,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4275977323940636e-07,
+      "loss": 0.6726,
+      "step": 19880
+    },
+    {
+      "epoch": 1.694162760971453,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4268180095006357e-07,
+      "loss": 0.7951,
+      "step": 19881
+    },
+    {
+      "epoch": 1.6942479761397529,
+      "grad_norm": 13.5,
+      "learning_rate": 1.4260384867107416e-07,
+      "loss": 0.6943,
+      "step": 19882
+    },
+    {
+      "epoch": 1.6943331913080528,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4252591640384713e-07,
+      "loss": 0.7309,
+      "step": 19883
+    },
+    {
+      "epoch": 1.6944184064763528,
+      "grad_norm": 19.375,
+      "learning_rate": 1.4244800414979e-07,
+      "loss": 0.8296,
+      "step": 19884
+    },
+    {
+      "epoch": 1.6945036216446527,
+      "grad_norm": 60.25,
+      "learning_rate": 1.4237011191031158e-07,
+      "loss": 1.2402,
+      "step": 19885
+    },
+    {
+      "epoch": 1.6945888368129527,
+      "grad_norm": 20.125,
+      "learning_rate": 1.422922396868187e-07,
+      "loss": 0.7189,
+      "step": 19886
+    },
+    {
+      "epoch": 1.6946740519812526,
+      "grad_norm": 18.375,
+      "learning_rate": 1.42214387480719e-07,
+      "loss": 0.6563,
+      "step": 19887
+    },
+    {
+      "epoch": 1.6947592671495526,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4213655529341952e-07,
+      "loss": 0.9295,
+      "step": 19888
+    },
+    {
+      "epoch": 1.6948444823178526,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.4205874312632624e-07,
+      "loss": 0.313,
+      "step": 19889
+    },
+    {
+      "epoch": 1.6949296974861525,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.4198095098084563e-07,
+      "loss": 0.5131,
+      "step": 19890
+    },
+    {
+      "epoch": 1.6950149126544525,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4190317885838343e-07,
+      "loss": 0.9255,
+      "step": 19891
+    },
+    {
+      "epoch": 1.6951001278227524,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.4182542676034526e-07,
+      "loss": 0.2048,
+      "step": 19892
+    },
+    {
+      "epoch": 1.6951853429910524,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4174769468813577e-07,
+      "loss": 0.569,
+      "step": 19893
+    },
+    {
+      "epoch": 1.6952705581593523,
+      "grad_norm": 16.625,
+      "learning_rate": 1.4166998264316015e-07,
+      "loss": 0.6468,
+      "step": 19894
+    },
+    {
+      "epoch": 1.6953557733276523,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4159229062682218e-07,
+      "loss": 0.7157,
+      "step": 19895
+    },
+    {
+      "epoch": 1.6954409884959523,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.415146186405264e-07,
+      "loss": 0.4289,
+      "step": 19896
+    },
+    {
+      "epoch": 1.6955262036642522,
+      "grad_norm": 13.875,
+      "learning_rate": 1.4143696668567588e-07,
+      "loss": 0.7149,
+      "step": 19897
+    },
+    {
+      "epoch": 1.6956114188325522,
+      "grad_norm": 12.875,
+      "learning_rate": 1.413593347636741e-07,
+      "loss": 0.3794,
+      "step": 19898
+    },
+    {
+      "epoch": 1.6956966340008521,
+      "grad_norm": 14.125,
+      "learning_rate": 1.412817228759239e-07,
+      "loss": 0.5518,
+      "step": 19899
+    },
+    {
+      "epoch": 1.695781849169152,
+      "grad_norm": 17.25,
+      "learning_rate": 1.412041310238281e-07,
+      "loss": 0.5134,
+      "step": 19900
+    },
+    {
+      "epoch": 1.695867064337452,
+      "grad_norm": 15.5,
+      "learning_rate": 1.411265592087885e-07,
+      "loss": 0.685,
+      "step": 19901
+    },
+    {
+      "epoch": 1.695952279505752,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.4104900743220725e-07,
+      "loss": 0.4813,
+      "step": 19902
+    },
+    {
+      "epoch": 1.696037494674052,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4097147569548525e-07,
+      "loss": 0.8228,
+      "step": 19903
+    },
+    {
+      "epoch": 1.696122709842352,
+      "grad_norm": 13.875,
+      "learning_rate": 1.4089396400002397e-07,
+      "loss": 0.467,
+      "step": 19904
+    },
+    {
+      "epoch": 1.6962079250106519,
+      "grad_norm": 17.25,
+      "learning_rate": 1.4081647234722427e-07,
+      "loss": 0.93,
+      "step": 19905
+    },
+    {
+      "epoch": 1.6962931401789518,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.407390007384861e-07,
+      "loss": 0.7149,
+      "step": 19906
+    },
+    {
+      "epoch": 1.6963783553472518,
+      "grad_norm": 16.5,
+      "learning_rate": 1.406615491752096e-07,
+      "loss": 0.5214,
+      "step": 19907
+    },
+    {
+      "epoch": 1.6964635705155517,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4058411765879436e-07,
+      "loss": 0.4017,
+      "step": 19908
+    },
+    {
+      "epoch": 1.6965487856838517,
+      "grad_norm": 16.875,
+      "learning_rate": 1.405067061906401e-07,
+      "loss": 0.2838,
+      "step": 19909
+    },
+    {
+      "epoch": 1.6966340008521517,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.4042931477214528e-07,
+      "loss": 0.5578,
+      "step": 19910
+    },
+    {
+      "epoch": 1.6967192160204516,
+      "grad_norm": 13.375,
+      "learning_rate": 1.403519434047082e-07,
+      "loss": 0.5497,
+      "step": 19911
+    },
+    {
+      "epoch": 1.6968044311887516,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4027459208972734e-07,
+      "loss": 0.3461,
+      "step": 19912
+    },
+    {
+      "epoch": 1.6968896463570515,
+      "grad_norm": 15.625,
+      "learning_rate": 1.401972608286009e-07,
+      "loss": 0.5493,
+      "step": 19913
+    },
+    {
+      "epoch": 1.6969748615253515,
+      "grad_norm": 23.25,
+      "learning_rate": 1.4011994962272553e-07,
+      "loss": 0.5026,
+      "step": 19914
+    },
+    {
+      "epoch": 1.6970600766936514,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.4004265847349886e-07,
+      "loss": 0.6886,
+      "step": 19915
+    },
+    {
+      "epoch": 1.6971452918619514,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3996538738231742e-07,
+      "loss": 0.5768,
+      "step": 19916
+    },
+    {
+      "epoch": 1.6972305070302514,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.3988813635057802e-07,
+      "loss": 0.3552,
+      "step": 19917
+    },
+    {
+      "epoch": 1.6973157221985513,
+      "grad_norm": 11.125,
+      "learning_rate": 1.3981090537967617e-07,
+      "loss": 0.2984,
+      "step": 19918
+    },
+    {
+      "epoch": 1.6974009373668513,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3973369447100747e-07,
+      "loss": 1.0995,
+      "step": 19919
+    },
+    {
+      "epoch": 1.6974861525351512,
+      "grad_norm": 17.25,
+      "learning_rate": 1.396565036259673e-07,
+      "loss": 0.4521,
+      "step": 19920
+    },
+    {
+      "epoch": 1.6975713677034512,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.3957933284595078e-07,
+      "loss": 0.5068,
+      "step": 19921
+    },
+    {
+      "epoch": 1.6976565828717511,
+      "grad_norm": 17.0,
+      "learning_rate": 1.395021821323525e-07,
+      "loss": 0.6094,
+      "step": 19922
+    },
+    {
+      "epoch": 1.697741798040051,
+      "grad_norm": 21.0,
+      "learning_rate": 1.394250514865661e-07,
+      "loss": 0.8253,
+      "step": 19923
+    },
+    {
+      "epoch": 1.697827013208351,
+      "grad_norm": 15.125,
+      "learning_rate": 1.39347940909986e-07,
+      "loss": 0.3387,
+      "step": 19924
+    },
+    {
+      "epoch": 1.697912228376651,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3927085040400538e-07,
+      "loss": 0.7344,
+      "step": 19925
+    },
+    {
+      "epoch": 1.697997443544951,
+      "grad_norm": 15.625,
+      "learning_rate": 1.3919377997001786e-07,
+      "loss": 0.8281,
+      "step": 19926
+    },
+    {
+      "epoch": 1.698082658713251,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3911672960941523e-07,
+      "loss": 0.3442,
+      "step": 19927
+    },
+    {
+      "epoch": 1.6981678738815509,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3903969932359054e-07,
+      "loss": 0.4838,
+      "step": 19928
+    },
+    {
+      "epoch": 1.6982530890498508,
+      "grad_norm": 34.75,
+      "learning_rate": 1.389626891139355e-07,
+      "loss": 0.6817,
+      "step": 19929
+    },
+    {
+      "epoch": 1.6983383042181508,
+      "grad_norm": 15.625,
+      "learning_rate": 1.388856989818421e-07,
+      "loss": 0.4962,
+      "step": 19930
+    },
+    {
+      "epoch": 1.6984235193864508,
+      "grad_norm": 16.875,
+      "learning_rate": 1.388087289287013e-07,
+      "loss": 0.833,
+      "step": 19931
+    },
+    {
+      "epoch": 1.6985087345547507,
+      "grad_norm": 16.875,
+      "learning_rate": 1.387317789559041e-07,
+      "loss": 0.9758,
+      "step": 19932
+    },
+    {
+      "epoch": 1.6985939497230507,
+      "grad_norm": 22.25,
+      "learning_rate": 1.3865484906484122e-07,
+      "loss": 0.8172,
+      "step": 19933
+    },
+    {
+      "epoch": 1.6986791648913506,
+      "grad_norm": 18.625,
+      "learning_rate": 1.3857793925690282e-07,
+      "loss": 0.4925,
+      "step": 19934
+    },
+    {
+      "epoch": 1.6987643800596506,
+      "grad_norm": 19.25,
+      "learning_rate": 1.385010495334789e-07,
+      "loss": 0.6085,
+      "step": 19935
+    },
+    {
+      "epoch": 1.6988495952279505,
+      "grad_norm": 20.5,
+      "learning_rate": 1.3842417989595832e-07,
+      "loss": 0.9952,
+      "step": 19936
+    },
+    {
+      "epoch": 1.6989348103962505,
+      "grad_norm": 15.5,
+      "learning_rate": 1.3834733034573055e-07,
+      "loss": 0.6747,
+      "step": 19937
+    },
+    {
+      "epoch": 1.6990200255645505,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3827050088418447e-07,
+      "loss": 0.8391,
+      "step": 19938
+    },
+    {
+      "epoch": 1.6991052407328504,
+      "grad_norm": 14.125,
+      "learning_rate": 1.3819369151270856e-07,
+      "loss": 0.4728,
+      "step": 19939
+    },
+    {
+      "epoch": 1.6991904559011504,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3811690223269044e-07,
+      "loss": 0.2329,
+      "step": 19940
+    },
+    {
+      "epoch": 1.6992756710694503,
+      "grad_norm": 10.75,
+      "learning_rate": 1.3804013304551806e-07,
+      "loss": 0.4992,
+      "step": 19941
+    },
+    {
+      "epoch": 1.6993608862377503,
+      "grad_norm": 20.0,
+      "learning_rate": 1.3796338395257875e-07,
+      "loss": 0.8773,
+      "step": 19942
+    },
+    {
+      "epoch": 1.6994461014060502,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3788665495525934e-07,
+      "loss": 0.435,
+      "step": 19943
+    },
+    {
+      "epoch": 1.6995313165743502,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3780994605494618e-07,
+      "loss": 0.6134,
+      "step": 19944
+    },
+    {
+      "epoch": 1.6996165317426501,
+      "grad_norm": 18.375,
+      "learning_rate": 1.3773325725302572e-07,
+      "loss": 0.9821,
+      "step": 19945
+    },
+    {
+      "epoch": 1.69970174691095,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.3765658855088377e-07,
+      "loss": 0.7667,
+      "step": 19946
+    },
+    {
+      "epoch": 1.69978696207925,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.37579939949906e-07,
+      "loss": 0.6059,
+      "step": 19947
+    },
+    {
+      "epoch": 1.69987217724755,
+      "grad_norm": 17.625,
+      "learning_rate": 1.375033114514772e-07,
+      "loss": 0.5839,
+      "step": 19948
+    },
+    {
+      "epoch": 1.69995739241585,
+      "grad_norm": 15.125,
+      "learning_rate": 1.3742670305698217e-07,
+      "loss": 0.5672,
+      "step": 19949
+    },
+    {
+      "epoch": 1.70004260758415,
+      "grad_norm": 14.75,
+      "learning_rate": 1.3735011476780565e-07,
+      "loss": 0.5276,
+      "step": 19950
+    },
+    {
+      "epoch": 1.70012782275245,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3727354658533114e-07,
+      "loss": 0.6352,
+      "step": 19951
+    },
+    {
+      "epoch": 1.7002130379207498,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3719699851094295e-07,
+      "loss": 0.9877,
+      "step": 19952
+    },
+    {
+      "epoch": 1.7002982530890498,
+      "grad_norm": 22.875,
+      "learning_rate": 1.371204705460237e-07,
+      "loss": 1.067,
+      "step": 19953
+    },
+    {
+      "epoch": 1.7003834682573498,
+      "grad_norm": 15.625,
+      "learning_rate": 1.370439626919566e-07,
+      "loss": 0.4755,
+      "step": 19954
+    },
+    {
+      "epoch": 1.7004686834256497,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3696747495012417e-07,
+      "loss": 0.7257,
+      "step": 19955
+    },
+    {
+      "epoch": 1.7005538985939497,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.3689100732190916e-07,
+      "loss": 0.3891,
+      "step": 19956
+    },
+    {
+      "epoch": 1.7006391137622496,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3681455980869257e-07,
+      "loss": 0.6662,
+      "step": 19957
+    },
+    {
+      "epoch": 1.7007243289305496,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3673813241185634e-07,
+      "loss": 0.5472,
+      "step": 19958
+    },
+    {
+      "epoch": 1.7008095440988495,
+      "grad_norm": 11.875,
+      "learning_rate": 1.3666172513278175e-07,
+      "loss": 0.3451,
+      "step": 19959
+    },
+    {
+      "epoch": 1.7008947592671495,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.3658533797284935e-07,
+      "loss": 0.344,
+      "step": 19960
+    },
+    {
+      "epoch": 1.7009799744354495,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3650897093343927e-07,
+      "loss": 0.6649,
+      "step": 19961
+    },
+    {
+      "epoch": 1.7010651896037494,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3643262401593166e-07,
+      "loss": 0.5444,
+      "step": 19962
+    },
+    {
+      "epoch": 1.7011504047720494,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.3635629722170653e-07,
+      "loss": 0.3413,
+      "step": 19963
+    },
+    {
+      "epoch": 1.7012356199403493,
+      "grad_norm": 27.0,
+      "learning_rate": 1.3627999055214308e-07,
+      "loss": 0.6247,
+      "step": 19964
+    },
+    {
+      "epoch": 1.7013208351086493,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3620370400861992e-07,
+      "loss": 0.6945,
+      "step": 19965
+    },
+    {
+      "epoch": 1.7014060502769492,
+      "grad_norm": 13.5,
+      "learning_rate": 1.3612743759251575e-07,
+      "loss": 0.4382,
+      "step": 19966
+    },
+    {
+      "epoch": 1.7014912654452492,
+      "grad_norm": 22.0,
+      "learning_rate": 1.3605119130520927e-07,
+      "loss": 0.7941,
+      "step": 19967
+    },
+    {
+      "epoch": 1.7015764806135492,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3597496514807754e-07,
+      "loss": 0.4649,
+      "step": 19968
+    },
+    {
+      "epoch": 1.7016616957818491,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.3589875912249874e-07,
+      "loss": 0.3829,
+      "step": 19969
+    },
+    {
+      "epoch": 1.701746910950149,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3582257322984946e-07,
+      "loss": 0.4994,
+      "step": 19970
+    },
+    {
+      "epoch": 1.701832126118449,
+      "grad_norm": 19.125,
+      "learning_rate": 1.357464074715066e-07,
+      "loss": 0.8673,
+      "step": 19971
+    },
+    {
+      "epoch": 1.701917341286749,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.3567026184884702e-07,
+      "loss": 0.5102,
+      "step": 19972
+    },
+    {
+      "epoch": 1.702002556455049,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3559413636324598e-07,
+      "loss": 0.7309,
+      "step": 19973
+    },
+    {
+      "epoch": 1.702087771623349,
+      "grad_norm": 19.125,
+      "learning_rate": 1.355180310160796e-07,
+      "loss": 0.6753,
+      "step": 19974
+    },
+    {
+      "epoch": 1.7021729867916489,
+      "grad_norm": 15.75,
+      "learning_rate": 1.354419458087232e-07,
+      "loss": 0.6555,
+      "step": 19975
+    },
+    {
+      "epoch": 1.7022582019599488,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3536588074255147e-07,
+      "loss": 0.6903,
+      "step": 19976
+    },
+    {
+      "epoch": 1.7023434171282488,
+      "grad_norm": 23.625,
+      "learning_rate": 1.3528983581893946e-07,
+      "loss": 0.7415,
+      "step": 19977
+    },
+    {
+      "epoch": 1.7024286322965487,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3521381103926062e-07,
+      "loss": 0.5858,
+      "step": 19978
+    },
+    {
+      "epoch": 1.7025138474648487,
+      "grad_norm": 13.125,
+      "learning_rate": 1.351378064048893e-07,
+      "loss": 0.3025,
+      "step": 19979
+    },
+    {
+      "epoch": 1.7025990626331486,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.3506182191719884e-07,
+      "loss": 0.5214,
+      "step": 19980
+    },
+    {
+      "epoch": 1.7026842778014486,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3498585757756283e-07,
+      "loss": 0.5411,
+      "step": 19981
+    },
+    {
+      "epoch": 1.7027694929697486,
+      "grad_norm": 15.5,
+      "learning_rate": 1.3490991338735313e-07,
+      "loss": 0.5804,
+      "step": 19982
+    },
+    {
+      "epoch": 1.7028547081380485,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.3483398934794307e-07,
+      "loss": 0.4496,
+      "step": 19983
+    },
+    {
+      "epoch": 1.7029399233063485,
+      "grad_norm": 20.75,
+      "learning_rate": 1.3475808546070377e-07,
+      "loss": 0.8078,
+      "step": 19984
+    },
+    {
+      "epoch": 1.7030251384746484,
+      "grad_norm": 22.875,
+      "learning_rate": 1.3468220172700745e-07,
+      "loss": 0.6641,
+      "step": 19985
+    },
+    {
+      "epoch": 1.7031103536429484,
+      "grad_norm": 22.125,
+      "learning_rate": 1.3460633814822557e-07,
+      "loss": 0.9287,
+      "step": 19986
+    },
+    {
+      "epoch": 1.7031955688112483,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3453049472572837e-07,
+      "loss": 0.2935,
+      "step": 19987
+    },
+    {
+      "epoch": 1.7032807839795483,
+      "grad_norm": 19.625,
+      "learning_rate": 1.3445467146088687e-07,
+      "loss": 0.6302,
+      "step": 19988
+    },
+    {
+      "epoch": 1.7033659991478483,
+      "grad_norm": 18.0,
+      "learning_rate": 1.3437886835507162e-07,
+      "loss": 0.6597,
+      "step": 19989
+    },
+    {
+      "epoch": 1.7034512143161482,
+      "grad_norm": 20.625,
+      "learning_rate": 1.3430308540965165e-07,
+      "loss": 1.0542,
+      "step": 19990
+    },
+    {
+      "epoch": 1.7035364294844482,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3422732262599714e-07,
+      "loss": 0.7308,
+      "step": 19991
+    },
+    {
+      "epoch": 1.7036216446527481,
+      "grad_norm": 16.25,
+      "learning_rate": 1.341515800054767e-07,
+      "loss": 0.8165,
+      "step": 19992
+    },
+    {
+      "epoch": 1.703706859821048,
+      "grad_norm": 15.25,
+      "learning_rate": 1.3407585754945926e-07,
+      "loss": 0.4318,
+      "step": 19993
+    },
+    {
+      "epoch": 1.703792074989348,
+      "grad_norm": 13.5,
+      "learning_rate": 1.3400015525931343e-07,
+      "loss": 0.3637,
+      "step": 19994
+    },
+    {
+      "epoch": 1.703877290157648,
+      "grad_norm": 15.5,
+      "learning_rate": 1.3392447313640684e-07,
+      "loss": 0.6671,
+      "step": 19995
+    },
+    {
+      "epoch": 1.703962505325948,
+      "grad_norm": 38.0,
+      "learning_rate": 1.338488111821072e-07,
+      "loss": 0.6747,
+      "step": 19996
+    },
+    {
+      "epoch": 1.704047720494248,
+      "grad_norm": 38.75,
+      "learning_rate": 1.3377316939778202e-07,
+      "loss": 0.8417,
+      "step": 19997
+    },
+    {
+      "epoch": 1.7041329356625479,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3369754778479835e-07,
+      "loss": 0.7618,
+      "step": 19998
+    },
+    {
+      "epoch": 1.7042181508308478,
+      "grad_norm": 13.875,
+      "learning_rate": 1.336219463445225e-07,
+      "loss": 0.3476,
+      "step": 19999
+    },
+    {
+      "epoch": 1.7043033659991478,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.3354636507832045e-07,
+      "loss": 0.6015,
+      "step": 20000
+    },
+    {
+      "epoch": 1.7043885811674477,
+      "grad_norm": 16.0,
+      "learning_rate": 1.3347080398755833e-07,
+      "loss": 1.1305,
+      "step": 20001
+    },
+    {
+      "epoch": 1.7044737963357477,
+      "grad_norm": 11.5,
+      "learning_rate": 1.333952630736017e-07,
+      "loss": 0.4524,
+      "step": 20002
+    },
+    {
+      "epoch": 1.7045590115040476,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3331974233781534e-07,
+      "loss": 0.465,
+      "step": 20003
+    },
+    {
+      "epoch": 1.7046442266723476,
+      "grad_norm": 19.75,
+      "learning_rate": 1.3324424178156408e-07,
+      "loss": 1.0665,
+      "step": 20004
+    },
+    {
+      "epoch": 1.7047294418406476,
+      "grad_norm": 17.25,
+      "learning_rate": 1.331687614062123e-07,
+      "loss": 0.3595,
+      "step": 20005
+    },
+    {
+      "epoch": 1.7048146570089475,
+      "grad_norm": 22.0,
+      "learning_rate": 1.3309330121312441e-07,
+      "loss": 0.9665,
+      "step": 20006
+    },
+    {
+      "epoch": 1.7048998721772475,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3301786120366368e-07,
+      "loss": 0.6113,
+      "step": 20007
+    },
+    {
+      "epoch": 1.7049850873455474,
+      "grad_norm": 21.5,
+      "learning_rate": 1.3294244137919313e-07,
+      "loss": 0.7563,
+      "step": 20008
+    },
+    {
+      "epoch": 1.7050703025138474,
+      "grad_norm": 10.3125,
+      "learning_rate": 1.3286704174107602e-07,
+      "loss": 0.3996,
+      "step": 20009
+    },
+    {
+      "epoch": 1.7051555176821473,
+      "grad_norm": 17.5,
+      "learning_rate": 1.327916622906747e-07,
+      "loss": 0.3564,
+      "step": 20010
+    },
+    {
+      "epoch": 1.7052407328504473,
+      "grad_norm": 22.25,
+      "learning_rate": 1.3271630302935186e-07,
+      "loss": 1.038,
+      "step": 20011
+    },
+    {
+      "epoch": 1.7053259480187473,
+      "grad_norm": 18.5,
+      "learning_rate": 1.326409639584686e-07,
+      "loss": 0.9324,
+      "step": 20012
+    },
+    {
+      "epoch": 1.7054111631870472,
+      "grad_norm": 19.875,
+      "learning_rate": 1.3256564507938677e-07,
+      "loss": 0.5797,
+      "step": 20013
+    },
+    {
+      "epoch": 1.7054963783553472,
+      "grad_norm": 11.75,
+      "learning_rate": 1.3249034639346737e-07,
+      "loss": 0.3314,
+      "step": 20014
+    },
+    {
+      "epoch": 1.7055815935236471,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.3241506790207136e-07,
+      "loss": 0.4046,
+      "step": 20015
+    },
+    {
+      "epoch": 1.705666808691947,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.3233980960655894e-07,
+      "loss": 0.5557,
+      "step": 20016
+    },
+    {
+      "epoch": 1.705752023860247,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.3226457150828987e-07,
+      "loss": 0.6009,
+      "step": 20017
+    },
+    {
+      "epoch": 1.705837239028547,
+      "grad_norm": 17.25,
+      "learning_rate": 1.32189353608624e-07,
+      "loss": 0.8705,
+      "step": 20018
+    },
+    {
+      "epoch": 1.705922454196847,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.3211415590892064e-07,
+      "loss": 0.4909,
+      "step": 20019
+    },
+    {
+      "epoch": 1.706007669365147,
+      "grad_norm": 21.0,
+      "learning_rate": 1.3203897841053845e-07,
+      "loss": 0.6456,
+      "step": 20020
+    },
+    {
+      "epoch": 1.7060928845334469,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3196382111483608e-07,
+      "loss": 0.7092,
+      "step": 20021
+    },
+    {
+      "epoch": 1.7061780997017468,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.3188868402317174e-07,
+      "loss": 0.3371,
+      "step": 20022
+    },
+    {
+      "epoch": 1.7062633148700468,
+      "grad_norm": 17.625,
+      "learning_rate": 1.3181356713690348e-07,
+      "loss": 0.8493,
+      "step": 20023
+    },
+    {
+      "epoch": 1.7063485300383467,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3173847045738847e-07,
+      "loss": 0.4968,
+      "step": 20024
+    },
+    {
+      "epoch": 1.7064337452066467,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3166339398598362e-07,
+      "loss": 0.4469,
+      "step": 20025
+    },
+    {
+      "epoch": 1.7065189603749467,
+      "grad_norm": 20.5,
+      "learning_rate": 1.3158833772404567e-07,
+      "loss": 1.081,
+      "step": 20026
+    },
+    {
+      "epoch": 1.7066041755432466,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.3151330167293128e-07,
+      "loss": 0.3628,
+      "step": 20027
+    },
+    {
+      "epoch": 1.7066893907115466,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3143828583399647e-07,
+      "loss": 0.7509,
+      "step": 20028
+    },
+    {
+      "epoch": 1.7067746058798465,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3136329020859624e-07,
+      "loss": 0.3011,
+      "step": 20029
+    },
+    {
+      "epoch": 1.7068598210481465,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.312883147980863e-07,
+      "loss": 0.6595,
+      "step": 20030
+    },
+    {
+      "epoch": 1.7069450362164464,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3121335960382147e-07,
+      "loss": 0.6328,
+      "step": 20031
+    },
+    {
+      "epoch": 1.7070302513847464,
+      "grad_norm": 18.25,
+      "learning_rate": 1.3113842462715687e-07,
+      "loss": 0.4145,
+      "step": 20032
+    },
+    {
+      "epoch": 1.7071154665530464,
+      "grad_norm": 12.25,
+      "learning_rate": 1.3106350986944545e-07,
+      "loss": 0.3674,
+      "step": 20033
+    },
+    {
+      "epoch": 1.7072006817213463,
+      "grad_norm": 11.125,
+      "learning_rate": 1.309886153320415e-07,
+      "loss": 0.348,
+      "step": 20034
+    },
+    {
+      "epoch": 1.7072858968896463,
+      "grad_norm": 23.125,
+      "learning_rate": 1.3091374101629876e-07,
+      "loss": 0.7777,
+      "step": 20035
+    },
+    {
+      "epoch": 1.7073711120579462,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3083888692357008e-07,
+      "loss": 0.5751,
+      "step": 20036
+    },
+    {
+      "epoch": 1.7074563272262462,
+      "grad_norm": 11.625,
+      "learning_rate": 1.3076405305520789e-07,
+      "loss": 0.2671,
+      "step": 20037
+    },
+    {
+      "epoch": 1.7075415423945461,
+      "grad_norm": 19.25,
+      "learning_rate": 1.3068923941256487e-07,
+      "loss": 1.0301,
+      "step": 20038
+    },
+    {
+      "epoch": 1.707626757562846,
+      "grad_norm": 15.75,
+      "learning_rate": 1.306144459969927e-07,
+      "loss": 0.5997,
+      "step": 20039
+    },
+    {
+      "epoch": 1.707711972731146,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3053967280984347e-07,
+      "loss": 0.6087,
+      "step": 20040
+    },
+    {
+      "epoch": 1.707797187899446,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.3046491985246796e-07,
+      "loss": 0.3466,
+      "step": 20041
+    },
+    {
+      "epoch": 1.707882403067746,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3039018712621697e-07,
+      "loss": 0.4123,
+      "step": 20042
+    },
+    {
+      "epoch": 1.707967618236046,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3031547463244116e-07,
+      "loss": 0.4959,
+      "step": 20043
+    },
+    {
+      "epoch": 1.7080528334043459,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.3024078237249064e-07,
+      "loss": 0.5744,
+      "step": 20044
+    },
+    {
+      "epoch": 1.7081380485726458,
+      "grad_norm": 36.75,
+      "learning_rate": 1.3016611034771531e-07,
+      "loss": 0.6897,
+      "step": 20045
+    },
+    {
+      "epoch": 1.7082232637409458,
+      "grad_norm": 15.0,
+      "learning_rate": 1.300914585594644e-07,
+      "loss": 0.555,
+      "step": 20046
+    },
+    {
+      "epoch": 1.7083084789092458,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.300168270090868e-07,
+      "loss": 0.5285,
+      "step": 20047
+    },
+    {
+      "epoch": 1.7083936940775457,
+      "grad_norm": 11.0,
+      "learning_rate": 1.2994221569793172e-07,
+      "loss": 0.2655,
+      "step": 20048
+    },
+    {
+      "epoch": 1.7084789092458457,
+      "grad_norm": 12.625,
+      "learning_rate": 1.298676246273471e-07,
+      "loss": 0.3018,
+      "step": 20049
+    },
+    {
+      "epoch": 1.7085641244141456,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2979305379868066e-07,
+      "loss": 0.4826,
+      "step": 20050
+    },
+    {
+      "epoch": 1.7086493395824456,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2971850321328014e-07,
+      "loss": 0.7242,
+      "step": 20051
+    },
+    {
+      "epoch": 1.7087345547507455,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2964397287249282e-07,
+      "loss": 0.5074,
+      "step": 20052
+    },
+    {
+      "epoch": 1.7088197699190455,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.295694627776657e-07,
+      "loss": 0.2861,
+      "step": 20053
+    },
+    {
+      "epoch": 1.7089049850873455,
+      "grad_norm": 12.5,
+      "learning_rate": 1.2949497293014492e-07,
+      "loss": 0.509,
+      "step": 20054
+    },
+    {
+      "epoch": 1.7089902002556454,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2942050333127676e-07,
+      "loss": 0.8834,
+      "step": 20055
+    },
+    {
+      "epoch": 1.7090754154239454,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2934605398240708e-07,
+      "loss": 0.6043,
+      "step": 20056
+    },
+    {
+      "epoch": 1.7091606305922453,
+      "grad_norm": 16.75,
+      "learning_rate": 1.292716248848809e-07,
+      "loss": 0.4089,
+      "step": 20057
+    },
+    {
+      "epoch": 1.7092458457605453,
+      "grad_norm": 14.625,
+      "learning_rate": 1.291972160400437e-07,
+      "loss": 0.3871,
+      "step": 20058
+    },
+    {
+      "epoch": 1.7093310609288452,
+      "grad_norm": 20.625,
+      "learning_rate": 1.2912282744923953e-07,
+      "loss": 0.4721,
+      "step": 20059
+    },
+    {
+      "epoch": 1.7094162760971452,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2904845911381304e-07,
+      "loss": 0.7479,
+      "step": 20060
+    },
+    {
+      "epoch": 1.7095014912654451,
+      "grad_norm": 16.75,
+      "learning_rate": 1.289741110351081e-07,
+      "loss": 0.9102,
+      "step": 20061
+    },
+    {
+      "epoch": 1.709586706433745,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.2889978321446853e-07,
+      "loss": 0.8046,
+      "step": 20062
+    },
+    {
+      "epoch": 1.709671921602045,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.28825475653237e-07,
+      "loss": 0.2122,
+      "step": 20063
+    },
+    {
+      "epoch": 1.709757136770345,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2875118835275675e-07,
+      "loss": 0.921,
+      "step": 20064
+    },
+    {
+      "epoch": 1.709842351938645,
+      "grad_norm": 18.25,
+      "learning_rate": 1.2867692131436989e-07,
+      "loss": 0.5957,
+      "step": 20065
+    },
+    {
+      "epoch": 1.709927567106945,
+      "grad_norm": 10.375,
+      "learning_rate": 1.286026745394188e-07,
+      "loss": 0.2149,
+      "step": 20066
+    },
+    {
+      "epoch": 1.710012782275245,
+      "grad_norm": 16.25,
+      "learning_rate": 1.285284480292448e-07,
+      "loss": 0.5347,
+      "step": 20067
+    },
+    {
+      "epoch": 1.7100979974435448,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2845424178518945e-07,
+      "loss": 0.2482,
+      "step": 20068
+    },
+    {
+      "epoch": 1.7101832126118448,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2838005580859375e-07,
+      "loss": 0.4436,
+      "step": 20069
+    },
+    {
+      "epoch": 1.7102684277801448,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2830589010079857e-07,
+      "loss": 0.3525,
+      "step": 20070
+    },
+    {
+      "epoch": 1.7103536429484447,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.2823174466314369e-07,
+      "loss": 0.6067,
+      "step": 20071
+    },
+    {
+      "epoch": 1.7104388581167447,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2815761949696942e-07,
+      "loss": 0.7096,
+      "step": 20072
+    },
+    {
+      "epoch": 1.7105240732850446,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2808351460361482e-07,
+      "loss": 0.8955,
+      "step": 20073
+    },
+    {
+      "epoch": 1.7106092884533446,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2800942998441924e-07,
+      "loss": 0.7419,
+      "step": 20074
+    },
+    {
+      "epoch": 1.7106945036216445,
+      "grad_norm": 11.625,
+      "learning_rate": 1.2793536564072186e-07,
+      "loss": 0.3938,
+      "step": 20075
+    },
+    {
+      "epoch": 1.7107797187899445,
+      "grad_norm": 18.125,
+      "learning_rate": 1.278613215738604e-07,
+      "loss": 0.7994,
+      "step": 20076
+    },
+    {
+      "epoch": 1.7108649339582445,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2778729778517339e-07,
+      "loss": 0.3486,
+      "step": 20077
+    },
+    {
+      "epoch": 1.7109501491265444,
+      "grad_norm": 23.5,
+      "learning_rate": 1.2771329427599828e-07,
+      "loss": 0.8718,
+      "step": 20078
+    },
+    {
+      "epoch": 1.7110353642948444,
+      "grad_norm": 12.625,
+      "learning_rate": 1.2763931104767272e-07,
+      "loss": 0.259,
+      "step": 20079
+    },
+    {
+      "epoch": 1.7111205794631443,
+      "grad_norm": 25.25,
+      "learning_rate": 1.2756534810153348e-07,
+      "loss": 1.097,
+      "step": 20080
+    },
+    {
+      "epoch": 1.7112057946314443,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2749140543891675e-07,
+      "loss": 0.5999,
+      "step": 20081
+    },
+    {
+      "epoch": 1.7112910097997442,
+      "grad_norm": 21.125,
+      "learning_rate": 1.2741748306115923e-07,
+      "loss": 0.5775,
+      "step": 20082
+    },
+    {
+      "epoch": 1.7113762249680442,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2734358096959678e-07,
+      "loss": 0.9197,
+      "step": 20083
+    },
+    {
+      "epoch": 1.7114614401363442,
+      "grad_norm": 17.125,
+      "learning_rate": 1.272696991655646e-07,
+      "loss": 0.4731,
+      "step": 20084
+    },
+    {
+      "epoch": 1.7115466553046441,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.271958376503979e-07,
+      "loss": 0.6585,
+      "step": 20085
+    },
+    {
+      "epoch": 1.711631870472944,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.271219964254315e-07,
+      "loss": 0.5869,
+      "step": 20086
+    },
+    {
+      "epoch": 1.711717085641244,
+      "grad_norm": 18.625,
+      "learning_rate": 1.27048175492e-07,
+      "loss": 0.7434,
+      "step": 20087
+    },
+    {
+      "epoch": 1.711802300809544,
+      "grad_norm": 17.25,
+      "learning_rate": 1.269743748514371e-07,
+      "loss": 0.4438,
+      "step": 20088
+    },
+    {
+      "epoch": 1.711887515977844,
+      "grad_norm": 30.5,
+      "learning_rate": 1.2690059450507637e-07,
+      "loss": 0.9487,
+      "step": 20089
+    },
+    {
+      "epoch": 1.711972731146144,
+      "grad_norm": 15.0,
+      "learning_rate": 1.2682683445425128e-07,
+      "loss": 0.2973,
+      "step": 20090
+    },
+    {
+      "epoch": 1.7120579463144439,
+      "grad_norm": 15.0,
+      "learning_rate": 1.267530947002947e-07,
+      "loss": 0.6495,
+      "step": 20091
+    },
+    {
+      "epoch": 1.7121431614827438,
+      "grad_norm": 18.625,
+      "learning_rate": 1.2667937524453945e-07,
+      "loss": 0.9474,
+      "step": 20092
+    },
+    {
+      "epoch": 1.7122283766510438,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.266056760883172e-07,
+      "loss": 0.4801,
+      "step": 20093
+    },
+    {
+      "epoch": 1.7123135918193437,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.265319972329601e-07,
+      "loss": 0.3763,
+      "step": 20094
+    },
+    {
+      "epoch": 1.7123988069876437,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.2645833867979975e-07,
+      "loss": 0.8136,
+      "step": 20095
+    },
+    {
+      "epoch": 1.7124840221559436,
+      "grad_norm": 19.5,
+      "learning_rate": 1.2638470043016674e-07,
+      "loss": 0.8341,
+      "step": 20096
+    },
+    {
+      "epoch": 1.7125692373242436,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2631108248539247e-07,
+      "loss": 0.6708,
+      "step": 20097
+    },
+    {
+      "epoch": 1.7126544524925436,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2623748484680647e-07,
+      "loss": 0.6029,
+      "step": 20098
+    },
+    {
+      "epoch": 1.7127396676608435,
+      "grad_norm": 23.375,
+      "learning_rate": 1.2616390751573935e-07,
+      "loss": 0.6127,
+      "step": 20099
+    },
+    {
+      "epoch": 1.7128248828291435,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2609035049352056e-07,
+      "loss": 0.452,
+      "step": 20100
+    },
+    {
+      "epoch": 1.7129100979974434,
+      "grad_norm": 16.0,
+      "learning_rate": 1.260168137814792e-07,
+      "loss": 0.6542,
+      "step": 20101
+    },
+    {
+      "epoch": 1.7129953131657434,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.2594329738094422e-07,
+      "loss": 0.6103,
+      "step": 20102
+    },
+    {
+      "epoch": 1.7130805283340433,
+      "grad_norm": 14.875,
+      "learning_rate": 1.2586980129324426e-07,
+      "loss": 0.4936,
+      "step": 20103
+    },
+    {
+      "epoch": 1.7131657435023433,
+      "grad_norm": 24.375,
+      "learning_rate": 1.2579632551970756e-07,
+      "loss": 0.5636,
+      "step": 20104
+    },
+    {
+      "epoch": 1.7132509586706433,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.257228700616618e-07,
+      "loss": 0.5469,
+      "step": 20105
+    },
+    {
+      "epoch": 1.7133361738389432,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2564943492043402e-07,
+      "loss": 0.6071,
+      "step": 20106
+    },
+    {
+      "epoch": 1.7134213890072432,
+      "grad_norm": 22.125,
+      "learning_rate": 1.255760200973516e-07,
+      "loss": 0.7089,
+      "step": 20107
+    },
+    {
+      "epoch": 1.7135066041755431,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.255026255937411e-07,
+      "loss": 0.6959,
+      "step": 20108
+    },
+    {
+      "epoch": 1.713591819343843,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2542925141092926e-07,
+      "loss": 0.4797,
+      "step": 20109
+    },
+    {
+      "epoch": 1.713677034512143,
+      "grad_norm": 18.0,
+      "learning_rate": 1.2535589755024141e-07,
+      "loss": 1.0135,
+      "step": 20110
+    },
+    {
+      "epoch": 1.713762249680443,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.2528256401300342e-07,
+      "loss": 0.5365,
+      "step": 20111
+    },
+    {
+      "epoch": 1.713847464848743,
+      "grad_norm": 27.0,
+      "learning_rate": 1.2520925080054076e-07,
+      "loss": 0.8062,
+      "step": 20112
+    },
+    {
+      "epoch": 1.713932680017043,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.2513595791417796e-07,
+      "loss": 0.5805,
+      "step": 20113
+    },
+    {
+      "epoch": 1.7140178951853429,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2506268535523933e-07,
+      "loss": 0.5623,
+      "step": 20114
+    },
+    {
+      "epoch": 1.7141031103536428,
+      "grad_norm": 18.875,
+      "learning_rate": 1.2498943312504924e-07,
+      "loss": 0.6646,
+      "step": 20115
+    },
+    {
+      "epoch": 1.7141883255219428,
+      "grad_norm": 12.5,
+      "learning_rate": 1.2491620122493137e-07,
+      "loss": 0.5012,
+      "step": 20116
+    },
+    {
+      "epoch": 1.7142735406902427,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2484298965620922e-07,
+      "loss": 0.5747,
+      "step": 20117
+    },
+    {
+      "epoch": 1.7143587558585427,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.247697984202055e-07,
+      "loss": 0.3942,
+      "step": 20118
+    },
+    {
+      "epoch": 1.7144439710268427,
+      "grad_norm": 19.25,
+      "learning_rate": 1.2469662751824305e-07,
+      "loss": 0.8065,
+      "step": 20119
+    },
+    {
+      "epoch": 1.7145291861951426,
+      "grad_norm": 16.375,
+      "learning_rate": 1.246234769516441e-07,
+      "loss": 0.5845,
+      "step": 20120
+    },
+    {
+      "epoch": 1.7146144013634426,
+      "grad_norm": 14.875,
+      "learning_rate": 1.2455034672173086e-07,
+      "loss": 0.5019,
+      "step": 20121
+    },
+    {
+      "epoch": 1.7146996165317425,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2447723682982457e-07,
+      "loss": 0.4346,
+      "step": 20122
+    },
+    {
+      "epoch": 1.7147848317000425,
+      "grad_norm": 19.5,
+      "learning_rate": 1.2440414727724615e-07,
+      "loss": 0.7161,
+      "step": 20123
+    },
+    {
+      "epoch": 1.7148700468683424,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.243310780653166e-07,
+      "loss": 0.5796,
+      "step": 20124
+    },
+    {
+      "epoch": 1.7149552620366424,
+      "grad_norm": 19.25,
+      "learning_rate": 1.2425802919535671e-07,
+      "loss": 0.6949,
+      "step": 20125
+    },
+    {
+      "epoch": 1.7150404772049423,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2418500066868608e-07,
+      "loss": 0.8475,
+      "step": 20126
+    },
+    {
+      "epoch": 1.7151256923732423,
+      "grad_norm": 18.625,
+      "learning_rate": 1.241119924866245e-07,
+      "loss": 0.8156,
+      "step": 20127
+    },
+    {
+      "epoch": 1.7152109075415423,
+      "grad_norm": 12.25,
+      "learning_rate": 1.2403900465049148e-07,
+      "loss": 0.3294,
+      "step": 20128
+    },
+    {
+      "epoch": 1.7152961227098422,
+      "grad_norm": 16.0,
+      "learning_rate": 1.2396603716160596e-07,
+      "loss": 0.391,
+      "step": 20129
+    },
+    {
+      "epoch": 1.7153813378781422,
+      "grad_norm": 21.75,
+      "learning_rate": 1.2389309002128662e-07,
+      "loss": 0.743,
+      "step": 20130
+    },
+    {
+      "epoch": 1.7154665530464421,
+      "grad_norm": 21.125,
+      "learning_rate": 1.2382016323085125e-07,
+      "loss": 0.5649,
+      "step": 20131
+    },
+    {
+      "epoch": 1.715551768214742,
+      "grad_norm": 15.125,
+      "learning_rate": 1.23747256791618e-07,
+      "loss": 0.3936,
+      "step": 20132
+    },
+    {
+      "epoch": 1.715636983383042,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2367437070490428e-07,
+      "loss": 0.481,
+      "step": 20133
+    },
+    {
+      "epoch": 1.715722198551342,
+      "grad_norm": 18.125,
+      "learning_rate": 1.2360150497202753e-07,
+      "loss": 0.5826,
+      "step": 20134
+    },
+    {
+      "epoch": 1.715807413719642,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.2352865959430415e-07,
+      "loss": 0.3104,
+      "step": 20135
+    },
+    {
+      "epoch": 1.715892628887942,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.234558345730505e-07,
+      "loss": 0.6593,
+      "step": 20136
+    },
+    {
+      "epoch": 1.7159778440562419,
+      "grad_norm": 26.0,
+      "learning_rate": 1.2338302990958311e-07,
+      "loss": 0.4735,
+      "step": 20137
+    },
+    {
+      "epoch": 1.7160630592245418,
+      "grad_norm": 13.875,
+      "learning_rate": 1.2331024560521695e-07,
+      "loss": 0.4946,
+      "step": 20138
+    },
+    {
+      "epoch": 1.7161482743928418,
+      "grad_norm": 17.5,
+      "learning_rate": 1.232374816612679e-07,
+      "loss": 0.4564,
+      "step": 20139
+    },
+    {
+      "epoch": 1.7162334895611417,
+      "grad_norm": 14.875,
+      "learning_rate": 1.231647380790503e-07,
+      "loss": 0.3257,
+      "step": 20140
+    },
+    {
+      "epoch": 1.7163187047294417,
+      "grad_norm": 19.875,
+      "learning_rate": 1.230920148598791e-07,
+      "loss": 1.0537,
+      "step": 20141
+    },
+    {
+      "epoch": 1.7164039198977417,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.2301931200506864e-07,
+      "loss": 0.4197,
+      "step": 20142
+    },
+    {
+      "epoch": 1.7164891350660416,
+      "grad_norm": 18.375,
+      "learning_rate": 1.229466295159322e-07,
+      "loss": 0.8462,
+      "step": 20143
+    },
+    {
+      "epoch": 1.7165743502343416,
+      "grad_norm": 20.875,
+      "learning_rate": 1.2287396739378343e-07,
+      "loss": 0.7014,
+      "step": 20144
+    },
+    {
+      "epoch": 1.7166595654026415,
+      "grad_norm": 15.375,
+      "learning_rate": 1.2280132563993575e-07,
+      "loss": 0.6376,
+      "step": 20145
+    },
+    {
+      "epoch": 1.7167447805709415,
+      "grad_norm": 20.125,
+      "learning_rate": 1.227287042557014e-07,
+      "loss": 0.6549,
+      "step": 20146
+    },
+    {
+      "epoch": 1.7168299957392414,
+      "grad_norm": 13.125,
+      "learning_rate": 1.2265610324239313e-07,
+      "loss": 0.4254,
+      "step": 20147
+    },
+    {
+      "epoch": 1.7169152109075414,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.225835226013225e-07,
+      "loss": 0.2966,
+      "step": 20148
+    },
+    {
+      "epoch": 1.7170004260758414,
+      "grad_norm": 19.25,
+      "learning_rate": 1.225109623338014e-07,
+      "loss": 0.9581,
+      "step": 20149
+    },
+    {
+      "epoch": 1.7170856412441413,
+      "grad_norm": 12.375,
+      "learning_rate": 1.2243842244114082e-07,
+      "loss": 0.3737,
+      "step": 20150
+    },
+    {
+      "epoch": 1.7171708564124413,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2236590292465212e-07,
+      "loss": 0.7292,
+      "step": 20151
+    },
+    {
+      "epoch": 1.7172560715807412,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.2229340378564532e-07,
+      "loss": 0.5515,
+      "step": 20152
+    },
+    {
+      "epoch": 1.7173412867490412,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2222092502543083e-07,
+      "loss": 0.4632,
+      "step": 20153
+    },
+    {
+      "epoch": 1.7174265019173411,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2214846664531807e-07,
+      "loss": 0.6727,
+      "step": 20154
+    },
+    {
+      "epoch": 1.717511717085641,
+      "grad_norm": 18.0,
+      "learning_rate": 1.2207602864661676e-07,
+      "loss": 0.8742,
+      "step": 20155
+    },
+    {
+      "epoch": 1.717596932253941,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2200361103063594e-07,
+      "loss": 0.3635,
+      "step": 20156
+    },
+    {
+      "epoch": 1.717682147422241,
+      "grad_norm": 18.125,
+      "learning_rate": 1.219312137986839e-07,
+      "loss": 0.6539,
+      "step": 20157
+    },
+    {
+      "epoch": 1.717767362590541,
+      "grad_norm": 12.0,
+      "learning_rate": 1.218588369520693e-07,
+      "loss": 0.3792,
+      "step": 20158
+    },
+    {
+      "epoch": 1.717852577758841,
+      "grad_norm": 22.125,
+      "learning_rate": 1.2178648049210012e-07,
+      "loss": 0.9524,
+      "step": 20159
+    },
+    {
+      "epoch": 1.7179377929271409,
+      "grad_norm": 25.0,
+      "learning_rate": 1.2171414442008352e-07,
+      "loss": 0.828,
+      "step": 20160
+    },
+    {
+      "epoch": 1.7180230080954408,
+      "grad_norm": 18.625,
+      "learning_rate": 1.2164182873732694e-07,
+      "loss": 0.7211,
+      "step": 20161
+    },
+    {
+      "epoch": 1.7181082232637408,
+      "grad_norm": 16.25,
+      "learning_rate": 1.2156953344513736e-07,
+      "loss": 0.7031,
+      "step": 20162
+    },
+    {
+      "epoch": 1.7181934384320408,
+      "grad_norm": 18.0,
+      "learning_rate": 1.2149725854482083e-07,
+      "loss": 0.7063,
+      "step": 20163
+    },
+    {
+      "epoch": 1.7182786536003407,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.214250040376838e-07,
+      "loss": 0.3333,
+      "step": 20164
+    },
+    {
+      "epoch": 1.7183638687686407,
+      "grad_norm": 14.125,
+      "learning_rate": 1.213527699250318e-07,
+      "loss": 0.6548,
+      "step": 20165
+    },
+    {
+      "epoch": 1.7184490839369406,
+      "grad_norm": 13.875,
+      "learning_rate": 1.2128055620816996e-07,
+      "loss": 0.4636,
+      "step": 20166
+    },
+    {
+      "epoch": 1.7185342991052406,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2120836288840352e-07,
+      "loss": 0.6505,
+      "step": 20167
+    },
+    {
+      "epoch": 1.7186195142735405,
+      "grad_norm": 17.5,
+      "learning_rate": 1.211361899670374e-07,
+      "loss": 0.9041,
+      "step": 20168
+    },
+    {
+      "epoch": 1.7187047294418405,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2106403744537517e-07,
+      "loss": 0.6368,
+      "step": 20169
+    },
+    {
+      "epoch": 1.7187899446101405,
+      "grad_norm": 19.125,
+      "learning_rate": 1.209919053247212e-07,
+      "loss": 0.4667,
+      "step": 20170
+    },
+    {
+      "epoch": 1.7188751597784404,
+      "grad_norm": 21.125,
+      "learning_rate": 1.209197936063787e-07,
+      "loss": 0.4623,
+      "step": 20171
+    },
+    {
+      "epoch": 1.7189603749467404,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.20847702291651e-07,
+      "loss": 0.5971,
+      "step": 20172
+    },
+    {
+      "epoch": 1.7190455901150403,
+      "grad_norm": 24.875,
+      "learning_rate": 1.207756313818406e-07,
+      "loss": 1.0578,
+      "step": 20173
+    },
+    {
+      "epoch": 1.7191308052833403,
+      "grad_norm": 17.0,
+      "learning_rate": 1.2070358087825e-07,
+      "loss": 0.6266,
+      "step": 20174
+    },
+    {
+      "epoch": 1.7192160204516402,
+      "grad_norm": 19.625,
+      "learning_rate": 1.2063155078218148e-07,
+      "loss": 0.5084,
+      "step": 20175
+    },
+    {
+      "epoch": 1.7193012356199402,
+      "grad_norm": 24.125,
+      "learning_rate": 1.2055954109493658e-07,
+      "loss": 0.5791,
+      "step": 20176
+    },
+    {
+      "epoch": 1.7193864507882402,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.2048755181781627e-07,
+      "loss": 0.4565,
+      "step": 20177
+    },
+    {
+      "epoch": 1.71947166595654,
+      "grad_norm": 15.0,
+      "learning_rate": 1.2041558295212197e-07,
+      "loss": 0.5942,
+      "step": 20178
+    },
+    {
+      "epoch": 1.71955688112484,
+      "grad_norm": 13.625,
+      "learning_rate": 1.2034363449915377e-07,
+      "loss": 0.4087,
+      "step": 20179
+    },
+    {
+      "epoch": 1.71964209629314,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.2027170646021214e-07,
+      "loss": 0.7813,
+      "step": 20180
+    },
+    {
+      "epoch": 1.71972731146144,
+      "grad_norm": 16.5,
+      "learning_rate": 1.2019979883659688e-07,
+      "loss": 0.5385,
+      "step": 20181
+    },
+    {
+      "epoch": 1.71981252662974,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.2012791162960724e-07,
+      "loss": 0.3076,
+      "step": 20182
+    },
+    {
+      "epoch": 1.71989774179804,
+      "grad_norm": 11.0,
+      "learning_rate": 1.2005604484054232e-07,
+      "loss": 0.3014,
+      "step": 20183
+    },
+    {
+      "epoch": 1.7199829569663398,
+      "grad_norm": 22.0,
+      "learning_rate": 1.1998419847070092e-07,
+      "loss": 1.0147,
+      "step": 20184
+    },
+    {
+      "epoch": 1.7200681721346398,
+      "grad_norm": 18.5,
+      "learning_rate": 1.199123725213816e-07,
+      "loss": 1.0917,
+      "step": 20185
+    },
+    {
+      "epoch": 1.7201533873029398,
+      "grad_norm": 16.25,
+      "learning_rate": 1.198405669938822e-07,
+      "loss": 0.4462,
+      "step": 20186
+    },
+    {
+      "epoch": 1.7202386024712397,
+      "grad_norm": 20.5,
+      "learning_rate": 1.1976878188949992e-07,
+      "loss": 0.755,
+      "step": 20187
+    },
+    {
+      "epoch": 1.7203238176395397,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.1969701720953229e-07,
+      "loss": 0.5837,
+      "step": 20188
+    },
+    {
+      "epoch": 1.7204090328078396,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.1962527295527631e-07,
+      "loss": 0.2258,
+      "step": 20189
+    },
+    {
+      "epoch": 1.7204942479761396,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1955354912802823e-07,
+      "loss": 0.5737,
+      "step": 20190
+    },
+    {
+      "epoch": 1.7205794631444395,
+      "grad_norm": 14.5,
+      "learning_rate": 1.1948184572908417e-07,
+      "loss": 0.4786,
+      "step": 20191
+    },
+    {
+      "epoch": 1.7206646783127395,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1941016275974007e-07,
+      "loss": 0.5263,
+      "step": 20192
+    },
+    {
+      "epoch": 1.7207498934810395,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1933850022129143e-07,
+      "loss": 0.5117,
+      "step": 20193
+    },
+    {
+      "epoch": 1.7208351086493394,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.1926685811503302e-07,
+      "loss": 0.5417,
+      "step": 20194
+    },
+    {
+      "epoch": 1.7209203238176394,
+      "grad_norm": 19.625,
+      "learning_rate": 1.1919523644225939e-07,
+      "loss": 0.7597,
+      "step": 20195
+    },
+    {
+      "epoch": 1.7210055389859393,
+      "grad_norm": 21.5,
+      "learning_rate": 1.1912363520426476e-07,
+      "loss": 0.6601,
+      "step": 20196
+    },
+    {
+      "epoch": 1.7210907541542393,
+      "grad_norm": 21.125,
+      "learning_rate": 1.1905205440234339e-07,
+      "loss": 0.4605,
+      "step": 20197
+    },
+    {
+      "epoch": 1.7211759693225395,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.189804940377888e-07,
+      "loss": 0.4952,
+      "step": 20198
+    },
+    {
+      "epoch": 1.7212611844908394,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1890895411189387e-07,
+      "loss": 0.4862,
+      "step": 20199
+    },
+    {
+      "epoch": 1.7213463996591394,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1883743462595147e-07,
+      "loss": 0.8428,
+      "step": 20200
+    },
+    {
+      "epoch": 1.7214316148274393,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1876593558125412e-07,
+      "loss": 0.5378,
+      "step": 20201
+    },
+    {
+      "epoch": 1.7215168299957393,
+      "grad_norm": 20.5,
+      "learning_rate": 1.1869445697909432e-07,
+      "loss": 0.5653,
+      "step": 20202
+    },
+    {
+      "epoch": 1.7216020451640393,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.1862299882076281e-07,
+      "loss": 0.555,
+      "step": 20203
+    },
+    {
+      "epoch": 1.7216872603323392,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.1855156110755148e-07,
+      "loss": 0.2329,
+      "step": 20204
+    },
+    {
+      "epoch": 1.7217724755006392,
+      "grad_norm": 13.375,
+      "learning_rate": 1.184801438407511e-07,
+      "loss": 0.4022,
+      "step": 20205
+    },
+    {
+      "epoch": 1.7218576906689391,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1840874702165259e-07,
+      "loss": 0.6301,
+      "step": 20206
+    },
+    {
+      "epoch": 1.721942905837239,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1833737065154574e-07,
+      "loss": 0.4992,
+      "step": 20207
+    },
+    {
+      "epoch": 1.722028121005539,
+      "grad_norm": 12.375,
+      "learning_rate": 1.1826601473172064e-07,
+      "loss": 0.6199,
+      "step": 20208
+    },
+    {
+      "epoch": 1.722113336173839,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1819467926346654e-07,
+      "loss": 0.6434,
+      "step": 20209
+    },
+    {
+      "epoch": 1.722198551342139,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1812336424807308e-07,
+      "loss": 0.6117,
+      "step": 20210
+    },
+    {
+      "epoch": 1.722283766510439,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1805206968682855e-07,
+      "loss": 0.4356,
+      "step": 20211
+    },
+    {
+      "epoch": 1.7223689816787389,
+      "grad_norm": 12.25,
+      "learning_rate": 1.179807955810211e-07,
+      "loss": 0.496,
+      "step": 20212
+    },
+    {
+      "epoch": 1.7224541968470388,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1790954193193912e-07,
+      "loss": 0.3543,
+      "step": 20213
+    },
+    {
+      "epoch": 1.7225394120153388,
+      "grad_norm": 19.875,
+      "learning_rate": 1.1783830874087008e-07,
+      "loss": 0.806,
+      "step": 20214
+    },
+    {
+      "epoch": 1.7226246271836387,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.1776709600910152e-07,
+      "loss": 0.8668,
+      "step": 20215
+    },
+    {
+      "epoch": 1.7227098423519387,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1769590373791981e-07,
+      "loss": 0.7513,
+      "step": 20216
+    },
+    {
+      "epoch": 1.7227950575202386,
+      "grad_norm": 18.75,
+      "learning_rate": 1.1762473192861168e-07,
+      "loss": 0.6213,
+      "step": 20217
+    },
+    {
+      "epoch": 1.7228802726885386,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1755358058246375e-07,
+      "loss": 0.6169,
+      "step": 20218
+    },
+    {
+      "epoch": 1.7229654878568386,
+      "grad_norm": 15.875,
+      "learning_rate": 1.1748244970076123e-07,
+      "loss": 0.801,
+      "step": 20219
+    },
+    {
+      "epoch": 1.7230507030251385,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1741133928478949e-07,
+      "loss": 0.6383,
+      "step": 20220
+    },
+    {
+      "epoch": 1.7231359181934385,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1734024933583373e-07,
+      "loss": 0.6675,
+      "step": 20221
+    },
+    {
+      "epoch": 1.7232211333617384,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1726917985517865e-07,
+      "loss": 0.4613,
+      "step": 20222
+    },
+    {
+      "epoch": 1.7233063485300384,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.1719813084410877e-07,
+      "loss": 0.5021,
+      "step": 20223
+    },
+    {
+      "epoch": 1.7233915636983383,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1712710230390764e-07,
+      "loss": 0.6141,
+      "step": 20224
+    },
+    {
+      "epoch": 1.7234767788666383,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1705609423585881e-07,
+      "loss": 0.6159,
+      "step": 20225
+    },
+    {
+      "epoch": 1.7235619940349383,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.1698510664124599e-07,
+      "loss": 0.3312,
+      "step": 20226
+    },
+    {
+      "epoch": 1.7236472092032382,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1691413952135133e-07,
+      "loss": 0.6655,
+      "step": 20227
+    },
+    {
+      "epoch": 1.7237324243715382,
+      "grad_norm": 20.0,
+      "learning_rate": 1.1684319287745785e-07,
+      "loss": 0.9766,
+      "step": 20228
+    },
+    {
+      "epoch": 1.7238176395398381,
+      "grad_norm": 22.375,
+      "learning_rate": 1.1677226671084716e-07,
+      "loss": 0.6145,
+      "step": 20229
+    },
+    {
+      "epoch": 1.723902854708138,
+      "grad_norm": 23.375,
+      "learning_rate": 1.16701361022801e-07,
+      "loss": 0.9781,
+      "step": 20230
+    },
+    {
+      "epoch": 1.723988069876438,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1663047581460099e-07,
+      "loss": 0.6627,
+      "step": 20231
+    },
+    {
+      "epoch": 1.724073285044738,
+      "grad_norm": 26.75,
+      "learning_rate": 1.165596110875282e-07,
+      "loss": 0.7677,
+      "step": 20232
+    },
+    {
+      "epoch": 1.724158500213038,
+      "grad_norm": 13.0,
+      "learning_rate": 1.164887668428627e-07,
+      "loss": 0.4149,
+      "step": 20233
+    },
+    {
+      "epoch": 1.724243715381338,
+      "grad_norm": 14.25,
+      "learning_rate": 1.16417943081885e-07,
+      "loss": 0.5992,
+      "step": 20234
+    },
+    {
+      "epoch": 1.7243289305496379,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1634713980587534e-07,
+      "loss": 0.9113,
+      "step": 20235
+    },
+    {
+      "epoch": 1.7244141457179378,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1627635701611282e-07,
+      "loss": 0.825,
+      "step": 20236
+    },
+    {
+      "epoch": 1.7244993608862378,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1620559471387627e-07,
+      "loss": 0.6716,
+      "step": 20237
+    },
+    {
+      "epoch": 1.7245845760545377,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1613485290044468e-07,
+      "loss": 0.7599,
+      "step": 20238
+    },
+    {
+      "epoch": 1.7246697912228377,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1606413157709662e-07,
+      "loss": 1.1758,
+      "step": 20239
+    },
+    {
+      "epoch": 1.7247550063911377,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1599343074511008e-07,
+      "loss": 0.4573,
+      "step": 20240
+    },
+    {
+      "epoch": 1.7248402215594376,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1592275040576251e-07,
+      "loss": 0.7164,
+      "step": 20241
+    },
+    {
+      "epoch": 1.7249254367277376,
+      "grad_norm": 15.125,
+      "learning_rate": 1.1585209056033108e-07,
+      "loss": 0.6129,
+      "step": 20242
+    },
+    {
+      "epoch": 1.7250106518960375,
+      "grad_norm": 14.0,
+      "learning_rate": 1.1578145121009312e-07,
+      "loss": 0.4169,
+      "step": 20243
+    },
+    {
+      "epoch": 1.7250958670643375,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1571083235632468e-07,
+      "loss": 0.4133,
+      "step": 20244
+    },
+    {
+      "epoch": 1.7251810822326374,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1564023400030239e-07,
+      "loss": 0.6,
+      "step": 20245
+    },
+    {
+      "epoch": 1.7252662974009374,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1556965614330146e-07,
+      "loss": 0.7857,
+      "step": 20246
+    },
+    {
+      "epoch": 1.7253515125692374,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.1549909878659754e-07,
+      "loss": 0.3227,
+      "step": 20247
+    },
+    {
+      "epoch": 1.7254367277375373,
+      "grad_norm": 27.875,
+      "learning_rate": 1.1542856193146587e-07,
+      "loss": 0.7216,
+      "step": 20248
+    },
+    {
+      "epoch": 1.7255219429058373,
+      "grad_norm": 27.25,
+      "learning_rate": 1.1535804557918112e-07,
+      "loss": 0.8416,
+      "step": 20249
+    },
+    {
+      "epoch": 1.7256071580741372,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.1528754973101728e-07,
+      "loss": 0.462,
+      "step": 20250
+    },
+    {
+      "epoch": 1.7256923732424372,
+      "grad_norm": 18.625,
+      "learning_rate": 1.1521707438824872e-07,
+      "loss": 0.678,
+      "step": 20251
+    },
+    {
+      "epoch": 1.7257775884107371,
+      "grad_norm": 19.875,
+      "learning_rate": 1.1514661955214834e-07,
+      "loss": 0.779,
+      "step": 20252
+    },
+    {
+      "epoch": 1.725862803579037,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1507618522399011e-07,
+      "loss": 0.4549,
+      "step": 20253
+    },
+    {
+      "epoch": 1.725948018747337,
+      "grad_norm": 10.0,
+      "learning_rate": 1.1500577140504621e-07,
+      "loss": 0.2901,
+      "step": 20254
+    },
+    {
+      "epoch": 1.726033233915637,
+      "grad_norm": 17.125,
+      "learning_rate": 1.1493537809658936e-07,
+      "loss": 0.6195,
+      "step": 20255
+    },
+    {
+      "epoch": 1.726118449083937,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1486500529989161e-07,
+      "loss": 0.8804,
+      "step": 20256
+    },
+    {
+      "epoch": 1.726203664252237,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1479465301622486e-07,
+      "loss": 0.709,
+      "step": 20257
+    },
+    {
+      "epoch": 1.7262888794205369,
+      "grad_norm": 18.25,
+      "learning_rate": 1.1472432124686017e-07,
+      "loss": 0.6032,
+      "step": 20258
+    },
+    {
+      "epoch": 1.7263740945888368,
+      "grad_norm": 20.5,
+      "learning_rate": 1.1465400999306875e-07,
+      "loss": 0.8568,
+      "step": 20259
+    },
+    {
+      "epoch": 1.7264593097571368,
+      "grad_norm": 18.625,
+      "learning_rate": 1.1458371925612083e-07,
+      "loss": 0.5294,
+      "step": 20260
+    },
+    {
+      "epoch": 1.7265445249254368,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1451344903728679e-07,
+      "loss": 0.6295,
+      "step": 20261
+    },
+    {
+      "epoch": 1.7266297400937367,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1444319933783685e-07,
+      "loss": 0.4918,
+      "step": 20262
+    },
+    {
+      "epoch": 1.7267149552620367,
+      "grad_norm": 15.0,
+      "learning_rate": 1.1437297015903987e-07,
+      "loss": 0.6205,
+      "step": 20263
+    },
+    {
+      "epoch": 1.7268001704303366,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1430276150216538e-07,
+      "loss": 0.5765,
+      "step": 20264
+    },
+    {
+      "epoch": 1.7268853855986366,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1423257336848223e-07,
+      "loss": 0.5456,
+      "step": 20265
+    },
+    {
+      "epoch": 1.7269706007669365,
+      "grad_norm": 14.5,
+      "learning_rate": 1.141624057592583e-07,
+      "loss": 0.553,
+      "step": 20266
+    },
+    {
+      "epoch": 1.7270558159352365,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1409225867576215e-07,
+      "loss": 0.5272,
+      "step": 20267
+    },
+    {
+      "epoch": 1.7271410311035365,
+      "grad_norm": 15.125,
+      "learning_rate": 1.1402213211926083e-07,
+      "loss": 0.5327,
+      "step": 20268
+    },
+    {
+      "epoch": 1.7272262462718364,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1395202609102194e-07,
+      "loss": 0.6694,
+      "step": 20269
+    },
+    {
+      "epoch": 1.7273114614401364,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.1388194059231264e-07,
+      "loss": 0.387,
+      "step": 20270
+    },
+    {
+      "epoch": 1.7273966766084363,
+      "grad_norm": 15.75,
+      "learning_rate": 1.1381187562439874e-07,
+      "loss": 0.6186,
+      "step": 20271
+    },
+    {
+      "epoch": 1.7274818917767363,
+      "grad_norm": 12.125,
+      "learning_rate": 1.1374183118854687e-07,
+      "loss": 0.3256,
+      "step": 20272
+    },
+    {
+      "epoch": 1.7275671069450362,
+      "grad_norm": 14.5,
+      "learning_rate": 1.1367180728602267e-07,
+      "loss": 0.6508,
+      "step": 20273
+    },
+    {
+      "epoch": 1.7276523221133362,
+      "grad_norm": 17.5,
+      "learning_rate": 1.1360180391809181e-07,
+      "loss": 0.7319,
+      "step": 20274
+    },
+    {
+      "epoch": 1.7277375372816361,
+      "grad_norm": 22.625,
+      "learning_rate": 1.135318210860191e-07,
+      "loss": 0.7254,
+      "step": 20275
+    },
+    {
+      "epoch": 1.727822752449936,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.1346185879106908e-07,
+      "loss": 0.6313,
+      "step": 20276
+    },
+    {
+      "epoch": 1.727907967618236,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.1339191703450602e-07,
+      "loss": 0.5576,
+      "step": 20277
+    },
+    {
+      "epoch": 1.727993182786536,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1332199581759393e-07,
+      "loss": 0.7117,
+      "step": 20278
+    },
+    {
+      "epoch": 1.728078397954836,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1325209514159677e-07,
+      "loss": 0.6357,
+      "step": 20279
+    },
+    {
+      "epoch": 1.728163613123136,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1318221500777702e-07,
+      "loss": 0.7452,
+      "step": 20280
+    },
+    {
+      "epoch": 1.728248828291436,
+      "grad_norm": 12.125,
+      "learning_rate": 1.1311235541739768e-07,
+      "loss": 0.7665,
+      "step": 20281
+    },
+    {
+      "epoch": 1.7283340434597358,
+      "grad_norm": 15.5,
+      "learning_rate": 1.1304251637172165e-07,
+      "loss": 0.8261,
+      "step": 20282
+    },
+    {
+      "epoch": 1.7284192586280358,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1297269787201068e-07,
+      "loss": 0.5681,
+      "step": 20283
+    },
+    {
+      "epoch": 1.7285044737963358,
+      "grad_norm": 20.625,
+      "learning_rate": 1.1290289991952627e-07,
+      "loss": 0.7,
+      "step": 20284
+    },
+    {
+      "epoch": 1.7285896889646357,
+      "grad_norm": 11.25,
+      "learning_rate": 1.1283312251552978e-07,
+      "loss": 0.4455,
+      "step": 20285
+    },
+    {
+      "epoch": 1.7286749041329357,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.1276336566128226e-07,
+      "loss": 0.6668,
+      "step": 20286
+    },
+    {
+      "epoch": 1.7287601193012356,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1269362935804454e-07,
+      "loss": 0.8941,
+      "step": 20287
+    },
+    {
+      "epoch": 1.7288453344695356,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1262391360707642e-07,
+      "loss": 0.5162,
+      "step": 20288
+    },
+    {
+      "epoch": 1.7289305496378355,
+      "grad_norm": 17.625,
+      "learning_rate": 1.1255421840963788e-07,
+      "loss": 0.7474,
+      "step": 20289
+    },
+    {
+      "epoch": 1.7290157648061355,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1248454376698844e-07,
+      "loss": 0.5555,
+      "step": 20290
+    },
+    {
+      "epoch": 1.7291009799744355,
+      "grad_norm": 12.375,
+      "learning_rate": 1.1241488968038727e-07,
+      "loss": 0.3473,
+      "step": 20291
+    },
+    {
+      "epoch": 1.7291861951427354,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.1234525615109307e-07,
+      "loss": 0.2393,
+      "step": 20292
+    },
+    {
+      "epoch": 1.7292714103110354,
+      "grad_norm": 24.625,
+      "learning_rate": 1.1227564318036385e-07,
+      "loss": 0.5151,
+      "step": 20293
+    },
+    {
+      "epoch": 1.7293566254793353,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1220605076945778e-07,
+      "loss": 0.5041,
+      "step": 20294
+    },
+    {
+      "epoch": 1.7294418406476353,
+      "grad_norm": 14.125,
+      "learning_rate": 1.1213647891963276e-07,
+      "loss": 0.3513,
+      "step": 20295
+    },
+    {
+      "epoch": 1.7295270558159352,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1206692763214553e-07,
+      "loss": 0.4617,
+      "step": 20296
+    },
+    {
+      "epoch": 1.7296122709842352,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1199739690825329e-07,
+      "loss": 0.7602,
+      "step": 20297
+    },
+    {
+      "epoch": 1.7296974861525352,
+      "grad_norm": 13.0,
+      "learning_rate": 1.119278867492124e-07,
+      "loss": 0.3296,
+      "step": 20298
+    },
+    {
+      "epoch": 1.7297827013208351,
+      "grad_norm": 20.875,
+      "learning_rate": 1.1185839715627923e-07,
+      "loss": 0.7227,
+      "step": 20299
+    },
+    {
+      "epoch": 1.729867916489135,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1178892813070927e-07,
+      "loss": 0.5868,
+      "step": 20300
+    },
+    {
+      "epoch": 1.729953131657435,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1171947967375779e-07,
+      "loss": 0.5576,
+      "step": 20301
+    },
+    {
+      "epoch": 1.730038346825735,
+      "grad_norm": 18.5,
+      "learning_rate": 1.116500517866799e-07,
+      "loss": 1.0257,
+      "step": 20302
+    },
+    {
+      "epoch": 1.730123561994035,
+      "grad_norm": 28.0,
+      "learning_rate": 1.1158064447073028e-07,
+      "loss": 0.7797,
+      "step": 20303
+    },
+    {
+      "epoch": 1.730208777162335,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.1151125772716334e-07,
+      "loss": 0.555,
+      "step": 20304
+    },
+    {
+      "epoch": 1.7302939923306349,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1144189155723267e-07,
+      "loss": 0.5972,
+      "step": 20305
+    },
+    {
+      "epoch": 1.7303792074989348,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.1137254596219186e-07,
+      "loss": 0.4612,
+      "step": 20306
+    },
+    {
+      "epoch": 1.7304644226672348,
+      "grad_norm": 12.25,
+      "learning_rate": 1.1130322094329418e-07,
+      "loss": 0.3243,
+      "step": 20307
+    },
+    {
+      "epoch": 1.7305496378355347,
+      "grad_norm": 16.0,
+      "learning_rate": 1.1123391650179255e-07,
+      "loss": 0.5448,
+      "step": 20308
+    },
+    {
+      "epoch": 1.7306348530038347,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1116463263893915e-07,
+      "loss": 0.7366,
+      "step": 20309
+    },
+    {
+      "epoch": 1.7307200681721346,
+      "grad_norm": 13.875,
+      "learning_rate": 1.1109536935598589e-07,
+      "loss": 0.6221,
+      "step": 20310
+    },
+    {
+      "epoch": 1.7308052833404346,
+      "grad_norm": 14.0,
+      "learning_rate": 1.1102612665418441e-07,
+      "loss": 0.2628,
+      "step": 20311
+    },
+    {
+      "epoch": 1.7308904985087346,
+      "grad_norm": 20.0,
+      "learning_rate": 1.109569045347865e-07,
+      "loss": 0.9918,
+      "step": 20312
+    },
+    {
+      "epoch": 1.7309757136770345,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1088770299904252e-07,
+      "loss": 0.7479,
+      "step": 20313
+    },
+    {
+      "epoch": 1.7310609288453345,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1081852204820317e-07,
+      "loss": 0.5524,
+      "step": 20314
+    },
+    {
+      "epoch": 1.7311461440136344,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1074936168351855e-07,
+      "loss": 0.6101,
+      "step": 20315
+    },
+    {
+      "epoch": 1.7312313591819344,
+      "grad_norm": 14.125,
+      "learning_rate": 1.106802219062389e-07,
+      "loss": 0.6001,
+      "step": 20316
+    },
+    {
+      "epoch": 1.7313165743502343,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1061110271761335e-07,
+      "loss": 0.5876,
+      "step": 20317
+    },
+    {
+      "epoch": 1.7314017895185343,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1054200411889066e-07,
+      "loss": 0.7039,
+      "step": 20318
+    },
+    {
+      "epoch": 1.7314870046868343,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.1047292611131966e-07,
+      "loss": 0.55,
+      "step": 20319
+    },
+    {
+      "epoch": 1.7315722198551342,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.1040386869614894e-07,
+      "loss": 0.4629,
+      "step": 20320
+    },
+    {
+      "epoch": 1.7316574350234342,
+      "grad_norm": 16.5,
+      "learning_rate": 1.103348318746264e-07,
+      "loss": 0.7466,
+      "step": 20321
+    },
+    {
+      "epoch": 1.7317426501917341,
+      "grad_norm": 19.375,
+      "learning_rate": 1.1026581564799923e-07,
+      "loss": 0.8077,
+      "step": 20322
+    },
+    {
+      "epoch": 1.731827865360034,
+      "grad_norm": 15.625,
+      "learning_rate": 1.1019682001751492e-07,
+      "loss": 0.7528,
+      "step": 20323
+    },
+    {
+      "epoch": 1.731913080528334,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.1012784498442036e-07,
+      "loss": 0.6486,
+      "step": 20324
+    },
+    {
+      "epoch": 1.731998295696634,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1005889054996194e-07,
+      "loss": 0.3865,
+      "step": 20325
+    },
+    {
+      "epoch": 1.732083510864934,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.0998995671538545e-07,
+      "loss": 0.5874,
+      "step": 20326
+    },
+    {
+      "epoch": 1.732168726033234,
+      "grad_norm": 19.375,
+      "learning_rate": 1.0992104348193686e-07,
+      "loss": 0.7047,
+      "step": 20327
+    },
+    {
+      "epoch": 1.7322539412015339,
+      "grad_norm": 12.25,
+      "learning_rate": 1.098521508508614e-07,
+      "loss": 0.4578,
+      "step": 20328
+    },
+    {
+      "epoch": 1.7323391563698338,
+      "grad_norm": 16.75,
+      "learning_rate": 1.0978327882340434e-07,
+      "loss": 0.5496,
+      "step": 20329
+    },
+    {
+      "epoch": 1.7324243715381338,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0971442740080968e-07,
+      "loss": 0.2163,
+      "step": 20330
+    },
+    {
+      "epoch": 1.7325095867064337,
+      "grad_norm": 21.875,
+      "learning_rate": 1.0964559658432211e-07,
+      "loss": 0.7502,
+      "step": 20331
+    },
+    {
+      "epoch": 1.7325948018747337,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0957678637518538e-07,
+      "loss": 0.3934,
+      "step": 20332
+    },
+    {
+      "epoch": 1.7326800170430336,
+      "grad_norm": 22.5,
+      "learning_rate": 1.0950799677464277e-07,
+      "loss": 0.6742,
+      "step": 20333
+    },
+    {
+      "epoch": 1.7327652322113336,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.0943922778393776e-07,
+      "loss": 0.6318,
+      "step": 20334
+    },
+    {
+      "epoch": 1.7328504473796336,
+      "grad_norm": 14.125,
+      "learning_rate": 1.0937047940431253e-07,
+      "loss": 0.5712,
+      "step": 20335
+    },
+    {
+      "epoch": 1.7329356625479335,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0930175163700985e-07,
+      "loss": 0.4628,
+      "step": 20336
+    },
+    {
+      "epoch": 1.7330208777162335,
+      "grad_norm": 16.5,
+      "learning_rate": 1.092330444832715e-07,
+      "loss": 0.4205,
+      "step": 20337
+    },
+    {
+      "epoch": 1.7331060928845334,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0916435794433939e-07,
+      "loss": 0.5404,
+      "step": 20338
+    },
+    {
+      "epoch": 1.7331913080528334,
+      "grad_norm": 16.625,
+      "learning_rate": 1.0909569202145435e-07,
+      "loss": 0.8362,
+      "step": 20339
+    },
+    {
+      "epoch": 1.7332765232211333,
+      "grad_norm": 17.125,
+      "learning_rate": 1.090270467158576e-07,
+      "loss": 0.8025,
+      "step": 20340
+    },
+    {
+      "epoch": 1.7333617383894333,
+      "grad_norm": 18.625,
+      "learning_rate": 1.0895842202878911e-07,
+      "loss": 0.7452,
+      "step": 20341
+    },
+    {
+      "epoch": 1.7334469535577333,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0888981796148973e-07,
+      "loss": 0.6066,
+      "step": 20342
+    },
+    {
+      "epoch": 1.7335321687260332,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.0882123451519857e-07,
+      "loss": 0.5634,
+      "step": 20343
+    },
+    {
+      "epoch": 1.7336173838943332,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0875267169115523e-07,
+      "loss": 0.6802,
+      "step": 20344
+    },
+    {
+      "epoch": 1.7337025990626331,
+      "grad_norm": 20.375,
+      "learning_rate": 1.0868412949059868e-07,
+      "loss": 0.6934,
+      "step": 20345
+    },
+    {
+      "epoch": 1.733787814230933,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0861560791476794e-07,
+      "loss": 0.5575,
+      "step": 20346
+    },
+    {
+      "epoch": 1.733873029399233,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0854710696490064e-07,
+      "loss": 0.3875,
+      "step": 20347
+    },
+    {
+      "epoch": 1.733958244567533,
+      "grad_norm": 15.75,
+      "learning_rate": 1.084786266422351e-07,
+      "loss": 0.3892,
+      "step": 20348
+    },
+    {
+      "epoch": 1.734043459735833,
+      "grad_norm": 25.5,
+      "learning_rate": 1.0841016694800838e-07,
+      "loss": 1.094,
+      "step": 20349
+    },
+    {
+      "epoch": 1.734128674904133,
+      "grad_norm": 20.875,
+      "learning_rate": 1.083417278834581e-07,
+      "loss": 1.0756,
+      "step": 20350
+    },
+    {
+      "epoch": 1.7342138900724329,
+      "grad_norm": 15.625,
+      "learning_rate": 1.082733094498209e-07,
+      "loss": 0.5394,
+      "step": 20351
+    },
+    {
+      "epoch": 1.7342991052407328,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0820491164833288e-07,
+      "loss": 0.845,
+      "step": 20352
+    },
+    {
+      "epoch": 1.7343843204090328,
+      "grad_norm": 17.125,
+      "learning_rate": 1.081365344802303e-07,
+      "loss": 0.727,
+      "step": 20353
+    },
+    {
+      "epoch": 1.7344695355773327,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0806817794674878e-07,
+      "loss": 0.4667,
+      "step": 20354
+    },
+    {
+      "epoch": 1.7345547507456327,
+      "grad_norm": 22.25,
+      "learning_rate": 1.0799984204912378e-07,
+      "loss": 0.8753,
+      "step": 20355
+    },
+    {
+      "epoch": 1.7346399659139327,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0793152678859011e-07,
+      "loss": 0.7515,
+      "step": 20356
+    },
+    {
+      "epoch": 1.7347251810822326,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0786323216638181e-07,
+      "loss": 0.6915,
+      "step": 20357
+    },
+    {
+      "epoch": 1.7348103962505326,
+      "grad_norm": 18.25,
+      "learning_rate": 1.0779495818373356e-07,
+      "loss": 0.7628,
+      "step": 20358
+    },
+    {
+      "epoch": 1.7348956114188325,
+      "grad_norm": 19.0,
+      "learning_rate": 1.077267048418791e-07,
+      "loss": 0.6592,
+      "step": 20359
+    },
+    {
+      "epoch": 1.7349808265871325,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0765847214205149e-07,
+      "loss": 0.5753,
+      "step": 20360
+    },
+    {
+      "epoch": 1.7350660417554324,
+      "grad_norm": 15.625,
+      "learning_rate": 1.0759026008548404e-07,
+      "loss": 0.2975,
+      "step": 20361
+    },
+    {
+      "epoch": 1.7351512569237324,
+      "grad_norm": 27.75,
+      "learning_rate": 1.0752206867340951e-07,
+      "loss": 1.1362,
+      "step": 20362
+    },
+    {
+      "epoch": 1.7352364720920324,
+      "grad_norm": 20.875,
+      "learning_rate": 1.074538979070601e-07,
+      "loss": 0.876,
+      "step": 20363
+    },
+    {
+      "epoch": 1.7353216872603323,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.0738574778766747e-07,
+      "loss": 0.3301,
+      "step": 20364
+    },
+    {
+      "epoch": 1.7354069024286323,
+      "grad_norm": 12.375,
+      "learning_rate": 1.073176183164637e-07,
+      "loss": 0.3144,
+      "step": 20365
+    },
+    {
+      "epoch": 1.7354921175969322,
+      "grad_norm": 20.5,
+      "learning_rate": 1.0724950949467933e-07,
+      "loss": 0.5909,
+      "step": 20366
+    },
+    {
+      "epoch": 1.7355773327652322,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0718142132354531e-07,
+      "loss": 0.633,
+      "step": 20367
+    },
+    {
+      "epoch": 1.7356625479335321,
+      "grad_norm": 24.875,
+      "learning_rate": 1.071133538042926e-07,
+      "loss": 0.5368,
+      "step": 20368
+    },
+    {
+      "epoch": 1.735747763101832,
+      "grad_norm": 18.5,
+      "learning_rate": 1.0704530693815065e-07,
+      "loss": 0.5844,
+      "step": 20369
+    },
+    {
+      "epoch": 1.735832978270132,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.0697728072634916e-07,
+      "loss": 0.4528,
+      "step": 20370
+    },
+    {
+      "epoch": 1.735918193438432,
+      "grad_norm": 19.375,
+      "learning_rate": 1.069092751701177e-07,
+      "loss": 0.7777,
+      "step": 20371
+    },
+    {
+      "epoch": 1.736003408606732,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.0684129027068544e-07,
+      "loss": 0.8896,
+      "step": 20372
+    },
+    {
+      "epoch": 1.736088623775032,
+      "grad_norm": 13.625,
+      "learning_rate": 1.067733260292804e-07,
+      "loss": 0.5444,
+      "step": 20373
+    },
+    {
+      "epoch": 1.7361738389433319,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0670538244713092e-07,
+      "loss": 0.4681,
+      "step": 20374
+    },
+    {
+      "epoch": 1.7362590541116318,
+      "grad_norm": 10.875,
+      "learning_rate": 1.0663745952546475e-07,
+      "loss": 0.5546,
+      "step": 20375
+    },
+    {
+      "epoch": 1.7363442692799318,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0656955726550955e-07,
+      "loss": 0.4509,
+      "step": 20376
+    },
+    {
+      "epoch": 1.7364294844482318,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0650167566849223e-07,
+      "loss": 0.7451,
+      "step": 20377
+    },
+    {
+      "epoch": 1.7365146996165317,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0643381473563946e-07,
+      "loss": 0.8219,
+      "step": 20378
+    },
+    {
+      "epoch": 1.7365999147848317,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0636597446817748e-07,
+      "loss": 0.6184,
+      "step": 20379
+    },
+    {
+      "epoch": 1.7366851299531316,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0629815486733267e-07,
+      "loss": 0.5796,
+      "step": 20380
+    },
+    {
+      "epoch": 1.7367703451214316,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.0623035593433029e-07,
+      "loss": 0.8514,
+      "step": 20381
+    },
+    {
+      "epoch": 1.7368555602897315,
+      "grad_norm": 19.125,
+      "learning_rate": 1.0616257767039534e-07,
+      "loss": 0.6903,
+      "step": 20382
+    },
+    {
+      "epoch": 1.7369407754580315,
+      "grad_norm": 15.75,
+      "learning_rate": 1.0609482007675268e-07,
+      "loss": 0.7132,
+      "step": 20383
+    },
+    {
+      "epoch": 1.7370259906263315,
+      "grad_norm": 13.625,
+      "learning_rate": 1.0602708315462701e-07,
+      "loss": 0.5968,
+      "step": 20384
+    },
+    {
+      "epoch": 1.7371112057946314,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0595936690524263e-07,
+      "loss": 0.8211,
+      "step": 20385
+    },
+    {
+      "epoch": 1.7371964209629314,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.0589167132982247e-07,
+      "loss": 0.6062,
+      "step": 20386
+    },
+    {
+      "epoch": 1.7372816361312313,
+      "grad_norm": 18.25,
+      "learning_rate": 1.0582399642959041e-07,
+      "loss": 0.808,
+      "step": 20387
+    },
+    {
+      "epoch": 1.7373668512995313,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.057563422057696e-07,
+      "loss": 0.2836,
+      "step": 20388
+    },
+    {
+      "epoch": 1.7374520664678312,
+      "grad_norm": 14.5,
+      "learning_rate": 1.0568870865958216e-07,
+      "loss": 0.3815,
+      "step": 20389
+    },
+    {
+      "epoch": 1.7375372816361312,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.056210957922503e-07,
+      "loss": 0.457,
+      "step": 20390
+    },
+    {
+      "epoch": 1.7376224968044311,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0555350360499595e-07,
+      "loss": 0.4201,
+      "step": 20391
+    },
+    {
+      "epoch": 1.737707711972731,
+      "grad_norm": 15.5,
+      "learning_rate": 1.0548593209904076e-07,
+      "loss": 0.5519,
+      "step": 20392
+    },
+    {
+      "epoch": 1.737792927141031,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0541838127560572e-07,
+      "loss": 0.8545,
+      "step": 20393
+    },
+    {
+      "epoch": 1.737878142309331,
+      "grad_norm": 21.25,
+      "learning_rate": 1.0535085113591136e-07,
+      "loss": 0.6859,
+      "step": 20394
+    },
+    {
+      "epoch": 1.737963357477631,
+      "grad_norm": 15.0,
+      "learning_rate": 1.0528334168117812e-07,
+      "loss": 0.4384,
+      "step": 20395
+    },
+    {
+      "epoch": 1.738048572645931,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0521585291262612e-07,
+      "loss": 0.9514,
+      "step": 20396
+    },
+    {
+      "epoch": 1.738133787814231,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.0514838483147494e-07,
+      "loss": 0.4935,
+      "step": 20397
+    },
+    {
+      "epoch": 1.7382190029825308,
+      "grad_norm": 27.0,
+      "learning_rate": 1.0508093743894373e-07,
+      "loss": 0.7809,
+      "step": 20398
+    },
+    {
+      "epoch": 1.7383042181508308,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0501351073625099e-07,
+      "loss": 0.8966,
+      "step": 20399
+    },
+    {
+      "epoch": 1.7383894333191308,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.0494610472461544e-07,
+      "loss": 0.4087,
+      "step": 20400
+    },
+    {
+      "epoch": 1.7384746484874307,
+      "grad_norm": 14.25,
+      "learning_rate": 1.0487871940525543e-07,
+      "loss": 0.4796,
+      "step": 20401
+    },
+    {
+      "epoch": 1.7385598636557307,
+      "grad_norm": 20.0,
+      "learning_rate": 1.0481135477938844e-07,
+      "loss": 0.6553,
+      "step": 20402
+    },
+    {
+      "epoch": 1.7386450788240306,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0474401084823185e-07,
+      "loss": 0.7148,
+      "step": 20403
+    },
+    {
+      "epoch": 1.7387302939923306,
+      "grad_norm": 14.75,
+      "learning_rate": 1.0467668761300245e-07,
+      "loss": 0.6115,
+      "step": 20404
+    },
+    {
+      "epoch": 1.7388155091606305,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0460938507491733e-07,
+      "loss": 0.5769,
+      "step": 20405
+    },
+    {
+      "epoch": 1.7389007243289305,
+      "grad_norm": 13.75,
+      "learning_rate": 1.0454210323519245e-07,
+      "loss": 0.6073,
+      "step": 20406
+    },
+    {
+      "epoch": 1.7389859394972305,
+      "grad_norm": 18.875,
+      "learning_rate": 1.0447484209504325e-07,
+      "loss": 0.5419,
+      "step": 20407
+    },
+    {
+      "epoch": 1.7390711546655304,
+      "grad_norm": 15.375,
+      "learning_rate": 1.0440760165568568e-07,
+      "loss": 0.5887,
+      "step": 20408
+    },
+    {
+      "epoch": 1.7391563698338304,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0434038191833474e-07,
+      "loss": 0.604,
+      "step": 20409
+    },
+    {
+      "epoch": 1.7392415850021303,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.0427318288420529e-07,
+      "loss": 0.5071,
+      "step": 20410
+    },
+    {
+      "epoch": 1.7393268001704303,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0420600455451138e-07,
+      "loss": 0.3532,
+      "step": 20411
+    },
+    {
+      "epoch": 1.7394120153387302,
+      "grad_norm": 23.625,
+      "learning_rate": 1.0413884693046703e-07,
+      "loss": 0.3544,
+      "step": 20412
+    },
+    {
+      "epoch": 1.7394972305070302,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0407171001328639e-07,
+      "loss": 0.5435,
+      "step": 20413
+    },
+    {
+      "epoch": 1.7395824456753302,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0400459380418184e-07,
+      "loss": 0.2353,
+      "step": 20414
+    },
+    {
+      "epoch": 1.7396676608436301,
+      "grad_norm": 16.0,
+      "learning_rate": 1.03937498304367e-07,
+      "loss": 0.7209,
+      "step": 20415
+    },
+    {
+      "epoch": 1.73975287601193,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.038704235150538e-07,
+      "loss": 0.4017,
+      "step": 20416
+    },
+    {
+      "epoch": 1.73983809118023,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0380336943745462e-07,
+      "loss": 0.565,
+      "step": 20417
+    },
+    {
+      "epoch": 1.73992330634853,
+      "grad_norm": 16.875,
+      "learning_rate": 1.037363360727811e-07,
+      "loss": 0.7592,
+      "step": 20418
+    },
+    {
+      "epoch": 1.74000852151683,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.036693234222448e-07,
+      "loss": 0.4036,
+      "step": 20419
+    },
+    {
+      "epoch": 1.74009373668513,
+      "grad_norm": 24.75,
+      "learning_rate": 1.0360233148705643e-07,
+      "loss": 0.6585,
+      "step": 20420
+    },
+    {
+      "epoch": 1.7401789518534299,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0353536026842693e-07,
+      "loss": 0.6857,
+      "step": 20421
+    },
+    {
+      "epoch": 1.7402641670217298,
+      "grad_norm": 21.0,
+      "learning_rate": 1.0346840976756604e-07,
+      "loss": 0.6703,
+      "step": 20422
+    },
+    {
+      "epoch": 1.7403493821900298,
+      "grad_norm": 13.0,
+      "learning_rate": 1.0340147998568417e-07,
+      "loss": 0.5531,
+      "step": 20423
+    },
+    {
+      "epoch": 1.74043459735833,
+      "grad_norm": 15.75,
+      "learning_rate": 1.0333457092399024e-07,
+      "loss": 0.7951,
+      "step": 20424
+    },
+    {
+      "epoch": 1.74051981252663,
+      "grad_norm": 14.25,
+      "learning_rate": 1.0326768258369383e-07,
+      "loss": 0.4648,
+      "step": 20425
+    },
+    {
+      "epoch": 1.7406050276949299,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.032008149660034e-07,
+      "loss": 0.9563,
+      "step": 20426
+    },
+    {
+      "epoch": 1.7406902428632298,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.0313396807212772e-07,
+      "loss": 0.3751,
+      "step": 20427
+    },
+    {
+      "epoch": 1.7407754580315298,
+      "grad_norm": 11.625,
+      "learning_rate": 1.030671419032743e-07,
+      "loss": 0.4556,
+      "step": 20428
+    },
+    {
+      "epoch": 1.7408606731998297,
+      "grad_norm": 21.25,
+      "learning_rate": 1.0300033646065119e-07,
+      "loss": 1.0432,
+      "step": 20429
+    },
+    {
+      "epoch": 1.7409458883681297,
+      "grad_norm": 9.875,
+      "learning_rate": 1.0293355174546507e-07,
+      "loss": 0.1894,
+      "step": 20430
+    },
+    {
+      "epoch": 1.7410311035364296,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.0286678775892318e-07,
+      "loss": 0.41,
+      "step": 20431
+    },
+    {
+      "epoch": 1.7411163187047296,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.028000445022323e-07,
+      "loss": 0.7088,
+      "step": 20432
+    },
+    {
+      "epoch": 1.7412015338730296,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.0273332197659788e-07,
+      "loss": 0.5009,
+      "step": 20433
+    },
+    {
+      "epoch": 1.7412867490413295,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0266662018322604e-07,
+      "loss": 0.4311,
+      "step": 20434
+    },
+    {
+      "epoch": 1.7413719642096295,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.0259993912332233e-07,
+      "loss": 0.5274,
+      "step": 20435
+    },
+    {
+      "epoch": 1.7414571793779294,
+      "grad_norm": 18.0,
+      "learning_rate": 1.0253327879809122e-07,
+      "loss": 0.6928,
+      "step": 20436
+    },
+    {
+      "epoch": 1.7415423945462294,
+      "grad_norm": 12.25,
+      "learning_rate": 1.024666392087377e-07,
+      "loss": 0.5453,
+      "step": 20437
+    },
+    {
+      "epoch": 1.7416276097145293,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.024000203564661e-07,
+      "loss": 0.6398,
+      "step": 20438
+    },
+    {
+      "epoch": 1.7417128248828293,
+      "grad_norm": 24.625,
+      "learning_rate": 1.0233342224248002e-07,
+      "loss": 0.5252,
+      "step": 20439
+    },
+    {
+      "epoch": 1.7417980400511293,
+      "grad_norm": 33.5,
+      "learning_rate": 1.0226684486798325e-07,
+      "loss": 0.7633,
+      "step": 20440
+    },
+    {
+      "epoch": 1.7418832552194292,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.022002882341784e-07,
+      "loss": 0.513,
+      "step": 20441
+    },
+    {
+      "epoch": 1.7419684703877292,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.0213375234226871e-07,
+      "loss": 0.6458,
+      "step": 20442
+    },
+    {
+      "epoch": 1.7420536855560291,
+      "grad_norm": 19.5,
+      "learning_rate": 1.0206723719345626e-07,
+      "loss": 0.6199,
+      "step": 20443
+    },
+    {
+      "epoch": 1.742138900724329,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.0200074278894342e-07,
+      "loss": 0.3249,
+      "step": 20444
+    },
+    {
+      "epoch": 1.742224115892629,
+      "grad_norm": 18.5,
+      "learning_rate": 1.0193426912993145e-07,
+      "loss": 0.9058,
+      "step": 20445
+    },
+    {
+      "epoch": 1.742309331060929,
+      "grad_norm": 14.125,
+      "learning_rate": 1.0186781621762176e-07,
+      "loss": 0.6742,
+      "step": 20446
+    },
+    {
+      "epoch": 1.742394546229229,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0180138405321505e-07,
+      "loss": 0.749,
+      "step": 20447
+    },
+    {
+      "epoch": 1.742479761397529,
+      "grad_norm": 13.625,
+      "learning_rate": 1.0173497263791188e-07,
+      "loss": 0.4413,
+      "step": 20448
+    },
+    {
+      "epoch": 1.7425649765658289,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0166858197291268e-07,
+      "loss": 0.4983,
+      "step": 20449
+    },
+    {
+      "epoch": 1.7426501917341288,
+      "grad_norm": 25.625,
+      "learning_rate": 1.0160221205941665e-07,
+      "loss": 0.4441,
+      "step": 20450
+    },
+    {
+      "epoch": 1.7427354069024288,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.0153586289862349e-07,
+      "loss": 0.4698,
+      "step": 20451
+    },
+    {
+      "epoch": 1.7428206220707287,
+      "grad_norm": 15.875,
+      "learning_rate": 1.0146953449173242e-07,
+      "loss": 0.5969,
+      "step": 20452
+    },
+    {
+      "epoch": 1.7429058372390287,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.0140322683994161e-07,
+      "loss": 0.5445,
+      "step": 20453
+    },
+    {
+      "epoch": 1.7429910524073287,
+      "grad_norm": 14.375,
+      "learning_rate": 1.0133693994444971e-07,
+      "loss": 0.605,
+      "step": 20454
+    },
+    {
+      "epoch": 1.7430762675756286,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0127067380645409e-07,
+      "loss": 0.494,
+      "step": 20455
+    },
+    {
+      "epoch": 1.7431614827439286,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0120442842715252e-07,
+      "loss": 0.6719,
+      "step": 20456
+    },
+    {
+      "epoch": 1.7432466979122285,
+      "grad_norm": 11.0,
+      "learning_rate": 1.0113820380774242e-07,
+      "loss": 0.2613,
+      "step": 20457
+    },
+    {
+      "epoch": 1.7433319130805285,
+      "grad_norm": 21.5,
+      "learning_rate": 1.0107199994941988e-07,
+      "loss": 0.5414,
+      "step": 20458
+    },
+    {
+      "epoch": 1.7434171282488284,
+      "grad_norm": 27.625,
+      "learning_rate": 1.0100581685338173e-07,
+      "loss": 0.8083,
+      "step": 20459
+    },
+    {
+      "epoch": 1.7435023434171284,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.0093965452082397e-07,
+      "loss": 0.505,
+      "step": 20460
+    },
+    {
+      "epoch": 1.7435875585854284,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0087351295294229e-07,
+      "loss": 0.4654,
+      "step": 20461
+    },
+    {
+      "epoch": 1.7436727737537283,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.0080739215093172e-07,
+      "loss": 0.4282,
+      "step": 20462
+    },
+    {
+      "epoch": 1.7437579889220283,
+      "grad_norm": 23.75,
+      "learning_rate": 1.0074129211598684e-07,
+      "loss": 0.7353,
+      "step": 20463
+    },
+    {
+      "epoch": 1.7438432040903282,
+      "grad_norm": 23.75,
+      "learning_rate": 1.0067521284930271e-07,
+      "loss": 0.6236,
+      "step": 20464
+    },
+    {
+      "epoch": 1.7439284192586282,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0060915435207333e-07,
+      "loss": 0.8028,
+      "step": 20465
+    },
+    {
+      "epoch": 1.7440136344269281,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.0054311662549208e-07,
+      "loss": 0.4277,
+      "step": 20466
+    },
+    {
+      "epoch": 1.744098849595228,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.004770996707527e-07,
+      "loss": 0.6807,
+      "step": 20467
+    },
+    {
+      "epoch": 1.744184064763528,
+      "grad_norm": 18.375,
+      "learning_rate": 1.0041110348904787e-07,
+      "loss": 0.6957,
+      "step": 20468
+    },
+    {
+      "epoch": 1.744269279931828,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0034512808157079e-07,
+      "loss": 0.8879,
+      "step": 20469
+    },
+    {
+      "epoch": 1.744354495100128,
+      "grad_norm": 17.375,
+      "learning_rate": 1.0027917344951329e-07,
+      "loss": 0.662,
+      "step": 20470
+    },
+    {
+      "epoch": 1.744439710268428,
+      "grad_norm": 14.25,
+      "learning_rate": 1.0021323959406692e-07,
+      "loss": 0.5011,
+      "step": 20471
+    },
+    {
+      "epoch": 1.7445249254367279,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0014732651642365e-07,
+      "loss": 0.8457,
+      "step": 20472
+    },
+    {
+      "epoch": 1.7446101406050278,
+      "grad_norm": 21.75,
+      "learning_rate": 1.0008143421777444e-07,
+      "loss": 0.9203,
+      "step": 20473
+    },
+    {
+      "epoch": 1.7446953557733278,
+      "grad_norm": 11.75,
+      "learning_rate": 1.0001556269931018e-07,
+      "loss": 0.3661,
+      "step": 20474
+    },
+    {
+      "epoch": 1.7447805709416278,
+      "grad_norm": 18.125,
+      "learning_rate": 9.9949711962221e-08,
+      "loss": 0.7605,
+      "step": 20475
+    },
+    {
+      "epoch": 1.7448657861099277,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.988388200769694e-08,
+      "loss": 0.2155,
+      "step": 20476
+    },
+    {
+      "epoch": 1.7449510012782277,
+      "grad_norm": 15.5,
+      "learning_rate": 9.98180728369276e-08,
+      "loss": 0.4357,
+      "step": 20477
+    },
+    {
+      "epoch": 1.7450362164465276,
+      "grad_norm": 15.125,
+      "learning_rate": 9.975228445110258e-08,
+      "loss": 0.6236,
+      "step": 20478
+    },
+    {
+      "epoch": 1.7451214316148276,
+      "grad_norm": 18.75,
+      "learning_rate": 9.968651685141038e-08,
+      "loss": 0.787,
+      "step": 20479
+    },
+    {
+      "epoch": 1.7452066467831275,
+      "grad_norm": 10.25,
+      "learning_rate": 9.962077003903933e-08,
+      "loss": 0.1847,
+      "step": 20480
+    },
+    {
+      "epoch": 1.7452918619514275,
+      "grad_norm": 22.625,
+      "learning_rate": 9.955504401517782e-08,
+      "loss": 0.5763,
+      "step": 20481
+    },
+    {
+      "epoch": 1.7453770771197274,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.948933878101378e-08,
+      "loss": 0.5129,
+      "step": 20482
+    },
+    {
+      "epoch": 1.7454622922880274,
+      "grad_norm": 17.875,
+      "learning_rate": 9.942365433773401e-08,
+      "loss": 0.5203,
+      "step": 20483
+    },
+    {
+      "epoch": 1.7455475074563274,
+      "grad_norm": 16.0,
+      "learning_rate": 9.93579906865258e-08,
+      "loss": 0.655,
+      "step": 20484
+    },
+    {
+      "epoch": 1.7456327226246273,
+      "grad_norm": 18.25,
+      "learning_rate": 9.92923478285758e-08,
+      "loss": 0.8075,
+      "step": 20485
+    },
+    {
+      "epoch": 1.7457179377929273,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.922672576507045e-08,
+      "loss": 0.5332,
+      "step": 20486
+    },
+    {
+      "epoch": 1.7458031529612272,
+      "grad_norm": 18.875,
+      "learning_rate": 9.916112449719517e-08,
+      "loss": 0.52,
+      "step": 20487
+    },
+    {
+      "epoch": 1.7458883681295272,
+      "grad_norm": 13.625,
+      "learning_rate": 9.909554402613555e-08,
+      "loss": 0.3861,
+      "step": 20488
+    },
+    {
+      "epoch": 1.7459735832978271,
+      "grad_norm": 13.5,
+      "learning_rate": 9.90299843530769e-08,
+      "loss": 0.3637,
+      "step": 20489
+    },
+    {
+      "epoch": 1.746058798466127,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.896444547920367e-08,
+      "loss": 0.5383,
+      "step": 20490
+    },
+    {
+      "epoch": 1.746144013634427,
+      "grad_norm": 18.0,
+      "learning_rate": 9.889892740570062e-08,
+      "loss": 0.662,
+      "step": 20491
+    },
+    {
+      "epoch": 1.746229228802727,
+      "grad_norm": 17.5,
+      "learning_rate": 9.883343013375123e-08,
+      "loss": 0.9331,
+      "step": 20492
+    },
+    {
+      "epoch": 1.746314443971027,
+      "grad_norm": 22.125,
+      "learning_rate": 9.876795366453932e-08,
+      "loss": 0.613,
+      "step": 20493
+    },
+    {
+      "epoch": 1.746399659139327,
+      "grad_norm": 23.75,
+      "learning_rate": 9.870249799924847e-08,
+      "loss": 0.5122,
+      "step": 20494
+    },
+    {
+      "epoch": 1.7464848743076269,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.863706313906111e-08,
+      "loss": 0.4902,
+      "step": 20495
+    },
+    {
+      "epoch": 1.7465700894759268,
+      "grad_norm": 13.3125,
+      "learning_rate": 9.85716490851596e-08,
+      "loss": 0.8337,
+      "step": 20496
+    },
+    {
+      "epoch": 1.7466553046442268,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.85062558387262e-08,
+      "loss": 0.5055,
+      "step": 20497
+    },
+    {
+      "epoch": 1.7467405198125268,
+      "grad_norm": 16.0,
+      "learning_rate": 9.844088340094274e-08,
+      "loss": 0.5628,
+      "step": 20498
+    },
+    {
+      "epoch": 1.7468257349808267,
+      "grad_norm": 21.0,
+      "learning_rate": 9.837553177299063e-08,
+      "loss": 0.5968,
+      "step": 20499
+    },
+    {
+      "epoch": 1.7469109501491267,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.831020095605062e-08,
+      "loss": 0.8684,
+      "step": 20500
+    },
+    {
+      "epoch": 1.7469961653174266,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.82448909513034e-08,
+      "loss": 0.2963,
+      "step": 20501
+    },
+    {
+      "epoch": 1.7470813804857266,
+      "grad_norm": 17.625,
+      "learning_rate": 9.817960175992902e-08,
+      "loss": 0.7024,
+      "step": 20502
+    },
+    {
+      "epoch": 1.7471665956540265,
+      "grad_norm": 15.625,
+      "learning_rate": 9.811433338310778e-08,
+      "loss": 0.5766,
+      "step": 20503
+    },
+    {
+      "epoch": 1.7472518108223265,
+      "grad_norm": 20.25,
+      "learning_rate": 9.804908582201886e-08,
+      "loss": 0.8173,
+      "step": 20504
+    },
+    {
+      "epoch": 1.7473370259906265,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.79838590778412e-08,
+      "loss": 0.5356,
+      "step": 20505
+    },
+    {
+      "epoch": 1.7474222411589264,
+      "grad_norm": 16.125,
+      "learning_rate": 9.791865315175355e-08,
+      "loss": 0.7171,
+      "step": 20506
+    },
+    {
+      "epoch": 1.7475074563272264,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.785346804493432e-08,
+      "loss": 0.4671,
+      "step": 20507
+    },
+    {
+      "epoch": 1.7475926714955263,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.778830375856168e-08,
+      "loss": 0.7069,
+      "step": 20508
+    },
+    {
+      "epoch": 1.7476778866638263,
+      "grad_norm": 15.875,
+      "learning_rate": 9.772316029381293e-08,
+      "loss": 0.7917,
+      "step": 20509
+    },
+    {
+      "epoch": 1.7477631018321262,
+      "grad_norm": 14.1875,
+      "learning_rate": 9.76580376518653e-08,
+      "loss": 0.5675,
+      "step": 20510
+    },
+    {
+      "epoch": 1.7478483170004262,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.759293583389603e-08,
+      "loss": 0.4669,
+      "step": 20511
+    },
+    {
+      "epoch": 1.7479335321687262,
+      "grad_norm": 19.5,
+      "learning_rate": 9.752785484108115e-08,
+      "loss": 0.88,
+      "step": 20512
+    },
+    {
+      "epoch": 1.7480187473370261,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.746279467459665e-08,
+      "loss": 0.7422,
+      "step": 20513
+    },
+    {
+      "epoch": 1.748103962505326,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.73977553356184e-08,
+      "loss": 0.4824,
+      "step": 20514
+    },
+    {
+      "epoch": 1.748189177673626,
+      "grad_norm": 14.875,
+      "learning_rate": 9.733273682532172e-08,
+      "loss": 0.8472,
+      "step": 20515
+    },
+    {
+      "epoch": 1.748274392841926,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.726773914488174e-08,
+      "loss": 0.5523,
+      "step": 20516
+    },
+    {
+      "epoch": 1.748359608010226,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.720276229547271e-08,
+      "loss": 0.4846,
+      "step": 20517
+    },
+    {
+      "epoch": 1.748444823178526,
+      "grad_norm": 18.875,
+      "learning_rate": 9.713780627826908e-08,
+      "loss": 0.7862,
+      "step": 20518
+    },
+    {
+      "epoch": 1.7485300383468259,
+      "grad_norm": 16.625,
+      "learning_rate": 9.707287109444478e-08,
+      "loss": 0.729,
+      "step": 20519
+    },
+    {
+      "epoch": 1.7486152535151258,
+      "grad_norm": 9.625,
+      "learning_rate": 9.700795674517274e-08,
+      "loss": 0.8156,
+      "step": 20520
+    },
+    {
+      "epoch": 1.7487004686834258,
+      "grad_norm": 12.0,
+      "learning_rate": 9.694306323162664e-08,
+      "loss": 0.1883,
+      "step": 20521
+    },
+    {
+      "epoch": 1.7487856838517257,
+      "grad_norm": 14.3125,
+      "learning_rate": 9.687819055497857e-08,
+      "loss": 0.4699,
+      "step": 20522
+    },
+    {
+      "epoch": 1.7488708990200257,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.681333871640123e-08,
+      "loss": 0.6269,
+      "step": 20523
+    },
+    {
+      "epoch": 1.7489561141883256,
+      "grad_norm": 13.125,
+      "learning_rate": 9.674850771706657e-08,
+      "loss": 0.5829,
+      "step": 20524
+    },
+    {
+      "epoch": 1.7490413293566256,
+      "grad_norm": 21.125,
+      "learning_rate": 9.668369755814616e-08,
+      "loss": 1.0002,
+      "step": 20525
+    },
+    {
+      "epoch": 1.7491265445249256,
+      "grad_norm": 13.875,
+      "learning_rate": 9.661890824081088e-08,
+      "loss": 0.5371,
+      "step": 20526
+    },
+    {
+      "epoch": 1.7492117596932255,
+      "grad_norm": 17.625,
+      "learning_rate": 9.655413976623187e-08,
+      "loss": 0.5174,
+      "step": 20527
+    },
+    {
+      "epoch": 1.7492969748615255,
+      "grad_norm": 18.0,
+      "learning_rate": 9.648939213557931e-08,
+      "loss": 0.7669,
+      "step": 20528
+    },
+    {
+      "epoch": 1.7493821900298254,
+      "grad_norm": 15.0,
+      "learning_rate": 9.642466535002365e-08,
+      "loss": 0.6562,
+      "step": 20529
+    },
+    {
+      "epoch": 1.7494674051981254,
+      "grad_norm": 22.0,
+      "learning_rate": 9.635995941073395e-08,
+      "loss": 0.5418,
+      "step": 20530
+    },
+    {
+      "epoch": 1.7495526203664253,
+      "grad_norm": 14.625,
+      "learning_rate": 9.629527431887997e-08,
+      "loss": 0.5061,
+      "step": 20531
+    },
+    {
+      "epoch": 1.7496378355347253,
+      "grad_norm": 17.5,
+      "learning_rate": 9.62306100756305e-08,
+      "loss": 0.8293,
+      "step": 20532
+    },
+    {
+      "epoch": 1.7497230507030253,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.616596668215433e-08,
+      "loss": 0.3131,
+      "step": 20533
+    },
+    {
+      "epoch": 1.7498082658713252,
+      "grad_norm": 13.375,
+      "learning_rate": 9.610134413961914e-08,
+      "loss": 0.441,
+      "step": 20534
+    },
+    {
+      "epoch": 1.7498934810396252,
+      "grad_norm": 14.0,
+      "learning_rate": 9.603674244919328e-08,
+      "loss": 0.6186,
+      "step": 20535
+    },
+    {
+      "epoch": 1.7499786962079251,
+      "grad_norm": 20.625,
+      "learning_rate": 9.597216161204362e-08,
+      "loss": 0.7141,
+      "step": 20536
+    },
+    {
+      "epoch": 1.750063911376225,
+      "grad_norm": 17.125,
+      "learning_rate": 9.59076016293374e-08,
+      "loss": 0.5162,
+      "step": 20537
+    },
+    {
+      "epoch": 1.750149126544525,
+      "grad_norm": 27.75,
+      "learning_rate": 9.584306250224163e-08,
+      "loss": 0.5631,
+      "step": 20538
+    },
+    {
+      "epoch": 1.750234341712825,
+      "grad_norm": 31.0,
+      "learning_rate": 9.577854423192215e-08,
+      "loss": 1.4732,
+      "step": 20539
+    },
+    {
+      "epoch": 1.750319556881125,
+      "grad_norm": 13.0625,
+      "learning_rate": 9.5714046819545e-08,
+      "loss": 0.43,
+      "step": 20540
+    },
+    {
+      "epoch": 1.750404772049425,
+      "grad_norm": 14.6875,
+      "learning_rate": 9.564957026627575e-08,
+      "loss": 0.5671,
+      "step": 20541
+    },
+    {
+      "epoch": 1.7504899872177249,
+      "grad_norm": 12.625,
+      "learning_rate": 9.558511457327974e-08,
+      "loss": 0.3444,
+      "step": 20542
+    },
+    {
+      "epoch": 1.7505752023860248,
+      "grad_norm": 34.75,
+      "learning_rate": 9.552067974172158e-08,
+      "loss": 0.75,
+      "step": 20543
+    },
+    {
+      "epoch": 1.7506604175543248,
+      "grad_norm": 16.0,
+      "learning_rate": 9.545626577276548e-08,
+      "loss": 0.3507,
+      "step": 20544
+    },
+    {
+      "epoch": 1.7507456327226247,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.539187266757566e-08,
+      "loss": 0.6438,
+      "step": 20545
+    },
+    {
+      "epoch": 1.7508308478909247,
+      "grad_norm": 17.125,
+      "learning_rate": 9.53275004273159e-08,
+      "loss": 0.6065,
+      "step": 20546
+    },
+    {
+      "epoch": 1.7509160630592246,
+      "grad_norm": 19.5,
+      "learning_rate": 9.526314905314916e-08,
+      "loss": 0.4324,
+      "step": 20547
+    },
+    {
+      "epoch": 1.7510012782275246,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.519881854623838e-08,
+      "loss": 0.3667,
+      "step": 20548
+    },
+    {
+      "epoch": 1.7510864933958246,
+      "grad_norm": 17.75,
+      "learning_rate": 9.513450890774628e-08,
+      "loss": 0.8786,
+      "step": 20549
+    },
+    {
+      "epoch": 1.7511717085641245,
+      "grad_norm": 21.75,
+      "learning_rate": 9.50702201388351e-08,
+      "loss": 1.016,
+      "step": 20550
+    },
+    {
+      "epoch": 1.7512569237324245,
+      "grad_norm": 14.375,
+      "learning_rate": 9.500595224066642e-08,
+      "loss": 0.4673,
+      "step": 20551
+    },
+    {
+      "epoch": 1.7513421389007244,
+      "grad_norm": 12.375,
+      "learning_rate": 9.494170521440138e-08,
+      "loss": 0.3984,
+      "step": 20552
+    },
+    {
+      "epoch": 1.7514273540690244,
+      "grad_norm": 11.6875,
+      "learning_rate": 9.48774790612013e-08,
+      "loss": 0.521,
+      "step": 20553
+    },
+    {
+      "epoch": 1.7515125692373243,
+      "grad_norm": 20.5,
+      "learning_rate": 9.481327378222674e-08,
+      "loss": 1.1127,
+      "step": 20554
+    },
+    {
+      "epoch": 1.7515977844056243,
+      "grad_norm": 16.375,
+      "learning_rate": 9.474908937863805e-08,
+      "loss": 0.6917,
+      "step": 20555
+    },
+    {
+      "epoch": 1.7516829995739243,
+      "grad_norm": 17.625,
+      "learning_rate": 9.468492585159486e-08,
+      "loss": 0.7381,
+      "step": 20556
+    },
+    {
+      "epoch": 1.7517682147422242,
+      "grad_norm": 15.125,
+      "learning_rate": 9.462078320225693e-08,
+      "loss": 0.7665,
+      "step": 20557
+    },
+    {
+      "epoch": 1.7518534299105242,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.455666143178332e-08,
+      "loss": 0.7582,
+      "step": 20558
+    },
+    {
+      "epoch": 1.7519386450788241,
+      "grad_norm": 13.6875,
+      "learning_rate": 9.449256054133286e-08,
+      "loss": 0.6289,
+      "step": 20559
+    },
+    {
+      "epoch": 1.752023860247124,
+      "grad_norm": 15.0625,
+      "learning_rate": 9.442848053206347e-08,
+      "loss": 0.5717,
+      "step": 20560
+    },
+    {
+      "epoch": 1.752109075415424,
+      "grad_norm": 19.875,
+      "learning_rate": 9.436442140513341e-08,
+      "loss": 0.7492,
+      "step": 20561
+    },
+    {
+      "epoch": 1.752194290583724,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.430038316170051e-08,
+      "loss": 0.9068,
+      "step": 20562
+    },
+    {
+      "epoch": 1.752279505752024,
+      "grad_norm": 14.375,
+      "learning_rate": 9.423636580292189e-08,
+      "loss": 0.3273,
+      "step": 20563
+    },
+    {
+      "epoch": 1.752364720920324,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.417236932995413e-08,
+      "loss": 0.34,
+      "step": 20564
+    },
+    {
+      "epoch": 1.7524499360886239,
+      "grad_norm": 12.75,
+      "learning_rate": 9.410839374395409e-08,
+      "loss": 0.559,
+      "step": 20565
+    },
+    {
+      "epoch": 1.7525351512569238,
+      "grad_norm": 16.125,
+      "learning_rate": 9.404443904607764e-08,
+      "loss": 0.7444,
+      "step": 20566
+    },
+    {
+      "epoch": 1.7526203664252238,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.398050523748065e-08,
+      "loss": 0.6814,
+      "step": 20567
+    },
+    {
+      "epoch": 1.7527055815935237,
+      "grad_norm": 16.875,
+      "learning_rate": 9.39165923193186e-08,
+      "loss": 0.7218,
+      "step": 20568
+    },
+    {
+      "epoch": 1.7527907967618237,
+      "grad_norm": 15.75,
+      "learning_rate": 9.385270029274598e-08,
+      "loss": 0.6058,
+      "step": 20569
+    },
+    {
+      "epoch": 1.7528760119301237,
+      "grad_norm": 19.125,
+      "learning_rate": 9.378882915891758e-08,
+      "loss": 0.8717,
+      "step": 20570
+    },
+    {
+      "epoch": 1.7529612270984236,
+      "grad_norm": 13.625,
+      "learning_rate": 9.372497891898771e-08,
+      "loss": 0.3192,
+      "step": 20571
+    },
+    {
+      "epoch": 1.7530464422667236,
+      "grad_norm": 14.375,
+      "learning_rate": 9.366114957411049e-08,
+      "loss": 0.7445,
+      "step": 20572
+    },
+    {
+      "epoch": 1.7531316574350235,
+      "grad_norm": 18.5,
+      "learning_rate": 9.359734112543886e-08,
+      "loss": 0.7718,
+      "step": 20573
+    },
+    {
+      "epoch": 1.7532168726033235,
+      "grad_norm": 30.25,
+      "learning_rate": 9.353355357412608e-08,
+      "loss": 0.8057,
+      "step": 20574
+    },
+    {
+      "epoch": 1.7533020877716234,
+      "grad_norm": 20.75,
+      "learning_rate": 9.346978692132525e-08,
+      "loss": 0.8301,
+      "step": 20575
+    },
+    {
+      "epoch": 1.7533873029399234,
+      "grad_norm": 14.5,
+      "learning_rate": 9.340604116818821e-08,
+      "loss": 0.7265,
+      "step": 20576
+    },
+    {
+      "epoch": 1.7534725181082234,
+      "grad_norm": 14.25,
+      "learning_rate": 9.334231631586684e-08,
+      "loss": 0.4403,
+      "step": 20577
+    },
+    {
+      "epoch": 1.7535577332765233,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.3278612365513e-08,
+      "loss": 0.6983,
+      "step": 20578
+    },
+    {
+      "epoch": 1.7536429484448233,
+      "grad_norm": 14.9375,
+      "learning_rate": 9.32149293182777e-08,
+      "loss": 0.4973,
+      "step": 20579
+    },
+    {
+      "epoch": 1.7537281636131232,
+      "grad_norm": 13.9375,
+      "learning_rate": 9.315126717531222e-08,
+      "loss": 0.5061,
+      "step": 20580
+    },
+    {
+      "epoch": 1.7538133787814232,
+      "grad_norm": 11.1875,
+      "learning_rate": 9.308762593776638e-08,
+      "loss": 0.301,
+      "step": 20581
+    },
+    {
+      "epoch": 1.7538985939497231,
+      "grad_norm": 16.75,
+      "learning_rate": 9.302400560679048e-08,
+      "loss": 0.5514,
+      "step": 20582
+    },
+    {
+      "epoch": 1.753983809118023,
+      "grad_norm": 20.125,
+      "learning_rate": 9.296040618353417e-08,
+      "loss": 0.6677,
+      "step": 20583
+    },
+    {
+      "epoch": 1.754069024286323,
+      "grad_norm": 15.375,
+      "learning_rate": 9.289682766914706e-08,
+      "loss": 0.651,
+      "step": 20584
+    },
+    {
+      "epoch": 1.754154239454623,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.283327006477786e-08,
+      "loss": 0.3943,
+      "step": 20585
+    },
+    {
+      "epoch": 1.754239454622923,
+      "grad_norm": 18.125,
+      "learning_rate": 9.27697333715749e-08,
+      "loss": 0.5301,
+      "step": 20586
+    },
+    {
+      "epoch": 1.754324669791223,
+      "grad_norm": 16.5,
+      "learning_rate": 9.270621759068649e-08,
+      "loss": 0.9207,
+      "step": 20587
+    },
+    {
+      "epoch": 1.7544098849595229,
+      "grad_norm": 22.625,
+      "learning_rate": 9.264272272326083e-08,
+      "loss": 0.7703,
+      "step": 20588
+    },
+    {
+      "epoch": 1.7544951001278228,
+      "grad_norm": 14.625,
+      "learning_rate": 9.257924877044466e-08,
+      "loss": 0.6064,
+      "step": 20589
+    },
+    {
+      "epoch": 1.7545803152961228,
+      "grad_norm": 12.375,
+      "learning_rate": 9.251579573338526e-08,
+      "loss": 0.7159,
+      "step": 20590
+    },
+    {
+      "epoch": 1.7546655304644228,
+      "grad_norm": 23.0,
+      "learning_rate": 9.245236361322948e-08,
+      "loss": 0.8068,
+      "step": 20591
+    },
+    {
+      "epoch": 1.7547507456327227,
+      "grad_norm": 22.375,
+      "learning_rate": 9.238895241112363e-08,
+      "loss": 0.4809,
+      "step": 20592
+    },
+    {
+      "epoch": 1.7548359608010227,
+      "grad_norm": 12.9375,
+      "learning_rate": 9.232556212821362e-08,
+      "loss": 0.4832,
+      "step": 20593
+    },
+    {
+      "epoch": 1.7549211759693226,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.226219276564447e-08,
+      "loss": 0.6109,
+      "step": 20594
+    },
+    {
+      "epoch": 1.7550063911376226,
+      "grad_norm": 13.5,
+      "learning_rate": 9.21988443245618e-08,
+      "loss": 0.5394,
+      "step": 20595
+    },
+    {
+      "epoch": 1.7550916063059225,
+      "grad_norm": 16.375,
+      "learning_rate": 9.213551680611013e-08,
+      "loss": 0.4728,
+      "step": 20596
+    },
+    {
+      "epoch": 1.7551768214742225,
+      "grad_norm": 16.625,
+      "learning_rate": 9.207221021143437e-08,
+      "loss": 0.7177,
+      "step": 20597
+    },
+    {
+      "epoch": 1.7552620366425224,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.200892454167789e-08,
+      "loss": 0.3143,
+      "step": 20598
+    },
+    {
+      "epoch": 1.7553472518108224,
+      "grad_norm": 20.625,
+      "learning_rate": 9.194565979798453e-08,
+      "loss": 0.87,
+      "step": 20599
+    },
+    {
+      "epoch": 1.7554324669791224,
+      "grad_norm": 18.0,
+      "learning_rate": 9.18824159814978e-08,
+      "loss": 0.8074,
+      "step": 20600
+    },
+    {
+      "epoch": 1.7555176821474223,
+      "grad_norm": 13.625,
+      "learning_rate": 9.181919309336027e-08,
+      "loss": 0.5505,
+      "step": 20601
+    },
+    {
+      "epoch": 1.7556028973157223,
+      "grad_norm": 13.125,
+      "learning_rate": 9.175599113471475e-08,
+      "loss": 0.5577,
+      "step": 20602
+    },
+    {
+      "epoch": 1.7556881124840222,
+      "grad_norm": 14.8125,
+      "learning_rate": 9.1692810106703e-08,
+      "loss": 0.7381,
+      "step": 20603
+    },
+    {
+      "epoch": 1.7557733276523222,
+      "grad_norm": 18.375,
+      "learning_rate": 9.162965001046686e-08,
+      "loss": 0.5871,
+      "step": 20604
+    },
+    {
+      "epoch": 1.7558585428206221,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.156651084714808e-08,
+      "loss": 0.4642,
+      "step": 20605
+    },
+    {
+      "epoch": 1.755943757988922,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.15033926178871e-08,
+      "loss": 0.3324,
+      "step": 20606
+    },
+    {
+      "epoch": 1.756028973157222,
+      "grad_norm": 13.8125,
+      "learning_rate": 9.144029532382471e-08,
+      "loss": 0.7006,
+      "step": 20607
+    },
+    {
+      "epoch": 1.756114188325522,
+      "grad_norm": 14.5,
+      "learning_rate": 9.137721896610149e-08,
+      "loss": 0.4925,
+      "step": 20608
+    },
+    {
+      "epoch": 1.756199403493822,
+      "grad_norm": 16.375,
+      "learning_rate": 9.131416354585685e-08,
+      "loss": 0.6276,
+      "step": 20609
+    },
+    {
+      "epoch": 1.756284618662122,
+      "grad_norm": 20.0,
+      "learning_rate": 9.125112906423053e-08,
+      "loss": 0.9773,
+      "step": 20610
+    },
+    {
+      "epoch": 1.756369833830422,
+      "grad_norm": 15.1875,
+      "learning_rate": 9.118811552236124e-08,
+      "loss": 0.4553,
+      "step": 20611
+    },
+    {
+      "epoch": 1.7564550489987218,
+      "grad_norm": 22.625,
+      "learning_rate": 9.112512292138817e-08,
+      "loss": 0.8546,
+      "step": 20612
+    },
+    {
+      "epoch": 1.7565402641670218,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.106215126244932e-08,
+      "loss": 0.6129,
+      "step": 20613
+    },
+    {
+      "epoch": 1.7566254793353218,
+      "grad_norm": 18.25,
+      "learning_rate": 9.09992005466831e-08,
+      "loss": 0.662,
+      "step": 20614
+    },
+    {
+      "epoch": 1.7567106945036217,
+      "grad_norm": 10.0625,
+      "learning_rate": 9.093627077522648e-08,
+      "loss": 0.2943,
+      "step": 20615
+    },
+    {
+      "epoch": 1.7567959096719217,
+      "grad_norm": 12.25,
+      "learning_rate": 9.087336194921734e-08,
+      "loss": 0.4908,
+      "step": 20616
+    },
+    {
+      "epoch": 1.7568811248402216,
+      "grad_norm": 12.625,
+      "learning_rate": 9.081047406979183e-08,
+      "loss": 0.5093,
+      "step": 20617
+    },
+    {
+      "epoch": 1.7569663400085216,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.074760713808694e-08,
+      "loss": 0.7204,
+      "step": 20618
+    },
+    {
+      "epoch": 1.7570515551768215,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.068476115523819e-08,
+      "loss": 0.3845,
+      "step": 20619
+    },
+    {
+      "epoch": 1.7571367703451215,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.062193612238174e-08,
+      "loss": 0.6329,
+      "step": 20620
+    },
+    {
+      "epoch": 1.7572219855134215,
+      "grad_norm": 14.875,
+      "learning_rate": 9.055913204065264e-08,
+      "loss": 0.5527,
+      "step": 20621
+    },
+    {
+      "epoch": 1.7573072006817214,
+      "grad_norm": 14.5625,
+      "learning_rate": 9.049634891118625e-08,
+      "loss": 0.7298,
+      "step": 20622
+    },
+    {
+      "epoch": 1.7573924158500214,
+      "grad_norm": 12.0,
+      "learning_rate": 9.043358673511652e-08,
+      "loss": 0.4839,
+      "step": 20623
+    },
+    {
+      "epoch": 1.7574776310183213,
+      "grad_norm": 19.5,
+      "learning_rate": 9.037084551357809e-08,
+      "loss": 0.713,
+      "step": 20624
+    },
+    {
+      "epoch": 1.7575628461866213,
+      "grad_norm": 11.9375,
+      "learning_rate": 9.03081252477045e-08,
+      "loss": 0.4115,
+      "step": 20625
+    },
+    {
+      "epoch": 1.7576480613549212,
+      "grad_norm": 15.75,
+      "learning_rate": 9.024542593862915e-08,
+      "loss": 0.4372,
+      "step": 20626
+    },
+    {
+      "epoch": 1.7577332765232212,
+      "grad_norm": 17.75,
+      "learning_rate": 9.018274758748543e-08,
+      "loss": 0.6959,
+      "step": 20627
+    },
+    {
+      "epoch": 1.7578184916915212,
+      "grad_norm": 13.875,
+      "learning_rate": 9.012009019540552e-08,
+      "loss": 0.6051,
+      "step": 20628
+    },
+    {
+      "epoch": 1.7579037068598211,
+      "grad_norm": 13.1875,
+      "learning_rate": 9.005745376352182e-08,
+      "loss": 0.5824,
+      "step": 20629
+    },
+    {
+      "epoch": 1.757988922028121,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.999483829296649e-08,
+      "loss": 0.7291,
+      "step": 20630
+    },
+    {
+      "epoch": 1.758074137196421,
+      "grad_norm": 22.625,
+      "learning_rate": 8.993224378487111e-08,
+      "loss": 0.6835,
+      "step": 20631
+    },
+    {
+      "epoch": 1.758159352364721,
+      "grad_norm": 24.125,
+      "learning_rate": 8.98696702403666e-08,
+      "loss": 0.7427,
+      "step": 20632
+    },
+    {
+      "epoch": 1.758244567533021,
+      "grad_norm": 16.75,
+      "learning_rate": 8.980711766058345e-08,
+      "loss": 0.6333,
+      "step": 20633
+    },
+    {
+      "epoch": 1.758329782701321,
+      "grad_norm": 21.875,
+      "learning_rate": 8.97445860466524e-08,
+      "loss": 0.735,
+      "step": 20634
+    },
+    {
+      "epoch": 1.7584149978696209,
+      "grad_norm": 12.75,
+      "learning_rate": 8.968207539970367e-08,
+      "loss": 0.4113,
+      "step": 20635
+    },
+    {
+      "epoch": 1.7585002130379208,
+      "grad_norm": 17.5,
+      "learning_rate": 8.961958572086635e-08,
+      "loss": 0.4289,
+      "step": 20636
+    },
+    {
+      "epoch": 1.7585854282062208,
+      "grad_norm": 15.375,
+      "learning_rate": 8.955711701127009e-08,
+      "loss": 0.7262,
+      "step": 20637
+    },
+    {
+      "epoch": 1.7586706433745207,
+      "grad_norm": 14.125,
+      "learning_rate": 8.949466927204358e-08,
+      "loss": 0.5647,
+      "step": 20638
+    },
+    {
+      "epoch": 1.7587558585428207,
+      "grad_norm": 13.25,
+      "learning_rate": 8.943224250431551e-08,
+      "loss": 0.5478,
+      "step": 20639
+    },
+    {
+      "epoch": 1.7588410737111206,
+      "grad_norm": 13.625,
+      "learning_rate": 8.93698367092137e-08,
+      "loss": 0.6549,
+      "step": 20640
+    },
+    {
+      "epoch": 1.7589262888794206,
+      "grad_norm": 23.25,
+      "learning_rate": 8.93074518878663e-08,
+      "loss": 0.8717,
+      "step": 20641
+    },
+    {
+      "epoch": 1.7590115040477206,
+      "grad_norm": 14.9375,
+      "learning_rate": 8.924508804140017e-08,
+      "loss": 0.5585,
+      "step": 20642
+    },
+    {
+      "epoch": 1.7590967192160205,
+      "grad_norm": 16.125,
+      "learning_rate": 8.918274517094247e-08,
+      "loss": 0.6258,
+      "step": 20643
+    },
+    {
+      "epoch": 1.7591819343843205,
+      "grad_norm": 16.375,
+      "learning_rate": 8.912042327762008e-08,
+      "loss": 0.4924,
+      "step": 20644
+    },
+    {
+      "epoch": 1.7592671495526204,
+      "grad_norm": 17.125,
+      "learning_rate": 8.90581223625589e-08,
+      "loss": 0.6396,
+      "step": 20645
+    },
+    {
+      "epoch": 1.7593523647209204,
+      "grad_norm": 20.5,
+      "learning_rate": 8.89958424268847e-08,
+      "loss": 0.6827,
+      "step": 20646
+    },
+    {
+      "epoch": 1.7594375798892203,
+      "grad_norm": 14.5,
+      "learning_rate": 8.893358347172324e-08,
+      "loss": 0.5003,
+      "step": 20647
+    },
+    {
+      "epoch": 1.7595227950575203,
+      "grad_norm": 10.1875,
+      "learning_rate": 8.887134549819959e-08,
+      "loss": 0.1869,
+      "step": 20648
+    },
+    {
+      "epoch": 1.7596080102258203,
+      "grad_norm": 29.625,
+      "learning_rate": 8.880912850743842e-08,
+      "loss": 1.0387,
+      "step": 20649
+    },
+    {
+      "epoch": 1.7596932253941202,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.874693250056369e-08,
+      "loss": 0.6402,
+      "step": 20650
+    },
+    {
+      "epoch": 1.7597784405624202,
+      "grad_norm": 14.625,
+      "learning_rate": 8.868475747869962e-08,
+      "loss": 0.4866,
+      "step": 20651
+    },
+    {
+      "epoch": 1.7598636557307201,
+      "grad_norm": 10.75,
+      "learning_rate": 8.862260344297006e-08,
+      "loss": 0.2299,
+      "step": 20652
+    },
+    {
+      "epoch": 1.75994887089902,
+      "grad_norm": 19.0,
+      "learning_rate": 8.856047039449755e-08,
+      "loss": 0.7385,
+      "step": 20653
+    },
+    {
+      "epoch": 1.76003408606732,
+      "grad_norm": 13.625,
+      "learning_rate": 8.84983583344054e-08,
+      "loss": 0.5489,
+      "step": 20654
+    },
+    {
+      "epoch": 1.76011930123562,
+      "grad_norm": 18.125,
+      "learning_rate": 8.843626726381588e-08,
+      "loss": 0.9173,
+      "step": 20655
+    },
+    {
+      "epoch": 1.76020451640392,
+      "grad_norm": 25.25,
+      "learning_rate": 8.837419718385116e-08,
+      "loss": 0.7126,
+      "step": 20656
+    },
+    {
+      "epoch": 1.76028973157222,
+      "grad_norm": 10.375,
+      "learning_rate": 8.831214809563296e-08,
+      "loss": 0.2193,
+      "step": 20657
+    },
+    {
+      "epoch": 1.7603749467405199,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.825012000028207e-08,
+      "loss": 0.3984,
+      "step": 20658
+    },
+    {
+      "epoch": 1.7604601619088198,
+      "grad_norm": 16.125,
+      "learning_rate": 8.818811289891982e-08,
+      "loss": 0.7087,
+      "step": 20659
+    },
+    {
+      "epoch": 1.7605453770771198,
+      "grad_norm": 17.25,
+      "learning_rate": 8.812612679266655e-08,
+      "loss": 0.8959,
+      "step": 20660
+    },
+    {
+      "epoch": 1.7606305922454197,
+      "grad_norm": 16.0,
+      "learning_rate": 8.80641616826429e-08,
+      "loss": 0.7694,
+      "step": 20661
+    },
+    {
+      "epoch": 1.7607158074137197,
+      "grad_norm": 25.25,
+      "learning_rate": 8.800221756996796e-08,
+      "loss": 0.9626,
+      "step": 20662
+    },
+    {
+      "epoch": 1.7608010225820196,
+      "grad_norm": 19.375,
+      "learning_rate": 8.794029445576142e-08,
+      "loss": 0.9632,
+      "step": 20663
+    },
+    {
+      "epoch": 1.7608862377503196,
+      "grad_norm": 18.375,
+      "learning_rate": 8.787839234114234e-08,
+      "loss": 0.6846,
+      "step": 20664
+    },
+    {
+      "epoch": 1.7609714529186196,
+      "grad_norm": 15.25,
+      "learning_rate": 8.78165112272296e-08,
+      "loss": 0.3888,
+      "step": 20665
+    },
+    {
+      "epoch": 1.7610566680869195,
+      "grad_norm": 14.9375,
+      "learning_rate": 8.775465111514087e-08,
+      "loss": 0.5192,
+      "step": 20666
+    },
+    {
+      "epoch": 1.7611418832552195,
+      "grad_norm": 21.625,
+      "learning_rate": 8.76928120059943e-08,
+      "loss": 0.5423,
+      "step": 20667
+    },
+    {
+      "epoch": 1.7612270984235194,
+      "grad_norm": 15.3125,
+      "learning_rate": 8.763099390090749e-08,
+      "loss": 0.6201,
+      "step": 20668
+    },
+    {
+      "epoch": 1.7613123135918194,
+      "grad_norm": 19.375,
+      "learning_rate": 8.756919680099757e-08,
+      "loss": 0.7506,
+      "step": 20669
+    },
+    {
+      "epoch": 1.7613975287601193,
+      "grad_norm": 12.375,
+      "learning_rate": 8.750742070738103e-08,
+      "loss": 0.3427,
+      "step": 20670
+    },
+    {
+      "epoch": 1.7614827439284193,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.744566562117432e-08,
+      "loss": 0.3936,
+      "step": 20671
+    },
+    {
+      "epoch": 1.7615679590967193,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.73839315434935e-08,
+      "loss": 0.6321,
+      "step": 20672
+    },
+    {
+      "epoch": 1.7616531742650192,
+      "grad_norm": 19.75,
+      "learning_rate": 8.732221847545435e-08,
+      "loss": 0.8804,
+      "step": 20673
+    },
+    {
+      "epoch": 1.7617383894333192,
+      "grad_norm": 13.625,
+      "learning_rate": 8.726052641817182e-08,
+      "loss": 0.4666,
+      "step": 20674
+    },
+    {
+      "epoch": 1.7618236046016191,
+      "grad_norm": 15.125,
+      "learning_rate": 8.719885537276069e-08,
+      "loss": 0.5986,
+      "step": 20675
+    },
+    {
+      "epoch": 1.761908819769919,
+      "grad_norm": 9.9375,
+      "learning_rate": 8.713720534033548e-08,
+      "loss": 0.2478,
+      "step": 20676
+    },
+    {
+      "epoch": 1.761994034938219,
+      "grad_norm": 26.625,
+      "learning_rate": 8.707557632201032e-08,
+      "loss": 0.7374,
+      "step": 20677
+    },
+    {
+      "epoch": 1.762079250106519,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.701396831889919e-08,
+      "loss": 0.3632,
+      "step": 20678
+    },
+    {
+      "epoch": 1.762164465274819,
+      "grad_norm": 16.25,
+      "learning_rate": 8.695238133211478e-08,
+      "loss": 0.7837,
+      "step": 20679
+    },
+    {
+      "epoch": 1.762249680443119,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.689081536277052e-08,
+      "loss": 0.3609,
+      "step": 20680
+    },
+    {
+      "epoch": 1.7623348956114189,
+      "grad_norm": 17.125,
+      "learning_rate": 8.682927041197886e-08,
+      "loss": 0.667,
+      "step": 20681
+    },
+    {
+      "epoch": 1.7624201107797188,
+      "grad_norm": 12.125,
+      "learning_rate": 8.67677464808521e-08,
+      "loss": 0.3665,
+      "step": 20682
+    },
+    {
+      "epoch": 1.7625053259480188,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.670624357050156e-08,
+      "loss": 0.6295,
+      "step": 20683
+    },
+    {
+      "epoch": 1.7625905411163187,
+      "grad_norm": 19.25,
+      "learning_rate": 8.664476168203915e-08,
+      "loss": 0.9638,
+      "step": 20684
+    },
+    {
+      "epoch": 1.7626757562846187,
+      "grad_norm": 17.125,
+      "learning_rate": 8.658330081657562e-08,
+      "loss": 0.4027,
+      "step": 20685
+    },
+    {
+      "epoch": 1.7627609714529187,
+      "grad_norm": 17.0,
+      "learning_rate": 8.652186097522191e-08,
+      "loss": 0.4941,
+      "step": 20686
+    },
+    {
+      "epoch": 1.7628461866212186,
+      "grad_norm": 13.8125,
+      "learning_rate": 8.646044215908796e-08,
+      "loss": 0.4105,
+      "step": 20687
+    },
+    {
+      "epoch": 1.7629314017895186,
+      "grad_norm": 17.125,
+      "learning_rate": 8.639904436928387e-08,
+      "loss": 0.6597,
+      "step": 20688
+    },
+    {
+      "epoch": 1.7630166169578185,
+      "grad_norm": 15.1875,
+      "learning_rate": 8.633766760691942e-08,
+      "loss": 0.639,
+      "step": 20689
+    },
+    {
+      "epoch": 1.7631018321261185,
+      "grad_norm": 13.875,
+      "learning_rate": 8.627631187310318e-08,
+      "loss": 0.6132,
+      "step": 20690
+    },
+    {
+      "epoch": 1.7631870472944184,
+      "grad_norm": 25.75,
+      "learning_rate": 8.621497716894439e-08,
+      "loss": 0.6651,
+      "step": 20691
+    },
+    {
+      "epoch": 1.7632722624627184,
+      "grad_norm": 11.0,
+      "learning_rate": 8.615366349555107e-08,
+      "loss": 0.2936,
+      "step": 20692
+    },
+    {
+      "epoch": 1.7633574776310184,
+      "grad_norm": 18.75,
+      "learning_rate": 8.609237085403135e-08,
+      "loss": 1.0493,
+      "step": 20693
+    },
+    {
+      "epoch": 1.7634426927993183,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.603109924549296e-08,
+      "loss": 0.6845,
+      "step": 20694
+    },
+    {
+      "epoch": 1.7635279079676183,
+      "grad_norm": 12.0625,
+      "learning_rate": 8.596984867104322e-08,
+      "loss": 0.3444,
+      "step": 20695
+    },
+    {
+      "epoch": 1.7636131231359182,
+      "grad_norm": 21.875,
+      "learning_rate": 8.590861913178859e-08,
+      "loss": 1.1164,
+      "step": 20696
+    },
+    {
+      "epoch": 1.7636983383042182,
+      "grad_norm": 8.625,
+      "learning_rate": 8.584741062883609e-08,
+      "loss": 0.4481,
+      "step": 20697
+    },
+    {
+      "epoch": 1.7637835534725181,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.578622316329125e-08,
+      "loss": 0.2588,
+      "step": 20698
+    },
+    {
+      "epoch": 1.763868768640818,
+      "grad_norm": 13.875,
+      "learning_rate": 8.572505673626025e-08,
+      "loss": 0.566,
+      "step": 20699
+    },
+    {
+      "epoch": 1.763953983809118,
+      "grad_norm": 23.0,
+      "learning_rate": 8.566391134884819e-08,
+      "loss": 0.6733,
+      "step": 20700
+    },
+    {
+      "epoch": 1.764039198977418,
+      "grad_norm": 14.3125,
+      "learning_rate": 8.560278700216001e-08,
+      "loss": 0.9824,
+      "step": 20701
+    },
+    {
+      "epoch": 1.764124414145718,
+      "grad_norm": 19.375,
+      "learning_rate": 8.554168369730026e-08,
+      "loss": 1.0365,
+      "step": 20702
+    },
+    {
+      "epoch": 1.764209629314018,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.548060143537359e-08,
+      "loss": 0.4187,
+      "step": 20703
+    },
+    {
+      "epoch": 1.7642948444823179,
+      "grad_norm": 13.875,
+      "learning_rate": 8.541954021748317e-08,
+      "loss": 0.3648,
+      "step": 20704
+    },
+    {
+      "epoch": 1.7643800596506178,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.535850004473267e-08,
+      "loss": 0.5229,
+      "step": 20705
+    },
+    {
+      "epoch": 1.7644652748189178,
+      "grad_norm": 19.25,
+      "learning_rate": 8.529748091822554e-08,
+      "loss": 0.5447,
+      "step": 20706
+    },
+    {
+      "epoch": 1.7645504899872178,
+      "grad_norm": 11.25,
+      "learning_rate": 8.523648283906394e-08,
+      "loss": 0.3352,
+      "step": 20707
+    },
+    {
+      "epoch": 1.7646357051555177,
+      "grad_norm": 16.75,
+      "learning_rate": 8.517550580835046e-08,
+      "loss": 0.6091,
+      "step": 20708
+    },
+    {
+      "epoch": 1.7647209203238177,
+      "grad_norm": 17.125,
+      "learning_rate": 8.511454982718672e-08,
+      "loss": 0.561,
+      "step": 20709
+    },
+    {
+      "epoch": 1.7648061354921176,
+      "grad_norm": 22.375,
+      "learning_rate": 8.50536148966745e-08,
+      "loss": 0.9402,
+      "step": 20710
+    },
+    {
+      "epoch": 1.7648913506604176,
+      "grad_norm": 26.75,
+      "learning_rate": 8.499270101791496e-08,
+      "loss": 0.8066,
+      "step": 20711
+    },
+    {
+      "epoch": 1.7649765658287175,
+      "grad_norm": 15.125,
+      "learning_rate": 8.493180819200891e-08,
+      "loss": 0.3866,
+      "step": 20712
+    },
+    {
+      "epoch": 1.7650617809970175,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.487093642005645e-08,
+      "loss": 0.3368,
+      "step": 20713
+    },
+    {
+      "epoch": 1.7651469961653175,
+      "grad_norm": 9.875,
+      "learning_rate": 8.481008570315807e-08,
+      "loss": 0.2732,
+      "step": 20714
+    },
+    {
+      "epoch": 1.7652322113336174,
+      "grad_norm": 14.875,
+      "learning_rate": 8.474925604241291e-08,
+      "loss": 0.4356,
+      "step": 20715
+    },
+    {
+      "epoch": 1.7653174265019174,
+      "grad_norm": 20.125,
+      "learning_rate": 8.468844743892063e-08,
+      "loss": 0.8607,
+      "step": 20716
+    },
+    {
+      "epoch": 1.7654026416702173,
+      "grad_norm": 16.875,
+      "learning_rate": 8.462765989377967e-08,
+      "loss": 0.5252,
+      "step": 20717
+    },
+    {
+      "epoch": 1.7654878568385173,
+      "grad_norm": 16.375,
+      "learning_rate": 8.456689340808885e-08,
+      "loss": 0.3658,
+      "step": 20718
+    },
+    {
+      "epoch": 1.7655730720068172,
+      "grad_norm": 30.875,
+      "learning_rate": 8.450614798294607e-08,
+      "loss": 0.6712,
+      "step": 20719
+    },
+    {
+      "epoch": 1.7656582871751172,
+      "grad_norm": 16.5,
+      "learning_rate": 8.44454236194496e-08,
+      "loss": 0.8908,
+      "step": 20720
+    },
+    {
+      "epoch": 1.7657435023434171,
+      "grad_norm": 25.375,
+      "learning_rate": 8.438472031869607e-08,
+      "loss": 0.9415,
+      "step": 20721
+    },
+    {
+      "epoch": 1.765828717511717,
+      "grad_norm": 19.625,
+      "learning_rate": 8.432403808178294e-08,
+      "loss": 0.7745,
+      "step": 20722
+    },
+    {
+      "epoch": 1.765913932680017,
+      "grad_norm": 17.75,
+      "learning_rate": 8.42633769098064e-08,
+      "loss": 0.7632,
+      "step": 20723
+    },
+    {
+      "epoch": 1.765999147848317,
+      "grad_norm": 17.0,
+      "learning_rate": 8.420273680386296e-08,
+      "loss": 0.716,
+      "step": 20724
+    },
+    {
+      "epoch": 1.766084363016617,
+      "grad_norm": 16.875,
+      "learning_rate": 8.414211776504852e-08,
+      "loss": 0.5491,
+      "step": 20725
+    },
+    {
+      "epoch": 1.766169578184917,
+      "grad_norm": 12.0625,
+      "learning_rate": 8.408151979445805e-08,
+      "loss": 0.2904,
+      "step": 20726
+    },
+    {
+      "epoch": 1.766254793353217,
+      "grad_norm": 18.0,
+      "learning_rate": 8.402094289318707e-08,
+      "loss": 0.8032,
+      "step": 20727
+    },
+    {
+      "epoch": 1.7663400085215168,
+      "grad_norm": 19.25,
+      "learning_rate": 8.39603870623304e-08,
+      "loss": 0.7255,
+      "step": 20728
+    },
+    {
+      "epoch": 1.7664252236898168,
+      "grad_norm": 13.0,
+      "learning_rate": 8.389985230298172e-08,
+      "loss": 0.5369,
+      "step": 20729
+    },
+    {
+      "epoch": 1.7665104388581168,
+      "grad_norm": 25.125,
+      "learning_rate": 8.383933861623562e-08,
+      "loss": 0.7596,
+      "step": 20730
+    },
+    {
+      "epoch": 1.7665956540264167,
+      "grad_norm": 16.125,
+      "learning_rate": 8.37788460031852e-08,
+      "loss": 0.3899,
+      "step": 20731
+    },
+    {
+      "epoch": 1.7666808691947167,
+      "grad_norm": 16.375,
+      "learning_rate": 8.371837446492364e-08,
+      "loss": 0.4507,
+      "step": 20732
+    },
+    {
+      "epoch": 1.7667660843630166,
+      "grad_norm": 19.75,
+      "learning_rate": 8.36579240025441e-08,
+      "loss": 0.838,
+      "step": 20733
+    },
+    {
+      "epoch": 1.7668512995313166,
+      "grad_norm": 16.0,
+      "learning_rate": 8.359749461713849e-08,
+      "loss": 0.4838,
+      "step": 20734
+    },
+    {
+      "epoch": 1.7669365146996165,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.353708630979895e-08,
+      "loss": 0.4212,
+      "step": 20735
+    },
+    {
+      "epoch": 1.7670217298679165,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.347669908161743e-08,
+      "loss": 0.5375,
+      "step": 20736
+    },
+    {
+      "epoch": 1.7671069450362165,
+      "grad_norm": 17.125,
+      "learning_rate": 8.341633293368512e-08,
+      "loss": 0.5857,
+      "step": 20737
+    },
+    {
+      "epoch": 1.7671921602045164,
+      "grad_norm": 17.75,
+      "learning_rate": 8.335598786709267e-08,
+      "loss": 0.7174,
+      "step": 20738
+    },
+    {
+      "epoch": 1.7672773753728164,
+      "grad_norm": 13.4375,
+      "learning_rate": 8.329566388293062e-08,
+      "loss": 0.3059,
+      "step": 20739
+    },
+    {
+      "epoch": 1.7673625905411163,
+      "grad_norm": 12.1875,
+      "learning_rate": 8.323536098228905e-08,
+      "loss": 0.3925,
+      "step": 20740
+    },
+    {
+      "epoch": 1.7674478057094163,
+      "grad_norm": 12.875,
+      "learning_rate": 8.317507916625778e-08,
+      "loss": 0.4589,
+      "step": 20741
+    },
+    {
+      "epoch": 1.7675330208777162,
+      "grad_norm": 15.75,
+      "learning_rate": 8.311481843592637e-08,
+      "loss": 0.8686,
+      "step": 20742
+    },
+    {
+      "epoch": 1.7676182360460162,
+      "grad_norm": 23.75,
+      "learning_rate": 8.305457879238338e-08,
+      "loss": 1.1705,
+      "step": 20743
+    },
+    {
+      "epoch": 1.7677034512143162,
+      "grad_norm": 19.375,
+      "learning_rate": 8.299436023671754e-08,
+      "loss": 1.0137,
+      "step": 20744
+    },
+    {
+      "epoch": 1.7677886663826161,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.29341627700174e-08,
+      "loss": 0.388,
+      "step": 20745
+    },
+    {
+      "epoch": 1.767873881550916,
+      "grad_norm": 12.3125,
+      "learning_rate": 8.287398639337058e-08,
+      "loss": 0.3021,
+      "step": 20746
+    },
+    {
+      "epoch": 1.767959096719216,
+      "grad_norm": 13.6875,
+      "learning_rate": 8.281383110786411e-08,
+      "loss": 0.3623,
+      "step": 20747
+    },
+    {
+      "epoch": 1.768044311887516,
+      "grad_norm": 16.375,
+      "learning_rate": 8.275369691458562e-08,
+      "loss": 0.8024,
+      "step": 20748
+    },
+    {
+      "epoch": 1.768129527055816,
+      "grad_norm": 18.25,
+      "learning_rate": 8.269358381462145e-08,
+      "loss": 0.956,
+      "step": 20749
+    },
+    {
+      "epoch": 1.768214742224116,
+      "grad_norm": 13.3125,
+      "learning_rate": 8.263349180905821e-08,
+      "loss": 0.5184,
+      "step": 20750
+    },
+    {
+      "epoch": 1.7682999573924159,
+      "grad_norm": 31.0,
+      "learning_rate": 8.257342089898146e-08,
+      "loss": 0.8549,
+      "step": 20751
+    },
+    {
+      "epoch": 1.7683851725607158,
+      "grad_norm": 11.9375,
+      "learning_rate": 8.251337108547697e-08,
+      "loss": 0.223,
+      "step": 20752
+    },
+    {
+      "epoch": 1.7684703877290158,
+      "grad_norm": 14.75,
+      "learning_rate": 8.245334236962985e-08,
+      "loss": 0.609,
+      "step": 20753
+    },
+    {
+      "epoch": 1.7685556028973157,
+      "grad_norm": 12.625,
+      "learning_rate": 8.239333475252506e-08,
+      "loss": 0.3914,
+      "step": 20754
+    },
+    {
+      "epoch": 1.7686408180656157,
+      "grad_norm": 21.25,
+      "learning_rate": 8.233334823524689e-08,
+      "loss": 0.4608,
+      "step": 20755
+    },
+    {
+      "epoch": 1.7687260332339156,
+      "grad_norm": 28.25,
+      "learning_rate": 8.227338281887904e-08,
+      "loss": 0.7798,
+      "step": 20756
+    },
+    {
+      "epoch": 1.7688112484022156,
+      "grad_norm": 18.125,
+      "learning_rate": 8.22134385045055e-08,
+      "loss": 0.626,
+      "step": 20757
+    },
+    {
+      "epoch": 1.7688964635705156,
+      "grad_norm": 10.5625,
+      "learning_rate": 8.215351529320958e-08,
+      "loss": 0.2872,
+      "step": 20758
+    },
+    {
+      "epoch": 1.7689816787388155,
+      "grad_norm": 13.25,
+      "learning_rate": 8.209361318607376e-08,
+      "loss": 0.4227,
+      "step": 20759
+    },
+    {
+      "epoch": 1.7690668939071155,
+      "grad_norm": 16.875,
+      "learning_rate": 8.203373218418076e-08,
+      "loss": 0.83,
+      "step": 20760
+    },
+    {
+      "epoch": 1.7691521090754154,
+      "grad_norm": 11.125,
+      "learning_rate": 8.197387228861278e-08,
+      "loss": 0.309,
+      "step": 20761
+    },
+    {
+      "epoch": 1.7692373242437154,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.19140335004516e-08,
+      "loss": 0.3202,
+      "step": 20762
+    },
+    {
+      "epoch": 1.7693225394120153,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.185421582077843e-08,
+      "loss": 0.3868,
+      "step": 20763
+    },
+    {
+      "epoch": 1.7694077545803153,
+      "grad_norm": 13.125,
+      "learning_rate": 8.179441925067408e-08,
+      "loss": 0.4064,
+      "step": 20764
+    },
+    {
+      "epoch": 1.7694929697486153,
+      "grad_norm": 12.625,
+      "learning_rate": 8.17346437912192e-08,
+      "loss": 0.3826,
+      "step": 20765
+    },
+    {
+      "epoch": 1.7695781849169152,
+      "grad_norm": 18.875,
+      "learning_rate": 8.16748894434942e-08,
+      "loss": 0.4824,
+      "step": 20766
+    },
+    {
+      "epoch": 1.7696634000852152,
+      "grad_norm": 10.75,
+      "learning_rate": 8.161515620857902e-08,
+      "loss": 0.3649,
+      "step": 20767
+    },
+    {
+      "epoch": 1.7697486152535151,
+      "grad_norm": 19.25,
+      "learning_rate": 8.155544408755253e-08,
+      "loss": 0.3717,
+      "step": 20768
+    },
+    {
+      "epoch": 1.769833830421815,
+      "grad_norm": 12.0625,
+      "learning_rate": 8.149575308149429e-08,
+      "loss": 0.3922,
+      "step": 20769
+    },
+    {
+      "epoch": 1.769919045590115,
+      "grad_norm": 10.8125,
+      "learning_rate": 8.143608319148274e-08,
+      "loss": 0.2052,
+      "step": 20770
+    },
+    {
+      "epoch": 1.770004260758415,
+      "grad_norm": 14.0,
+      "learning_rate": 8.137643441859632e-08,
+      "loss": 0.4265,
+      "step": 20771
+    },
+    {
+      "epoch": 1.770089475926715,
+      "grad_norm": 22.375,
+      "learning_rate": 8.131680676391307e-08,
+      "loss": 1.121,
+      "step": 20772
+    },
+    {
+      "epoch": 1.770174691095015,
+      "grad_norm": 15.9375,
+      "learning_rate": 8.125720022851003e-08,
+      "loss": 0.7936,
+      "step": 20773
+    },
+    {
+      "epoch": 1.7702599062633149,
+      "grad_norm": 15.6875,
+      "learning_rate": 8.119761481346466e-08,
+      "loss": 0.5297,
+      "step": 20774
+    },
+    {
+      "epoch": 1.7703451214316148,
+      "grad_norm": 16.75,
+      "learning_rate": 8.11380505198539e-08,
+      "loss": 0.6804,
+      "step": 20775
+    },
+    {
+      "epoch": 1.7704303365999148,
+      "grad_norm": 13.75,
+      "learning_rate": 8.10785073487537e-08,
+      "loss": 0.4745,
+      "step": 20776
+    },
+    {
+      "epoch": 1.7705155517682147,
+      "grad_norm": 19.75,
+      "learning_rate": 8.101898530124039e-08,
+      "loss": 0.6839,
+      "step": 20777
+    },
+    {
+      "epoch": 1.7706007669365147,
+      "grad_norm": 23.0,
+      "learning_rate": 8.095948437838952e-08,
+      "loss": 0.705,
+      "step": 20778
+    },
+    {
+      "epoch": 1.7706859821048146,
+      "grad_norm": 13.0,
+      "learning_rate": 8.090000458127647e-08,
+      "loss": 0.3285,
+      "step": 20779
+    },
+    {
+      "epoch": 1.7707711972731146,
+      "grad_norm": 17.125,
+      "learning_rate": 8.084054591097609e-08,
+      "loss": 0.676,
+      "step": 20780
+    },
+    {
+      "epoch": 1.7708564124414146,
+      "grad_norm": 11.1875,
+      "learning_rate": 8.078110836856237e-08,
+      "loss": 0.3991,
+      "step": 20781
+    },
+    {
+      "epoch": 1.7709416276097145,
+      "grad_norm": 46.25,
+      "learning_rate": 8.072169195510987e-08,
+      "loss": 0.4596,
+      "step": 20782
+    },
+    {
+      "epoch": 1.7710268427780145,
+      "grad_norm": 30.625,
+      "learning_rate": 8.066229667169218e-08,
+      "loss": 0.8189,
+      "step": 20783
+    },
+    {
+      "epoch": 1.7711120579463144,
+      "grad_norm": 17.5,
+      "learning_rate": 8.060292251938273e-08,
+      "loss": 0.8743,
+      "step": 20784
+    },
+    {
+      "epoch": 1.7711972731146144,
+      "grad_norm": 21.5,
+      "learning_rate": 8.054356949925429e-08,
+      "loss": 1.1527,
+      "step": 20785
+    },
+    {
+      "epoch": 1.7712824882829143,
+      "grad_norm": 21.625,
+      "learning_rate": 8.048423761237947e-08,
+      "loss": 0.6426,
+      "step": 20786
+    },
+    {
+      "epoch": 1.7713677034512143,
+      "grad_norm": 19.25,
+      "learning_rate": 8.042492685983074e-08,
+      "loss": 0.8161,
+      "step": 20787
+    },
+    {
+      "epoch": 1.7714529186195143,
+      "grad_norm": 15.125,
+      "learning_rate": 8.036563724267976e-08,
+      "loss": 0.393,
+      "step": 20788
+    },
+    {
+      "epoch": 1.7715381337878142,
+      "grad_norm": 21.875,
+      "learning_rate": 8.03063687619976e-08,
+      "loss": 0.5933,
+      "step": 20789
+    },
+    {
+      "epoch": 1.7716233489561142,
+      "grad_norm": 12.8125,
+      "learning_rate": 8.024712141885563e-08,
+      "loss": 0.5284,
+      "step": 20790
+    },
+    {
+      "epoch": 1.7717085641244141,
+      "grad_norm": 18.0,
+      "learning_rate": 8.018789521432438e-08,
+      "loss": 0.7969,
+      "step": 20791
+    },
+    {
+      "epoch": 1.771793779292714,
+      "grad_norm": 24.125,
+      "learning_rate": 8.012869014947453e-08,
+      "loss": 0.585,
+      "step": 20792
+    },
+    {
+      "epoch": 1.771878994461014,
+      "grad_norm": 14.625,
+      "learning_rate": 8.006950622537535e-08,
+      "loss": 0.524,
+      "step": 20793
+    },
+    {
+      "epoch": 1.771964209629314,
+      "grad_norm": 16.75,
+      "learning_rate": 8.001034344309655e-08,
+      "loss": 0.8657,
+      "step": 20794
+    },
+    {
+      "epoch": 1.772049424797614,
+      "grad_norm": 13.5,
+      "learning_rate": 7.995120180370769e-08,
+      "loss": 0.5076,
+      "step": 20795
+    },
+    {
+      "epoch": 1.772134639965914,
+      "grad_norm": 18.0,
+      "learning_rate": 7.989208130827694e-08,
+      "loss": 0.5252,
+      "step": 20796
+    },
+    {
+      "epoch": 1.7722198551342139,
+      "grad_norm": 13.5625,
+      "learning_rate": 7.983298195787317e-08,
+      "loss": 0.5297,
+      "step": 20797
+    },
+    {
+      "epoch": 1.7723050703025138,
+      "grad_norm": 15.625,
+      "learning_rate": 7.977390375356387e-08,
+      "loss": 0.5128,
+      "step": 20798
+    },
+    {
+      "epoch": 1.7723902854708138,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.97148466964169e-08,
+      "loss": 0.3772,
+      "step": 20799
+    },
+    {
+      "epoch": 1.7724755006391137,
+      "grad_norm": 14.5,
+      "learning_rate": 7.965581078749938e-08,
+      "loss": 0.5202,
+      "step": 20800
+    },
+    {
+      "epoch": 1.7725607158074137,
+      "grad_norm": 18.625,
+      "learning_rate": 7.959679602787846e-08,
+      "loss": 0.6212,
+      "step": 20801
+    },
+    {
+      "epoch": 1.7726459309757137,
+      "grad_norm": 12.4375,
+      "learning_rate": 7.953780241862024e-08,
+      "loss": 0.5291,
+      "step": 20802
+    },
+    {
+      "epoch": 1.7727311461440136,
+      "grad_norm": 13.9375,
+      "learning_rate": 7.947882996079112e-08,
+      "loss": 0.5405,
+      "step": 20803
+    },
+    {
+      "epoch": 1.7728163613123136,
+      "grad_norm": 31.0,
+      "learning_rate": 7.941987865545633e-08,
+      "loss": 0.9705,
+      "step": 20804
+    },
+    {
+      "epoch": 1.7729015764806135,
+      "grad_norm": 17.625,
+      "learning_rate": 7.93609485036817e-08,
+      "loss": 0.5886,
+      "step": 20805
+    },
+    {
+      "epoch": 1.7729867916489135,
+      "grad_norm": 11.1875,
+      "learning_rate": 7.930203950653165e-08,
+      "loss": 0.1844,
+      "step": 20806
+    },
+    {
+      "epoch": 1.7730720068172134,
+      "grad_norm": 16.5,
+      "learning_rate": 7.924315166507102e-08,
+      "loss": 0.7846,
+      "step": 20807
+    },
+    {
+      "epoch": 1.7731572219855134,
+      "grad_norm": 24.5,
+      "learning_rate": 7.918428498036395e-08,
+      "loss": 0.853,
+      "step": 20808
+    },
+    {
+      "epoch": 1.7732424371538134,
+      "grad_norm": 20.875,
+      "learning_rate": 7.912543945347434e-08,
+      "loss": 0.4868,
+      "step": 20809
+    },
+    {
+      "epoch": 1.7733276523221133,
+      "grad_norm": 16.25,
+      "learning_rate": 7.906661508546521e-08,
+      "loss": 0.4852,
+      "step": 20810
+    },
+    {
+      "epoch": 1.7734128674904133,
+      "grad_norm": 14.0,
+      "learning_rate": 7.900781187740001e-08,
+      "loss": 0.5662,
+      "step": 20811
+    },
+    {
+      "epoch": 1.7734980826587132,
+      "grad_norm": 17.625,
+      "learning_rate": 7.894902983034108e-08,
+      "loss": 0.6124,
+      "step": 20812
+    },
+    {
+      "epoch": 1.7735832978270132,
+      "grad_norm": 15.125,
+      "learning_rate": 7.889026894535065e-08,
+      "loss": 0.2674,
+      "step": 20813
+    },
+    {
+      "epoch": 1.7736685129953131,
+      "grad_norm": 19.0,
+      "learning_rate": 7.883152922349091e-08,
+      "loss": 0.9484,
+      "step": 20814
+    },
+    {
+      "epoch": 1.773753728163613,
+      "grad_norm": 14.625,
+      "learning_rate": 7.877281066582295e-08,
+      "loss": 0.4141,
+      "step": 20815
+    },
+    {
+      "epoch": 1.773838943331913,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.871411327340817e-08,
+      "loss": 0.6194,
+      "step": 20816
+    },
+    {
+      "epoch": 1.773924158500213,
+      "grad_norm": 16.5,
+      "learning_rate": 7.865543704730708e-08,
+      "loss": 0.6546,
+      "step": 20817
+    },
+    {
+      "epoch": 1.774009373668513,
+      "grad_norm": 13.0625,
+      "learning_rate": 7.859678198858037e-08,
+      "loss": 0.3428,
+      "step": 20818
+    },
+    {
+      "epoch": 1.774094588836813,
+      "grad_norm": 18.75,
+      "learning_rate": 7.853814809828775e-08,
+      "loss": 0.8422,
+      "step": 20819
+    },
+    {
+      "epoch": 1.7741798040051129,
+      "grad_norm": 13.125,
+      "learning_rate": 7.847953537748851e-08,
+      "loss": 0.8881,
+      "step": 20820
+    },
+    {
+      "epoch": 1.7742650191734128,
+      "grad_norm": 13.0625,
+      "learning_rate": 7.842094382724222e-08,
+      "loss": 0.5303,
+      "step": 20821
+    },
+    {
+      "epoch": 1.7743502343417128,
+      "grad_norm": 13.75,
+      "learning_rate": 7.836237344860775e-08,
+      "loss": 0.5254,
+      "step": 20822
+    },
+    {
+      "epoch": 1.7744354495100128,
+      "grad_norm": 16.25,
+      "learning_rate": 7.830382424264316e-08,
+      "loss": 0.7206,
+      "step": 20823
+    },
+    {
+      "epoch": 1.7745206646783127,
+      "grad_norm": 19.625,
+      "learning_rate": 7.824529621040661e-08,
+      "loss": 0.6966,
+      "step": 20824
+    },
+    {
+      "epoch": 1.7746058798466127,
+      "grad_norm": 14.875,
+      "learning_rate": 7.818678935295601e-08,
+      "loss": 0.6469,
+      "step": 20825
+    },
+    {
+      "epoch": 1.7746910950149126,
+      "grad_norm": 17.75,
+      "learning_rate": 7.812830367134857e-08,
+      "loss": 0.785,
+      "step": 20826
+    },
+    {
+      "epoch": 1.7747763101832126,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.80698391666411e-08,
+      "loss": 0.6313,
+      "step": 20827
+    },
+    {
+      "epoch": 1.7748615253515125,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.801139583988996e-08,
+      "loss": 0.5272,
+      "step": 20828
+    },
+    {
+      "epoch": 1.7749467405198125,
+      "grad_norm": 14.25,
+      "learning_rate": 7.795297369215124e-08,
+      "loss": 0.413,
+      "step": 20829
+    },
+    {
+      "epoch": 1.7750319556881125,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.789457272448106e-08,
+      "loss": 0.6186,
+      "step": 20830
+    },
+    {
+      "epoch": 1.7751171708564124,
+      "grad_norm": 14.625,
+      "learning_rate": 7.783619293793482e-08,
+      "loss": 0.5406,
+      "step": 20831
+    },
+    {
+      "epoch": 1.7752023860247124,
+      "grad_norm": 10.875,
+      "learning_rate": 7.777783433356695e-08,
+      "loss": 0.3311,
+      "step": 20832
+    },
+    {
+      "epoch": 1.7752876011930123,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.771949691243244e-08,
+      "loss": 0.512,
+      "step": 20833
+    },
+    {
+      "epoch": 1.7753728163613123,
+      "grad_norm": 20.5,
+      "learning_rate": 7.766118067558545e-08,
+      "loss": 0.3655,
+      "step": 20834
+    },
+    {
+      "epoch": 1.7754580315296122,
+      "grad_norm": 11.9375,
+      "learning_rate": 7.760288562408014e-08,
+      "loss": 0.7185,
+      "step": 20835
+    },
+    {
+      "epoch": 1.7755432466979122,
+      "grad_norm": 21.75,
+      "learning_rate": 7.754461175896926e-08,
+      "loss": 0.8764,
+      "step": 20836
+    },
+    {
+      "epoch": 1.7756284618662121,
+      "grad_norm": 15.375,
+      "learning_rate": 7.748635908130616e-08,
+      "loss": 0.6555,
+      "step": 20837
+    },
+    {
+      "epoch": 1.775713677034512,
+      "grad_norm": 17.5,
+      "learning_rate": 7.742812759214375e-08,
+      "loss": 0.4821,
+      "step": 20838
+    },
+    {
+      "epoch": 1.775798892202812,
+      "grad_norm": 17.125,
+      "learning_rate": 7.736991729253434e-08,
+      "loss": 0.7515,
+      "step": 20839
+    },
+    {
+      "epoch": 1.775884107371112,
+      "grad_norm": 16.125,
+      "learning_rate": 7.73117281835295e-08,
+      "loss": 0.7898,
+      "step": 20840
+    },
+    {
+      "epoch": 1.775969322539412,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.725356026618086e-08,
+      "loss": 0.7704,
+      "step": 20841
+    },
+    {
+      "epoch": 1.776054537707712,
+      "grad_norm": 18.125,
+      "learning_rate": 7.719541354153981e-08,
+      "loss": 0.5937,
+      "step": 20842
+    },
+    {
+      "epoch": 1.776139752876012,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.713728801065718e-08,
+      "loss": 0.419,
+      "step": 20843
+    },
+    {
+      "epoch": 1.7762249680443118,
+      "grad_norm": 12.125,
+      "learning_rate": 7.707918367458309e-08,
+      "loss": 0.3469,
+      "step": 20844
+    },
+    {
+      "epoch": 1.7763101832126118,
+      "grad_norm": 10.9375,
+      "learning_rate": 7.70211005343674e-08,
+      "loss": 0.4084,
+      "step": 20845
+    },
+    {
+      "epoch": 1.7763953983809118,
+      "grad_norm": 16.375,
+      "learning_rate": 7.696303859105997e-08,
+      "loss": 0.5205,
+      "step": 20846
+    },
+    {
+      "epoch": 1.7764806135492117,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.690499784570995e-08,
+      "loss": 0.5272,
+      "step": 20847
+    },
+    {
+      "epoch": 1.7765658287175117,
+      "grad_norm": 22.25,
+      "learning_rate": 7.684697829936652e-08,
+      "loss": 0.8031,
+      "step": 20848
+    },
+    {
+      "epoch": 1.7766510438858116,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.678897995307772e-08,
+      "loss": 0.5615,
+      "step": 20849
+    },
+    {
+      "epoch": 1.7767362590541116,
+      "grad_norm": 16.75,
+      "learning_rate": 7.673100280789187e-08,
+      "loss": 0.4769,
+      "step": 20850
+    },
+    {
+      "epoch": 1.7768214742224115,
+      "grad_norm": 17.625,
+      "learning_rate": 7.667304686485661e-08,
+      "loss": 0.4673,
+      "step": 20851
+    },
+    {
+      "epoch": 1.7769066893907115,
+      "grad_norm": 18.5,
+      "learning_rate": 7.661511212501943e-08,
+      "loss": 0.8326,
+      "step": 20852
+    },
+    {
+      "epoch": 1.7769919045590115,
+      "grad_norm": 17.125,
+      "learning_rate": 7.655719858942687e-08,
+      "loss": 0.7633,
+      "step": 20853
+    },
+    {
+      "epoch": 1.7770771197273114,
+      "grad_norm": 18.875,
+      "learning_rate": 7.649930625912572e-08,
+      "loss": 0.6106,
+      "step": 20854
+    },
+    {
+      "epoch": 1.7771623348956114,
+      "grad_norm": 29.25,
+      "learning_rate": 7.644143513516209e-08,
+      "loss": 0.9202,
+      "step": 20855
+    },
+    {
+      "epoch": 1.7772475500639113,
+      "grad_norm": 13.625,
+      "learning_rate": 7.638358521858207e-08,
+      "loss": 0.4029,
+      "step": 20856
+    },
+    {
+      "epoch": 1.7773327652322113,
+      "grad_norm": 31.75,
+      "learning_rate": 7.63257565104307e-08,
+      "loss": 0.6591,
+      "step": 20857
+    },
+    {
+      "epoch": 1.7774179804005112,
+      "grad_norm": 27.625,
+      "learning_rate": 7.626794901175294e-08,
+      "loss": 0.7683,
+      "step": 20858
+    },
+    {
+      "epoch": 1.7775031955688112,
+      "grad_norm": 14.5,
+      "learning_rate": 7.62101627235938e-08,
+      "loss": 0.6206,
+      "step": 20859
+    },
+    {
+      "epoch": 1.7775884107371112,
+      "grad_norm": 19.25,
+      "learning_rate": 7.615239764699744e-08,
+      "loss": 0.8193,
+      "step": 20860
+    },
+    {
+      "epoch": 1.7776736259054111,
+      "grad_norm": 18.875,
+      "learning_rate": 7.609465378300776e-08,
+      "loss": 0.684,
+      "step": 20861
+    },
+    {
+      "epoch": 1.777758841073711,
+      "grad_norm": 12.375,
+      "learning_rate": 7.603693113266793e-08,
+      "loss": 0.418,
+      "step": 20862
+    },
+    {
+      "epoch": 1.777844056242011,
+      "grad_norm": 25.875,
+      "learning_rate": 7.597922969702118e-08,
+      "loss": 0.8458,
+      "step": 20863
+    },
+    {
+      "epoch": 1.777929271410311,
+      "grad_norm": 12.5,
+      "learning_rate": 7.59215494771104e-08,
+      "loss": 0.4495,
+      "step": 20864
+    },
+    {
+      "epoch": 1.778014486578611,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.586389047397794e-08,
+      "loss": 0.7082,
+      "step": 20865
+    },
+    {
+      "epoch": 1.778099701746911,
+      "grad_norm": 14.875,
+      "learning_rate": 7.580625268866562e-08,
+      "loss": 0.6233,
+      "step": 20866
+    },
+    {
+      "epoch": 1.7781849169152109,
+      "grad_norm": 14.4375,
+      "learning_rate": 7.574863612221497e-08,
+      "loss": 0.6831,
+      "step": 20867
+    },
+    {
+      "epoch": 1.7782701320835108,
+      "grad_norm": 14.5,
+      "learning_rate": 7.569104077566739e-08,
+      "loss": 0.5957,
+      "step": 20868
+    },
+    {
+      "epoch": 1.7783553472518108,
+      "grad_norm": 15.8125,
+      "learning_rate": 7.563346665006369e-08,
+      "loss": 0.4627,
+      "step": 20869
+    },
+    {
+      "epoch": 1.7784405624201107,
+      "grad_norm": 15.125,
+      "learning_rate": 7.557591374644388e-08,
+      "loss": 0.5541,
+      "step": 20870
+    },
+    {
+      "epoch": 1.7785257775884107,
+      "grad_norm": 20.375,
+      "learning_rate": 7.551838206584838e-08,
+      "loss": 0.3962,
+      "step": 20871
+    },
+    {
+      "epoch": 1.7786109927567106,
+      "grad_norm": 17.375,
+      "learning_rate": 7.546087160931664e-08,
+      "loss": 0.5385,
+      "step": 20872
+    },
+    {
+      "epoch": 1.7786962079250106,
+      "grad_norm": 15.9375,
+      "learning_rate": 7.540338237788825e-08,
+      "loss": 0.645,
+      "step": 20873
+    },
+    {
+      "epoch": 1.7787814230933106,
+      "grad_norm": 14.8125,
+      "learning_rate": 7.53459143726018e-08,
+      "loss": 0.7579,
+      "step": 20874
+    },
+    {
+      "epoch": 1.7788666382616105,
+      "grad_norm": 62.5,
+      "learning_rate": 7.528846759449578e-08,
+      "loss": 0.7933,
+      "step": 20875
+    },
+    {
+      "epoch": 1.7789518534299105,
+      "grad_norm": 20.0,
+      "learning_rate": 7.523104204460869e-08,
+      "loss": 0.9118,
+      "step": 20876
+    },
+    {
+      "epoch": 1.7790370685982104,
+      "grad_norm": 14.375,
+      "learning_rate": 7.517363772397771e-08,
+      "loss": 0.4295,
+      "step": 20877
+    },
+    {
+      "epoch": 1.7791222837665104,
+      "grad_norm": 19.625,
+      "learning_rate": 7.511625463364064e-08,
+      "loss": 0.6884,
+      "step": 20878
+    },
+    {
+      "epoch": 1.7792074989348103,
+      "grad_norm": 15.5,
+      "learning_rate": 7.505889277463416e-08,
+      "loss": 0.8537,
+      "step": 20879
+    },
+    {
+      "epoch": 1.7792927141031103,
+      "grad_norm": 12.0,
+      "learning_rate": 7.500155214799507e-08,
+      "loss": 0.3129,
+      "step": 20880
+    },
+    {
+      "epoch": 1.7793779292714103,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.494423275475949e-08,
+      "loss": 0.4777,
+      "step": 20881
+    },
+    {
+      "epoch": 1.7794631444397102,
+      "grad_norm": 11.8125,
+      "learning_rate": 7.488693459596325e-08,
+      "loss": 0.3009,
+      "step": 20882
+    },
+    {
+      "epoch": 1.7795483596080102,
+      "grad_norm": 15.3125,
+      "learning_rate": 7.482965767264164e-08,
+      "loss": 0.4861,
+      "step": 20883
+    },
+    {
+      "epoch": 1.7796335747763101,
+      "grad_norm": 17.0,
+      "learning_rate": 7.477240198583008e-08,
+      "loss": 0.8164,
+      "step": 20884
+    },
+    {
+      "epoch": 1.77971878994461,
+      "grad_norm": 12.6875,
+      "learning_rate": 7.471516753656275e-08,
+      "loss": 0.3189,
+      "step": 20885
+    },
+    {
+      "epoch": 1.77980400511291,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.465795432587436e-08,
+      "loss": 0.688,
+      "step": 20886
+    },
+    {
+      "epoch": 1.77988922028121,
+      "grad_norm": 16.25,
+      "learning_rate": 7.460076235479855e-08,
+      "loss": 0.7007,
+      "step": 20887
+    },
+    {
+      "epoch": 1.77997443544951,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.454359162436877e-08,
+      "loss": 0.3728,
+      "step": 20888
+    },
+    {
+      "epoch": 1.78005965061781,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.44864421356184e-08,
+      "loss": 0.5299,
+      "step": 20889
+    },
+    {
+      "epoch": 1.7801448657861099,
+      "grad_norm": 15.6875,
+      "learning_rate": 7.442931388958033e-08,
+      "loss": 0.5082,
+      "step": 20890
+    },
+    {
+      "epoch": 1.7802300809544098,
+      "grad_norm": 15.75,
+      "learning_rate": 7.437220688728638e-08,
+      "loss": 0.5812,
+      "step": 20891
+    },
+    {
+      "epoch": 1.7803152961227098,
+      "grad_norm": 17.875,
+      "learning_rate": 7.431512112976908e-08,
+      "loss": 0.4528,
+      "step": 20892
+    },
+    {
+      "epoch": 1.7804005112910097,
+      "grad_norm": 8.8125,
+      "learning_rate": 7.425805661805966e-08,
+      "loss": 0.2022,
+      "step": 20893
+    },
+    {
+      "epoch": 1.7804857264593097,
+      "grad_norm": 14.5,
+      "learning_rate": 7.420101335318938e-08,
+      "loss": 0.2984,
+      "step": 20894
+    },
+    {
+      "epoch": 1.7805709416276096,
+      "grad_norm": 12.875,
+      "learning_rate": 7.414399133618924e-08,
+      "loss": 0.26,
+      "step": 20895
+    },
+    {
+      "epoch": 1.7806561567959096,
+      "grad_norm": 18.375,
+      "learning_rate": 7.408699056808952e-08,
+      "loss": 0.798,
+      "step": 20896
+    },
+    {
+      "epoch": 1.7807413719642096,
+      "grad_norm": 18.125,
+      "learning_rate": 7.403001104992036e-08,
+      "loss": 0.7421,
+      "step": 20897
+    },
+    {
+      "epoch": 1.7808265871325095,
+      "grad_norm": 23.375,
+      "learning_rate": 7.397305278271163e-08,
+      "loss": 0.5364,
+      "step": 20898
+    },
+    {
+      "epoch": 1.7809118023008095,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.391611576749224e-08,
+      "loss": 0.3358,
+      "step": 20899
+    },
+    {
+      "epoch": 1.7809970174691094,
+      "grad_norm": 13.5,
+      "learning_rate": 7.385920000529137e-08,
+      "loss": 0.8672,
+      "step": 20900
+    },
+    {
+      "epoch": 1.7810822326374094,
+      "grad_norm": 11.875,
+      "learning_rate": 7.380230549713734e-08,
+      "loss": 0.3468,
+      "step": 20901
+    },
+    {
+      "epoch": 1.7811674478057093,
+      "grad_norm": 13.75,
+      "learning_rate": 7.374543224405839e-08,
+      "loss": 0.5031,
+      "step": 20902
+    },
+    {
+      "epoch": 1.7812526629740093,
+      "grad_norm": 12.3125,
+      "learning_rate": 7.368858024708256e-08,
+      "loss": 0.3944,
+      "step": 20903
+    },
+    {
+      "epoch": 1.7813378781423093,
+      "grad_norm": 13.75,
+      "learning_rate": 7.363174950723668e-08,
+      "loss": 0.5372,
+      "step": 20904
+    },
+    {
+      "epoch": 1.7814230933106092,
+      "grad_norm": 18.75,
+      "learning_rate": 7.35749400255481e-08,
+      "loss": 0.7873,
+      "step": 20905
+    },
+    {
+      "epoch": 1.7815083084789092,
+      "grad_norm": 19.75,
+      "learning_rate": 7.351815180304326e-08,
+      "loss": 0.5804,
+      "step": 20906
+    },
+    {
+      "epoch": 1.7815935236472091,
+      "grad_norm": 19.625,
+      "learning_rate": 7.346138484074867e-08,
+      "loss": 0.5995,
+      "step": 20907
+    },
+    {
+      "epoch": 1.781678738815509,
+      "grad_norm": 21.625,
+      "learning_rate": 7.340463913968976e-08,
+      "loss": 0.5808,
+      "step": 20908
+    },
+    {
+      "epoch": 1.781763953983809,
+      "grad_norm": 10.9375,
+      "learning_rate": 7.334791470089239e-08,
+      "loss": 0.3275,
+      "step": 20909
+    },
+    {
+      "epoch": 1.781849169152109,
+      "grad_norm": 13.125,
+      "learning_rate": 7.329121152538116e-08,
+      "loss": 0.5764,
+      "step": 20910
+    },
+    {
+      "epoch": 1.781934384320409,
+      "grad_norm": 18.875,
+      "learning_rate": 7.32345296141812e-08,
+      "loss": 0.7965,
+      "step": 20911
+    },
+    {
+      "epoch": 1.782019599488709,
+      "grad_norm": 14.5,
+      "learning_rate": 7.317786896831658e-08,
+      "loss": 0.4579,
+      "step": 20912
+    },
+    {
+      "epoch": 1.7821048146570089,
+      "grad_norm": 35.0,
+      "learning_rate": 7.312122958881119e-08,
+      "loss": 0.7094,
+      "step": 20913
+    },
+    {
+      "epoch": 1.7821900298253088,
+      "grad_norm": 33.25,
+      "learning_rate": 7.306461147668851e-08,
+      "loss": 0.7936,
+      "step": 20914
+    },
+    {
+      "epoch": 1.7822752449936088,
+      "grad_norm": 17.5,
+      "learning_rate": 7.300801463297202e-08,
+      "loss": 0.3258,
+      "step": 20915
+    },
+    {
+      "epoch": 1.7823604601619087,
+      "grad_norm": 15.0,
+      "learning_rate": 7.295143905868413e-08,
+      "loss": 0.8004,
+      "step": 20916
+    },
+    {
+      "epoch": 1.7824456753302087,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.289488475484761e-08,
+      "loss": 0.4927,
+      "step": 20917
+    },
+    {
+      "epoch": 1.7825308904985087,
+      "grad_norm": 13.4375,
+      "learning_rate": 7.283835172248386e-08,
+      "loss": 0.5715,
+      "step": 20918
+    },
+    {
+      "epoch": 1.7826161056668086,
+      "grad_norm": 13.9375,
+      "learning_rate": 7.278183996261498e-08,
+      "loss": 0.4183,
+      "step": 20919
+    },
+    {
+      "epoch": 1.7827013208351086,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.272534947626211e-08,
+      "loss": 0.6403,
+      "step": 20920
+    },
+    {
+      "epoch": 1.7827865360034085,
+      "grad_norm": 16.375,
+      "learning_rate": 7.266888026444582e-08,
+      "loss": 0.7423,
+      "step": 20921
+    },
+    {
+      "epoch": 1.7828717511717085,
+      "grad_norm": 12.625,
+      "learning_rate": 7.261243232818683e-08,
+      "loss": 0.424,
+      "step": 20922
+    },
+    {
+      "epoch": 1.7829569663400084,
+      "grad_norm": 16.125,
+      "learning_rate": 7.255600566850512e-08,
+      "loss": 0.658,
+      "step": 20923
+    },
+    {
+      "epoch": 1.7830421815083084,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.249960028642047e-08,
+      "loss": 0.6631,
+      "step": 20924
+    },
+    {
+      "epoch": 1.7831273966766084,
+      "grad_norm": 18.5,
+      "learning_rate": 7.244321618295233e-08,
+      "loss": 0.9254,
+      "step": 20925
+    },
+    {
+      "epoch": 1.7832126118449083,
+      "grad_norm": 32.0,
+      "learning_rate": 7.238685335911904e-08,
+      "loss": 1.0861,
+      "step": 20926
+    },
+    {
+      "epoch": 1.7832978270132083,
+      "grad_norm": 17.125,
+      "learning_rate": 7.233051181593953e-08,
+      "loss": 0.7111,
+      "step": 20927
+    },
+    {
+      "epoch": 1.7833830421815082,
+      "grad_norm": 21.0,
+      "learning_rate": 7.227419155443213e-08,
+      "loss": 0.7134,
+      "step": 20928
+    },
+    {
+      "epoch": 1.7834682573498082,
+      "grad_norm": 14.4375,
+      "learning_rate": 7.221789257561423e-08,
+      "loss": 0.6712,
+      "step": 20929
+    },
+    {
+      "epoch": 1.7835534725181081,
+      "grad_norm": 13.0625,
+      "learning_rate": 7.216161488050321e-08,
+      "loss": 0.4143,
+      "step": 20930
+    },
+    {
+      "epoch": 1.783638687686408,
+      "grad_norm": 16.75,
+      "learning_rate": 7.210535847011632e-08,
+      "loss": 0.6517,
+      "step": 20931
+    },
+    {
+      "epoch": 1.783723902854708,
+      "grad_norm": 20.25,
+      "learning_rate": 7.204912334547023e-08,
+      "loss": 0.9461,
+      "step": 20932
+    },
+    {
+      "epoch": 1.783809118023008,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.19929095075811e-08,
+      "loss": 0.4668,
+      "step": 20933
+    },
+    {
+      "epoch": 1.783894333191308,
+      "grad_norm": 19.125,
+      "learning_rate": 7.193671695746446e-08,
+      "loss": 0.745,
+      "step": 20934
+    },
+    {
+      "epoch": 1.783979548359608,
+      "grad_norm": 15.875,
+      "learning_rate": 7.188054569613592e-08,
+      "loss": 0.8223,
+      "step": 20935
+    },
+    {
+      "epoch": 1.7840647635279079,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.18243957246105e-08,
+      "loss": 0.7636,
+      "step": 20936
+    },
+    {
+      "epoch": 1.7841499786962078,
+      "grad_norm": 12.875,
+      "learning_rate": 7.176826704390336e-08,
+      "loss": 0.3382,
+      "step": 20937
+    },
+    {
+      "epoch": 1.7842351938645078,
+      "grad_norm": 19.75,
+      "learning_rate": 7.171215965502825e-08,
+      "loss": 0.3358,
+      "step": 20938
+    },
+    {
+      "epoch": 1.7843204090328078,
+      "grad_norm": 23.25,
+      "learning_rate": 7.165607355899912e-08,
+      "loss": 0.8789,
+      "step": 20939
+    },
+    {
+      "epoch": 1.7844056242011077,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.160000875682985e-08,
+      "loss": 0.7682,
+      "step": 20940
+    },
+    {
+      "epoch": 1.7844908393694077,
+      "grad_norm": 21.0,
+      "learning_rate": 7.154396524953367e-08,
+      "loss": 0.9422,
+      "step": 20941
+    },
+    {
+      "epoch": 1.7845760545377076,
+      "grad_norm": 15.375,
+      "learning_rate": 7.148794303812268e-08,
+      "loss": 0.4914,
+      "step": 20942
+    },
+    {
+      "epoch": 1.7846612697060076,
+      "grad_norm": 16.75,
+      "learning_rate": 7.143194212360968e-08,
+      "loss": 0.7679,
+      "step": 20943
+    },
+    {
+      "epoch": 1.7847464848743075,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.137596250700666e-08,
+      "loss": 0.4968,
+      "step": 20944
+    },
+    {
+      "epoch": 1.7848317000426075,
+      "grad_norm": 22.125,
+      "learning_rate": 7.132000418932528e-08,
+      "loss": 0.8453,
+      "step": 20945
+    },
+    {
+      "epoch": 1.7849169152109075,
+      "grad_norm": 19.375,
+      "learning_rate": 7.126406717157656e-08,
+      "loss": 0.7881,
+      "step": 20946
+    },
+    {
+      "epoch": 1.7850021303792074,
+      "grad_norm": 19.125,
+      "learning_rate": 7.120815145477148e-08,
+      "loss": 0.626,
+      "step": 20947
+    },
+    {
+      "epoch": 1.7850873455475074,
+      "grad_norm": 12.9375,
+      "learning_rate": 7.115225703992035e-08,
+      "loss": 0.3385,
+      "step": 20948
+    },
+    {
+      "epoch": 1.7851725607158073,
+      "grad_norm": 9.8125,
+      "learning_rate": 7.109638392803361e-08,
+      "loss": 0.2594,
+      "step": 20949
+    },
+    {
+      "epoch": 1.7852577758841073,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.104053212012074e-08,
+      "loss": 0.4388,
+      "step": 20950
+    },
+    {
+      "epoch": 1.7853429910524072,
+      "grad_norm": 17.125,
+      "learning_rate": 7.098470161719079e-08,
+      "loss": 0.7642,
+      "step": 20951
+    },
+    {
+      "epoch": 1.7854282062207072,
+      "grad_norm": 11.0,
+      "learning_rate": 7.092889242025281e-08,
+      "loss": 0.3661,
+      "step": 20952
+    },
+    {
+      "epoch": 1.7855134213890071,
+      "grad_norm": 15.125,
+      "learning_rate": 7.087310453031543e-08,
+      "loss": 0.414,
+      "step": 20953
+    },
+    {
+      "epoch": 1.785598636557307,
+      "grad_norm": 13.8125,
+      "learning_rate": 7.081733794838703e-08,
+      "loss": 0.4447,
+      "step": 20954
+    },
+    {
+      "epoch": 1.785683851725607,
+      "grad_norm": 15.125,
+      "learning_rate": 7.076159267547484e-08,
+      "loss": 0.6418,
+      "step": 20955
+    },
+    {
+      "epoch": 1.785769066893907,
+      "grad_norm": 14.6875,
+      "learning_rate": 7.07058687125864e-08,
+      "loss": 0.5832,
+      "step": 20956
+    },
+    {
+      "epoch": 1.785854282062207,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.065016606072908e-08,
+      "loss": 0.8401,
+      "step": 20957
+    },
+    {
+      "epoch": 1.785939497230507,
+      "grad_norm": 24.0,
+      "learning_rate": 7.059448472090918e-08,
+      "loss": 0.4693,
+      "step": 20958
+    },
+    {
+      "epoch": 1.786024712398807,
+      "grad_norm": 9.9375,
+      "learning_rate": 7.05388246941327e-08,
+      "loss": 0.2564,
+      "step": 20959
+    },
+    {
+      "epoch": 1.7861099275671068,
+      "grad_norm": 15.75,
+      "learning_rate": 7.048318598140564e-08,
+      "loss": 0.5743,
+      "step": 20960
+    },
+    {
+      "epoch": 1.7861951427354068,
+      "grad_norm": 19.125,
+      "learning_rate": 7.042756858373356e-08,
+      "loss": 1.1321,
+      "step": 20961
+    },
+    {
+      "epoch": 1.7862803579037068,
+      "grad_norm": 13.375,
+      "learning_rate": 7.037197250212165e-08,
+      "loss": 0.2654,
+      "step": 20962
+    },
+    {
+      "epoch": 1.7863655730720067,
+      "grad_norm": 16.75,
+      "learning_rate": 7.031639773757426e-08,
+      "loss": 0.5771,
+      "step": 20963
+    },
+    {
+      "epoch": 1.7864507882403067,
+      "grad_norm": 12.75,
+      "learning_rate": 7.026084429109584e-08,
+      "loss": 0.4544,
+      "step": 20964
+    },
+    {
+      "epoch": 1.7865360034086066,
+      "grad_norm": 15.6875,
+      "learning_rate": 7.020531216369045e-08,
+      "loss": 0.8182,
+      "step": 20965
+    },
+    {
+      "epoch": 1.7866212185769066,
+      "grad_norm": 20.5,
+      "learning_rate": 7.014980135636118e-08,
+      "loss": 0.7598,
+      "step": 20966
+    },
+    {
+      "epoch": 1.7867064337452065,
+      "grad_norm": 15.625,
+      "learning_rate": 7.009431187011168e-08,
+      "loss": 0.6849,
+      "step": 20967
+    },
+    {
+      "epoch": 1.7867916489135065,
+      "grad_norm": 13.5625,
+      "learning_rate": 7.003884370594433e-08,
+      "loss": 0.5311,
+      "step": 20968
+    },
+    {
+      "epoch": 1.7868768640818065,
+      "grad_norm": 11.25,
+      "learning_rate": 6.998339686486153e-08,
+      "loss": 0.2957,
+      "step": 20969
+    },
+    {
+      "epoch": 1.7869620792501064,
+      "grad_norm": 16.25,
+      "learning_rate": 6.992797134786525e-08,
+      "loss": 0.6898,
+      "step": 20970
+    },
+    {
+      "epoch": 1.7870472944184064,
+      "grad_norm": 10.3125,
+      "learning_rate": 6.987256715595761e-08,
+      "loss": 0.2829,
+      "step": 20971
+    },
+    {
+      "epoch": 1.7871325095867063,
+      "grad_norm": 20.0,
+      "learning_rate": 6.981718429013907e-08,
+      "loss": 0.9671,
+      "step": 20972
+    },
+    {
+      "epoch": 1.7872177247550063,
+      "grad_norm": 28.0,
+      "learning_rate": 6.976182275141075e-08,
+      "loss": 1.1709,
+      "step": 20973
+    },
+    {
+      "epoch": 1.7873029399233062,
+      "grad_norm": 14.8125,
+      "learning_rate": 6.970648254077325e-08,
+      "loss": 0.7531,
+      "step": 20974
+    },
+    {
+      "epoch": 1.7873881550916062,
+      "grad_norm": 21.0,
+      "learning_rate": 6.965116365922661e-08,
+      "loss": 0.5028,
+      "step": 20975
+    },
+    {
+      "epoch": 1.7874733702599062,
+      "grad_norm": 15.3125,
+      "learning_rate": 6.959586610777016e-08,
+      "loss": 0.8341,
+      "step": 20976
+    },
+    {
+      "epoch": 1.7875585854282061,
+      "grad_norm": 12.125,
+      "learning_rate": 6.954058988740339e-08,
+      "loss": 0.4628,
+      "step": 20977
+    },
+    {
+      "epoch": 1.787643800596506,
+      "grad_norm": 14.75,
+      "learning_rate": 6.948533499912521e-08,
+      "loss": 0.5426,
+      "step": 20978
+    },
+    {
+      "epoch": 1.787729015764806,
+      "grad_norm": 14.3125,
+      "learning_rate": 6.943010144393427e-08,
+      "loss": 0.6175,
+      "step": 20979
+    },
+    {
+      "epoch": 1.787814230933106,
+      "grad_norm": 13.5,
+      "learning_rate": 6.93748892228284e-08,
+      "loss": 0.5818,
+      "step": 20980
+    },
+    {
+      "epoch": 1.787899446101406,
+      "grad_norm": 18.625,
+      "learning_rate": 6.931969833680552e-08,
+      "loss": 0.5297,
+      "step": 20981
+    },
+    {
+      "epoch": 1.787984661269706,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.926452878686305e-08,
+      "loss": 0.6702,
+      "step": 20982
+    },
+    {
+      "epoch": 1.7880698764380059,
+      "grad_norm": 23.875,
+      "learning_rate": 6.920938057399782e-08,
+      "loss": 0.6081,
+      "step": 20983
+    },
+    {
+      "epoch": 1.7881550916063058,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.915425369920654e-08,
+      "loss": 0.3354,
+      "step": 20984
+    },
+    {
+      "epoch": 1.7882403067746058,
+      "grad_norm": 26.625,
+      "learning_rate": 6.90991481634852e-08,
+      "loss": 0.8031,
+      "step": 20985
+    },
+    {
+      "epoch": 1.7883255219429057,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.90440639678297e-08,
+      "loss": 0.3551,
+      "step": 20986
+    },
+    {
+      "epoch": 1.7884107371112057,
+      "grad_norm": 14.6875,
+      "learning_rate": 6.898900111323548e-08,
+      "loss": 0.488,
+      "step": 20987
+    },
+    {
+      "epoch": 1.7884959522795056,
+      "grad_norm": 13.1875,
+      "learning_rate": 6.893395960069784e-08,
+      "loss": 0.6641,
+      "step": 20988
+    },
+    {
+      "epoch": 1.7885811674478056,
+      "grad_norm": 18.75,
+      "learning_rate": 6.887893943121102e-08,
+      "loss": 1.0079,
+      "step": 20989
+    },
+    {
+      "epoch": 1.7886663826161056,
+      "grad_norm": 19.25,
+      "learning_rate": 6.88239406057696e-08,
+      "loss": 1.0023,
+      "step": 20990
+    },
+    {
+      "epoch": 1.7887515977844055,
+      "grad_norm": 24.625,
+      "learning_rate": 6.876896312536727e-08,
+      "loss": 0.7635,
+      "step": 20991
+    },
+    {
+      "epoch": 1.7888368129527055,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.871400699099764e-08,
+      "loss": 0.5046,
+      "step": 20992
+    },
+    {
+      "epoch": 1.7889220281210054,
+      "grad_norm": 13.25,
+      "learning_rate": 6.86590722036537e-08,
+      "loss": 0.4322,
+      "step": 20993
+    },
+    {
+      "epoch": 1.7890072432893054,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.860415876432824e-08,
+      "loss": 0.7393,
+      "step": 20994
+    },
+    {
+      "epoch": 1.7890924584576053,
+      "grad_norm": 11.8125,
+      "learning_rate": 6.854926667401368e-08,
+      "loss": 0.6324,
+      "step": 20995
+    },
+    {
+      "epoch": 1.7891776736259053,
+      "grad_norm": 15.1875,
+      "learning_rate": 6.8494395933702e-08,
+      "loss": 0.5619,
+      "step": 20996
+    },
+    {
+      "epoch": 1.7892628887942053,
+      "grad_norm": 13.625,
+      "learning_rate": 6.843954654438448e-08,
+      "loss": 0.6032,
+      "step": 20997
+    },
+    {
+      "epoch": 1.7893481039625052,
+      "grad_norm": 16.625,
+      "learning_rate": 6.83847185070527e-08,
+      "loss": 0.9303,
+      "step": 20998
+    },
+    {
+      "epoch": 1.7894333191308052,
+      "grad_norm": 16.875,
+      "learning_rate": 6.832991182269711e-08,
+      "loss": 0.6181,
+      "step": 20999
+    },
+    {
+      "epoch": 1.7895185342991051,
+      "grad_norm": 17.375,
+      "learning_rate": 6.827512649230831e-08,
+      "loss": 0.3765,
+      "step": 21000
+    },
+    {
+      "epoch": 1.789603749467405,
+      "grad_norm": 12.3125,
+      "learning_rate": 6.822036251687633e-08,
+      "loss": 0.3982,
+      "step": 21001
+    },
+    {
+      "epoch": 1.789688964635705,
+      "grad_norm": 14.75,
+      "learning_rate": 6.816561989739068e-08,
+      "loss": 0.5757,
+      "step": 21002
+    },
+    {
+      "epoch": 1.789774179804005,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.811089863484083e-08,
+      "loss": 0.3254,
+      "step": 21003
+    },
+    {
+      "epoch": 1.789859394972305,
+      "grad_norm": 16.875,
+      "learning_rate": 6.80561987302153e-08,
+      "loss": 0.6562,
+      "step": 21004
+    },
+    {
+      "epoch": 1.789944610140605,
+      "grad_norm": 16.25,
+      "learning_rate": 6.800152018450315e-08,
+      "loss": 0.5072,
+      "step": 21005
+    },
+    {
+      "epoch": 1.7900298253089049,
+      "grad_norm": 12.0,
+      "learning_rate": 6.794686299869207e-08,
+      "loss": 0.455,
+      "step": 21006
+    },
+    {
+      "epoch": 1.7901150404772048,
+      "grad_norm": 15.9375,
+      "learning_rate": 6.789222717376962e-08,
+      "loss": 0.826,
+      "step": 21007
+    },
+    {
+      "epoch": 1.7902002556455048,
+      "grad_norm": 17.25,
+      "learning_rate": 6.783761271072331e-08,
+      "loss": 0.8436,
+      "step": 21008
+    },
+    {
+      "epoch": 1.7902854708138047,
+      "grad_norm": 39.25,
+      "learning_rate": 6.778301961054028e-08,
+      "loss": 0.7903,
+      "step": 21009
+    },
+    {
+      "epoch": 1.7903706859821047,
+      "grad_norm": 12.25,
+      "learning_rate": 6.772844787420671e-08,
+      "loss": 0.4044,
+      "step": 21010
+    },
+    {
+      "epoch": 1.7904559011504046,
+      "grad_norm": 13.0,
+      "learning_rate": 6.7673897502709e-08,
+      "loss": 0.5063,
+      "step": 21011
+    },
+    {
+      "epoch": 1.7905411163187046,
+      "grad_norm": 14.0,
+      "learning_rate": 6.76193684970329e-08,
+      "loss": 0.3535,
+      "step": 21012
+    },
+    {
+      "epoch": 1.7906263314870046,
+      "grad_norm": 12.1875,
+      "learning_rate": 6.756486085816388e-08,
+      "loss": 0.2655,
+      "step": 21013
+    },
+    {
+      "epoch": 1.7907115466553045,
+      "grad_norm": 15.125,
+      "learning_rate": 6.751037458708684e-08,
+      "loss": 0.6746,
+      "step": 21014
+    },
+    {
+      "epoch": 1.7907967618236045,
+      "grad_norm": 11.625,
+      "learning_rate": 6.745590968478627e-08,
+      "loss": 0.3738,
+      "step": 21015
+    },
+    {
+      "epoch": 1.7908819769919044,
+      "grad_norm": 19.5,
+      "learning_rate": 6.740146615224666e-08,
+      "loss": 0.8112,
+      "step": 21016
+    },
+    {
+      "epoch": 1.7909671921602044,
+      "grad_norm": 28.5,
+      "learning_rate": 6.734704399045166e-08,
+      "loss": 0.7921,
+      "step": 21017
+    },
+    {
+      "epoch": 1.7910524073285043,
+      "grad_norm": 13.375,
+      "learning_rate": 6.729264320038495e-08,
+      "loss": 0.3311,
+      "step": 21018
+    },
+    {
+      "epoch": 1.7911376224968043,
+      "grad_norm": 13.375,
+      "learning_rate": 6.723826378302947e-08,
+      "loss": 0.572,
+      "step": 21019
+    },
+    {
+      "epoch": 1.7912228376651043,
+      "grad_norm": 15.875,
+      "learning_rate": 6.718390573936778e-08,
+      "loss": 0.8561,
+      "step": 21020
+    },
+    {
+      "epoch": 1.7913080528334042,
+      "grad_norm": 14.25,
+      "learning_rate": 6.712956907038257e-08,
+      "loss": 0.5327,
+      "step": 21021
+    },
+    {
+      "epoch": 1.7913932680017042,
+      "grad_norm": 25.5,
+      "learning_rate": 6.70752537770554e-08,
+      "loss": 0.6842,
+      "step": 21022
+    },
+    {
+      "epoch": 1.7914784831700041,
+      "grad_norm": 18.0,
+      "learning_rate": 6.702095986036785e-08,
+      "loss": 0.732,
+      "step": 21023
+    },
+    {
+      "epoch": 1.791563698338304,
+      "grad_norm": 15.125,
+      "learning_rate": 6.69666873213011e-08,
+      "loss": 0.4281,
+      "step": 21024
+    },
+    {
+      "epoch": 1.791648913506604,
+      "grad_norm": 16.25,
+      "learning_rate": 6.691243616083601e-08,
+      "loss": 0.5987,
+      "step": 21025
+    },
+    {
+      "epoch": 1.791734128674904,
+      "grad_norm": 17.0,
+      "learning_rate": 6.685820637995292e-08,
+      "loss": 0.8103,
+      "step": 21026
+    },
+    {
+      "epoch": 1.791819343843204,
+      "grad_norm": 24.25,
+      "learning_rate": 6.680399797963172e-08,
+      "loss": 0.4756,
+      "step": 21027
+    },
+    {
+      "epoch": 1.791904559011504,
+      "grad_norm": 15.25,
+      "learning_rate": 6.674981096085206e-08,
+      "loss": 0.7943,
+      "step": 21028
+    },
+    {
+      "epoch": 1.7919897741798039,
+      "grad_norm": 19.25,
+      "learning_rate": 6.669564532459316e-08,
+      "loss": 0.7504,
+      "step": 21029
+    },
+    {
+      "epoch": 1.7920749893481038,
+      "grad_norm": 18.875,
+      "learning_rate": 6.664150107183395e-08,
+      "loss": 0.6268,
+      "step": 21030
+    },
+    {
+      "epoch": 1.7921602045164038,
+      "grad_norm": 13.9375,
+      "learning_rate": 6.658737820355296e-08,
+      "loss": 0.6233,
+      "step": 21031
+    },
+    {
+      "epoch": 1.7922454196847037,
+      "grad_norm": 10.4375,
+      "learning_rate": 6.653327672072773e-08,
+      "loss": 0.3469,
+      "step": 21032
+    },
+    {
+      "epoch": 1.7923306348530037,
+      "grad_norm": 12.25,
+      "learning_rate": 6.647919662433625e-08,
+      "loss": 0.409,
+      "step": 21033
+    },
+    {
+      "epoch": 1.7924158500213037,
+      "grad_norm": 12.625,
+      "learning_rate": 6.64251379153559e-08,
+      "loss": 0.3213,
+      "step": 21034
+    },
+    {
+      "epoch": 1.7925010651896036,
+      "grad_norm": 12.3125,
+      "learning_rate": 6.637110059476357e-08,
+      "loss": 0.6079,
+      "step": 21035
+    },
+    {
+      "epoch": 1.7925862803579036,
+      "grad_norm": 18.875,
+      "learning_rate": 6.631708466353554e-08,
+      "loss": 0.5788,
+      "step": 21036
+    },
+    {
+      "epoch": 1.7926714955262035,
+      "grad_norm": 16.0,
+      "learning_rate": 6.626309012264812e-08,
+      "loss": 0.5411,
+      "step": 21037
+    },
+    {
+      "epoch": 1.7927567106945035,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.62091169730772e-08,
+      "loss": 0.6758,
+      "step": 21038
+    },
+    {
+      "epoch": 1.7928419258628034,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.615516521579782e-08,
+      "loss": 0.8004,
+      "step": 21039
+    },
+    {
+      "epoch": 1.7929271410311034,
+      "grad_norm": 20.5,
+      "learning_rate": 6.61012348517849e-08,
+      "loss": 1.1118,
+      "step": 21040
+    },
+    {
+      "epoch": 1.7930123561994034,
+      "grad_norm": 15.25,
+      "learning_rate": 6.604732588201323e-08,
+      "loss": 0.6572,
+      "step": 21041
+    },
+    {
+      "epoch": 1.7930975713677033,
+      "grad_norm": 15.75,
+      "learning_rate": 6.599343830745688e-08,
+      "loss": 0.4054,
+      "step": 21042
+    },
+    {
+      "epoch": 1.7931827865360033,
+      "grad_norm": 16.625,
+      "learning_rate": 6.593957212908994e-08,
+      "loss": 0.5916,
+      "step": 21043
+    },
+    {
+      "epoch": 1.7932680017043032,
+      "grad_norm": 16.875,
+      "learning_rate": 6.588572734788537e-08,
+      "loss": 0.765,
+      "step": 21044
+    },
+    {
+      "epoch": 1.7933532168726032,
+      "grad_norm": 18.0,
+      "learning_rate": 6.583190396481642e-08,
+      "loss": 0.2296,
+      "step": 21045
+    },
+    {
+      "epoch": 1.7934384320409031,
+      "grad_norm": 11.125,
+      "learning_rate": 6.577810198085579e-08,
+      "loss": 0.311,
+      "step": 21046
+    },
+    {
+      "epoch": 1.793523647209203,
+      "grad_norm": 16.0,
+      "learning_rate": 6.572432139697577e-08,
+      "loss": 0.6527,
+      "step": 21047
+    },
+    {
+      "epoch": 1.793608862377503,
+      "grad_norm": 16.625,
+      "learning_rate": 6.567056221414833e-08,
+      "loss": 0.9143,
+      "step": 21048
+    },
+    {
+      "epoch": 1.793694077545803,
+      "grad_norm": 22.625,
+      "learning_rate": 6.561682443334438e-08,
+      "loss": 0.4783,
+      "step": 21049
+    },
+    {
+      "epoch": 1.793779292714103,
+      "grad_norm": 12.375,
+      "learning_rate": 6.55631080555355e-08,
+      "loss": 0.3378,
+      "step": 21050
+    },
+    {
+      "epoch": 1.793864507882403,
+      "grad_norm": 13.875,
+      "learning_rate": 6.550941308169245e-08,
+      "loss": 0.5618,
+      "step": 21051
+    },
+    {
+      "epoch": 1.7939497230507029,
+      "grad_norm": 19.75,
+      "learning_rate": 6.545573951278528e-08,
+      "loss": 0.779,
+      "step": 21052
+    },
+    {
+      "epoch": 1.7940349382190028,
+      "grad_norm": 12.0,
+      "learning_rate": 6.540208734978403e-08,
+      "loss": 0.3878,
+      "step": 21053
+    },
+    {
+      "epoch": 1.7941201533873028,
+      "grad_norm": 15.25,
+      "learning_rate": 6.534845659365823e-08,
+      "loss": 0.7297,
+      "step": 21054
+    },
+    {
+      "epoch": 1.7942053685556028,
+      "grad_norm": 13.875,
+      "learning_rate": 6.529484724537724e-08,
+      "loss": 0.3871,
+      "step": 21055
+    },
+    {
+      "epoch": 1.7942905837239027,
+      "grad_norm": 13.625,
+      "learning_rate": 6.524125930590972e-08,
+      "loss": 0.2855,
+      "step": 21056
+    },
+    {
+      "epoch": 1.7943757988922027,
+      "grad_norm": 14.4375,
+      "learning_rate": 6.518769277622378e-08,
+      "loss": 0.4363,
+      "step": 21057
+    },
+    {
+      "epoch": 1.7944610140605026,
+      "grad_norm": 23.875,
+      "learning_rate": 6.513414765728781e-08,
+      "loss": 0.689,
+      "step": 21058
+    },
+    {
+      "epoch": 1.7945462292288026,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.508062395006911e-08,
+      "loss": 0.538,
+      "step": 21059
+    },
+    {
+      "epoch": 1.7946314443971025,
+      "grad_norm": 14.875,
+      "learning_rate": 6.502712165553524e-08,
+      "loss": 0.3263,
+      "step": 21060
+    },
+    {
+      "epoch": 1.7947166595654025,
+      "grad_norm": 19.875,
+      "learning_rate": 6.497364077465277e-08,
+      "loss": 0.508,
+      "step": 21061
+    },
+    {
+      "epoch": 1.7948018747337025,
+      "grad_norm": 18.625,
+      "learning_rate": 6.492018130838815e-08,
+      "loss": 0.7944,
+      "step": 21062
+    },
+    {
+      "epoch": 1.7948870899020024,
+      "grad_norm": 21.375,
+      "learning_rate": 6.486674325770783e-08,
+      "loss": 1.1572,
+      "step": 21063
+    },
+    {
+      "epoch": 1.7949723050703024,
+      "grad_norm": 14.75,
+      "learning_rate": 6.481332662357703e-08,
+      "loss": 0.7825,
+      "step": 21064
+    },
+    {
+      "epoch": 1.7950575202386023,
+      "grad_norm": 15.0,
+      "learning_rate": 6.475993140696121e-08,
+      "loss": 0.6307,
+      "step": 21065
+    },
+    {
+      "epoch": 1.7951427354069023,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.47065576088253e-08,
+      "loss": 0.5511,
+      "step": 21066
+    },
+    {
+      "epoch": 1.7952279505752022,
+      "grad_norm": 16.375,
+      "learning_rate": 6.465320523013366e-08,
+      "loss": 0.3282,
+      "step": 21067
+    },
+    {
+      "epoch": 1.7953131657435022,
+      "grad_norm": 14.125,
+      "learning_rate": 6.459987427185094e-08,
+      "loss": 0.4313,
+      "step": 21068
+    },
+    {
+      "epoch": 1.7953983809118021,
+      "grad_norm": 17.5,
+      "learning_rate": 6.454656473494011e-08,
+      "loss": 0.6278,
+      "step": 21069
+    },
+    {
+      "epoch": 1.795483596080102,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.4493276620365e-08,
+      "loss": 0.8363,
+      "step": 21070
+    },
+    {
+      "epoch": 1.795568811248402,
+      "grad_norm": 14.375,
+      "learning_rate": 6.444000992908871e-08,
+      "loss": 0.5306,
+      "step": 21071
+    },
+    {
+      "epoch": 1.795654026416702,
+      "grad_norm": 11.25,
+      "learning_rate": 6.438676466207341e-08,
+      "loss": 0.523,
+      "step": 21072
+    },
+    {
+      "epoch": 1.795739241585002,
+      "grad_norm": 13.5,
+      "learning_rate": 6.433354082028165e-08,
+      "loss": 0.296,
+      "step": 21073
+    },
+    {
+      "epoch": 1.795824456753302,
+      "grad_norm": 13.75,
+      "learning_rate": 6.428033840467487e-08,
+      "loss": 0.3835,
+      "step": 21074
+    },
+    {
+      "epoch": 1.795909671921602,
+      "grad_norm": 18.25,
+      "learning_rate": 6.422715741621482e-08,
+      "loss": 0.6709,
+      "step": 21075
+    },
+    {
+      "epoch": 1.7959948870899018,
+      "grad_norm": 16.625,
+      "learning_rate": 6.41739978558624e-08,
+      "loss": 0.5212,
+      "step": 21076
+    },
+    {
+      "epoch": 1.7960801022582018,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.412085972457849e-08,
+      "loss": 0.5237,
+      "step": 21077
+    },
+    {
+      "epoch": 1.7961653174265018,
+      "grad_norm": 35.75,
+      "learning_rate": 6.406774302332289e-08,
+      "loss": 0.6927,
+      "step": 21078
+    },
+    {
+      "epoch": 1.7962505325948017,
+      "grad_norm": 17.875,
+      "learning_rate": 6.401464775305579e-08,
+      "loss": 0.7802,
+      "step": 21079
+    },
+    {
+      "epoch": 1.7963357477631017,
+      "grad_norm": 21.75,
+      "learning_rate": 6.39615739147366e-08,
+      "loss": 0.7312,
+      "step": 21080
+    },
+    {
+      "epoch": 1.7964209629314016,
+      "grad_norm": 12.375,
+      "learning_rate": 6.39085215093245e-08,
+      "loss": 0.4967,
+      "step": 21081
+    },
+    {
+      "epoch": 1.7965061780997016,
+      "grad_norm": 12.375,
+      "learning_rate": 6.385549053777807e-08,
+      "loss": 0.3718,
+      "step": 21082
+    },
+    {
+      "epoch": 1.7965913932680015,
+      "grad_norm": 16.5,
+      "learning_rate": 6.380248100105554e-08,
+      "loss": 0.4894,
+      "step": 21083
+    },
+    {
+      "epoch": 1.7966766084363015,
+      "grad_norm": 26.875,
+      "learning_rate": 6.374949290011507e-08,
+      "loss": 1.0531,
+      "step": 21084
+    },
+    {
+      "epoch": 1.7967618236046015,
+      "grad_norm": 17.75,
+      "learning_rate": 6.369652623591433e-08,
+      "loss": 0.5618,
+      "step": 21085
+    },
+    {
+      "epoch": 1.7968470387729014,
+      "grad_norm": 16.125,
+      "learning_rate": 6.364358100941007e-08,
+      "loss": 0.581,
+      "step": 21086
+    },
+    {
+      "epoch": 1.7969322539412014,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.359065722155944e-08,
+      "loss": 0.475,
+      "step": 21087
+    },
+    {
+      "epoch": 1.7970174691095013,
+      "grad_norm": 18.375,
+      "learning_rate": 6.353775487331847e-08,
+      "loss": 0.5578,
+      "step": 21088
+    },
+    {
+      "epoch": 1.7971026842778013,
+      "grad_norm": 14.375,
+      "learning_rate": 6.348487396564352e-08,
+      "loss": 0.6529,
+      "step": 21089
+    },
+    {
+      "epoch": 1.7971878994461012,
+      "grad_norm": 14.4375,
+      "learning_rate": 6.343201449949001e-08,
+      "loss": 0.5464,
+      "step": 21090
+    },
+    {
+      "epoch": 1.7972731146144012,
+      "grad_norm": 16.5,
+      "learning_rate": 6.337917647581307e-08,
+      "loss": 0.632,
+      "step": 21091
+    },
+    {
+      "epoch": 1.7973583297827012,
+      "grad_norm": 19.5,
+      "learning_rate": 6.332635989556759e-08,
+      "loss": 0.9232,
+      "step": 21092
+    },
+    {
+      "epoch": 1.7974435449510011,
+      "grad_norm": 16.75,
+      "learning_rate": 6.327356475970811e-08,
+      "loss": 0.5285,
+      "step": 21093
+    },
+    {
+      "epoch": 1.797528760119301,
+      "grad_norm": 16.5,
+      "learning_rate": 6.322079106918886e-08,
+      "loss": 0.7075,
+      "step": 21094
+    },
+    {
+      "epoch": 1.797613975287601,
+      "grad_norm": 15.75,
+      "learning_rate": 6.316803882496336e-08,
+      "loss": 0.6207,
+      "step": 21095
+    },
+    {
+      "epoch": 1.797699190455901,
+      "grad_norm": 10.4375,
+      "learning_rate": 6.311530802798463e-08,
+      "loss": 0.3289,
+      "step": 21096
+    },
+    {
+      "epoch": 1.797784405624201,
+      "grad_norm": 16.375,
+      "learning_rate": 6.306259867920578e-08,
+      "loss": 0.5936,
+      "step": 21097
+    },
+    {
+      "epoch": 1.797869620792501,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.300991077957952e-08,
+      "loss": 0.4197,
+      "step": 21098
+    },
+    {
+      "epoch": 1.7979548359608009,
+      "grad_norm": 11.6875,
+      "learning_rate": 6.29572443300576e-08,
+      "loss": 0.261,
+      "step": 21099
+    },
+    {
+      "epoch": 1.7980400511291008,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.290459933159202e-08,
+      "loss": 0.7058,
+      "step": 21100
+    },
+    {
+      "epoch": 1.798125266297401,
+      "grad_norm": 16.625,
+      "learning_rate": 6.285197578513396e-08,
+      "loss": 0.6482,
+      "step": 21101
+    },
+    {
+      "epoch": 1.798210481465701,
+      "grad_norm": 15.875,
+      "learning_rate": 6.279937369163464e-08,
+      "loss": 0.6408,
+      "step": 21102
+    },
+    {
+      "epoch": 1.798295696634001,
+      "grad_norm": 16.75,
+      "learning_rate": 6.274679305204437e-08,
+      "loss": 0.5156,
+      "step": 21103
+    },
+    {
+      "epoch": 1.7983809118023009,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.269423386731338e-08,
+      "loss": 0.4094,
+      "step": 21104
+    },
+    {
+      "epoch": 1.7984661269706008,
+      "grad_norm": 14.75,
+      "learning_rate": 6.264169613839147e-08,
+      "loss": 0.6403,
+      "step": 21105
+    },
+    {
+      "epoch": 1.7985513421389008,
+      "grad_norm": 17.125,
+      "learning_rate": 6.258917986622815e-08,
+      "loss": 0.5696,
+      "step": 21106
+    },
+    {
+      "epoch": 1.7986365573072007,
+      "grad_norm": 19.0,
+      "learning_rate": 6.253668505177251e-08,
+      "loss": 0.6961,
+      "step": 21107
+    },
+    {
+      "epoch": 1.7987217724755007,
+      "grad_norm": 13.875,
+      "learning_rate": 6.248421169597286e-08,
+      "loss": 0.5665,
+      "step": 21108
+    },
+    {
+      "epoch": 1.7988069876438006,
+      "grad_norm": 17.25,
+      "learning_rate": 6.243175979977769e-08,
+      "loss": 0.6573,
+      "step": 21109
+    },
+    {
+      "epoch": 1.7988922028121006,
+      "grad_norm": 31.625,
+      "learning_rate": 6.237932936413476e-08,
+      "loss": 0.3568,
+      "step": 21110
+    },
+    {
+      "epoch": 1.7989774179804006,
+      "grad_norm": 14.375,
+      "learning_rate": 6.232692038999177e-08,
+      "loss": 0.3256,
+      "step": 21111
+    },
+    {
+      "epoch": 1.7990626331487005,
+      "grad_norm": 16.75,
+      "learning_rate": 6.227453287829575e-08,
+      "loss": 0.6166,
+      "step": 21112
+    },
+    {
+      "epoch": 1.7991478483170005,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.222216682999299e-08,
+      "loss": 0.6015,
+      "step": 21113
+    },
+    {
+      "epoch": 1.7992330634853004,
+      "grad_norm": 21.375,
+      "learning_rate": 6.216982224603e-08,
+      "loss": 0.7397,
+      "step": 21114
+    },
+    {
+      "epoch": 1.7993182786536004,
+      "grad_norm": 13.6875,
+      "learning_rate": 6.211749912735307e-08,
+      "loss": 0.5298,
+      "step": 21115
+    },
+    {
+      "epoch": 1.7994034938219003,
+      "grad_norm": 12.125,
+      "learning_rate": 6.20651974749073e-08,
+      "loss": 0.3061,
+      "step": 21116
+    },
+    {
+      "epoch": 1.7994887089902003,
+      "grad_norm": 12.875,
+      "learning_rate": 6.20129172896379e-08,
+      "loss": 0.3986,
+      "step": 21117
+    },
+    {
+      "epoch": 1.7995739241585003,
+      "grad_norm": 18.5,
+      "learning_rate": 6.196065857248968e-08,
+      "loss": 0.6202,
+      "step": 21118
+    },
+    {
+      "epoch": 1.7996591393268002,
+      "grad_norm": 16.0,
+      "learning_rate": 6.190842132440727e-08,
+      "loss": 0.5267,
+      "step": 21119
+    },
+    {
+      "epoch": 1.7997443544951002,
+      "grad_norm": 14.75,
+      "learning_rate": 6.185620554633426e-08,
+      "loss": 0.5679,
+      "step": 21120
+    },
+    {
+      "epoch": 1.7998295696634001,
+      "grad_norm": 30.375,
+      "learning_rate": 6.180401123921445e-08,
+      "loss": 1.1426,
+      "step": 21121
+    },
+    {
+      "epoch": 1.7999147848317,
+      "grad_norm": 23.5,
+      "learning_rate": 6.175183840399071e-08,
+      "loss": 0.6732,
+      "step": 21122
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 22.625,
+      "learning_rate": 6.16996870416063e-08,
+      "loss": 0.4845,
+      "step": 21123
+    },
+    {
+      "epoch": 1.8000852151683,
+      "grad_norm": 10.875,
+      "learning_rate": 6.16475571530037e-08,
+      "loss": 0.3375,
+      "step": 21124
+    },
+    {
+      "epoch": 1.8001704303366,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.159544873912446e-08,
+      "loss": 0.7536,
+      "step": 21125
+    },
+    {
+      "epoch": 1.8002556455049,
+      "grad_norm": 25.125,
+      "learning_rate": 6.154336180091036e-08,
+      "loss": 0.8698,
+      "step": 21126
+    },
+    {
+      "epoch": 1.8003408606731999,
+      "grad_norm": 22.5,
+      "learning_rate": 6.149129633930301e-08,
+      "loss": 0.3568,
+      "step": 21127
+    },
+    {
+      "epoch": 1.8004260758414998,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.143925235524331e-08,
+      "loss": 0.4095,
+      "step": 21128
+    },
+    {
+      "epoch": 1.8005112910097998,
+      "grad_norm": 24.375,
+      "learning_rate": 6.138722984967122e-08,
+      "loss": 0.8236,
+      "step": 21129
+    },
+    {
+      "epoch": 1.8005965061780997,
+      "grad_norm": 21.75,
+      "learning_rate": 6.133522882352707e-08,
+      "loss": 0.3936,
+      "step": 21130
+    },
+    {
+      "epoch": 1.8006817213463997,
+      "grad_norm": 11.5,
+      "learning_rate": 6.12832492777507e-08,
+      "loss": 0.3036,
+      "step": 21131
+    },
+    {
+      "epoch": 1.8007669365146997,
+      "grad_norm": 20.625,
+      "learning_rate": 6.123129121328148e-08,
+      "loss": 0.7993,
+      "step": 21132
+    },
+    {
+      "epoch": 1.8008521516829996,
+      "grad_norm": 17.75,
+      "learning_rate": 6.11793546310581e-08,
+      "loss": 0.6057,
+      "step": 21133
+    },
+    {
+      "epoch": 1.8009373668512996,
+      "grad_norm": 15.4375,
+      "learning_rate": 6.112743953201924e-08,
+      "loss": 0.5835,
+      "step": 21134
+    },
+    {
+      "epoch": 1.8010225820195995,
+      "grad_norm": 13.25,
+      "learning_rate": 6.107554591710294e-08,
+      "loss": 0.3504,
+      "step": 21135
+    },
+    {
+      "epoch": 1.8011077971878995,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.10236737872473e-08,
+      "loss": 0.5919,
+      "step": 21136
+    },
+    {
+      "epoch": 1.8011930123561994,
+      "grad_norm": 18.875,
+      "learning_rate": 6.097182314338951e-08,
+      "loss": 0.7687,
+      "step": 21137
+    },
+    {
+      "epoch": 1.8012782275244994,
+      "grad_norm": 21.375,
+      "learning_rate": 6.091999398646645e-08,
+      "loss": 0.8587,
+      "step": 21138
+    },
+    {
+      "epoch": 1.8013634426927994,
+      "grad_norm": 14.25,
+      "learning_rate": 6.086818631741473e-08,
+      "loss": 0.6482,
+      "step": 21139
+    },
+    {
+      "epoch": 1.8014486578610993,
+      "grad_norm": 17.25,
+      "learning_rate": 6.081640013717069e-08,
+      "loss": 0.9081,
+      "step": 21140
+    },
+    {
+      "epoch": 1.8015338730293993,
+      "grad_norm": 22.0,
+      "learning_rate": 6.076463544667024e-08,
+      "loss": 0.571,
+      "step": 21141
+    },
+    {
+      "epoch": 1.8016190881976992,
+      "grad_norm": 13.375,
+      "learning_rate": 6.071289224684848e-08,
+      "loss": 0.4691,
+      "step": 21142
+    },
+    {
+      "epoch": 1.8017043033659992,
+      "grad_norm": 13.5625,
+      "learning_rate": 6.066117053864076e-08,
+      "loss": 0.4281,
+      "step": 21143
+    },
+    {
+      "epoch": 1.8017895185342991,
+      "grad_norm": 12.75,
+      "learning_rate": 6.060947032298176e-08,
+      "loss": 0.4088,
+      "step": 21144
+    },
+    {
+      "epoch": 1.801874733702599,
+      "grad_norm": 17.75,
+      "learning_rate": 6.055779160080571e-08,
+      "loss": 0.5262,
+      "step": 21145
+    },
+    {
+      "epoch": 1.801959948870899,
+      "grad_norm": 15.625,
+      "learning_rate": 6.05061343730462e-08,
+      "loss": 0.5465,
+      "step": 21146
+    },
+    {
+      "epoch": 1.802045164039199,
+      "grad_norm": 11.9375,
+      "learning_rate": 6.045449864063704e-08,
+      "loss": 0.4282,
+      "step": 21147
+    },
+    {
+      "epoch": 1.802130379207499,
+      "grad_norm": 13.375,
+      "learning_rate": 6.040288440451126e-08,
+      "loss": 0.4883,
+      "step": 21148
+    },
+    {
+      "epoch": 1.802215594375799,
+      "grad_norm": 15.5,
+      "learning_rate": 6.035129166560183e-08,
+      "loss": 0.601,
+      "step": 21149
+    },
+    {
+      "epoch": 1.8023008095440989,
+      "grad_norm": 18.0,
+      "learning_rate": 6.029972042484053e-08,
+      "loss": 0.6285,
+      "step": 21150
+    },
+    {
+      "epoch": 1.8023860247123988,
+      "grad_norm": 15.5,
+      "learning_rate": 6.024817068315966e-08,
+      "loss": 0.3969,
+      "step": 21151
+    },
+    {
+      "epoch": 1.8024712398806988,
+      "grad_norm": 21.0,
+      "learning_rate": 6.019664244149098e-08,
+      "loss": 0.8103,
+      "step": 21152
+    },
+    {
+      "epoch": 1.8025564550489988,
+      "grad_norm": 18.75,
+      "learning_rate": 6.014513570076527e-08,
+      "loss": 0.9013,
+      "step": 21153
+    },
+    {
+      "epoch": 1.8026416702172987,
+      "grad_norm": 16.125,
+      "learning_rate": 6.009365046191345e-08,
+      "loss": 0.5859,
+      "step": 21154
+    },
+    {
+      "epoch": 1.8027268853855987,
+      "grad_norm": 20.5,
+      "learning_rate": 6.004218672586589e-08,
+      "loss": 0.6017,
+      "step": 21155
+    },
+    {
+      "epoch": 1.8028121005538986,
+      "grad_norm": 14.5,
+      "learning_rate": 5.999074449355255e-08,
+      "loss": 0.423,
+      "step": 21156
+    },
+    {
+      "epoch": 1.8028973157221986,
+      "grad_norm": 12.0,
+      "learning_rate": 5.993932376590308e-08,
+      "loss": 0.4374,
+      "step": 21157
+    },
+    {
+      "epoch": 1.8029825308904985,
+      "grad_norm": 16.0,
+      "learning_rate": 5.98879245438469e-08,
+      "loss": 0.5996,
+      "step": 21158
+    },
+    {
+      "epoch": 1.8030677460587985,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.983654682831256e-08,
+      "loss": 0.5771,
+      "step": 21159
+    },
+    {
+      "epoch": 1.8031529612270984,
+      "grad_norm": 19.625,
+      "learning_rate": 5.978519062022875e-08,
+      "loss": 0.8037,
+      "step": 21160
+    },
+    {
+      "epoch": 1.8032381763953984,
+      "grad_norm": 24.375,
+      "learning_rate": 5.973385592052322e-08,
+      "loss": 0.5462,
+      "step": 21161
+    },
+    {
+      "epoch": 1.8033233915636984,
+      "grad_norm": 11.6875,
+      "learning_rate": 5.968254273012411e-08,
+      "loss": 0.3765,
+      "step": 21162
+    },
+    {
+      "epoch": 1.8034086067319983,
+      "grad_norm": 14.75,
+      "learning_rate": 5.963125104995804e-08,
+      "loss": 0.3729,
+      "step": 21163
+    },
+    {
+      "epoch": 1.8034938219002983,
+      "grad_norm": 26.0,
+      "learning_rate": 5.9579980880952456e-08,
+      "loss": 1.0825,
+      "step": 21164
+    },
+    {
+      "epoch": 1.8035790370685982,
+      "grad_norm": 17.75,
+      "learning_rate": 5.9528732224033574e-08,
+      "loss": 0.5638,
+      "step": 21165
+    },
+    {
+      "epoch": 1.8036642522368982,
+      "grad_norm": 23.5,
+      "learning_rate": 5.947750508012773e-08,
+      "loss": 0.7519,
+      "step": 21166
+    },
+    {
+      "epoch": 1.8037494674051981,
+      "grad_norm": 22.5,
+      "learning_rate": 5.942629945016043e-08,
+      "loss": 0.7844,
+      "step": 21167
+    },
+    {
+      "epoch": 1.803834682573498,
+      "grad_norm": 18.875,
+      "learning_rate": 5.937511533505732e-08,
+      "loss": 0.7357,
+      "step": 21168
+    },
+    {
+      "epoch": 1.803919897741798,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.9323952735742956e-08,
+      "loss": 0.4265,
+      "step": 21169
+    },
+    {
+      "epoch": 1.804005112910098,
+      "grad_norm": 21.25,
+      "learning_rate": 5.9272811653142e-08,
+      "loss": 0.7067,
+      "step": 21170
+    },
+    {
+      "epoch": 1.804090328078398,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.922169208817899e-08,
+      "loss": 0.778,
+      "step": 21171
+    },
+    {
+      "epoch": 1.804175543246698,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.917059404177722e-08,
+      "loss": 0.4801,
+      "step": 21172
+    },
+    {
+      "epoch": 1.804260758414998,
+      "grad_norm": 13.0625,
+      "learning_rate": 5.911951751486025e-08,
+      "loss": 0.7916,
+      "step": 21173
+    },
+    {
+      "epoch": 1.8043459735832978,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.906846250835138e-08,
+      "loss": 0.8196,
+      "step": 21174
+    },
+    {
+      "epoch": 1.8044311887515978,
+      "grad_norm": 23.0,
+      "learning_rate": 5.9017429023172644e-08,
+      "loss": 0.7831,
+      "step": 21175
+    },
+    {
+      "epoch": 1.8045164039198978,
+      "grad_norm": 15.375,
+      "learning_rate": 5.896641706024678e-08,
+      "loss": 0.6292,
+      "step": 21176
+    },
+    {
+      "epoch": 1.8046016190881977,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.89154266204954e-08,
+      "loss": 0.7866,
+      "step": 21177
+    },
+    {
+      "epoch": 1.8046868342564977,
+      "grad_norm": 14.25,
+      "learning_rate": 5.8864457704840005e-08,
+      "loss": 0.3858,
+      "step": 21178
+    },
+    {
+      "epoch": 1.8047720494247976,
+      "grad_norm": 13.5,
+      "learning_rate": 5.881351031420165e-08,
+      "loss": 0.4504,
+      "step": 21179
+    },
+    {
+      "epoch": 1.8048572645930976,
+      "grad_norm": 18.75,
+      "learning_rate": 5.8762584449500865e-08,
+      "loss": 0.7286,
+      "step": 21180
+    },
+    {
+      "epoch": 1.8049424797613975,
+      "grad_norm": 12.5,
+      "learning_rate": 5.871168011165815e-08,
+      "loss": 0.4054,
+      "step": 21181
+    },
+    {
+      "epoch": 1.8050276949296975,
+      "grad_norm": 11.1875,
+      "learning_rate": 5.8660797301593323e-08,
+      "loss": 0.1924,
+      "step": 21182
+    },
+    {
+      "epoch": 1.8051129100979975,
+      "grad_norm": 16.25,
+      "learning_rate": 5.8609936020225935e-08,
+      "loss": 0.8235,
+      "step": 21183
+    },
+    {
+      "epoch": 1.8051981252662974,
+      "grad_norm": 39.0,
+      "learning_rate": 5.8559096268474834e-08,
+      "loss": 0.7683,
+      "step": 21184
+    },
+    {
+      "epoch": 1.8052833404345974,
+      "grad_norm": 10.0,
+      "learning_rate": 5.850827804725928e-08,
+      "loss": 0.4351,
+      "step": 21185
+    },
+    {
+      "epoch": 1.8053685556028973,
+      "grad_norm": 14.875,
+      "learning_rate": 5.8457481357497015e-08,
+      "loss": 0.3737,
+      "step": 21186
+    },
+    {
+      "epoch": 1.8054537707711973,
+      "grad_norm": 11.3125,
+      "learning_rate": 5.840670620010619e-08,
+      "loss": 0.294,
+      "step": 21187
+    },
+    {
+      "epoch": 1.8055389859394972,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.835595257600468e-08,
+      "loss": 0.5477,
+      "step": 21188
+    },
+    {
+      "epoch": 1.8056242011077972,
+      "grad_norm": 10.8125,
+      "learning_rate": 5.830522048610926e-08,
+      "loss": 0.3038,
+      "step": 21189
+    },
+    {
+      "epoch": 1.8057094162760972,
+      "grad_norm": 19.25,
+      "learning_rate": 5.8254509931336826e-08,
+      "loss": 0.3682,
+      "step": 21190
+    },
+    {
+      "epoch": 1.8057946314443971,
+      "grad_norm": 11.8125,
+      "learning_rate": 5.820382091260388e-08,
+      "loss": 0.3793,
+      "step": 21191
+    },
+    {
+      "epoch": 1.805879846612697,
+      "grad_norm": 10.9375,
+      "learning_rate": 5.81531534308262e-08,
+      "loss": 0.273,
+      "step": 21192
+    },
+    {
+      "epoch": 1.805965061780997,
+      "grad_norm": 17.125,
+      "learning_rate": 5.810250748691973e-08,
+      "loss": 0.4601,
+      "step": 21193
+    },
+    {
+      "epoch": 1.806050276949297,
+      "grad_norm": 12.9375,
+      "learning_rate": 5.805188308179929e-08,
+      "loss": 0.5265,
+      "step": 21194
+    },
+    {
+      "epoch": 1.806135492117597,
+      "grad_norm": 19.0,
+      "learning_rate": 5.8001280216379994e-08,
+      "loss": 0.4475,
+      "step": 21195
+    },
+    {
+      "epoch": 1.806220707285897,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.795069889157637e-08,
+      "loss": 0.6086,
+      "step": 21196
+    },
+    {
+      "epoch": 1.8063059224541969,
+      "grad_norm": 18.125,
+      "learning_rate": 5.7900139108302e-08,
+      "loss": 1.2038,
+      "step": 21197
+    },
+    {
+      "epoch": 1.8063911376224968,
+      "grad_norm": 22.375,
+      "learning_rate": 5.7849600867471015e-08,
+      "loss": 0.8019,
+      "step": 21198
+    },
+    {
+      "epoch": 1.8064763527907968,
+      "grad_norm": 23.125,
+      "learning_rate": 5.7799084169996575e-08,
+      "loss": 0.7975,
+      "step": 21199
+    },
+    {
+      "epoch": 1.8065615679590967,
+      "grad_norm": 11.625,
+      "learning_rate": 5.774858901679156e-08,
+      "loss": 0.465,
+      "step": 21200
+    },
+    {
+      "epoch": 1.8066467831273967,
+      "grad_norm": 19.625,
+      "learning_rate": 5.7698115408768576e-08,
+      "loss": 0.6513,
+      "step": 21201
+    },
+    {
+      "epoch": 1.8067319982956966,
+      "grad_norm": 19.125,
+      "learning_rate": 5.764766334683924e-08,
+      "loss": 0.6677,
+      "step": 21202
+    },
+    {
+      "epoch": 1.8068172134639966,
+      "grad_norm": 15.25,
+      "learning_rate": 5.759723283191576e-08,
+      "loss": 0.7323,
+      "step": 21203
+    },
+    {
+      "epoch": 1.8069024286322966,
+      "grad_norm": 17.5,
+      "learning_rate": 5.7546823864909616e-08,
+      "loss": 0.4889,
+      "step": 21204
+    },
+    {
+      "epoch": 1.8069876438005965,
+      "grad_norm": 17.75,
+      "learning_rate": 5.74964364467312e-08,
+      "loss": 0.4655,
+      "step": 21205
+    },
+    {
+      "epoch": 1.8070728589688965,
+      "grad_norm": 19.125,
+      "learning_rate": 5.7446070578291296e-08,
+      "loss": 0.9548,
+      "step": 21206
+    },
+    {
+      "epoch": 1.8071580741371964,
+      "grad_norm": 10.6875,
+      "learning_rate": 5.7395726260500025e-08,
+      "loss": 0.2968,
+      "step": 21207
+    },
+    {
+      "epoch": 1.8072432893054964,
+      "grad_norm": 13.4375,
+      "learning_rate": 5.734540349426762e-08,
+      "loss": 0.5287,
+      "step": 21208
+    },
+    {
+      "epoch": 1.8073285044737963,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.7295102280502945e-08,
+      "loss": 0.4295,
+      "step": 21209
+    },
+    {
+      "epoch": 1.8074137196420963,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.7244822620114995e-08,
+      "loss": 0.292,
+      "step": 21210
+    },
+    {
+      "epoch": 1.8074989348103963,
+      "grad_norm": 24.75,
+      "learning_rate": 5.719456451401248e-08,
+      "loss": 0.7745,
+      "step": 21211
+    },
+    {
+      "epoch": 1.8075841499786962,
+      "grad_norm": 17.125,
+      "learning_rate": 5.7144327963103566e-08,
+      "loss": 0.6955,
+      "step": 21212
+    },
+    {
+      "epoch": 1.8076693651469962,
+      "grad_norm": 12.25,
+      "learning_rate": 5.7094112968296425e-08,
+      "loss": 0.3815,
+      "step": 21213
+    },
+    {
+      "epoch": 1.8077545803152961,
+      "grad_norm": 19.0,
+      "learning_rate": 5.704391953049809e-08,
+      "loss": 0.536,
+      "step": 21214
+    },
+    {
+      "epoch": 1.807839795483596,
+      "grad_norm": 18.375,
+      "learning_rate": 5.699374765061563e-08,
+      "loss": 0.7225,
+      "step": 21215
+    },
+    {
+      "epoch": 1.807925010651896,
+      "grad_norm": 28.625,
+      "learning_rate": 5.6943597329555953e-08,
+      "loss": 0.3959,
+      "step": 21216
+    },
+    {
+      "epoch": 1.808010225820196,
+      "grad_norm": 15.5625,
+      "learning_rate": 5.689346856822528e-08,
+      "loss": 0.8343,
+      "step": 21217
+    },
+    {
+      "epoch": 1.808095440988496,
+      "grad_norm": 14.625,
+      "learning_rate": 5.684336136752941e-08,
+      "loss": 0.6874,
+      "step": 21218
+    },
+    {
+      "epoch": 1.808180656156796,
+      "grad_norm": 14.875,
+      "learning_rate": 5.679327572837359e-08,
+      "loss": 0.5423,
+      "step": 21219
+    },
+    {
+      "epoch": 1.8082658713250959,
+      "grad_norm": 16.75,
+      "learning_rate": 5.6743211651663076e-08,
+      "loss": 0.5814,
+      "step": 21220
+    },
+    {
+      "epoch": 1.8083510864933958,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.669316913830297e-08,
+      "loss": 0.5748,
+      "step": 21221
+    },
+    {
+      "epoch": 1.8084363016616958,
+      "grad_norm": 15.5,
+      "learning_rate": 5.6643148189197e-08,
+      "loss": 0.6341,
+      "step": 21222
+    },
+    {
+      "epoch": 1.8085215168299957,
+      "grad_norm": 19.375,
+      "learning_rate": 5.659314880524944e-08,
+      "loss": 0.3313,
+      "step": 21223
+    },
+    {
+      "epoch": 1.8086067319982957,
+      "grad_norm": 11.125,
+      "learning_rate": 5.654317098736359e-08,
+      "loss": 0.2741,
+      "step": 21224
+    },
+    {
+      "epoch": 1.8086919471665956,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.649321473644304e-08,
+      "loss": 0.6152,
+      "step": 21225
+    },
+    {
+      "epoch": 1.8087771623348956,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.6443280053390255e-08,
+      "loss": 0.411,
+      "step": 21226
+    },
+    {
+      "epoch": 1.8088623775031956,
+      "grad_norm": 15.6875,
+      "learning_rate": 5.639336693910744e-08,
+      "loss": 0.6349,
+      "step": 21227
+    },
+    {
+      "epoch": 1.8089475926714955,
+      "grad_norm": 11.75,
+      "learning_rate": 5.634347539449664e-08,
+      "loss": 0.3545,
+      "step": 21228
+    },
+    {
+      "epoch": 1.8090328078397955,
+      "grad_norm": 14.625,
+      "learning_rate": 5.629360542045978e-08,
+      "loss": 0.4897,
+      "step": 21229
+    },
+    {
+      "epoch": 1.8091180230080954,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.62437570178978e-08,
+      "loss": 0.7002,
+      "step": 21230
+    },
+    {
+      "epoch": 1.8092032381763954,
+      "grad_norm": 16.875,
+      "learning_rate": 5.619393018771152e-08,
+      "loss": 0.2501,
+      "step": 21231
+    },
+    {
+      "epoch": 1.8092884533446953,
+      "grad_norm": 11.25,
+      "learning_rate": 5.614412493080132e-08,
+      "loss": 0.4144,
+      "step": 21232
+    },
+    {
+      "epoch": 1.8093736685129953,
+      "grad_norm": 16.625,
+      "learning_rate": 5.6094341248067454e-08,
+      "loss": 0.6416,
+      "step": 21233
+    },
+    {
+      "epoch": 1.8094588836812953,
+      "grad_norm": 19.0,
+      "learning_rate": 5.604457914040934e-08,
+      "loss": 0.38,
+      "step": 21234
+    },
+    {
+      "epoch": 1.8095440988495952,
+      "grad_norm": 13.0625,
+      "learning_rate": 5.599483860872626e-08,
+      "loss": 0.7383,
+      "step": 21235
+    },
+    {
+      "epoch": 1.8096293140178952,
+      "grad_norm": 11.0,
+      "learning_rate": 5.5945119653916943e-08,
+      "loss": 0.3988,
+      "step": 21236
+    },
+    {
+      "epoch": 1.8097145291861951,
+      "grad_norm": 25.875,
+      "learning_rate": 5.589542227688025e-08,
+      "loss": 0.7243,
+      "step": 21237
+    },
+    {
+      "epoch": 1.809799744354495,
+      "grad_norm": 18.375,
+      "learning_rate": 5.5845746478514066e-08,
+      "loss": 0.7721,
+      "step": 21238
+    },
+    {
+      "epoch": 1.809884959522795,
+      "grad_norm": 15.0,
+      "learning_rate": 5.579609225971588e-08,
+      "loss": 0.5977,
+      "step": 21239
+    },
+    {
+      "epoch": 1.809970174691095,
+      "grad_norm": 13.75,
+      "learning_rate": 5.574645962138331e-08,
+      "loss": 0.5725,
+      "step": 21240
+    },
+    {
+      "epoch": 1.810055389859395,
+      "grad_norm": 17.125,
+      "learning_rate": 5.569684856441312e-08,
+      "loss": 0.8047,
+      "step": 21241
+    },
+    {
+      "epoch": 1.810140605027695,
+      "grad_norm": 17.375,
+      "learning_rate": 5.56472590897017e-08,
+      "loss": 0.766,
+      "step": 21242
+    },
+    {
+      "epoch": 1.8102258201959949,
+      "grad_norm": 10.5,
+      "learning_rate": 5.559769119814567e-08,
+      "loss": 0.4415,
+      "step": 21243
+    },
+    {
+      "epoch": 1.8103110353642948,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.554814489064003e-08,
+      "loss": 0.4318,
+      "step": 21244
+    },
+    {
+      "epoch": 1.8103962505325948,
+      "grad_norm": 14.25,
+      "learning_rate": 5.549862016808072e-08,
+      "loss": 0.4174,
+      "step": 21245
+    },
+    {
+      "epoch": 1.8104814657008947,
+      "grad_norm": 13.5,
+      "learning_rate": 5.544911703136244e-08,
+      "loss": 0.4747,
+      "step": 21246
+    },
+    {
+      "epoch": 1.8105666808691947,
+      "grad_norm": 20.25,
+      "learning_rate": 5.539963548137989e-08,
+      "loss": 0.6882,
+      "step": 21247
+    },
+    {
+      "epoch": 1.8106518960374947,
+      "grad_norm": 21.125,
+      "learning_rate": 5.535017551902708e-08,
+      "loss": 0.5219,
+      "step": 21248
+    },
+    {
+      "epoch": 1.8107371112057946,
+      "grad_norm": 18.25,
+      "learning_rate": 5.5300737145198016e-08,
+      "loss": 0.6248,
+      "step": 21249
+    },
+    {
+      "epoch": 1.8108223263740946,
+      "grad_norm": 19.625,
+      "learning_rate": 5.5251320360786007e-08,
+      "loss": 0.847,
+      "step": 21250
+    },
+    {
+      "epoch": 1.8109075415423945,
+      "grad_norm": 13.5,
+      "learning_rate": 5.52019251666841e-08,
+      "loss": 0.5029,
+      "step": 21251
+    },
+    {
+      "epoch": 1.8109927567106945,
+      "grad_norm": 12.6875,
+      "learning_rate": 5.515255156378477e-08,
+      "loss": 0.5523,
+      "step": 21252
+    },
+    {
+      "epoch": 1.8110779718789944,
+      "grad_norm": 16.25,
+      "learning_rate": 5.510319955298035e-08,
+      "loss": 0.5491,
+      "step": 21253
+    },
+    {
+      "epoch": 1.8111631870472944,
+      "grad_norm": 18.25,
+      "learning_rate": 5.505386913516264e-08,
+      "loss": 0.9061,
+      "step": 21254
+    },
+    {
+      "epoch": 1.8112484022155944,
+      "grad_norm": 16.75,
+      "learning_rate": 5.5004560311223285e-08,
+      "loss": 0.5282,
+      "step": 21255
+    },
+    {
+      "epoch": 1.8113336173838943,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.495527308205295e-08,
+      "loss": 0.5139,
+      "step": 21256
+    },
+    {
+      "epoch": 1.8114188325521943,
+      "grad_norm": 13.875,
+      "learning_rate": 5.4906007448542606e-08,
+      "loss": 0.6256,
+      "step": 21257
+    },
+    {
+      "epoch": 1.8115040477204942,
+      "grad_norm": 22.125,
+      "learning_rate": 5.485676341158264e-08,
+      "loss": 1.0493,
+      "step": 21258
+    },
+    {
+      "epoch": 1.8115892628887942,
+      "grad_norm": 20.625,
+      "learning_rate": 5.4807540972062616e-08,
+      "loss": 0.7657,
+      "step": 21259
+    },
+    {
+      "epoch": 1.8116744780570941,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.475834013087239e-08,
+      "loss": 0.4227,
+      "step": 21260
+    },
+    {
+      "epoch": 1.811759693225394,
+      "grad_norm": 15.0,
+      "learning_rate": 5.470916088890055e-08,
+      "loss": 0.5205,
+      "step": 21261
+    },
+    {
+      "epoch": 1.811844908393694,
+      "grad_norm": 14.75,
+      "learning_rate": 5.4660003247036245e-08,
+      "loss": 0.7306,
+      "step": 21262
+    },
+    {
+      "epoch": 1.811930123561994,
+      "grad_norm": 19.375,
+      "learning_rate": 5.461086720616765e-08,
+      "loss": 0.7887,
+      "step": 21263
+    },
+    {
+      "epoch": 1.812015338730294,
+      "grad_norm": 14.875,
+      "learning_rate": 5.456175276718295e-08,
+      "loss": 0.6164,
+      "step": 21264
+    },
+    {
+      "epoch": 1.812100553898594,
+      "grad_norm": 17.25,
+      "learning_rate": 5.451265993096935e-08,
+      "loss": 0.9021,
+      "step": 21265
+    },
+    {
+      "epoch": 1.8121857690668939,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.44635886984142e-08,
+      "loss": 0.629,
+      "step": 21266
+    },
+    {
+      "epoch": 1.8122709842351938,
+      "grad_norm": 9.9375,
+      "learning_rate": 5.4414539070404144e-08,
+      "loss": 0.3214,
+      "step": 21267
+    },
+    {
+      "epoch": 1.8123561994034938,
+      "grad_norm": 15.0,
+      "learning_rate": 5.4365511047825695e-08,
+      "loss": 0.5054,
+      "step": 21268
+    },
+    {
+      "epoch": 1.8124414145717938,
+      "grad_norm": 18.25,
+      "learning_rate": 5.4316504631564544e-08,
+      "loss": 0.7605,
+      "step": 21269
+    },
+    {
+      "epoch": 1.8125266297400937,
+      "grad_norm": 16.125,
+      "learning_rate": 5.4267519822506634e-08,
+      "loss": 0.4561,
+      "step": 21270
+    },
+    {
+      "epoch": 1.8126118449083937,
+      "grad_norm": 17.875,
+      "learning_rate": 5.421855662153711e-08,
+      "loss": 0.7088,
+      "step": 21271
+    },
+    {
+      "epoch": 1.8126970600766936,
+      "grad_norm": 13.625,
+      "learning_rate": 5.416961502954066e-08,
+      "loss": 0.5098,
+      "step": 21272
+    },
+    {
+      "epoch": 1.8127822752449936,
+      "grad_norm": 13.625,
+      "learning_rate": 5.412069504740172e-08,
+      "loss": 0.4754,
+      "step": 21273
+    },
+    {
+      "epoch": 1.8128674904132935,
+      "grad_norm": 17.625,
+      "learning_rate": 5.40717966760046e-08,
+      "loss": 0.7427,
+      "step": 21274
+    },
+    {
+      "epoch": 1.8129527055815935,
+      "grad_norm": 13.25,
+      "learning_rate": 5.402291991623246e-08,
+      "loss": 0.4288,
+      "step": 21275
+    },
+    {
+      "epoch": 1.8130379207498935,
+      "grad_norm": 20.0,
+      "learning_rate": 5.3974064768968907e-08,
+      "loss": 0.8285,
+      "step": 21276
+    },
+    {
+      "epoch": 1.8131231359181934,
+      "grad_norm": 17.375,
+      "learning_rate": 5.392523123509671e-08,
+      "loss": 0.4472,
+      "step": 21277
+    },
+    {
+      "epoch": 1.8132083510864934,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.387641931549822e-08,
+      "loss": 0.4207,
+      "step": 21278
+    },
+    {
+      "epoch": 1.8132935662547933,
+      "grad_norm": 20.5,
+      "learning_rate": 5.382762901105565e-08,
+      "loss": 0.8046,
+      "step": 21279
+    },
+    {
+      "epoch": 1.8133787814230933,
+      "grad_norm": 14.375,
+      "learning_rate": 5.377886032265078e-08,
+      "loss": 0.5426,
+      "step": 21280
+    },
+    {
+      "epoch": 1.8134639965913932,
+      "grad_norm": 18.75,
+      "learning_rate": 5.373011325116487e-08,
+      "loss": 0.9003,
+      "step": 21281
+    },
+    {
+      "epoch": 1.8135492117596932,
+      "grad_norm": 12.4375,
+      "learning_rate": 5.3681387797478724e-08,
+      "loss": 0.5048,
+      "step": 21282
+    },
+    {
+      "epoch": 1.8136344269279931,
+      "grad_norm": 15.375,
+      "learning_rate": 5.363268396247276e-08,
+      "loss": 0.6907,
+      "step": 21283
+    },
+    {
+      "epoch": 1.813719642096293,
+      "grad_norm": 20.5,
+      "learning_rate": 5.3584001747027105e-08,
+      "loss": 0.6764,
+      "step": 21284
+    },
+    {
+      "epoch": 1.813804857264593,
+      "grad_norm": 20.0,
+      "learning_rate": 5.353534115202203e-08,
+      "loss": 0.6199,
+      "step": 21285
+    },
+    {
+      "epoch": 1.813890072432893,
+      "grad_norm": 16.375,
+      "learning_rate": 5.348670217833612e-08,
+      "loss": 0.7251,
+      "step": 21286
+    },
+    {
+      "epoch": 1.813975287601193,
+      "grad_norm": 15.9375,
+      "learning_rate": 5.343808482684884e-08,
+      "loss": 0.8319,
+      "step": 21287
+    },
+    {
+      "epoch": 1.814060502769493,
+      "grad_norm": 18.125,
+      "learning_rate": 5.33894890984385e-08,
+      "loss": 0.5676,
+      "step": 21288
+    },
+    {
+      "epoch": 1.814145717937793,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.3340914993983555e-08,
+      "loss": 0.6181,
+      "step": 21289
+    },
+    {
+      "epoch": 1.8142309331060928,
+      "grad_norm": 20.5,
+      "learning_rate": 5.329236251436165e-08,
+      "loss": 1.2265,
+      "step": 21290
+    },
+    {
+      "epoch": 1.8143161482743928,
+      "grad_norm": 13.125,
+      "learning_rate": 5.324383166044986e-08,
+      "loss": 0.5486,
+      "step": 21291
+    },
+    {
+      "epoch": 1.8144013634426928,
+      "grad_norm": 16.25,
+      "learning_rate": 5.319532243312553e-08,
+      "loss": 0.7016,
+      "step": 21292
+    },
+    {
+      "epoch": 1.8144865786109927,
+      "grad_norm": 17.25,
+      "learning_rate": 5.3146834833265326e-08,
+      "loss": 0.8644,
+      "step": 21293
+    },
+    {
+      "epoch": 1.8145717937792927,
+      "grad_norm": 16.75,
+      "learning_rate": 5.3098368861745355e-08,
+      "loss": 0.4157,
+      "step": 21294
+    },
+    {
+      "epoch": 1.8146570089475926,
+      "grad_norm": 20.375,
+      "learning_rate": 5.30499245194413e-08,
+      "loss": 0.8387,
+      "step": 21295
+    },
+    {
+      "epoch": 1.8147422241158926,
+      "grad_norm": 16.0,
+      "learning_rate": 5.300150180722871e-08,
+      "loss": 0.864,
+      "step": 21296
+    },
+    {
+      "epoch": 1.8148274392841925,
+      "grad_norm": 17.375,
+      "learning_rate": 5.2953100725982724e-08,
+      "loss": 0.6088,
+      "step": 21297
+    },
+    {
+      "epoch": 1.8149126544524925,
+      "grad_norm": 21.25,
+      "learning_rate": 5.2904721276578336e-08,
+      "loss": 1.0939,
+      "step": 21298
+    },
+    {
+      "epoch": 1.8149978696207925,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.285636345988887e-08,
+      "loss": 0.6003,
+      "step": 21299
+    },
+    {
+      "epoch": 1.8150830847890924,
+      "grad_norm": 18.75,
+      "learning_rate": 5.280802727678891e-08,
+      "loss": 0.7013,
+      "step": 21300
+    },
+    {
+      "epoch": 1.8151682999573924,
+      "grad_norm": 11.9375,
+      "learning_rate": 5.275971272815164e-08,
+      "loss": 0.2607,
+      "step": 21301
+    },
+    {
+      "epoch": 1.8152535151256923,
+      "grad_norm": 15.25,
+      "learning_rate": 5.271141981485053e-08,
+      "loss": 0.589,
+      "step": 21302
+    },
+    {
+      "epoch": 1.8153387302939923,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.2663148537757945e-08,
+      "loss": 0.586,
+      "step": 21303
+    },
+    {
+      "epoch": 1.8154239454622922,
+      "grad_norm": 16.125,
+      "learning_rate": 5.26148988977461e-08,
+      "loss": 0.7015,
+      "step": 21304
+    },
+    {
+      "epoch": 1.8155091606305922,
+      "grad_norm": 15.0625,
+      "learning_rate": 5.2566670895687205e-08,
+      "loss": 0.5755,
+      "step": 21305
+    },
+    {
+      "epoch": 1.8155943757988922,
+      "grad_norm": 14.4375,
+      "learning_rate": 5.2518464532452795e-08,
+      "loss": 0.6604,
+      "step": 21306
+    },
+    {
+      "epoch": 1.8156795909671921,
+      "grad_norm": 13.1875,
+      "learning_rate": 5.2470279808913975e-08,
+      "loss": 0.5393,
+      "step": 21307
+    },
+    {
+      "epoch": 1.815764806135492,
+      "grad_norm": 17.125,
+      "learning_rate": 5.2422116725941165e-08,
+      "loss": 0.7454,
+      "step": 21308
+    },
+    {
+      "epoch": 1.815850021303792,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.237397528440491e-08,
+      "loss": 0.7634,
+      "step": 21309
+    },
+    {
+      "epoch": 1.815935236472092,
+      "grad_norm": 16.625,
+      "learning_rate": 5.2325855485175355e-08,
+      "loss": 0.5728,
+      "step": 21310
+    },
+    {
+      "epoch": 1.816020451640392,
+      "grad_norm": 12.5625,
+      "learning_rate": 5.2277757329122075e-08,
+      "loss": 0.5788,
+      "step": 21311
+    },
+    {
+      "epoch": 1.816105666808692,
+      "grad_norm": 19.25,
+      "learning_rate": 5.2229680817113966e-08,
+      "loss": 0.7389,
+      "step": 21312
+    },
+    {
+      "epoch": 1.8161908819769919,
+      "grad_norm": 11.8125,
+      "learning_rate": 5.218162595001991e-08,
+      "loss": 0.252,
+      "step": 21313
+    },
+    {
+      "epoch": 1.8162760971452918,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.2133592728708524e-08,
+      "loss": 0.4949,
+      "step": 21314
+    },
+    {
+      "epoch": 1.8163613123135918,
+      "grad_norm": 40.25,
+      "learning_rate": 5.2085581154047714e-08,
+      "loss": 0.6069,
+      "step": 21315
+    },
+    {
+      "epoch": 1.8164465274818917,
+      "grad_norm": 12.625,
+      "learning_rate": 5.203759122690485e-08,
+      "loss": 0.5695,
+      "step": 21316
+    },
+    {
+      "epoch": 1.8165317426501917,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.198962294814744e-08,
+      "loss": 0.7667,
+      "step": 21317
+    },
+    {
+      "epoch": 1.8166169578184916,
+      "grad_norm": 17.0,
+      "learning_rate": 5.194167631864227e-08,
+      "loss": 0.7167,
+      "step": 21318
+    },
+    {
+      "epoch": 1.8167021729867916,
+      "grad_norm": 19.125,
+      "learning_rate": 5.189375133925575e-08,
+      "loss": 0.8102,
+      "step": 21319
+    },
+    {
+      "epoch": 1.8167873881550916,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.184584801085385e-08,
+      "loss": 0.4576,
+      "step": 21320
+    },
+    {
+      "epoch": 1.8168726033233915,
+      "grad_norm": 14.3125,
+      "learning_rate": 5.17979663343024e-08,
+      "loss": 0.5733,
+      "step": 21321
+    },
+    {
+      "epoch": 1.8169578184916915,
+      "grad_norm": 19.125,
+      "learning_rate": 5.17501063104664e-08,
+      "loss": 0.8762,
+      "step": 21322
+    },
+    {
+      "epoch": 1.8170430336599914,
+      "grad_norm": 14.875,
+      "learning_rate": 5.170226794021127e-08,
+      "loss": 0.5276,
+      "step": 21323
+    },
+    {
+      "epoch": 1.8171282488282914,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.165445122440091e-08,
+      "loss": 0.2523,
+      "step": 21324
+    },
+    {
+      "epoch": 1.8172134639965913,
+      "grad_norm": 15.5625,
+      "learning_rate": 5.160665616389962e-08,
+      "loss": 0.75,
+      "step": 21325
+    },
+    {
+      "epoch": 1.8172986791648913,
+      "grad_norm": 17.125,
+      "learning_rate": 5.155888275957116e-08,
+      "loss": 0.3432,
+      "step": 21326
+    },
+    {
+      "epoch": 1.8173838943331915,
+      "grad_norm": 14.0625,
+      "learning_rate": 5.1511131012278863e-08,
+      "loss": 0.6225,
+      "step": 21327
+    },
+    {
+      "epoch": 1.8174691095014914,
+      "grad_norm": 12.0625,
+      "learning_rate": 5.146340092288565e-08,
+      "loss": 0.4061,
+      "step": 21328
+    },
+    {
+      "epoch": 1.8175543246697914,
+      "grad_norm": 31.875,
+      "learning_rate": 5.1415692492253895e-08,
+      "loss": 0.6328,
+      "step": 21329
+    },
+    {
+      "epoch": 1.8176395398380913,
+      "grad_norm": 12.75,
+      "learning_rate": 5.1368005721245814e-08,
+      "loss": 0.3743,
+      "step": 21330
+    },
+    {
+      "epoch": 1.8177247550063913,
+      "grad_norm": 19.375,
+      "learning_rate": 5.1320340610723356e-08,
+      "loss": 0.8362,
+      "step": 21331
+    },
+    {
+      "epoch": 1.8178099701746913,
+      "grad_norm": 16.75,
+      "learning_rate": 5.1272697161547644e-08,
+      "loss": 0.7192,
+      "step": 21332
+    },
+    {
+      "epoch": 1.8178951853429912,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.122507537457966e-08,
+      "loss": 0.4297,
+      "step": 21333
+    },
+    {
+      "epoch": 1.8179804005112912,
+      "grad_norm": 17.125,
+      "learning_rate": 5.1177475250679955e-08,
+      "loss": 0.4724,
+      "step": 21334
+    },
+    {
+      "epoch": 1.8180656156795911,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.112989679070882e-08,
+      "loss": 0.3734,
+      "step": 21335
+    },
+    {
+      "epoch": 1.818150830847891,
+      "grad_norm": 11.0,
+      "learning_rate": 5.1082339995526114e-08,
+      "loss": 0.2715,
+      "step": 21336
+    },
+    {
+      "epoch": 1.818236046016191,
+      "grad_norm": 19.5,
+      "learning_rate": 5.1034804865991024e-08,
+      "loss": 0.5687,
+      "step": 21337
+    },
+    {
+      "epoch": 1.818321261184491,
+      "grad_norm": 13.875,
+      "learning_rate": 5.098729140296271e-08,
+      "loss": 0.5952,
+      "step": 21338
+    },
+    {
+      "epoch": 1.818406476352791,
+      "grad_norm": 13.6875,
+      "learning_rate": 5.09397996072998e-08,
+      "loss": 0.593,
+      "step": 21339
+    },
+    {
+      "epoch": 1.818491691521091,
+      "grad_norm": 14.0,
+      "learning_rate": 5.089232947986036e-08,
+      "loss": 0.4833,
+      "step": 21340
+    },
+    {
+      "epoch": 1.8185769066893909,
+      "grad_norm": 14.25,
+      "learning_rate": 5.084488102150245e-08,
+      "loss": 0.3318,
+      "step": 21341
+    },
+    {
+      "epoch": 1.8186621218576908,
+      "grad_norm": 22.25,
+      "learning_rate": 5.07974542330833e-08,
+      "loss": 0.8225,
+      "step": 21342
+    },
+    {
+      "epoch": 1.8187473370259908,
+      "grad_norm": 17.5,
+      "learning_rate": 5.075004911546e-08,
+      "loss": 0.7787,
+      "step": 21343
+    },
+    {
+      "epoch": 1.8188325521942907,
+      "grad_norm": 16.0,
+      "learning_rate": 5.0702665669489365e-08,
+      "loss": 0.553,
+      "step": 21344
+    },
+    {
+      "epoch": 1.8189177673625907,
+      "grad_norm": 33.5,
+      "learning_rate": 5.065530389602738e-08,
+      "loss": 1.2756,
+      "step": 21345
+    },
+    {
+      "epoch": 1.8190029825308907,
+      "grad_norm": 25.0,
+      "learning_rate": 5.060796379593016e-08,
+      "loss": 0.6956,
+      "step": 21346
+    },
+    {
+      "epoch": 1.8190881976991906,
+      "grad_norm": 16.125,
+      "learning_rate": 5.0560645370053145e-08,
+      "loss": 0.6878,
+      "step": 21347
+    },
+    {
+      "epoch": 1.8191734128674906,
+      "grad_norm": 18.375,
+      "learning_rate": 5.051334861925134e-08,
+      "loss": 0.6867,
+      "step": 21348
+    },
+    {
+      "epoch": 1.8192586280357905,
+      "grad_norm": 16.5,
+      "learning_rate": 5.046607354437974e-08,
+      "loss": 0.6802,
+      "step": 21349
+    },
+    {
+      "epoch": 1.8193438432040905,
+      "grad_norm": 21.0,
+      "learning_rate": 5.0418820146292124e-08,
+      "loss": 0.6712,
+      "step": 21350
+    },
+    {
+      "epoch": 1.8194290583723904,
+      "grad_norm": 16.625,
+      "learning_rate": 5.037158842584281e-08,
+      "loss": 0.7883,
+      "step": 21351
+    },
+    {
+      "epoch": 1.8195142735406904,
+      "grad_norm": 10.3125,
+      "learning_rate": 5.032437838388513e-08,
+      "loss": 0.2326,
+      "step": 21352
+    },
+    {
+      "epoch": 1.8195994887089904,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.0277190021272434e-08,
+      "loss": 0.397,
+      "step": 21353
+    },
+    {
+      "epoch": 1.8196847038772903,
+      "grad_norm": 20.375,
+      "learning_rate": 5.0230023338857236e-08,
+      "loss": 0.4478,
+      "step": 21354
+    },
+    {
+      "epoch": 1.8197699190455903,
+      "grad_norm": 27.5,
+      "learning_rate": 5.018287833749205e-08,
+      "loss": 0.6683,
+      "step": 21355
+    },
+    {
+      "epoch": 1.8198551342138902,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.01357550180287e-08,
+      "loss": 0.6098,
+      "step": 21356
+    },
+    {
+      "epoch": 1.8199403493821902,
+      "grad_norm": 23.125,
+      "learning_rate": 5.008865338131885e-08,
+      "loss": 0.8253,
+      "step": 21357
+    },
+    {
+      "epoch": 1.8200255645504901,
+      "grad_norm": 21.375,
+      "learning_rate": 5.004157342821364e-08,
+      "loss": 0.9422,
+      "step": 21358
+    },
+    {
+      "epoch": 1.82011077971879,
+      "grad_norm": 16.375,
+      "learning_rate": 4.9994515159563775e-08,
+      "loss": 0.774,
+      "step": 21359
+    },
+    {
+      "epoch": 1.82019599488709,
+      "grad_norm": 20.125,
+      "learning_rate": 4.9947478576219816e-08,
+      "loss": 1.0129,
+      "step": 21360
+    },
+    {
+      "epoch": 1.82028121005539,
+      "grad_norm": 14.875,
+      "learning_rate": 4.990046367903165e-08,
+      "loss": 0.5755,
+      "step": 21361
+    },
+    {
+      "epoch": 1.82036642522369,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.9853470468848864e-08,
+      "loss": 0.5216,
+      "step": 21362
+    },
+    {
+      "epoch": 1.82045164039199,
+      "grad_norm": 15.875,
+      "learning_rate": 4.980649894652079e-08,
+      "loss": 0.5713,
+      "step": 21363
+    },
+    {
+      "epoch": 1.8205368555602899,
+      "grad_norm": 13.75,
+      "learning_rate": 4.975954911289604e-08,
+      "loss": 0.6019,
+      "step": 21364
+    },
+    {
+      "epoch": 1.8206220707285898,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.971262096882326e-08,
+      "loss": 0.5116,
+      "step": 21365
+    },
+    {
+      "epoch": 1.8207072858968898,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.966571451515037e-08,
+      "loss": 0.8255,
+      "step": 21366
+    },
+    {
+      "epoch": 1.8207925010651898,
+      "grad_norm": 15.5,
+      "learning_rate": 4.961882975272503e-08,
+      "loss": 0.6264,
+      "step": 21367
+    },
+    {
+      "epoch": 1.8208777162334897,
+      "grad_norm": 13.625,
+      "learning_rate": 4.9571966682394476e-08,
+      "loss": 0.333,
+      "step": 21368
+    },
+    {
+      "epoch": 1.8209629314017897,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.9525125305005664e-08,
+      "loss": 0.6245,
+      "step": 21369
+    },
+    {
+      "epoch": 1.8210481465700896,
+      "grad_norm": 30.25,
+      "learning_rate": 4.947830562140501e-08,
+      "loss": 0.9868,
+      "step": 21370
+    },
+    {
+      "epoch": 1.8211333617383896,
+      "grad_norm": 12.3125,
+      "learning_rate": 4.9431507632438785e-08,
+      "loss": 0.39,
+      "step": 21371
+    },
+    {
+      "epoch": 1.8212185769066895,
+      "grad_norm": 23.25,
+      "learning_rate": 4.938473133895214e-08,
+      "loss": 1.0266,
+      "step": 21372
+    },
+    {
+      "epoch": 1.8213037920749895,
+      "grad_norm": 11.9375,
+      "learning_rate": 4.933797674179092e-08,
+      "loss": 0.4355,
+      "step": 21373
+    },
+    {
+      "epoch": 1.8213890072432894,
+      "grad_norm": 13.25,
+      "learning_rate": 4.929124384179973e-08,
+      "loss": 0.3653,
+      "step": 21374
+    },
+    {
+      "epoch": 1.8214742224115894,
+      "grad_norm": 12.3125,
+      "learning_rate": 4.9244532639823176e-08,
+      "loss": 0.3776,
+      "step": 21375
+    },
+    {
+      "epoch": 1.8215594375798894,
+      "grad_norm": 15.5,
+      "learning_rate": 4.919784313670531e-08,
+      "loss": 0.8436,
+      "step": 21376
+    },
+    {
+      "epoch": 1.8216446527481893,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.915117533329003e-08,
+      "loss": 0.3679,
+      "step": 21377
+    },
+    {
+      "epoch": 1.8217298679164893,
+      "grad_norm": 16.75,
+      "learning_rate": 4.910452923042056e-08,
+      "loss": 0.6665,
+      "step": 21378
+    },
+    {
+      "epoch": 1.8218150830847892,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.9057904828939825e-08,
+      "loss": 0.5479,
+      "step": 21379
+    },
+    {
+      "epoch": 1.8219002982530892,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.901130212969049e-08,
+      "loss": 0.6039,
+      "step": 21380
+    },
+    {
+      "epoch": 1.8219855134213891,
+      "grad_norm": 14.125,
+      "learning_rate": 4.8964721133514385e-08,
+      "loss": 0.3785,
+      "step": 21381
+    },
+    {
+      "epoch": 1.822070728589689,
+      "grad_norm": 17.0,
+      "learning_rate": 4.891816184125361e-08,
+      "loss": 1.0527,
+      "step": 21382
+    },
+    {
+      "epoch": 1.822155943757989,
+      "grad_norm": 14.375,
+      "learning_rate": 4.8871624253749586e-08,
+      "loss": 0.4976,
+      "step": 21383
+    },
+    {
+      "epoch": 1.822241158926289,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.8825108371842875e-08,
+      "loss": 0.7907,
+      "step": 21384
+    },
+    {
+      "epoch": 1.822326374094589,
+      "grad_norm": 27.75,
+      "learning_rate": 4.877861419637434e-08,
+      "loss": 0.9878,
+      "step": 21385
+    },
+    {
+      "epoch": 1.822411589262889,
+      "grad_norm": 18.75,
+      "learning_rate": 4.8732141728184277e-08,
+      "loss": 0.6157,
+      "step": 21386
+    },
+    {
+      "epoch": 1.8224968044311889,
+      "grad_norm": 14.25,
+      "learning_rate": 4.868569096811243e-08,
+      "loss": 0.5093,
+      "step": 21387
+    },
+    {
+      "epoch": 1.8225820195994888,
+      "grad_norm": 11.1875,
+      "learning_rate": 4.8639261916998276e-08,
+      "loss": 0.3687,
+      "step": 21388
+    },
+    {
+      "epoch": 1.8226672347677888,
+      "grad_norm": 12.125,
+      "learning_rate": 4.859285457568044e-08,
+      "loss": 0.3571,
+      "step": 21389
+    },
+    {
+      "epoch": 1.8227524499360888,
+      "grad_norm": 11.5,
+      "learning_rate": 4.854646894499798e-08,
+      "loss": 0.4654,
+      "step": 21390
+    },
+    {
+      "epoch": 1.8228376651043887,
+      "grad_norm": 22.875,
+      "learning_rate": 4.850010502578897e-08,
+      "loss": 0.7046,
+      "step": 21391
+    },
+    {
+      "epoch": 1.8229228802726887,
+      "grad_norm": 11.875,
+      "learning_rate": 4.845376281889122e-08,
+      "loss": 0.4469,
+      "step": 21392
+    },
+    {
+      "epoch": 1.8230080954409886,
+      "grad_norm": 12.8125,
+      "learning_rate": 4.8407442325142246e-08,
+      "loss": 0.6002,
+      "step": 21393
+    },
+    {
+      "epoch": 1.8230933106092886,
+      "grad_norm": 13.375,
+      "learning_rate": 4.836114354537902e-08,
+      "loss": 0.4496,
+      "step": 21394
+    },
+    {
+      "epoch": 1.8231785257775885,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.8314866480438375e-08,
+      "loss": 0.5457,
+      "step": 21395
+    },
+    {
+      "epoch": 1.8232637409458885,
+      "grad_norm": 18.375,
+      "learning_rate": 4.826861113115658e-08,
+      "loss": 0.9431,
+      "step": 21396
+    },
+    {
+      "epoch": 1.8233489561141885,
+      "grad_norm": 21.875,
+      "learning_rate": 4.822237749836922e-08,
+      "loss": 0.49,
+      "step": 21397
+    },
+    {
+      "epoch": 1.8234341712824884,
+      "grad_norm": 19.25,
+      "learning_rate": 4.8176165582912024e-08,
+      "loss": 0.7673,
+      "step": 21398
+    },
+    {
+      "epoch": 1.8235193864507884,
+      "grad_norm": 14.625,
+      "learning_rate": 4.8129975385620005e-08,
+      "loss": 0.5409,
+      "step": 21399
+    },
+    {
+      "epoch": 1.8236046016190883,
+      "grad_norm": 14.75,
+      "learning_rate": 4.8083806907328055e-08,
+      "loss": 0.5689,
+      "step": 21400
+    },
+    {
+      "epoch": 1.8236898167873883,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.80376601488701e-08,
+      "loss": 0.4984,
+      "step": 21401
+    },
+    {
+      "epoch": 1.8237750319556882,
+      "grad_norm": 16.625,
+      "learning_rate": 4.7991535111080454e-08,
+      "loss": 0.9298,
+      "step": 21402
+    },
+    {
+      "epoch": 1.8238602471239882,
+      "grad_norm": 19.625,
+      "learning_rate": 4.7945431794792354e-08,
+      "loss": 0.8878,
+      "step": 21403
+    },
+    {
+      "epoch": 1.8239454622922882,
+      "grad_norm": 17.0,
+      "learning_rate": 4.789935020083944e-08,
+      "loss": 0.6657,
+      "step": 21404
+    },
+    {
+      "epoch": 1.8240306774605881,
+      "grad_norm": 16.5,
+      "learning_rate": 4.785329033005368e-08,
+      "loss": 0.7463,
+      "step": 21405
+    },
+    {
+      "epoch": 1.824115892628888,
+      "grad_norm": 19.25,
+      "learning_rate": 4.7807252183267873e-08,
+      "loss": 0.7621,
+      "step": 21406
+    },
+    {
+      "epoch": 1.824201107797188,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.7761235761314e-08,
+      "loss": 0.4564,
+      "step": 21407
+    },
+    {
+      "epoch": 1.824286322965488,
+      "grad_norm": 20.875,
+      "learning_rate": 4.7715241065023624e-08,
+      "loss": 0.4771,
+      "step": 21408
+    },
+    {
+      "epoch": 1.824371538133788,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.7669268095227594e-08,
+      "loss": 0.6978,
+      "step": 21409
+    },
+    {
+      "epoch": 1.824456753302088,
+      "grad_norm": 13.75,
+      "learning_rate": 4.762331685275706e-08,
+      "loss": 0.5189,
+      "step": 21410
+    },
+    {
+      "epoch": 1.8245419684703879,
+      "grad_norm": 11.25,
+      "learning_rate": 4.7577387338442196e-08,
+      "loss": 0.2768,
+      "step": 21411
+    },
+    {
+      "epoch": 1.8246271836386878,
+      "grad_norm": 19.5,
+      "learning_rate": 4.7531479553113296e-08,
+      "loss": 0.5551,
+      "step": 21412
+    },
+    {
+      "epoch": 1.8247123988069878,
+      "grad_norm": 12.5,
+      "learning_rate": 4.748559349759957e-08,
+      "loss": 0.4111,
+      "step": 21413
+    },
+    {
+      "epoch": 1.8247976139752877,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.743972917273035e-08,
+      "loss": 0.5345,
+      "step": 21414
+    },
+    {
+      "epoch": 1.8248828291435877,
+      "grad_norm": 21.125,
+      "learning_rate": 4.7393886579334554e-08,
+      "loss": 1.0472,
+      "step": 21415
+    },
+    {
+      "epoch": 1.8249680443118876,
+      "grad_norm": 18.875,
+      "learning_rate": 4.734806571824055e-08,
+      "loss": 0.7113,
+      "step": 21416
+    },
+    {
+      "epoch": 1.8250532594801876,
+      "grad_norm": 16.25,
+      "learning_rate": 4.7302266590276424e-08,
+      "loss": 0.4739,
+      "step": 21417
+    },
+    {
+      "epoch": 1.8251384746484876,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.7256489196269574e-08,
+      "loss": 0.5918,
+      "step": 21418
+    },
+    {
+      "epoch": 1.8252236898167875,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.7210733537047525e-08,
+      "loss": 0.7054,
+      "step": 21419
+    },
+    {
+      "epoch": 1.8253089049850875,
+      "grad_norm": 19.0,
+      "learning_rate": 4.7164999613436986e-08,
+      "loss": 0.6973,
+      "step": 21420
+    },
+    {
+      "epoch": 1.8253941201533874,
+      "grad_norm": 24.875,
+      "learning_rate": 4.711928742626465e-08,
+      "loss": 0.8388,
+      "step": 21421
+    },
+    {
+      "epoch": 1.8254793353216874,
+      "grad_norm": 11.5625,
+      "learning_rate": 4.7073596976356103e-08,
+      "loss": 0.3528,
+      "step": 21422
+    },
+    {
+      "epoch": 1.8255645504899873,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.702792826453736e-08,
+      "loss": 0.5413,
+      "step": 21423
+    },
+    {
+      "epoch": 1.8256497656582873,
+      "grad_norm": 24.875,
+      "learning_rate": 4.69822812916336e-08,
+      "loss": 0.9731,
+      "step": 21424
+    },
+    {
+      "epoch": 1.8257349808265873,
+      "grad_norm": 20.25,
+      "learning_rate": 4.693665605846998e-08,
+      "loss": 0.9214,
+      "step": 21425
+    },
+    {
+      "epoch": 1.8258201959948872,
+      "grad_norm": 17.25,
+      "learning_rate": 4.689105256587043e-08,
+      "loss": 0.5614,
+      "step": 21426
+    },
+    {
+      "epoch": 1.8259054111631872,
+      "grad_norm": 19.75,
+      "learning_rate": 4.6845470814659564e-08,
+      "loss": 0.9505,
+      "step": 21427
+    },
+    {
+      "epoch": 1.8259906263314871,
+      "grad_norm": 10.9375,
+      "learning_rate": 4.67999108056609e-08,
+      "loss": 0.269,
+      "step": 21428
+    },
+    {
+      "epoch": 1.826075841499787,
+      "grad_norm": 11.875,
+      "learning_rate": 4.6754372539697666e-08,
+      "loss": 0.4099,
+      "step": 21429
+    },
+    {
+      "epoch": 1.826161056668087,
+      "grad_norm": 22.25,
+      "learning_rate": 4.670885601759309e-08,
+      "loss": 0.8499,
+      "step": 21430
+    },
+    {
+      "epoch": 1.826246271836387,
+      "grad_norm": 17.5,
+      "learning_rate": 4.666336124016929e-08,
+      "loss": 0.4865,
+      "step": 21431
+    },
+    {
+      "epoch": 1.826331487004687,
+      "grad_norm": 37.25,
+      "learning_rate": 4.6617888208248667e-08,
+      "loss": 1.4276,
+      "step": 21432
+    },
+    {
+      "epoch": 1.826416702172987,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.657243692265279e-08,
+      "loss": 0.6957,
+      "step": 21433
+    },
+    {
+      "epoch": 1.8265019173412869,
+      "grad_norm": 10.75,
+      "learning_rate": 4.652700738420335e-08,
+      "loss": 0.9205,
+      "step": 21434
+    },
+    {
+      "epoch": 1.8265871325095868,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.648159959372081e-08,
+      "loss": 0.6468,
+      "step": 21435
+    },
+    {
+      "epoch": 1.8266723476778868,
+      "grad_norm": 21.875,
+      "learning_rate": 4.643621355202632e-08,
+      "loss": 0.8943,
+      "step": 21436
+    },
+    {
+      "epoch": 1.8267575628461867,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.63908492599395e-08,
+      "loss": 0.5578,
+      "step": 21437
+    },
+    {
+      "epoch": 1.8268427780144867,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.634550671828039e-08,
+      "loss": 0.5269,
+      "step": 21438
+    },
+    {
+      "epoch": 1.8269279931827866,
+      "grad_norm": 19.75,
+      "learning_rate": 4.630018592786833e-08,
+      "loss": 0.8209,
+      "step": 21439
+    },
+    {
+      "epoch": 1.8270132083510866,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.625488688952226e-08,
+      "loss": 0.2515,
+      "step": 21440
+    },
+    {
+      "epoch": 1.8270984235193866,
+      "grad_norm": 17.625,
+      "learning_rate": 4.6209609604060957e-08,
+      "loss": 0.2577,
+      "step": 21441
+    },
+    {
+      "epoch": 1.8271836386876865,
+      "grad_norm": 13.875,
+      "learning_rate": 4.6164354072302517e-08,
+      "loss": 0.7431,
+      "step": 21442
+    },
+    {
+      "epoch": 1.8272688538559865,
+      "grad_norm": 16.5,
+      "learning_rate": 4.611912029506463e-08,
+      "loss": 0.8397,
+      "step": 21443
+    },
+    {
+      "epoch": 1.8273540690242864,
+      "grad_norm": 16.875,
+      "learning_rate": 4.6073908273164966e-08,
+      "loss": 0.785,
+      "step": 21444
+    },
+    {
+      "epoch": 1.8274392841925864,
+      "grad_norm": 16.75,
+      "learning_rate": 4.602871800742037e-08,
+      "loss": 0.4577,
+      "step": 21445
+    },
+    {
+      "epoch": 1.8275244993608863,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.598354949864756e-08,
+      "loss": 0.484,
+      "step": 21446
+    },
+    {
+      "epoch": 1.8276097145291863,
+      "grad_norm": 18.125,
+      "learning_rate": 4.5938402747662814e-08,
+      "loss": 0.6348,
+      "step": 21447
+    },
+    {
+      "epoch": 1.8276949296974863,
+      "grad_norm": 23.5,
+      "learning_rate": 4.589327775528174e-08,
+      "loss": 0.9853,
+      "step": 21448
+    },
+    {
+      "epoch": 1.8277801448657862,
+      "grad_norm": 16.375,
+      "learning_rate": 4.584817452231993e-08,
+      "loss": 0.6923,
+      "step": 21449
+    },
+    {
+      "epoch": 1.8278653600340862,
+      "grad_norm": 22.875,
+      "learning_rate": 4.580309304959257e-08,
+      "loss": 0.7489,
+      "step": 21450
+    },
+    {
+      "epoch": 1.8279505752023861,
+      "grad_norm": 17.0,
+      "learning_rate": 4.5758033337914284e-08,
+      "loss": 0.6064,
+      "step": 21451
+    },
+    {
+      "epoch": 1.828035790370686,
+      "grad_norm": 16.75,
+      "learning_rate": 4.571299538809914e-08,
+      "loss": 0.3532,
+      "step": 21452
+    },
+    {
+      "epoch": 1.828121005538986,
+      "grad_norm": 13.625,
+      "learning_rate": 4.566797920096136e-08,
+      "loss": 0.4172,
+      "step": 21453
+    },
+    {
+      "epoch": 1.828206220707286,
+      "grad_norm": 28.25,
+      "learning_rate": 4.562298477731417e-08,
+      "loss": 1.0465,
+      "step": 21454
+    },
+    {
+      "epoch": 1.828291435875586,
+      "grad_norm": 15.375,
+      "learning_rate": 4.55780121179708e-08,
+      "loss": 0.6456,
+      "step": 21455
+    },
+    {
+      "epoch": 1.828376651043886,
+      "grad_norm": 12.25,
+      "learning_rate": 4.553306122374382e-08,
+      "loss": 0.3024,
+      "step": 21456
+    },
+    {
+      "epoch": 1.8284618662121859,
+      "grad_norm": 12.5625,
+      "learning_rate": 4.548813209544548e-08,
+      "loss": 0.4537,
+      "step": 21457
+    },
+    {
+      "epoch": 1.8285470813804858,
+      "grad_norm": 11.875,
+      "learning_rate": 4.5443224733887905e-08,
+      "loss": 0.4194,
+      "step": 21458
+    },
+    {
+      "epoch": 1.8286322965487858,
+      "grad_norm": 14.0,
+      "learning_rate": 4.539833913988267e-08,
+      "loss": 0.687,
+      "step": 21459
+    },
+    {
+      "epoch": 1.8287175117170857,
+      "grad_norm": 11.0625,
+      "learning_rate": 4.535347531424067e-08,
+      "loss": 0.291,
+      "step": 21460
+    },
+    {
+      "epoch": 1.8288027268853857,
+      "grad_norm": 23.625,
+      "learning_rate": 4.53086332577729e-08,
+      "loss": 0.9811,
+      "step": 21461
+    },
+    {
+      "epoch": 1.8288879420536857,
+      "grad_norm": 11.875,
+      "learning_rate": 4.5263812971289286e-08,
+      "loss": 0.3369,
+      "step": 21462
+    },
+    {
+      "epoch": 1.8289731572219856,
+      "grad_norm": 17.0,
+      "learning_rate": 4.5219014455600004e-08,
+      "loss": 0.5036,
+      "step": 21463
+    },
+    {
+      "epoch": 1.8290583723902856,
+      "grad_norm": 21.125,
+      "learning_rate": 4.5174237711514966e-08,
+      "loss": 0.5683,
+      "step": 21464
+    },
+    {
+      "epoch": 1.8291435875585855,
+      "grad_norm": 26.75,
+      "learning_rate": 4.512948273984269e-08,
+      "loss": 0.7892,
+      "step": 21465
+    },
+    {
+      "epoch": 1.8292288027268855,
+      "grad_norm": 11.4375,
+      "learning_rate": 4.5084749541392245e-08,
+      "loss": 0.4789,
+      "step": 21466
+    },
+    {
+      "epoch": 1.8293140178951854,
+      "grad_norm": 11.625,
+      "learning_rate": 4.50400381169723e-08,
+      "loss": 0.2607,
+      "step": 21467
+    },
+    {
+      "epoch": 1.8293992330634854,
+      "grad_norm": 13.75,
+      "learning_rate": 4.499534846739026e-08,
+      "loss": 0.4236,
+      "step": 21468
+    },
+    {
+      "epoch": 1.8294844482317854,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.4950680593454226e-08,
+      "loss": 0.5723,
+      "step": 21469
+    },
+    {
+      "epoch": 1.8295696634000853,
+      "grad_norm": 16.875,
+      "learning_rate": 4.490603449597092e-08,
+      "loss": 0.5243,
+      "step": 21470
+    },
+    {
+      "epoch": 1.8296548785683853,
+      "grad_norm": 12.0,
+      "learning_rate": 4.4861410175747464e-08,
+      "loss": 0.4585,
+      "step": 21471
+    },
+    {
+      "epoch": 1.8297400937366852,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.481680763359031e-08,
+      "loss": 0.5439,
+      "step": 21472
+    },
+    {
+      "epoch": 1.8298253089049852,
+      "grad_norm": 17.25,
+      "learning_rate": 4.477222687030519e-08,
+      "loss": 0.4801,
+      "step": 21473
+    },
+    {
+      "epoch": 1.8299105240732851,
+      "grad_norm": 24.875,
+      "learning_rate": 4.4727667886697996e-08,
+      "loss": 0.6309,
+      "step": 21474
+    },
+    {
+      "epoch": 1.829995739241585,
+      "grad_norm": 13.625,
+      "learning_rate": 4.468313068357377e-08,
+      "loss": 0.4806,
+      "step": 21475
+    },
+    {
+      "epoch": 1.830080954409885,
+      "grad_norm": 15.0,
+      "learning_rate": 4.4638615261737436e-08,
+      "loss": 0.2977,
+      "step": 21476
+    },
+    {
+      "epoch": 1.830166169578185,
+      "grad_norm": 13.625,
+      "learning_rate": 4.4594121621993615e-08,
+      "loss": 0.6979,
+      "step": 21477
+    },
+    {
+      "epoch": 1.830251384746485,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.4549649765145834e-08,
+      "loss": 0.6475,
+      "step": 21478
+    },
+    {
+      "epoch": 1.830336599914785,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.450519969199818e-08,
+      "loss": 0.4471,
+      "step": 21479
+    },
+    {
+      "epoch": 1.8304218150830849,
+      "grad_norm": 16.375,
+      "learning_rate": 4.4460771403353615e-08,
+      "loss": 0.4798,
+      "step": 21480
+    },
+    {
+      "epoch": 1.8305070302513848,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.441636490001552e-08,
+      "loss": 0.6174,
+      "step": 21481
+    },
+    {
+      "epoch": 1.8305922454196848,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.437198018278577e-08,
+      "loss": 0.4413,
+      "step": 21482
+    },
+    {
+      "epoch": 1.8306774605879848,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.432761725246662e-08,
+      "loss": 0.5577,
+      "step": 21483
+    },
+    {
+      "epoch": 1.8307626757562847,
+      "grad_norm": 15.75,
+      "learning_rate": 4.4283276109860076e-08,
+      "loss": 0.5993,
+      "step": 21484
+    },
+    {
+      "epoch": 1.8308478909245847,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.423895675576717e-08,
+      "loss": 0.4681,
+      "step": 21485
+    },
+    {
+      "epoch": 1.8309331060928846,
+      "grad_norm": 13.0,
+      "learning_rate": 4.4194659190988645e-08,
+      "loss": 0.279,
+      "step": 21486
+    },
+    {
+      "epoch": 1.8310183212611846,
+      "grad_norm": 20.125,
+      "learning_rate": 4.4150383416325106e-08,
+      "loss": 0.6261,
+      "step": 21487
+    },
+    {
+      "epoch": 1.8311035364294845,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.410612943257675e-08,
+      "loss": 0.7691,
+      "step": 21488
+    },
+    {
+      "epoch": 1.8311887515977845,
+      "grad_norm": 20.0,
+      "learning_rate": 4.406189724054349e-08,
+      "loss": 0.6924,
+      "step": 21489
+    },
+    {
+      "epoch": 1.8312739667660844,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.4017686841024145e-08,
+      "loss": 0.6117,
+      "step": 21490
+    },
+    {
+      "epoch": 1.8313591819343844,
+      "grad_norm": 21.75,
+      "learning_rate": 4.397349823481792e-08,
+      "loss": 0.7554,
+      "step": 21491
+    },
+    {
+      "epoch": 1.8314443971026844,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.3929331422723345e-08,
+      "loss": 0.5587,
+      "step": 21492
+    },
+    {
+      "epoch": 1.8315296122709843,
+      "grad_norm": 13.75,
+      "learning_rate": 4.388518640553868e-08,
+      "loss": 0.5387,
+      "step": 21493
+    },
+    {
+      "epoch": 1.8316148274392843,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.3841063184061615e-08,
+      "loss": 0.4323,
+      "step": 21494
+    },
+    {
+      "epoch": 1.8317000426075842,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.379696175908929e-08,
+      "loss": 0.5678,
+      "step": 21495
+    },
+    {
+      "epoch": 1.8317852577758842,
+      "grad_norm": 18.625,
+      "learning_rate": 4.375288213141871e-08,
+      "loss": 0.5798,
+      "step": 21496
+    },
+    {
+      "epoch": 1.8318704729441841,
+      "grad_norm": 20.5,
+      "learning_rate": 4.3708824301846723e-08,
+      "loss": 0.7563,
+      "step": 21497
+    },
+    {
+      "epoch": 1.831955688112484,
+      "grad_norm": 20.125,
+      "learning_rate": 4.36647882711691e-08,
+      "loss": 0.7358,
+      "step": 21498
+    },
+    {
+      "epoch": 1.832040903280784,
+      "grad_norm": 20.625,
+      "learning_rate": 4.3620774040181865e-08,
+      "loss": 0.9437,
+      "step": 21499
+    },
+    {
+      "epoch": 1.832126118449084,
+      "grad_norm": 22.0,
+      "learning_rate": 4.3576781609680346e-08,
+      "loss": 0.7043,
+      "step": 21500
+    },
+    {
+      "epoch": 1.832211333617384,
+      "grad_norm": 11.6875,
+      "learning_rate": 4.353281098045961e-08,
+      "loss": 0.4967,
+      "step": 21501
+    },
+    {
+      "epoch": 1.832296548785684,
+      "grad_norm": 22.125,
+      "learning_rate": 4.3488862153314296e-08,
+      "loss": 0.8167,
+      "step": 21502
+    },
+    {
+      "epoch": 1.8323817639539839,
+      "grad_norm": 14.75,
+      "learning_rate": 4.3444935129038216e-08,
+      "loss": 0.6233,
+      "step": 21503
+    },
+    {
+      "epoch": 1.8324669791222838,
+      "grad_norm": 13.625,
+      "learning_rate": 4.3401029908425596e-08,
+      "loss": 0.5071,
+      "step": 21504
+    },
+    {
+      "epoch": 1.8325521942905838,
+      "grad_norm": 36.25,
+      "learning_rate": 4.3357146492269545e-08,
+      "loss": 0.7682,
+      "step": 21505
+    },
+    {
+      "epoch": 1.8326374094588838,
+      "grad_norm": 15.375,
+      "learning_rate": 4.331328488136347e-08,
+      "loss": 0.366,
+      "step": 21506
+    },
+    {
+      "epoch": 1.8327226246271837,
+      "grad_norm": 11.375,
+      "learning_rate": 4.3269445076499636e-08,
+      "loss": 0.3402,
+      "step": 21507
+    },
+    {
+      "epoch": 1.8328078397954837,
+      "grad_norm": 13.875,
+      "learning_rate": 4.322562707847033e-08,
+      "loss": 0.7265,
+      "step": 21508
+    },
+    {
+      "epoch": 1.8328930549637836,
+      "grad_norm": 19.0,
+      "learning_rate": 4.318183088806757e-08,
+      "loss": 0.7686,
+      "step": 21509
+    },
+    {
+      "epoch": 1.8329782701320836,
+      "grad_norm": 14.25,
+      "learning_rate": 4.313805650608252e-08,
+      "loss": 0.6415,
+      "step": 21510
+    },
+    {
+      "epoch": 1.8330634853003835,
+      "grad_norm": 22.0,
+      "learning_rate": 4.309430393330663e-08,
+      "loss": 0.7169,
+      "step": 21511
+    },
+    {
+      "epoch": 1.8331487004686835,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.3050573170529966e-08,
+      "loss": 0.506,
+      "step": 21512
+    },
+    {
+      "epoch": 1.8332339156369835,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.3006864218543274e-08,
+      "loss": 0.5407,
+      "step": 21513
+    },
+    {
+      "epoch": 1.8333191308052834,
+      "grad_norm": 17.75,
+      "learning_rate": 4.296317707813649e-08,
+      "loss": 0.768,
+      "step": 21514
+    },
+    {
+      "epoch": 1.8334043459735834,
+      "grad_norm": 18.0,
+      "learning_rate": 4.2919511750098566e-08,
+      "loss": 1.1035,
+      "step": 21515
+    },
+    {
+      "epoch": 1.8334895611418833,
+      "grad_norm": 16.75,
+      "learning_rate": 4.2875868235218995e-08,
+      "loss": 0.6447,
+      "step": 21516
+    },
+    {
+      "epoch": 1.8335747763101833,
+      "grad_norm": 11.4375,
+      "learning_rate": 4.283224653428619e-08,
+      "loss": 0.2259,
+      "step": 21517
+    },
+    {
+      "epoch": 1.8336599914784832,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.278864664808882e-08,
+      "loss": 0.3799,
+      "step": 21518
+    },
+    {
+      "epoch": 1.8337452066467832,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.274506857741459e-08,
+      "loss": 0.6836,
+      "step": 21519
+    },
+    {
+      "epoch": 1.8338304218150832,
+      "grad_norm": 19.5,
+      "learning_rate": 4.270151232305078e-08,
+      "loss": 0.8429,
+      "step": 21520
+    },
+    {
+      "epoch": 1.8339156369833831,
+      "grad_norm": 17.375,
+      "learning_rate": 4.2657977885784685e-08,
+      "loss": 0.7382,
+      "step": 21521
+    },
+    {
+      "epoch": 1.834000852151683,
+      "grad_norm": 18.125,
+      "learning_rate": 4.261446526640303e-08,
+      "loss": 0.7791,
+      "step": 21522
+    },
+    {
+      "epoch": 1.834086067319983,
+      "grad_norm": 11.1875,
+      "learning_rate": 4.257097446569228e-08,
+      "loss": 0.357,
+      "step": 21523
+    },
+    {
+      "epoch": 1.834171282488283,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.25275054844379e-08,
+      "loss": 0.508,
+      "step": 21524
+    },
+    {
+      "epoch": 1.834256497656583,
+      "grad_norm": 14.9375,
+      "learning_rate": 4.248405832342581e-08,
+      "loss": 0.4457,
+      "step": 21525
+    },
+    {
+      "epoch": 1.834341712824883,
+      "grad_norm": 12.125,
+      "learning_rate": 4.24406329834412e-08,
+      "loss": 0.3831,
+      "step": 21526
+    },
+    {
+      "epoch": 1.8344269279931829,
+      "grad_norm": 15.375,
+      "learning_rate": 4.2397229465268444e-08,
+      "loss": 0.5507,
+      "step": 21527
+    },
+    {
+      "epoch": 1.8345121431614828,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.23538477696922e-08,
+      "loss": 0.484,
+      "step": 21528
+    },
+    {
+      "epoch": 1.8345973583297828,
+      "grad_norm": 14.375,
+      "learning_rate": 4.2310487897496284e-08,
+      "loss": 0.7808,
+      "step": 21529
+    },
+    {
+      "epoch": 1.8346825734980827,
+      "grad_norm": 17.75,
+      "learning_rate": 4.226714984946423e-08,
+      "loss": 0.6765,
+      "step": 21530
+    },
+    {
+      "epoch": 1.8347677886663827,
+      "grad_norm": 14.625,
+      "learning_rate": 4.22238336263793e-08,
+      "loss": 0.7523,
+      "step": 21531
+    },
+    {
+      "epoch": 1.8348530038346826,
+      "grad_norm": 14.75,
+      "learning_rate": 4.218053922902421e-08,
+      "loss": 0.463,
+      "step": 21532
+    },
+    {
+      "epoch": 1.8349382190029826,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.213726665818124e-08,
+      "loss": 0.5986,
+      "step": 21533
+    },
+    {
+      "epoch": 1.8350234341712826,
+      "grad_norm": 16.75,
+      "learning_rate": 4.209401591463255e-08,
+      "loss": 0.7107,
+      "step": 21534
+    },
+    {
+      "epoch": 1.8351086493395825,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.2050786999159596e-08,
+      "loss": 0.44,
+      "step": 21535
+    },
+    {
+      "epoch": 1.8351938645078825,
+      "grad_norm": 12.0625,
+      "learning_rate": 4.20075799125437e-08,
+      "loss": 0.424,
+      "step": 21536
+    },
+    {
+      "epoch": 1.8352790796761824,
+      "grad_norm": 9.6875,
+      "learning_rate": 4.196439465556548e-08,
+      "loss": 0.188,
+      "step": 21537
+    },
+    {
+      "epoch": 1.8353642948444824,
+      "grad_norm": 21.0,
+      "learning_rate": 4.1921231229005435e-08,
+      "loss": 0.6415,
+      "step": 21538
+    },
+    {
+      "epoch": 1.8354495100127823,
+      "grad_norm": 20.625,
+      "learning_rate": 4.187808963364348e-08,
+      "loss": 0.8235,
+      "step": 21539
+    },
+    {
+      "epoch": 1.8355347251810823,
+      "grad_norm": 20.125,
+      "learning_rate": 4.183496987025942e-08,
+      "loss": 0.9157,
+      "step": 21540
+    },
+    {
+      "epoch": 1.8356199403493823,
+      "grad_norm": 17.125,
+      "learning_rate": 4.179187193963236e-08,
+      "loss": 0.6801,
+      "step": 21541
+    },
+    {
+      "epoch": 1.8357051555176822,
+      "grad_norm": 14.75,
+      "learning_rate": 4.17487958425411e-08,
+      "loss": 0.4658,
+      "step": 21542
+    },
+    {
+      "epoch": 1.8357903706859822,
+      "grad_norm": 16.0,
+      "learning_rate": 4.1705741579764065e-08,
+      "loss": 0.5891,
+      "step": 21543
+    },
+    {
+      "epoch": 1.8358755858542821,
+      "grad_norm": 10.5625,
+      "learning_rate": 4.166270915207937e-08,
+      "loss": 0.2203,
+      "step": 21544
+    },
+    {
+      "epoch": 1.835960801022582,
+      "grad_norm": 17.25,
+      "learning_rate": 4.16196985602646e-08,
+      "loss": 0.4637,
+      "step": 21545
+    },
+    {
+      "epoch": 1.836046016190882,
+      "grad_norm": 18.625,
+      "learning_rate": 4.15767098050969e-08,
+      "loss": 0.7447,
+      "step": 21546
+    },
+    {
+      "epoch": 1.836131231359182,
+      "grad_norm": 15.125,
+      "learning_rate": 4.153374288735315e-08,
+      "loss": 0.672,
+      "step": 21547
+    },
+    {
+      "epoch": 1.836216446527482,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.1490797807810096e-08,
+      "loss": 0.3447,
+      "step": 21548
+    },
+    {
+      "epoch": 1.836301661695782,
+      "grad_norm": 18.5,
+      "learning_rate": 4.144787456724336e-08,
+      "loss": 0.5308,
+      "step": 21549
+    },
+    {
+      "epoch": 1.8363868768640819,
+      "grad_norm": 14.875,
+      "learning_rate": 4.140497316642914e-08,
+      "loss": 0.7252,
+      "step": 21550
+    },
+    {
+      "epoch": 1.8364720920323818,
+      "grad_norm": 16.625,
+      "learning_rate": 4.1362093606142074e-08,
+      "loss": 0.5289,
+      "step": 21551
+    },
+    {
+      "epoch": 1.8365573072006818,
+      "grad_norm": 11.875,
+      "learning_rate": 4.131923588715739e-08,
+      "loss": 0.2394,
+      "step": 21552
+    },
+    {
+      "epoch": 1.8366425223689817,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.127640001024974e-08,
+      "loss": 0.3451,
+      "step": 21553
+    },
+    {
+      "epoch": 1.8367277375372817,
+      "grad_norm": 14.875,
+      "learning_rate": 4.123358597619281e-08,
+      "loss": 0.6578,
+      "step": 21554
+    },
+    {
+      "epoch": 1.8368129527055816,
+      "grad_norm": 11.5,
+      "learning_rate": 4.1190793785760574e-08,
+      "loss": 0.3202,
+      "step": 21555
+    },
+    {
+      "epoch": 1.8368981678738816,
+      "grad_norm": 26.625,
+      "learning_rate": 4.114802343972615e-08,
+      "loss": 0.6101,
+      "step": 21556
+    },
+    {
+      "epoch": 1.8369833830421816,
+      "grad_norm": 16.75,
+      "learning_rate": 4.1105274938862685e-08,
+      "loss": 0.6658,
+      "step": 21557
+    },
+    {
+      "epoch": 1.8370685982104815,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.10625482839426e-08,
+      "loss": 0.4154,
+      "step": 21558
+    },
+    {
+      "epoch": 1.8371538133787815,
+      "grad_norm": 12.75,
+      "learning_rate": 4.101984347573779e-08,
+      "loss": 0.4266,
+      "step": 21559
+    },
+    {
+      "epoch": 1.8372390285470814,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.097716051501999e-08,
+      "loss": 0.3439,
+      "step": 21560
+    },
+    {
+      "epoch": 1.8373242437153814,
+      "grad_norm": 17.5,
+      "learning_rate": 4.093449940256094e-08,
+      "loss": 1.0169,
+      "step": 21561
+    },
+    {
+      "epoch": 1.8374094588836813,
+      "grad_norm": 17.5,
+      "learning_rate": 4.089186013913116e-08,
+      "loss": 0.4537,
+      "step": 21562
+    },
+    {
+      "epoch": 1.8374946740519813,
+      "grad_norm": 15.9375,
+      "learning_rate": 4.084924272550139e-08,
+      "loss": 0.713,
+      "step": 21563
+    },
+    {
+      "epoch": 1.8375798892202813,
+      "grad_norm": 14.4375,
+      "learning_rate": 4.0806647162441596e-08,
+      "loss": 0.6854,
+      "step": 21564
+    },
+    {
+      "epoch": 1.8376651043885812,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.0764073450721697e-08,
+      "loss": 0.4051,
+      "step": 21565
+    },
+    {
+      "epoch": 1.8377503195568812,
+      "grad_norm": 20.125,
+      "learning_rate": 4.072152159111109e-08,
+      "loss": 0.8643,
+      "step": 21566
+    },
+    {
+      "epoch": 1.8378355347251811,
+      "grad_norm": 21.5,
+      "learning_rate": 4.067899158437846e-08,
+      "loss": 0.8252,
+      "step": 21567
+    },
+    {
+      "epoch": 1.837920749893481,
+      "grad_norm": 20.625,
+      "learning_rate": 4.0636483431292507e-08,
+      "loss": 0.7776,
+      "step": 21568
+    },
+    {
+      "epoch": 1.838005965061781,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.059399713262135e-08,
+      "loss": 0.5886,
+      "step": 21569
+    },
+    {
+      "epoch": 1.838091180230081,
+      "grad_norm": 14.5625,
+      "learning_rate": 4.0551532689133004e-08,
+      "loss": 0.528,
+      "step": 21570
+    },
+    {
+      "epoch": 1.838176395398381,
+      "grad_norm": 16.125,
+      "learning_rate": 4.0509090101594486e-08,
+      "loss": 1.032,
+      "step": 21571
+    },
+    {
+      "epoch": 1.838261610566681,
+      "grad_norm": 16.875,
+      "learning_rate": 4.046666937077309e-08,
+      "loss": 0.2717,
+      "step": 21572
+    },
+    {
+      "epoch": 1.8383468257349809,
+      "grad_norm": 11.5,
+      "learning_rate": 4.0424270497435165e-08,
+      "loss": 0.5418,
+      "step": 21573
+    },
+    {
+      "epoch": 1.8384320409032808,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.0381893482347166e-08,
+      "loss": 0.5199,
+      "step": 21574
+    },
+    {
+      "epoch": 1.8385172560715808,
+      "grad_norm": 10.25,
+      "learning_rate": 4.0339538326274604e-08,
+      "loss": 0.2337,
+      "step": 21575
+    },
+    {
+      "epoch": 1.8386024712398807,
+      "grad_norm": 14.625,
+      "learning_rate": 4.0297205029982825e-08,
+      "loss": 0.5854,
+      "step": 21576
+    },
+    {
+      "epoch": 1.8386876864081807,
+      "grad_norm": 11.5,
+      "learning_rate": 4.0254893594237064e-08,
+      "loss": 1.0806,
+      "step": 21577
+    },
+    {
+      "epoch": 1.8387729015764807,
+      "grad_norm": 19.625,
+      "learning_rate": 4.021260401980212e-08,
+      "loss": 0.6252,
+      "step": 21578
+    },
+    {
+      "epoch": 1.8388581167447806,
+      "grad_norm": 18.0,
+      "learning_rate": 4.017033630744169e-08,
+      "loss": 0.8195,
+      "step": 21579
+    },
+    {
+      "epoch": 1.8389433319130806,
+      "grad_norm": 14.1875,
+      "learning_rate": 4.012809045791988e-08,
+      "loss": 0.43,
+      "step": 21580
+    },
+    {
+      "epoch": 1.8390285470813805,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.0085866472000106e-08,
+      "loss": 0.4873,
+      "step": 21581
+    },
+    {
+      "epoch": 1.8391137622496805,
+      "grad_norm": 16.875,
+      "learning_rate": 4.004366435044566e-08,
+      "loss": 0.4081,
+      "step": 21582
+    },
+    {
+      "epoch": 1.8391989774179804,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.000148409401869e-08,
+      "loss": 0.4974,
+      "step": 21583
+    },
+    {
+      "epoch": 1.8392841925862804,
+      "grad_norm": 21.125,
+      "learning_rate": 3.9959325703481656e-08,
+      "loss": 0.9991,
+      "step": 21584
+    },
+    {
+      "epoch": 1.8393694077545804,
+      "grad_norm": 18.625,
+      "learning_rate": 3.9917189179596307e-08,
+      "loss": 1.0739,
+      "step": 21585
+    },
+    {
+      "epoch": 1.8394546229228803,
+      "grad_norm": 12.9375,
+      "learning_rate": 3.9875074523124255e-08,
+      "loss": 0.4647,
+      "step": 21586
+    },
+    {
+      "epoch": 1.8395398380911803,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.983298173482655e-08,
+      "loss": 0.3774,
+      "step": 21587
+    },
+    {
+      "epoch": 1.8396250532594802,
+      "grad_norm": 14.25,
+      "learning_rate": 3.979091081546371e-08,
+      "loss": 0.5796,
+      "step": 21588
+    },
+    {
+      "epoch": 1.8397102684277802,
+      "grad_norm": 19.625,
+      "learning_rate": 3.9748861765795956e-08,
+      "loss": 0.9681,
+      "step": 21589
+    },
+    {
+      "epoch": 1.8397954835960801,
+      "grad_norm": 16.5,
+      "learning_rate": 3.9706834586583365e-08,
+      "loss": 0.3252,
+      "step": 21590
+    },
+    {
+      "epoch": 1.83988069876438,
+      "grad_norm": 13.4375,
+      "learning_rate": 3.9664829278585616e-08,
+      "loss": 0.3255,
+      "step": 21591
+    },
+    {
+      "epoch": 1.83996591393268,
+      "grad_norm": 16.875,
+      "learning_rate": 3.9622845842561266e-08,
+      "loss": 0.6017,
+      "step": 21592
+    },
+    {
+      "epoch": 1.84005112910098,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.958088427926901e-08,
+      "loss": 0.5483,
+      "step": 21593
+    },
+    {
+      "epoch": 1.84013634426928,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.9538944589467555e-08,
+      "loss": 0.6672,
+      "step": 21594
+    },
+    {
+      "epoch": 1.84022155943758,
+      "grad_norm": 12.5,
+      "learning_rate": 3.949702677391462e-08,
+      "loss": 0.2627,
+      "step": 21595
+    },
+    {
+      "epoch": 1.8403067746058799,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.945513083336752e-08,
+      "loss": 0.5056,
+      "step": 21596
+    },
+    {
+      "epoch": 1.8403919897741798,
+      "grad_norm": 12.75,
+      "learning_rate": 3.941325676858357e-08,
+      "loss": 0.3087,
+      "step": 21597
+    },
+    {
+      "epoch": 1.8404772049424798,
+      "grad_norm": 16.625,
+      "learning_rate": 3.937140458031938e-08,
+      "loss": 0.5177,
+      "step": 21598
+    },
+    {
+      "epoch": 1.8405624201107798,
+      "grad_norm": 13.4375,
+      "learning_rate": 3.932957426933143e-08,
+      "loss": 0.388,
+      "step": 21599
+    },
+    {
+      "epoch": 1.8406476352790797,
+      "grad_norm": 16.75,
+      "learning_rate": 3.928776583637536e-08,
+      "loss": 0.5433,
+      "step": 21600
+    },
+    {
+      "epoch": 1.8407328504473797,
+      "grad_norm": 11.25,
+      "learning_rate": 3.924597928220683e-08,
+      "loss": 0.3943,
+      "step": 21601
+    },
+    {
+      "epoch": 1.8408180656156796,
+      "grad_norm": 20.5,
+      "learning_rate": 3.920421460758092e-08,
+      "loss": 0.8793,
+      "step": 21602
+    },
+    {
+      "epoch": 1.8409032807839796,
+      "grad_norm": 18.75,
+      "learning_rate": 3.9162471813252444e-08,
+      "loss": 0.7447,
+      "step": 21603
+    },
+    {
+      "epoch": 1.8409884959522795,
+      "grad_norm": 14.5,
+      "learning_rate": 3.91207508999758e-08,
+      "loss": 0.5494,
+      "step": 21604
+    },
+    {
+      "epoch": 1.8410737111205795,
+      "grad_norm": 10.25,
+      "learning_rate": 3.907905186850469e-08,
+      "loss": 0.3094,
+      "step": 21605
+    },
+    {
+      "epoch": 1.8411589262888794,
+      "grad_norm": 11.3125,
+      "learning_rate": 3.903737471959282e-08,
+      "loss": 0.3316,
+      "step": 21606
+    },
+    {
+      "epoch": 1.8412441414571794,
+      "grad_norm": 19.75,
+      "learning_rate": 3.899571945399347e-08,
+      "loss": 0.6314,
+      "step": 21607
+    },
+    {
+      "epoch": 1.8413293566254794,
+      "grad_norm": 17.5,
+      "learning_rate": 3.8954086072459096e-08,
+      "loss": 0.7562,
+      "step": 21608
+    },
+    {
+      "epoch": 1.8414145717937793,
+      "grad_norm": 16.5,
+      "learning_rate": 3.8912474575742155e-08,
+      "loss": 0.8842,
+      "step": 21609
+    },
+    {
+      "epoch": 1.8414997869620793,
+      "grad_norm": 12.75,
+      "learning_rate": 3.887088496459468e-08,
+      "loss": 0.3192,
+      "step": 21610
+    },
+    {
+      "epoch": 1.8415850021303792,
+      "grad_norm": 13.9375,
+      "learning_rate": 3.8829317239768017e-08,
+      "loss": 0.4573,
+      "step": 21611
+    },
+    {
+      "epoch": 1.8416702172986792,
+      "grad_norm": 18.75,
+      "learning_rate": 3.878777140201379e-08,
+      "loss": 0.7371,
+      "step": 21612
+    },
+    {
+      "epoch": 1.8417554324669791,
+      "grad_norm": 11.125,
+      "learning_rate": 3.874624745208236e-08,
+      "loss": 0.3375,
+      "step": 21613
+    },
+    {
+      "epoch": 1.841840647635279,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.8704745390724244e-08,
+      "loss": 0.4617,
+      "step": 21614
+    },
+    {
+      "epoch": 1.841925862803579,
+      "grad_norm": 14.25,
+      "learning_rate": 3.866326521868954e-08,
+      "loss": 0.4934,
+      "step": 21615
+    },
+    {
+      "epoch": 1.842011077971879,
+      "grad_norm": 12.6875,
+      "learning_rate": 3.862180693672765e-08,
+      "loss": 0.4485,
+      "step": 21616
+    },
+    {
+      "epoch": 1.842096293140179,
+      "grad_norm": 13.75,
+      "learning_rate": 3.858037054558797e-08,
+      "loss": 0.3848,
+      "step": 21617
+    },
+    {
+      "epoch": 1.842181508308479,
+      "grad_norm": 21.5,
+      "learning_rate": 3.853895604601893e-08,
+      "loss": 0.9219,
+      "step": 21618
+    },
+    {
+      "epoch": 1.842266723476779,
+      "grad_norm": 16.5,
+      "learning_rate": 3.849756343876923e-08,
+      "loss": 0.9358,
+      "step": 21619
+    },
+    {
+      "epoch": 1.8423519386450788,
+      "grad_norm": 15.375,
+      "learning_rate": 3.8456192724586895e-08,
+      "loss": 0.6992,
+      "step": 21620
+    },
+    {
+      "epoch": 1.8424371538133788,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.841484390421951e-08,
+      "loss": 0.7419,
+      "step": 21621
+    },
+    {
+      "epoch": 1.8425223689816788,
+      "grad_norm": 13.75,
+      "learning_rate": 3.8373516978414125e-08,
+      "loss": 0.5878,
+      "step": 21622
+    },
+    {
+      "epoch": 1.8426075841499787,
+      "grad_norm": 17.125,
+      "learning_rate": 3.833221194791778e-08,
+      "loss": 0.5279,
+      "step": 21623
+    },
+    {
+      "epoch": 1.8426927993182787,
+      "grad_norm": 16.375,
+      "learning_rate": 3.829092881347668e-08,
+      "loss": 0.5738,
+      "step": 21624
+    },
+    {
+      "epoch": 1.8427780144865786,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.824966757583704e-08,
+      "loss": 0.5105,
+      "step": 21625
+    },
+    {
+      "epoch": 1.8428632296548786,
+      "grad_norm": 16.75,
+      "learning_rate": 3.820842823574436e-08,
+      "loss": 0.6726,
+      "step": 21626
+    },
+    {
+      "epoch": 1.8429484448231785,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.8167210793943894e-08,
+      "loss": 0.5963,
+      "step": 21627
+    },
+    {
+      "epoch": 1.8430336599914785,
+      "grad_norm": 17.125,
+      "learning_rate": 3.8126015251180594e-08,
+      "loss": 0.5211,
+      "step": 21628
+    },
+    {
+      "epoch": 1.8431188751597785,
+      "grad_norm": 13.6875,
+      "learning_rate": 3.8084841608198864e-08,
+      "loss": 0.4285,
+      "step": 21629
+    },
+    {
+      "epoch": 1.8432040903280784,
+      "grad_norm": 16.75,
+      "learning_rate": 3.80436898657427e-08,
+      "loss": 0.6153,
+      "step": 21630
+    },
+    {
+      "epoch": 1.8432893054963784,
+      "grad_norm": 11.3125,
+      "learning_rate": 3.8002560024555794e-08,
+      "loss": 0.3088,
+      "step": 21631
+    },
+    {
+      "epoch": 1.8433745206646783,
+      "grad_norm": 18.0,
+      "learning_rate": 3.796145208538132e-08,
+      "loss": 0.9886,
+      "step": 21632
+    },
+    {
+      "epoch": 1.8434597358329783,
+      "grad_norm": 14.0,
+      "learning_rate": 3.792036604896229e-08,
+      "loss": 0.7846,
+      "step": 21633
+    },
+    {
+      "epoch": 1.8435449510012782,
+      "grad_norm": 16.25,
+      "learning_rate": 3.787930191604103e-08,
+      "loss": 0.6192,
+      "step": 21634
+    },
+    {
+      "epoch": 1.8436301661695782,
+      "grad_norm": 25.875,
+      "learning_rate": 3.783825968735971e-08,
+      "loss": 1.1105,
+      "step": 21635
+    },
+    {
+      "epoch": 1.8437153813378782,
+      "grad_norm": 20.5,
+      "learning_rate": 3.7797239363659836e-08,
+      "loss": 0.4376,
+      "step": 21636
+    },
+    {
+      "epoch": 1.8438005965061781,
+      "grad_norm": 24.0,
+      "learning_rate": 3.775624094568289e-08,
+      "loss": 0.7224,
+      "step": 21637
+    },
+    {
+      "epoch": 1.843885811674478,
+      "grad_norm": 11.625,
+      "learning_rate": 3.7715264434169675e-08,
+      "loss": 0.4368,
+      "step": 21638
+    },
+    {
+      "epoch": 1.843971026842778,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.7674309829860836e-08,
+      "loss": 0.8988,
+      "step": 21639
+    },
+    {
+      "epoch": 1.844056242011078,
+      "grad_norm": 25.125,
+      "learning_rate": 3.763337713349607e-08,
+      "loss": 1.1449,
+      "step": 21640
+    },
+    {
+      "epoch": 1.844141457179378,
+      "grad_norm": 18.875,
+      "learning_rate": 3.759246634581534e-08,
+      "loss": 0.8606,
+      "step": 21641
+    },
+    {
+      "epoch": 1.844226672347678,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.7551577467558056e-08,
+      "loss": 0.8007,
+      "step": 21642
+    },
+    {
+      "epoch": 1.8443118875159779,
+      "grad_norm": 17.0,
+      "learning_rate": 3.751071049946278e-08,
+      "loss": 0.7167,
+      "step": 21643
+    },
+    {
+      "epoch": 1.8443971026842778,
+      "grad_norm": 12.25,
+      "learning_rate": 3.7469865442268384e-08,
+      "loss": 0.3641,
+      "step": 21644
+    },
+    {
+      "epoch": 1.8444823178525778,
+      "grad_norm": 19.0,
+      "learning_rate": 3.7429042296712595e-08,
+      "loss": 0.7557,
+      "step": 21645
+    },
+    {
+      "epoch": 1.8445675330208777,
+      "grad_norm": 17.875,
+      "learning_rate": 3.738824106353373e-08,
+      "loss": 0.8048,
+      "step": 21646
+    },
+    {
+      "epoch": 1.8446527481891777,
+      "grad_norm": 17.625,
+      "learning_rate": 3.73474617434684e-08,
+      "loss": 0.6834,
+      "step": 21647
+    },
+    {
+      "epoch": 1.8447379633574776,
+      "grad_norm": 14.25,
+      "learning_rate": 3.7306704337254085e-08,
+      "loss": 0.5979,
+      "step": 21648
+    },
+    {
+      "epoch": 1.8448231785257776,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.726596884562686e-08,
+      "loss": 0.535,
+      "step": 21649
+    },
+    {
+      "epoch": 1.8449083936940776,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.722525526932308e-08,
+      "loss": 0.7474,
+      "step": 21650
+    },
+    {
+      "epoch": 1.8449936088623775,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.718456360907868e-08,
+      "loss": 0.5332,
+      "step": 21651
+    },
+    {
+      "epoch": 1.8450788240306775,
+      "grad_norm": 12.875,
+      "learning_rate": 3.7143893865628636e-08,
+      "loss": 0.3849,
+      "step": 21652
+    },
+    {
+      "epoch": 1.8451640391989774,
+      "grad_norm": 13.375,
+      "learning_rate": 3.7103246039708055e-08,
+      "loss": 0.4779,
+      "step": 21653
+    },
+    {
+      "epoch": 1.8452492543672774,
+      "grad_norm": 13.6875,
+      "learning_rate": 3.706262013205161e-08,
+      "loss": 0.6061,
+      "step": 21654
+    },
+    {
+      "epoch": 1.8453344695355773,
+      "grad_norm": 21.375,
+      "learning_rate": 3.702201614339318e-08,
+      "loss": 0.6873,
+      "step": 21655
+    },
+    {
+      "epoch": 1.8454196847038773,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.698143407446675e-08,
+      "loss": 0.4483,
+      "step": 21656
+    },
+    {
+      "epoch": 1.8455048998721773,
+      "grad_norm": 16.75,
+      "learning_rate": 3.6940873926005484e-08,
+      "loss": 0.568,
+      "step": 21657
+    },
+    {
+      "epoch": 1.8455901150404772,
+      "grad_norm": 19.875,
+      "learning_rate": 3.690033569874254e-08,
+      "loss": 0.6002,
+      "step": 21658
+    },
+    {
+      "epoch": 1.8456753302087772,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.685981939341041e-08,
+      "loss": 0.3991,
+      "step": 21659
+    },
+    {
+      "epoch": 1.8457605453770771,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.681932501074112e-08,
+      "loss": 0.8265,
+      "step": 21660
+    },
+    {
+      "epoch": 1.845845760545377,
+      "grad_norm": 11.4375,
+      "learning_rate": 3.6778852551466737e-08,
+      "loss": 0.4021,
+      "step": 21661
+    },
+    {
+      "epoch": 1.845930975713677,
+      "grad_norm": 20.625,
+      "learning_rate": 3.6738402016318345e-08,
+      "loss": 0.805,
+      "step": 21662
+    },
+    {
+      "epoch": 1.846016190881977,
+      "grad_norm": 10.875,
+      "learning_rate": 3.669797340602715e-08,
+      "loss": 0.3081,
+      "step": 21663
+    },
+    {
+      "epoch": 1.846101406050277,
+      "grad_norm": 16.875,
+      "learning_rate": 3.66575667213237e-08,
+      "loss": 0.7856,
+      "step": 21664
+    },
+    {
+      "epoch": 1.846186621218577,
+      "grad_norm": 18.875,
+      "learning_rate": 3.6617181962937956e-08,
+      "loss": 0.5548,
+      "step": 21665
+    },
+    {
+      "epoch": 1.8462718363868769,
+      "grad_norm": 36.25,
+      "learning_rate": 3.6576819131600024e-08,
+      "loss": 1.0543,
+      "step": 21666
+    },
+    {
+      "epoch": 1.8463570515551768,
+      "grad_norm": 12.125,
+      "learning_rate": 3.653647822803919e-08,
+      "loss": 0.4291,
+      "step": 21667
+    },
+    {
+      "epoch": 1.8464422667234768,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.64961592529843e-08,
+      "loss": 0.6597,
+      "step": 21668
+    },
+    {
+      "epoch": 1.8465274818917767,
+      "grad_norm": 13.375,
+      "learning_rate": 3.64558622071641e-08,
+      "loss": 0.3739,
+      "step": 21669
+    },
+    {
+      "epoch": 1.8466126970600767,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.641558709130674e-08,
+      "loss": 0.6106,
+      "step": 21670
+    },
+    {
+      "epoch": 1.8466979122283766,
+      "grad_norm": 16.625,
+      "learning_rate": 3.637533390614026e-08,
+      "loss": 0.4475,
+      "step": 21671
+    },
+    {
+      "epoch": 1.8467831273966766,
+      "grad_norm": 13.875,
+      "learning_rate": 3.6335102652391844e-08,
+      "loss": 0.8005,
+      "step": 21672
+    },
+    {
+      "epoch": 1.8468683425649766,
+      "grad_norm": 16.375,
+      "learning_rate": 3.629489333078842e-08,
+      "loss": 0.6222,
+      "step": 21673
+    },
+    {
+      "epoch": 1.8469535577332765,
+      "grad_norm": 23.25,
+      "learning_rate": 3.625470594205677e-08,
+      "loss": 1.2736,
+      "step": 21674
+    },
+    {
+      "epoch": 1.8470387729015765,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.6214540486923106e-08,
+      "loss": 0.51,
+      "step": 21675
+    },
+    {
+      "epoch": 1.8471239880698764,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.617439696611338e-08,
+      "loss": 0.3949,
+      "step": 21676
+    },
+    {
+      "epoch": 1.8472092032381764,
+      "grad_norm": 17.75,
+      "learning_rate": 3.6134275380352855e-08,
+      "loss": 0.7713,
+      "step": 21677
+    },
+    {
+      "epoch": 1.8472944184064763,
+      "grad_norm": 17.875,
+      "learning_rate": 3.60941757303665e-08,
+      "loss": 0.5781,
+      "step": 21678
+    },
+    {
+      "epoch": 1.8473796335747763,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.605409801687901e-08,
+      "loss": 0.6103,
+      "step": 21679
+    },
+    {
+      "epoch": 1.8474648487430763,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.601404224061494e-08,
+      "loss": 0.7507,
+      "step": 21680
+    },
+    {
+      "epoch": 1.8475500639113762,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.597400840229789e-08,
+      "loss": 0.6495,
+      "step": 21681
+    },
+    {
+      "epoch": 1.8476352790796762,
+      "grad_norm": 15.5,
+      "learning_rate": 3.593399650265103e-08,
+      "loss": 0.3275,
+      "step": 21682
+    },
+    {
+      "epoch": 1.8477204942479761,
+      "grad_norm": 19.5,
+      "learning_rate": 3.5894006542397805e-08,
+      "loss": 0.9388,
+      "step": 21683
+    },
+    {
+      "epoch": 1.847805709416276,
+      "grad_norm": 14.625,
+      "learning_rate": 3.5854038522260964e-08,
+      "loss": 0.5404,
+      "step": 21684
+    },
+    {
+      "epoch": 1.847890924584576,
+      "grad_norm": 12.6875,
+      "learning_rate": 3.58140924429623e-08,
+      "loss": 0.4406,
+      "step": 21685
+    },
+    {
+      "epoch": 1.847976139752876,
+      "grad_norm": 19.625,
+      "learning_rate": 3.577416830522401e-08,
+      "loss": 0.4623,
+      "step": 21686
+    },
+    {
+      "epoch": 1.848061354921176,
+      "grad_norm": 13.125,
+      "learning_rate": 3.573426610976746e-08,
+      "loss": 0.3937,
+      "step": 21687
+    },
+    {
+      "epoch": 1.848146570089476,
+      "grad_norm": 11.8125,
+      "learning_rate": 3.5694385857313885e-08,
+      "loss": 0.3808,
+      "step": 21688
+    },
+    {
+      "epoch": 1.8482317852577759,
+      "grad_norm": 17.625,
+      "learning_rate": 3.565452754858381e-08,
+      "loss": 0.7616,
+      "step": 21689
+    },
+    {
+      "epoch": 1.8483170004260758,
+      "grad_norm": 11.625,
+      "learning_rate": 3.5614691184297504e-08,
+      "loss": 0.4579,
+      "step": 21690
+    },
+    {
+      "epoch": 1.8484022155943758,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.5574876765174796e-08,
+      "loss": 0.3455,
+      "step": 21691
+    },
+    {
+      "epoch": 1.8484874307626757,
+      "grad_norm": 24.875,
+      "learning_rate": 3.553508429193539e-08,
+      "loss": 0.9388,
+      "step": 21692
+    },
+    {
+      "epoch": 1.8485726459309757,
+      "grad_norm": 13.875,
+      "learning_rate": 3.54953137652983e-08,
+      "loss": 0.6455,
+      "step": 21693
+    },
+    {
+      "epoch": 1.8486578610992757,
+      "grad_norm": 16.625,
+      "learning_rate": 3.545556518598198e-08,
+      "loss": 0.3097,
+      "step": 21694
+    },
+    {
+      "epoch": 1.8487430762675756,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.5415838554705015e-08,
+      "loss": 0.547,
+      "step": 21695
+    },
+    {
+      "epoch": 1.8488282914358756,
+      "grad_norm": 16.875,
+      "learning_rate": 3.5376133872185304e-08,
+      "loss": 0.6479,
+      "step": 21696
+    },
+    {
+      "epoch": 1.8489135066041755,
+      "grad_norm": 22.125,
+      "learning_rate": 3.533645113914019e-08,
+      "loss": 0.8056,
+      "step": 21697
+    },
+    {
+      "epoch": 1.8489987217724755,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.529679035628675e-08,
+      "loss": 0.632,
+      "step": 21698
+    },
+    {
+      "epoch": 1.8490839369407754,
+      "grad_norm": 16.125,
+      "learning_rate": 3.525715152434175e-08,
+      "loss": 0.5089,
+      "step": 21699
+    },
+    {
+      "epoch": 1.8491691521090754,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.5217534644021585e-08,
+      "loss": 0.6917,
+      "step": 21700
+    },
+    {
+      "epoch": 1.8492543672773754,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.517793971604233e-08,
+      "loss": 0.4252,
+      "step": 21701
+    },
+    {
+      "epoch": 1.8493395824456753,
+      "grad_norm": 22.0,
+      "learning_rate": 3.513836674111912e-08,
+      "loss": 0.8536,
+      "step": 21702
+    },
+    {
+      "epoch": 1.8494247976139753,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.5098815719967205e-08,
+      "loss": 0.43,
+      "step": 21703
+    },
+    {
+      "epoch": 1.8495100127822752,
+      "grad_norm": 13.0625,
+      "learning_rate": 3.505928665330158e-08,
+      "loss": 0.4086,
+      "step": 21704
+    },
+    {
+      "epoch": 1.8495952279505752,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.501977954183611e-08,
+      "loss": 0.4753,
+      "step": 21705
+    },
+    {
+      "epoch": 1.8496804431188751,
+      "grad_norm": 22.75,
+      "learning_rate": 3.498029438628523e-08,
+      "loss": 0.7643,
+      "step": 21706
+    },
+    {
+      "epoch": 1.849765658287175,
+      "grad_norm": 17.625,
+      "learning_rate": 3.4940831187362104e-08,
+      "loss": 0.5079,
+      "step": 21707
+    },
+    {
+      "epoch": 1.849850873455475,
+      "grad_norm": 18.5,
+      "learning_rate": 3.4901389945779933e-08,
+      "loss": 0.6738,
+      "step": 21708
+    },
+    {
+      "epoch": 1.849936088623775,
+      "grad_norm": 22.875,
+      "learning_rate": 3.4861970662251474e-08,
+      "loss": 0.7038,
+      "step": 21709
+    },
+    {
+      "epoch": 1.850021303792075,
+      "grad_norm": 12.125,
+      "learning_rate": 3.482257333748934e-08,
+      "loss": 0.4246,
+      "step": 21710
+    },
+    {
+      "epoch": 1.850106518960375,
+      "grad_norm": 12.25,
+      "learning_rate": 3.4783197972205194e-08,
+      "loss": 0.5588,
+      "step": 21711
+    },
+    {
+      "epoch": 1.8501917341286749,
+      "grad_norm": 17.75,
+      "learning_rate": 3.4743844567110693e-08,
+      "loss": 0.6004,
+      "step": 21712
+    },
+    {
+      "epoch": 1.8502769492969748,
+      "grad_norm": 11.375,
+      "learning_rate": 3.470451312291678e-08,
+      "loss": 0.2119,
+      "step": 21713
+    },
+    {
+      "epoch": 1.8503621644652748,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.466520364033457e-08,
+      "loss": 0.5964,
+      "step": 21714
+    },
+    {
+      "epoch": 1.8504473796335748,
+      "grad_norm": 13.875,
+      "learning_rate": 3.462591612007404e-08,
+      "loss": 0.4294,
+      "step": 21715
+    },
+    {
+      "epoch": 1.8505325948018747,
+      "grad_norm": 14.125,
+      "learning_rate": 3.4586650562845464e-08,
+      "loss": 0.4852,
+      "step": 21716
+    },
+    {
+      "epoch": 1.8506178099701747,
+      "grad_norm": 14.375,
+      "learning_rate": 3.4547406969358264e-08,
+      "loss": 0.5655,
+      "step": 21717
+    },
+    {
+      "epoch": 1.8507030251384746,
+      "grad_norm": 17.5,
+      "learning_rate": 3.450818534032188e-08,
+      "loss": 0.575,
+      "step": 21718
+    },
+    {
+      "epoch": 1.8507882403067746,
+      "grad_norm": 16.875,
+      "learning_rate": 3.4468985676444635e-08,
+      "loss": 0.2654,
+      "step": 21719
+    },
+    {
+      "epoch": 1.8508734554750745,
+      "grad_norm": 23.125,
+      "learning_rate": 3.442980797843512e-08,
+      "loss": 0.528,
+      "step": 21720
+    },
+    {
+      "epoch": 1.8509586706433745,
+      "grad_norm": 21.25,
+      "learning_rate": 3.439065224700139e-08,
+      "loss": 1.1141,
+      "step": 21721
+    },
+    {
+      "epoch": 1.8510438858116744,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.435151848285093e-08,
+      "loss": 0.5612,
+      "step": 21722
+    },
+    {
+      "epoch": 1.8511291009799744,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.4312406686691086e-08,
+      "loss": 0.4325,
+      "step": 21723
+    },
+    {
+      "epoch": 1.8512143161482744,
+      "grad_norm": 13.125,
+      "learning_rate": 3.427331685922838e-08,
+      "loss": 0.5176,
+      "step": 21724
+    },
+    {
+      "epoch": 1.8512995313165743,
+      "grad_norm": 22.5,
+      "learning_rate": 3.423424900116934e-08,
+      "loss": 0.5155,
+      "step": 21725
+    },
+    {
+      "epoch": 1.8513847464848743,
+      "grad_norm": 19.125,
+      "learning_rate": 3.419520311322005e-08,
+      "loss": 0.7926,
+      "step": 21726
+    },
+    {
+      "epoch": 1.8514699616531742,
+      "grad_norm": 18.5,
+      "learning_rate": 3.4156179196086075e-08,
+      "loss": 0.5331,
+      "step": 21727
+    },
+    {
+      "epoch": 1.8515551768214742,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.4117177250472394e-08,
+      "loss": 0.6379,
+      "step": 21728
+    },
+    {
+      "epoch": 1.8516403919897741,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.407819727708414e-08,
+      "loss": 0.8452,
+      "step": 21729
+    },
+    {
+      "epoch": 1.851725607158074,
+      "grad_norm": 13.9375,
+      "learning_rate": 3.4039239276625474e-08,
+      "loss": 0.6309,
+      "step": 21730
+    },
+    {
+      "epoch": 1.851810822326374,
+      "grad_norm": 23.125,
+      "learning_rate": 3.400030324980055e-08,
+      "loss": 0.945,
+      "step": 21731
+    },
+    {
+      "epoch": 1.851896037494674,
+      "grad_norm": 19.625,
+      "learning_rate": 3.396138919731298e-08,
+      "loss": 0.6985,
+      "step": 21732
+    },
+    {
+      "epoch": 1.851981252662974,
+      "grad_norm": 18.0,
+      "learning_rate": 3.392249711986581e-08,
+      "loss": 0.6468,
+      "step": 21733
+    },
+    {
+      "epoch": 1.852066467831274,
+      "grad_norm": 16.125,
+      "learning_rate": 3.3883627018161946e-08,
+      "loss": 0.6337,
+      "step": 21734
+    },
+    {
+      "epoch": 1.852151682999574,
+      "grad_norm": 16.125,
+      "learning_rate": 3.384477889290402e-08,
+      "loss": 0.5205,
+      "step": 21735
+    },
+    {
+      "epoch": 1.8522368981678738,
+      "grad_norm": 13.125,
+      "learning_rate": 3.3805952744793704e-08,
+      "loss": 0.4538,
+      "step": 21736
+    },
+    {
+      "epoch": 1.8523221133361738,
+      "grad_norm": 17.0,
+      "learning_rate": 3.376714857453292e-08,
+      "loss": 0.5704,
+      "step": 21737
+    },
+    {
+      "epoch": 1.8524073285044738,
+      "grad_norm": 16.375,
+      "learning_rate": 3.372836638282265e-08,
+      "loss": 0.6977,
+      "step": 21738
+    },
+    {
+      "epoch": 1.8524925436727737,
+      "grad_norm": 17.75,
+      "learning_rate": 3.368960617036399e-08,
+      "loss": 0.6494,
+      "step": 21739
+    },
+    {
+      "epoch": 1.8525777588410737,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.3650867937857215e-08,
+      "loss": 0.6914,
+      "step": 21740
+    },
+    {
+      "epoch": 1.8526629740093736,
+      "grad_norm": 14.625,
+      "learning_rate": 3.361215168600232e-08,
+      "loss": 0.4342,
+      "step": 21741
+    },
+    {
+      "epoch": 1.8527481891776736,
+      "grad_norm": 14.375,
+      "learning_rate": 3.3573457415499025e-08,
+      "loss": 0.6917,
+      "step": 21742
+    },
+    {
+      "epoch": 1.8528334043459735,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.35347851270465e-08,
+      "loss": 0.5003,
+      "step": 21743
+    },
+    {
+      "epoch": 1.8529186195142735,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.3496134821343894e-08,
+      "loss": 0.5738,
+      "step": 21744
+    },
+    {
+      "epoch": 1.8530038346825735,
+      "grad_norm": 9.0625,
+      "learning_rate": 3.345750649908927e-08,
+      "loss": 0.2187,
+      "step": 21745
+    },
+    {
+      "epoch": 1.8530890498508734,
+      "grad_norm": 15.375,
+      "learning_rate": 3.341890016098082e-08,
+      "loss": 0.785,
+      "step": 21746
+    },
+    {
+      "epoch": 1.8531742650191734,
+      "grad_norm": 22.875,
+      "learning_rate": 3.338031580771617e-08,
+      "loss": 0.9838,
+      "step": 21747
+    },
+    {
+      "epoch": 1.8532594801874733,
+      "grad_norm": 11.0,
+      "learning_rate": 3.3341753439992825e-08,
+      "loss": 0.3502,
+      "step": 21748
+    },
+    {
+      "epoch": 1.8533446953557733,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.3303213058507314e-08,
+      "loss": 0.9607,
+      "step": 21749
+    },
+    {
+      "epoch": 1.8534299105240732,
+      "grad_norm": 14.625,
+      "learning_rate": 3.3264694663956155e-08,
+      "loss": 0.5956,
+      "step": 21750
+    },
+    {
+      "epoch": 1.8535151256923732,
+      "grad_norm": 14.625,
+      "learning_rate": 3.322619825703546e-08,
+      "loss": 0.6904,
+      "step": 21751
+    },
+    {
+      "epoch": 1.8536003408606732,
+      "grad_norm": 10.375,
+      "learning_rate": 3.318772383844107e-08,
+      "loss": 0.2875,
+      "step": 21752
+    },
+    {
+      "epoch": 1.8536855560289731,
+      "grad_norm": 15.0,
+      "learning_rate": 3.3149271408868116e-08,
+      "loss": 0.7884,
+      "step": 21753
+    },
+    {
+      "epoch": 1.853770771197273,
+      "grad_norm": 15.0625,
+      "learning_rate": 3.311084096901132e-08,
+      "loss": 0.6617,
+      "step": 21754
+    },
+    {
+      "epoch": 1.853855986365573,
+      "grad_norm": 19.625,
+      "learning_rate": 3.307243251956541e-08,
+      "loss": 1.328,
+      "step": 21755
+    },
+    {
+      "epoch": 1.853941201533873,
+      "grad_norm": 17.25,
+      "learning_rate": 3.303404606122426e-08,
+      "loss": 0.6941,
+      "step": 21756
+    },
+    {
+      "epoch": 1.854026416702173,
+      "grad_norm": 20.0,
+      "learning_rate": 3.2995681594681775e-08,
+      "loss": 0.6445,
+      "step": 21757
+    },
+    {
+      "epoch": 1.854111631870473,
+      "grad_norm": 23.25,
+      "learning_rate": 3.295733912063101e-08,
+      "loss": 0.8694,
+      "step": 21758
+    },
+    {
+      "epoch": 1.8541968470387729,
+      "grad_norm": 12.3125,
+      "learning_rate": 3.2919018639765015e-08,
+      "loss": 0.4179,
+      "step": 21759
+    },
+    {
+      "epoch": 1.8542820622070728,
+      "grad_norm": 15.625,
+      "learning_rate": 3.2880720152776294e-08,
+      "loss": 0.4169,
+      "step": 21760
+    },
+    {
+      "epoch": 1.8543672773753728,
+      "grad_norm": 19.875,
+      "learning_rate": 3.28424436603568e-08,
+      "loss": 0.7681,
+      "step": 21761
+    },
+    {
+      "epoch": 1.8544524925436727,
+      "grad_norm": 18.0,
+      "learning_rate": 3.280418916319833e-08,
+      "loss": 0.5571,
+      "step": 21762
+    },
+    {
+      "epoch": 1.8545377077119727,
+      "grad_norm": 16.5,
+      "learning_rate": 3.2765956661992005e-08,
+      "loss": 0.7276,
+      "step": 21763
+    },
+    {
+      "epoch": 1.8546229228802726,
+      "grad_norm": 20.875,
+      "learning_rate": 3.2727746157428934e-08,
+      "loss": 0.7449,
+      "step": 21764
+    },
+    {
+      "epoch": 1.8547081380485726,
+      "grad_norm": 24.25,
+      "learning_rate": 3.268955765019968e-08,
+      "loss": 1.0629,
+      "step": 21765
+    },
+    {
+      "epoch": 1.8547933532168726,
+      "grad_norm": 26.875,
+      "learning_rate": 3.265139114099397e-08,
+      "loss": 0.667,
+      "step": 21766
+    },
+    {
+      "epoch": 1.8548785683851725,
+      "grad_norm": 16.375,
+      "learning_rate": 3.2613246630501946e-08,
+      "loss": 0.451,
+      "step": 21767
+    },
+    {
+      "epoch": 1.8549637835534725,
+      "grad_norm": 17.25,
+      "learning_rate": 3.257512411941263e-08,
+      "loss": 0.8483,
+      "step": 21768
+    },
+    {
+      "epoch": 1.8550489987217724,
+      "grad_norm": 15.4375,
+      "learning_rate": 3.253702360841521e-08,
+      "loss": 0.6194,
+      "step": 21769
+    },
+    {
+      "epoch": 1.8551342138900724,
+      "grad_norm": 16.875,
+      "learning_rate": 3.249894509819801e-08,
+      "loss": 0.8664,
+      "step": 21770
+    },
+    {
+      "epoch": 1.8552194290583723,
+      "grad_norm": 16.625,
+      "learning_rate": 3.246088858944896e-08,
+      "loss": 0.6467,
+      "step": 21771
+    },
+    {
+      "epoch": 1.8553046442266723,
+      "grad_norm": 16.75,
+      "learning_rate": 3.242285408285611e-08,
+      "loss": 0.4776,
+      "step": 21772
+    },
+    {
+      "epoch": 1.8553898593949723,
+      "grad_norm": 17.375,
+      "learning_rate": 3.238484157910657e-08,
+      "loss": 1.0107,
+      "step": 21773
+    },
+    {
+      "epoch": 1.8554750745632722,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.234685107888755e-08,
+      "loss": 0.6366,
+      "step": 21774
+    },
+    {
+      "epoch": 1.8555602897315722,
+      "grad_norm": 20.0,
+      "learning_rate": 3.230888258288517e-08,
+      "loss": 0.9838,
+      "step": 21775
+    },
+    {
+      "epoch": 1.8556455048998721,
+      "grad_norm": 12.75,
+      "learning_rate": 3.227093609178583e-08,
+      "loss": 0.2061,
+      "step": 21776
+    },
+    {
+      "epoch": 1.855730720068172,
+      "grad_norm": 13.125,
+      "learning_rate": 3.2233011606275236e-08,
+      "loss": 0.5362,
+      "step": 21777
+    },
+    {
+      "epoch": 1.855815935236472,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.21951091270388e-08,
+      "loss": 0.5551,
+      "step": 21778
+    },
+    {
+      "epoch": 1.855901150404772,
+      "grad_norm": 18.5,
+      "learning_rate": 3.215722865476112e-08,
+      "loss": 0.5603,
+      "step": 21779
+    },
+    {
+      "epoch": 1.855986365573072,
+      "grad_norm": 13.625,
+      "learning_rate": 3.2119370190127064e-08,
+      "loss": 0.625,
+      "step": 21780
+    },
+    {
+      "epoch": 1.856071580741372,
+      "grad_norm": 17.25,
+      "learning_rate": 3.208153373382053e-08,
+      "loss": 0.5408,
+      "step": 21781
+    },
+    {
+      "epoch": 1.8561567959096719,
+      "grad_norm": 17.125,
+      "learning_rate": 3.2043719286525556e-08,
+      "loss": 0.6354,
+      "step": 21782
+    },
+    {
+      "epoch": 1.8562420110779718,
+      "grad_norm": 13.375,
+      "learning_rate": 3.20059268489252e-08,
+      "loss": 0.503,
+      "step": 21783
+    },
+    {
+      "epoch": 1.8563272262462718,
+      "grad_norm": 13.625,
+      "learning_rate": 3.1968156421702535e-08,
+      "loss": 0.485,
+      "step": 21784
+    },
+    {
+      "epoch": 1.8564124414145717,
+      "grad_norm": 14.125,
+      "learning_rate": 3.1930408005540204e-08,
+      "loss": 0.4398,
+      "step": 21785
+    },
+    {
+      "epoch": 1.8564976565828717,
+      "grad_norm": 17.5,
+      "learning_rate": 3.18926816011203e-08,
+      "loss": 0.5902,
+      "step": 21786
+    },
+    {
+      "epoch": 1.8565828717511716,
+      "grad_norm": 22.375,
+      "learning_rate": 3.1854977209124506e-08,
+      "loss": 0.6216,
+      "step": 21787
+    },
+    {
+      "epoch": 1.8566680869194716,
+      "grad_norm": 13.75,
+      "learning_rate": 3.1817294830234074e-08,
+      "loss": 0.3478,
+      "step": 21788
+    },
+    {
+      "epoch": 1.8567533020877716,
+      "grad_norm": 12.375,
+      "learning_rate": 3.177963446513027e-08,
+      "loss": 0.4178,
+      "step": 21789
+    },
+    {
+      "epoch": 1.8568385172560715,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.174199611449338e-08,
+      "loss": 0.4947,
+      "step": 21790
+    },
+    {
+      "epoch": 1.8569237324243715,
+      "grad_norm": 16.125,
+      "learning_rate": 3.1704379779003703e-08,
+      "loss": 0.6826,
+      "step": 21791
+    },
+    {
+      "epoch": 1.8570089475926714,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.166678545934096e-08,
+      "loss": 0.6584,
+      "step": 21792
+    },
+    {
+      "epoch": 1.8570941627609714,
+      "grad_norm": 16.625,
+      "learning_rate": 3.1629213156184475e-08,
+      "loss": 0.5517,
+      "step": 21793
+    },
+    {
+      "epoch": 1.8571793779292713,
+      "grad_norm": 15.875,
+      "learning_rate": 3.159166287021343e-08,
+      "loss": 0.6834,
+      "step": 21794
+    },
+    {
+      "epoch": 1.8572645930975713,
+      "grad_norm": 16.625,
+      "learning_rate": 3.15541346021063e-08,
+      "loss": 0.6807,
+      "step": 21795
+    },
+    {
+      "epoch": 1.8573498082658713,
+      "grad_norm": 18.625,
+      "learning_rate": 3.151662835254102e-08,
+      "loss": 0.8298,
+      "step": 21796
+    },
+    {
+      "epoch": 1.8574350234341712,
+      "grad_norm": 37.5,
+      "learning_rate": 3.147914412219552e-08,
+      "loss": 0.6353,
+      "step": 21797
+    },
+    {
+      "epoch": 1.8575202386024712,
+      "grad_norm": 15.875,
+      "learning_rate": 3.1441681911747315e-08,
+      "loss": 0.6473,
+      "step": 21798
+    },
+    {
+      "epoch": 1.8576054537707711,
+      "grad_norm": 23.375,
+      "learning_rate": 3.140424172187323e-08,
+      "loss": 0.6477,
+      "step": 21799
+    },
+    {
+      "epoch": 1.857690668939071,
+      "grad_norm": 11.6875,
+      "learning_rate": 3.136682355324994e-08,
+      "loss": 0.5029,
+      "step": 21800
+    },
+    {
+      "epoch": 1.857775884107371,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.1329427406553424e-08,
+      "loss": 0.7623,
+      "step": 21801
+    },
+    {
+      "epoch": 1.857861099275671,
+      "grad_norm": 12.0,
+      "learning_rate": 3.129205328245982e-08,
+      "loss": 0.3784,
+      "step": 21802
+    },
+    {
+      "epoch": 1.857946314443971,
+      "grad_norm": 15.8125,
+      "learning_rate": 3.125470118164414e-08,
+      "loss": 0.8258,
+      "step": 21803
+    },
+    {
+      "epoch": 1.858031529612271,
+      "grad_norm": 16.75,
+      "learning_rate": 3.121737110478182e-08,
+      "loss": 0.8034,
+      "step": 21804
+    },
+    {
+      "epoch": 1.8581167447805709,
+      "grad_norm": 15.875,
+      "learning_rate": 3.1180063052546895e-08,
+      "loss": 0.3096,
+      "step": 21805
+    },
+    {
+      "epoch": 1.8582019599488708,
+      "grad_norm": 22.25,
+      "learning_rate": 3.1142777025613967e-08,
+      "loss": 0.7881,
+      "step": 21806
+    },
+    {
+      "epoch": 1.8582871751171708,
+      "grad_norm": 16.5,
+      "learning_rate": 3.1105513024656806e-08,
+      "loss": 0.7035,
+      "step": 21807
+    },
+    {
+      "epoch": 1.8583723902854707,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.1068271050348486e-08,
+      "loss": 0.8079,
+      "step": 21808
+    },
+    {
+      "epoch": 1.8584576054537707,
+      "grad_norm": 14.375,
+      "learning_rate": 3.1031051103362353e-08,
+      "loss": 0.7055,
+      "step": 21809
+    },
+    {
+      "epoch": 1.8585428206220707,
+      "grad_norm": 23.75,
+      "learning_rate": 3.099385318437106e-08,
+      "loss": 1.0617,
+      "step": 21810
+    },
+    {
+      "epoch": 1.8586280357903706,
+      "grad_norm": 22.375,
+      "learning_rate": 3.0956677294046446e-08,
+      "loss": 0.6819,
+      "step": 21811
+    },
+    {
+      "epoch": 1.8587132509586706,
+      "grad_norm": 12.5625,
+      "learning_rate": 3.0919523433060594e-08,
+      "loss": 0.525,
+      "step": 21812
+    },
+    {
+      "epoch": 1.8587984661269705,
+      "grad_norm": 19.125,
+      "learning_rate": 3.0882391602084784e-08,
+      "loss": 0.6218,
+      "step": 21813
+    },
+    {
+      "epoch": 1.8588836812952705,
+      "grad_norm": 11.3125,
+      "learning_rate": 3.084528180179014e-08,
+      "loss": 0.3098,
+      "step": 21814
+    },
+    {
+      "epoch": 1.8589688964635704,
+      "grad_norm": 20.5,
+      "learning_rate": 3.08081940328471e-08,
+      "loss": 0.5908,
+      "step": 21815
+    },
+    {
+      "epoch": 1.8590541116318704,
+      "grad_norm": 15.0,
+      "learning_rate": 3.07711282959261e-08,
+      "loss": 0.5866,
+      "step": 21816
+    },
+    {
+      "epoch": 1.8591393268001704,
+      "grad_norm": 17.0,
+      "learning_rate": 3.073408459169675e-08,
+      "loss": 0.5662,
+      "step": 21817
+    },
+    {
+      "epoch": 1.8592245419684703,
+      "grad_norm": 21.5,
+      "learning_rate": 3.0697062920828635e-08,
+      "loss": 0.6827,
+      "step": 21818
+    },
+    {
+      "epoch": 1.8593097571367703,
+      "grad_norm": 11.25,
+      "learning_rate": 3.066006328399068e-08,
+      "loss": 0.338,
+      "step": 21819
+    },
+    {
+      "epoch": 1.8593949723050702,
+      "grad_norm": 18.375,
+      "learning_rate": 3.0623085681851517e-08,
+      "loss": 0.7246,
+      "step": 21820
+    },
+    {
+      "epoch": 1.8594801874733702,
+      "grad_norm": 23.125,
+      "learning_rate": 3.0586130115079354e-08,
+      "loss": 0.5349,
+      "step": 21821
+    },
+    {
+      "epoch": 1.8595654026416701,
+      "grad_norm": 23.25,
+      "learning_rate": 3.054919658434199e-08,
+      "loss": 0.6655,
+      "step": 21822
+    },
+    {
+      "epoch": 1.85965061780997,
+      "grad_norm": 22.125,
+      "learning_rate": 3.051228509030696e-08,
+      "loss": 0.7001,
+      "step": 21823
+    },
+    {
+      "epoch": 1.85973583297827,
+      "grad_norm": 19.5,
+      "learning_rate": 3.047539563364135e-08,
+      "loss": 1.0705,
+      "step": 21824
+    },
+    {
+      "epoch": 1.85982104814657,
+      "grad_norm": 14.25,
+      "learning_rate": 3.0438528215011446e-08,
+      "loss": 0.4888,
+      "step": 21825
+    },
+    {
+      "epoch": 1.85990626331487,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.0401682835083934e-08,
+      "loss": 0.6391,
+      "step": 21826
+    },
+    {
+      "epoch": 1.85999147848317,
+      "grad_norm": 17.375,
+      "learning_rate": 3.036485949452425e-08,
+      "loss": 0.7349,
+      "step": 21827
+    },
+    {
+      "epoch": 1.8600766936514699,
+      "grad_norm": 14.625,
+      "learning_rate": 3.032805819399798e-08,
+      "loss": 0.5132,
+      "step": 21828
+    },
+    {
+      "epoch": 1.8601619088197698,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.0291278934170276e-08,
+      "loss": 0.4225,
+      "step": 21829
+    },
+    {
+      "epoch": 1.8602471239880698,
+      "grad_norm": 17.625,
+      "learning_rate": 3.0254521715705476e-08,
+      "loss": 0.5203,
+      "step": 21830
+    },
+    {
+      "epoch": 1.8603323391563698,
+      "grad_norm": 14.875,
+      "learning_rate": 3.021778653926818e-08,
+      "loss": 0.5024,
+      "step": 21831
+    },
+    {
+      "epoch": 1.8604175543246697,
+      "grad_norm": 25.625,
+      "learning_rate": 3.01810734055219e-08,
+      "loss": 0.9732,
+      "step": 21832
+    },
+    {
+      "epoch": 1.8605027694929697,
+      "grad_norm": 20.125,
+      "learning_rate": 3.014438231513053e-08,
+      "loss": 0.4581,
+      "step": 21833
+    },
+    {
+      "epoch": 1.8605879846612696,
+      "grad_norm": 16.75,
+      "learning_rate": 3.010771326875675e-08,
+      "loss": 0.5764,
+      "step": 21834
+    },
+    {
+      "epoch": 1.8606731998295696,
+      "grad_norm": 17.875,
+      "learning_rate": 3.0071066267063215e-08,
+      "loss": 0.6923,
+      "step": 21835
+    },
+    {
+      "epoch": 1.8607584149978695,
+      "grad_norm": 18.75,
+      "learning_rate": 3.0034441310712174e-08,
+      "loss": 0.629,
+      "step": 21836
+    },
+    {
+      "epoch": 1.8608436301661695,
+      "grad_norm": 21.75,
+      "learning_rate": 2.999783840036574e-08,
+      "loss": 0.8463,
+      "step": 21837
+    },
+    {
+      "epoch": 1.8609288453344695,
+      "grad_norm": 37.25,
+      "learning_rate": 2.996125753668505e-08,
+      "loss": 0.7086,
+      "step": 21838
+    },
+    {
+      "epoch": 1.8610140605027694,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.992469872033124e-08,
+      "loss": 0.3883,
+      "step": 21839
+    },
+    {
+      "epoch": 1.8610992756710694,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.9888161951965034e-08,
+      "loss": 0.3862,
+      "step": 21840
+    },
+    {
+      "epoch": 1.8611844908393693,
+      "grad_norm": 16.5,
+      "learning_rate": 2.985164723224673e-08,
+      "loss": 0.6201,
+      "step": 21841
+    },
+    {
+      "epoch": 1.8612697060076693,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.981515456183609e-08,
+      "loss": 0.3361,
+      "step": 21842
+    },
+    {
+      "epoch": 1.8613549211759692,
+      "grad_norm": 21.5,
+      "learning_rate": 2.9778683941392576e-08,
+      "loss": 0.6424,
+      "step": 21843
+    },
+    {
+      "epoch": 1.8614401363442692,
+      "grad_norm": 19.5,
+      "learning_rate": 2.974223537157525e-08,
+      "loss": 0.9167,
+      "step": 21844
+    },
+    {
+      "epoch": 1.8615253515125691,
+      "grad_norm": 13.625,
+      "learning_rate": 2.9705808853042752e-08,
+      "loss": 0.5752,
+      "step": 21845
+    },
+    {
+      "epoch": 1.861610566680869,
+      "grad_norm": 12.5,
+      "learning_rate": 2.9669404386453438e-08,
+      "loss": 0.472,
+      "step": 21846
+    },
+    {
+      "epoch": 1.861695781849169,
+      "grad_norm": 16.625,
+      "learning_rate": 2.963302197246512e-08,
+      "loss": 0.7601,
+      "step": 21847
+    },
+    {
+      "epoch": 1.861780997017469,
+      "grad_norm": 16.875,
+      "learning_rate": 2.959666161173533e-08,
+      "loss": 0.5139,
+      "step": 21848
+    },
+    {
+      "epoch": 1.861866212185769,
+      "grad_norm": 16.25,
+      "learning_rate": 2.9560323304920898e-08,
+      "loss": 0.7937,
+      "step": 21849
+    },
+    {
+      "epoch": 1.861951427354069,
+      "grad_norm": 13.375,
+      "learning_rate": 2.9524007052678937e-08,
+      "loss": 0.6047,
+      "step": 21850
+    },
+    {
+      "epoch": 1.862036642522369,
+      "grad_norm": 14.125,
+      "learning_rate": 2.9487712855665318e-08,
+      "loss": 0.3327,
+      "step": 21851
+    },
+    {
+      "epoch": 1.8621218576906688,
+      "grad_norm": 63.75,
+      "learning_rate": 2.9451440714536038e-08,
+      "loss": 0.2523,
+      "step": 21852
+    },
+    {
+      "epoch": 1.8622070728589688,
+      "grad_norm": 18.375,
+      "learning_rate": 2.941519062994655e-08,
+      "loss": 0.7093,
+      "step": 21853
+    },
+    {
+      "epoch": 1.8622922880272688,
+      "grad_norm": 18.125,
+      "learning_rate": 2.937896260255216e-08,
+      "loss": 0.5175,
+      "step": 21854
+    },
+    {
+      "epoch": 1.8623775031955687,
+      "grad_norm": 18.875,
+      "learning_rate": 2.9342756633007063e-08,
+      "loss": 0.6343,
+      "step": 21855
+    },
+    {
+      "epoch": 1.8624627183638687,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.930657272196588e-08,
+      "loss": 0.6965,
+      "step": 21856
+    },
+    {
+      "epoch": 1.8625479335321686,
+      "grad_norm": 28.0,
+      "learning_rate": 2.9270410870082532e-08,
+      "loss": 0.7711,
+      "step": 21857
+    },
+    {
+      "epoch": 1.8626331487004686,
+      "grad_norm": 12.75,
+      "learning_rate": 2.9234271078010384e-08,
+      "loss": 0.3663,
+      "step": 21858
+    },
+    {
+      "epoch": 1.8627183638687685,
+      "grad_norm": 13.0,
+      "learning_rate": 2.9198153346402664e-08,
+      "loss": 0.3477,
+      "step": 21859
+    },
+    {
+      "epoch": 1.8628035790370685,
+      "grad_norm": 22.5,
+      "learning_rate": 2.9162057675911627e-08,
+      "loss": 0.982,
+      "step": 21860
+    },
+    {
+      "epoch": 1.8628887942053685,
+      "grad_norm": 19.25,
+      "learning_rate": 2.9125984067189943e-08,
+      "loss": 0.6279,
+      "step": 21861
+    },
+    {
+      "epoch": 1.8629740093736684,
+      "grad_norm": 22.25,
+      "learning_rate": 2.9089932520889458e-08,
+      "loss": 1.2789,
+      "step": 21862
+    },
+    {
+      "epoch": 1.8630592245419684,
+      "grad_norm": 22.625,
+      "learning_rate": 2.905390303766173e-08,
+      "loss": 0.5285,
+      "step": 21863
+    },
+    {
+      "epoch": 1.8631444397102683,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.9017895618157626e-08,
+      "loss": 0.5708,
+      "step": 21864
+    },
+    {
+      "epoch": 1.8632296548785683,
+      "grad_norm": 20.625,
+      "learning_rate": 2.898191026302788e-08,
+      "loss": 0.8125,
+      "step": 21865
+    },
+    {
+      "epoch": 1.8633148700468682,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.89459469729228e-08,
+      "loss": 0.5551,
+      "step": 21866
+    },
+    {
+      "epoch": 1.8634000852151682,
+      "grad_norm": 16.375,
+      "learning_rate": 2.8910005748492566e-08,
+      "loss": 0.4955,
+      "step": 21867
+    },
+    {
+      "epoch": 1.8634853003834682,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.8874086590386234e-08,
+      "loss": 0.5985,
+      "step": 21868
+    },
+    {
+      "epoch": 1.8635705155517681,
+      "grad_norm": 19.375,
+      "learning_rate": 2.8838189499253154e-08,
+      "loss": 0.8152,
+      "step": 21869
+    },
+    {
+      "epoch": 1.863655730720068,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.880231447574197e-08,
+      "loss": 0.4431,
+      "step": 21870
+    },
+    {
+      "epoch": 1.863740945888368,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.8766461520500915e-08,
+      "loss": 0.4003,
+      "step": 21871
+    },
+    {
+      "epoch": 1.863826161056668,
+      "grad_norm": 23.125,
+      "learning_rate": 2.873063063417794e-08,
+      "loss": 0.6986,
+      "step": 21872
+    },
+    {
+      "epoch": 1.863911376224968,
+      "grad_norm": 16.25,
+      "learning_rate": 2.869482181742059e-08,
+      "loss": 0.6462,
+      "step": 21873
+    },
+    {
+      "epoch": 1.863996591393268,
+      "grad_norm": 12.0,
+      "learning_rate": 2.8659035070875975e-08,
+      "loss": 0.3209,
+      "step": 21874
+    },
+    {
+      "epoch": 1.8640818065615679,
+      "grad_norm": 17.5,
+      "learning_rate": 2.8623270395190667e-08,
+      "loss": 0.429,
+      "step": 21875
+    },
+    {
+      "epoch": 1.8641670217298678,
+      "grad_norm": 12.75,
+      "learning_rate": 2.8587527791011233e-08,
+      "loss": 0.4146,
+      "step": 21876
+    },
+    {
+      "epoch": 1.8642522368981678,
+      "grad_norm": 24.75,
+      "learning_rate": 2.8551807258983266e-08,
+      "loss": 0.6662,
+      "step": 21877
+    },
+    {
+      "epoch": 1.8643374520664677,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.8516108799752362e-08,
+      "loss": 0.5689,
+      "step": 21878
+    },
+    {
+      "epoch": 1.8644226672347677,
+      "grad_norm": 24.75,
+      "learning_rate": 2.848043241396356e-08,
+      "loss": 0.5794,
+      "step": 21879
+    },
+    {
+      "epoch": 1.8645078824030676,
+      "grad_norm": 21.75,
+      "learning_rate": 2.84447781022619e-08,
+      "loss": 0.7863,
+      "step": 21880
+    },
+    {
+      "epoch": 1.8645930975713676,
+      "grad_norm": 15.5,
+      "learning_rate": 2.840914586529145e-08,
+      "loss": 0.6148,
+      "step": 21881
+    },
+    {
+      "epoch": 1.8646783127396676,
+      "grad_norm": 16.875,
+      "learning_rate": 2.8373535703695998e-08,
+      "loss": 0.6959,
+      "step": 21882
+    },
+    {
+      "epoch": 1.8647635279079675,
+      "grad_norm": 23.25,
+      "learning_rate": 2.8337947618119337e-08,
+      "loss": 0.5256,
+      "step": 21883
+    },
+    {
+      "epoch": 1.8648487430762675,
+      "grad_norm": 36.0,
+      "learning_rate": 2.8302381609204426e-08,
+      "loss": 0.9293,
+      "step": 21884
+    },
+    {
+      "epoch": 1.8649339582445674,
+      "grad_norm": 18.625,
+      "learning_rate": 2.8266837677593944e-08,
+      "loss": 0.7706,
+      "step": 21885
+    },
+    {
+      "epoch": 1.8650191734128674,
+      "grad_norm": 19.875,
+      "learning_rate": 2.8231315823930154e-08,
+      "loss": 0.5775,
+      "step": 21886
+    },
+    {
+      "epoch": 1.8651043885811673,
+      "grad_norm": 20.5,
+      "learning_rate": 2.8195816048855184e-08,
+      "loss": 0.6981,
+      "step": 21887
+    },
+    {
+      "epoch": 1.8651896037494673,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.8160338353010464e-08,
+      "loss": 0.6661,
+      "step": 21888
+    },
+    {
+      "epoch": 1.8652748189177673,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.8124882737037008e-08,
+      "loss": 0.5212,
+      "step": 21889
+    },
+    {
+      "epoch": 1.8653600340860672,
+      "grad_norm": 12.125,
+      "learning_rate": 2.8089449201575553e-08,
+      "loss": 0.3896,
+      "step": 21890
+    },
+    {
+      "epoch": 1.8654452492543672,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.80540377472667e-08,
+      "loss": 0.6503,
+      "step": 21891
+    },
+    {
+      "epoch": 1.8655304644226671,
+      "grad_norm": 22.375,
+      "learning_rate": 2.8018648374749936e-08,
+      "loss": 0.5124,
+      "step": 21892
+    },
+    {
+      "epoch": 1.865615679590967,
+      "grad_norm": 12.625,
+      "learning_rate": 2.7983281084665165e-08,
+      "loss": 0.7948,
+      "step": 21893
+    },
+    {
+      "epoch": 1.865700894759267,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.7947935877651182e-08,
+      "loss": 0.2206,
+      "step": 21894
+    },
+    {
+      "epoch": 1.865786109927567,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.7912612754346925e-08,
+      "loss": 0.3345,
+      "step": 21895
+    },
+    {
+      "epoch": 1.865871325095867,
+      "grad_norm": 13.875,
+      "learning_rate": 2.7877311715390626e-08,
+      "loss": 0.4532,
+      "step": 21896
+    },
+    {
+      "epoch": 1.865956540264167,
+      "grad_norm": 24.625,
+      "learning_rate": 2.784203276142039e-08,
+      "loss": 0.5684,
+      "step": 21897
+    },
+    {
+      "epoch": 1.8660417554324669,
+      "grad_norm": 19.875,
+      "learning_rate": 2.7806775893073347e-08,
+      "loss": 0.857,
+      "step": 21898
+    },
+    {
+      "epoch": 1.8661269706007668,
+      "grad_norm": 16.125,
+      "learning_rate": 2.777154111098718e-08,
+      "loss": 0.7175,
+      "step": 21899
+    },
+    {
+      "epoch": 1.8662121857690668,
+      "grad_norm": 18.0,
+      "learning_rate": 2.773632841579804e-08,
+      "loss": 0.5918,
+      "step": 21900
+    },
+    {
+      "epoch": 1.8662974009373667,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.7701137808142652e-08,
+      "loss": 0.8391,
+      "step": 21901
+    },
+    {
+      "epoch": 1.8663826161056667,
+      "grad_norm": 15.0,
+      "learning_rate": 2.7665969288656753e-08,
+      "loss": 0.6167,
+      "step": 21902
+    },
+    {
+      "epoch": 1.8664678312739666,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.7630822857975942e-08,
+      "loss": 0.3063,
+      "step": 21903
+    },
+    {
+      "epoch": 1.8665530464422666,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.7595698516735408e-08,
+      "loss": 0.3097,
+      "step": 21904
+    },
+    {
+      "epoch": 1.8666382616105666,
+      "grad_norm": 19.25,
+      "learning_rate": 2.7560596265569924e-08,
+      "loss": 0.2733,
+      "step": 21905
+    },
+    {
+      "epoch": 1.8667234767788665,
+      "grad_norm": 24.125,
+      "learning_rate": 2.752551610511356e-08,
+      "loss": 0.6206,
+      "step": 21906
+    },
+    {
+      "epoch": 1.8668086919471665,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.7490458036000535e-08,
+      "loss": 0.5084,
+      "step": 21907
+    },
+    {
+      "epoch": 1.8668939071154664,
+      "grad_norm": 25.125,
+      "learning_rate": 2.745542205886409e-08,
+      "loss": 0.7136,
+      "step": 21908
+    },
+    {
+      "epoch": 1.8669791222837664,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.742040817433761e-08,
+      "loss": 0.4616,
+      "step": 21909
+    },
+    {
+      "epoch": 1.8670643374520663,
+      "grad_norm": 14.5,
+      "learning_rate": 2.738541638305392e-08,
+      "loss": 0.391,
+      "step": 21910
+    },
+    {
+      "epoch": 1.8671495526203663,
+      "grad_norm": 23.0,
+      "learning_rate": 2.7350446685645017e-08,
+      "loss": 0.8314,
+      "step": 21911
+    },
+    {
+      "epoch": 1.8672347677886663,
+      "grad_norm": 18.125,
+      "learning_rate": 2.731549908274289e-08,
+      "loss": 0.8527,
+      "step": 21912
+    },
+    {
+      "epoch": 1.8673199829569662,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.7280573574979402e-08,
+      "loss": 0.5991,
+      "step": 21913
+    },
+    {
+      "epoch": 1.8674051981252662,
+      "grad_norm": 13.125,
+      "learning_rate": 2.7245670162985433e-08,
+      "loss": 0.3721,
+      "step": 21914
+    },
+    {
+      "epoch": 1.8674904132935661,
+      "grad_norm": 17.375,
+      "learning_rate": 2.7210788847391867e-08,
+      "loss": 0.6904,
+      "step": 21915
+    },
+    {
+      "epoch": 1.867575628461866,
+      "grad_norm": 17.625,
+      "learning_rate": 2.7175929628828894e-08,
+      "loss": 0.522,
+      "step": 21916
+    },
+    {
+      "epoch": 1.867660843630166,
+      "grad_norm": 15.875,
+      "learning_rate": 2.714109250792643e-08,
+      "loss": 0.6886,
+      "step": 21917
+    },
+    {
+      "epoch": 1.867746058798466,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.7106277485314247e-08,
+      "loss": 0.3627,
+      "step": 21918
+    },
+    {
+      "epoch": 1.867831273966766,
+      "grad_norm": 23.5,
+      "learning_rate": 2.7071484561621146e-08,
+      "loss": 0.6687,
+      "step": 21919
+    },
+    {
+      "epoch": 1.867916489135066,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.703671373747621e-08,
+      "loss": 0.6993,
+      "step": 21920
+    },
+    {
+      "epoch": 1.8680017043033659,
+      "grad_norm": 18.125,
+      "learning_rate": 2.7001965013507546e-08,
+      "loss": 0.7583,
+      "step": 21921
+    },
+    {
+      "epoch": 1.8680869194716658,
+      "grad_norm": 14.0,
+      "learning_rate": 2.6967238390343403e-08,
+      "loss": 0.6386,
+      "step": 21922
+    },
+    {
+      "epoch": 1.8681721346399658,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.6932533868610917e-08,
+      "loss": 0.3819,
+      "step": 21923
+    },
+    {
+      "epoch": 1.8682573498082657,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.68978514489375e-08,
+      "loss": 0.5252,
+      "step": 21924
+    },
+    {
+      "epoch": 1.8683425649765657,
+      "grad_norm": 22.5,
+      "learning_rate": 2.6863191131949878e-08,
+      "loss": 0.9955,
+      "step": 21925
+    },
+    {
+      "epoch": 1.8684277801448657,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.6828552918274216e-08,
+      "loss": 0.3316,
+      "step": 21926
+    },
+    {
+      "epoch": 1.8685129953131656,
+      "grad_norm": 19.25,
+      "learning_rate": 2.6793936808536813e-08,
+      "loss": 0.6341,
+      "step": 21927
+    },
+    {
+      "epoch": 1.8685982104814656,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.6759342803362732e-08,
+      "loss": 0.4685,
+      "step": 21928
+    },
+    {
+      "epoch": 1.8686834256497655,
+      "grad_norm": 17.0,
+      "learning_rate": 2.6724770903377578e-08,
+      "loss": 0.8278,
+      "step": 21929
+    },
+    {
+      "epoch": 1.8687686408180655,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.6690221109205854e-08,
+      "loss": 0.5974,
+      "step": 21930
+    },
+    {
+      "epoch": 1.8688538559863654,
+      "grad_norm": 20.25,
+      "learning_rate": 2.6655693421472062e-08,
+      "loss": 0.7381,
+      "step": 21931
+    },
+    {
+      "epoch": 1.8689390711546654,
+      "grad_norm": 14.5,
+      "learning_rate": 2.6621187840800006e-08,
+      "loss": 0.4029,
+      "step": 21932
+    },
+    {
+      "epoch": 1.8690242863229654,
+      "grad_norm": 15.375,
+      "learning_rate": 2.6586704367813215e-08,
+      "loss": 0.4434,
+      "step": 21933
+    },
+    {
+      "epoch": 1.8691095014912653,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.655224300313494e-08,
+      "loss": 0.3934,
+      "step": 21934
+    },
+    {
+      "epoch": 1.8691947166595653,
+      "grad_norm": 21.75,
+      "learning_rate": 2.6517803747387882e-08,
+      "loss": 0.956,
+      "step": 21935
+    },
+    {
+      "epoch": 1.8692799318278652,
+      "grad_norm": 14.5,
+      "learning_rate": 2.6483386601194456e-08,
+      "loss": 0.4541,
+      "step": 21936
+    },
+    {
+      "epoch": 1.8693651469961652,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.6448991565176525e-08,
+      "loss": 0.4694,
+      "step": 21937
+    },
+    {
+      "epoch": 1.8694503621644651,
+      "grad_norm": 16.5,
+      "learning_rate": 2.6414618639955675e-08,
+      "loss": 0.4535,
+      "step": 21938
+    },
+    {
+      "epoch": 1.869535577332765,
+      "grad_norm": 12.375,
+      "learning_rate": 2.6380267826153216e-08,
+      "loss": 0.4347,
+      "step": 21939
+    },
+    {
+      "epoch": 1.869620792501065,
+      "grad_norm": 14.25,
+      "learning_rate": 2.6345939124389763e-08,
+      "loss": 0.4251,
+      "step": 21940
+    },
+    {
+      "epoch": 1.869706007669365,
+      "grad_norm": 12.6875,
+      "learning_rate": 2.6311632535285653e-08,
+      "loss": 0.3962,
+      "step": 21941
+    },
+    {
+      "epoch": 1.869791222837665,
+      "grad_norm": 16.75,
+      "learning_rate": 2.6277348059460667e-08,
+      "loss": 0.9155,
+      "step": 21942
+    },
+    {
+      "epoch": 1.869876438005965,
+      "grad_norm": 21.0,
+      "learning_rate": 2.6243085697534725e-08,
+      "loss": 0.8807,
+      "step": 21943
+    },
+    {
+      "epoch": 1.8699616531742649,
+      "grad_norm": 12.375,
+      "learning_rate": 2.6208845450126778e-08,
+      "loss": 0.4244,
+      "step": 21944
+    },
+    {
+      "epoch": 1.8700468683425648,
+      "grad_norm": 27.625,
+      "learning_rate": 2.61746273178555e-08,
+      "loss": 0.8291,
+      "step": 21945
+    },
+    {
+      "epoch": 1.8701320835108648,
+      "grad_norm": 17.75,
+      "learning_rate": 2.614043130133942e-08,
+      "loss": 0.6901,
+      "step": 21946
+    },
+    {
+      "epoch": 1.8702172986791648,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.6106257401196654e-08,
+      "loss": 0.6168,
+      "step": 21947
+    },
+    {
+      "epoch": 1.8703025138474647,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.6072105618044354e-08,
+      "loss": 0.5937,
+      "step": 21948
+    },
+    {
+      "epoch": 1.8703877290157647,
+      "grad_norm": 13.25,
+      "learning_rate": 2.6037975952499795e-08,
+      "loss": 0.5186,
+      "step": 21949
+    },
+    {
+      "epoch": 1.8704729441840646,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.6003868405179848e-08,
+      "loss": 0.3299,
+      "step": 21950
+    },
+    {
+      "epoch": 1.8705581593523646,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.596978297670083e-08,
+      "loss": 0.5085,
+      "step": 21951
+    },
+    {
+      "epoch": 1.8706433745206645,
+      "grad_norm": 20.125,
+      "learning_rate": 2.593571966767877e-08,
+      "loss": 0.6104,
+      "step": 21952
+    },
+    {
+      "epoch": 1.8707285896889645,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.5901678478729015e-08,
+      "loss": 0.4664,
+      "step": 21953
+    },
+    {
+      "epoch": 1.8708138048572645,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.58676594104669e-08,
+      "loss": 0.8993,
+      "step": 21954
+    },
+    {
+      "epoch": 1.8708990200255644,
+      "grad_norm": 24.375,
+      "learning_rate": 2.5833662463507076e-08,
+      "loss": 0.5503,
+      "step": 21955
+    },
+    {
+      "epoch": 1.8709842351938644,
+      "grad_norm": 14.25,
+      "learning_rate": 2.579968763846405e-08,
+      "loss": 0.4742,
+      "step": 21956
+    },
+    {
+      "epoch": 1.8710694503621643,
+      "grad_norm": 16.0,
+      "learning_rate": 2.5765734935951775e-08,
+      "loss": 0.5133,
+      "step": 21957
+    },
+    {
+      "epoch": 1.8711546655304643,
+      "grad_norm": 17.75,
+      "learning_rate": 2.573180435658365e-08,
+      "loss": 0.5561,
+      "step": 21958
+    },
+    {
+      "epoch": 1.8712398806987642,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.5697895900972934e-08,
+      "loss": 0.5137,
+      "step": 21959
+    },
+    {
+      "epoch": 1.8713250958670642,
+      "grad_norm": 18.875,
+      "learning_rate": 2.5664009569732334e-08,
+      "loss": 0.5218,
+      "step": 21960
+    },
+    {
+      "epoch": 1.8714103110353641,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.5630145363474274e-08,
+      "loss": 0.5542,
+      "step": 21961
+    },
+    {
+      "epoch": 1.871495526203664,
+      "grad_norm": 17.875,
+      "learning_rate": 2.5596303282810626e-08,
+      "loss": 0.4552,
+      "step": 21962
+    },
+    {
+      "epoch": 1.871580741371964,
+      "grad_norm": 13.25,
+      "learning_rate": 2.5562483328352983e-08,
+      "loss": 0.6426,
+      "step": 21963
+    },
+    {
+      "epoch": 1.871665956540264,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.5528685500712663e-08,
+      "loss": 0.524,
+      "step": 21964
+    },
+    {
+      "epoch": 1.871751171708564,
+      "grad_norm": 12.125,
+      "learning_rate": 2.5494909800500426e-08,
+      "loss": 0.4173,
+      "step": 21965
+    },
+    {
+      "epoch": 1.871836386876864,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.5461156228326207e-08,
+      "loss": 0.573,
+      "step": 21966
+    },
+    {
+      "epoch": 1.871921602045164,
+      "grad_norm": 16.5,
+      "learning_rate": 2.542742478480034e-08,
+      "loss": 0.6359,
+      "step": 21967
+    },
+    {
+      "epoch": 1.8720068172134638,
+      "grad_norm": 16.125,
+      "learning_rate": 2.539371547053235e-08,
+      "loss": 0.5192,
+      "step": 21968
+    },
+    {
+      "epoch": 1.8720920323817638,
+      "grad_norm": 18.125,
+      "learning_rate": 2.536002828613146e-08,
+      "loss": 0.4916,
+      "step": 21969
+    },
+    {
+      "epoch": 1.8721772475500638,
+      "grad_norm": 13.1875,
+      "learning_rate": 2.532636323220608e-08,
+      "loss": 0.366,
+      "step": 21970
+    },
+    {
+      "epoch": 1.8722624627183637,
+      "grad_norm": 23.625,
+      "learning_rate": 2.5292720309365027e-08,
+      "loss": 0.4947,
+      "step": 21971
+    },
+    {
+      "epoch": 1.8723476778866637,
+      "grad_norm": 15.125,
+      "learning_rate": 2.5259099518216008e-08,
+      "loss": 0.8987,
+      "step": 21972
+    },
+    {
+      "epoch": 1.8724328930549636,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.522550085936659e-08,
+      "loss": 0.2877,
+      "step": 21973
+    },
+    {
+      "epoch": 1.8725181082232636,
+      "grad_norm": 15.375,
+      "learning_rate": 2.519192433342407e-08,
+      "loss": 0.6671,
+      "step": 21974
+    },
+    {
+      "epoch": 1.8726033233915635,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.5158369940995048e-08,
+      "loss": 0.6552,
+      "step": 21975
+    },
+    {
+      "epoch": 1.8726885385598635,
+      "grad_norm": 12.25,
+      "learning_rate": 2.5124837682685976e-08,
+      "loss": 0.3795,
+      "step": 21976
+    },
+    {
+      "epoch": 1.8727737537281635,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.5091327559102762e-08,
+      "loss": 0.771,
+      "step": 21977
+    },
+    {
+      "epoch": 1.8728589688964634,
+      "grad_norm": 13.625,
+      "learning_rate": 2.5057839570851032e-08,
+      "loss": 0.5563,
+      "step": 21978
+    },
+    {
+      "epoch": 1.8729441840647634,
+      "grad_norm": 22.25,
+      "learning_rate": 2.502437371853586e-08,
+      "loss": 0.5065,
+      "step": 21979
+    },
+    {
+      "epoch": 1.8730293992330633,
+      "grad_norm": 16.0,
+      "learning_rate": 2.4990930002762177e-08,
+      "loss": 0.7986,
+      "step": 21980
+    },
+    {
+      "epoch": 1.8731146144013633,
+      "grad_norm": 14.0,
+      "learning_rate": 2.495750842413408e-08,
+      "loss": 0.5089,
+      "step": 21981
+    },
+    {
+      "epoch": 1.8731998295696632,
+      "grad_norm": 12.5,
+      "learning_rate": 2.4924108983255813e-08,
+      "loss": 0.491,
+      "step": 21982
+    },
+    {
+      "epoch": 1.8732850447379632,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4890731680730778e-08,
+      "loss": 0.6292,
+      "step": 21983
+    },
+    {
+      "epoch": 1.8733702599062632,
+      "grad_norm": 20.0,
+      "learning_rate": 2.4857376517162107e-08,
+      "loss": 0.8556,
+      "step": 21984
+    },
+    {
+      "epoch": 1.8734554750745631,
+      "grad_norm": 13.875,
+      "learning_rate": 2.482404349315265e-08,
+      "loss": 0.4589,
+      "step": 21985
+    },
+    {
+      "epoch": 1.873540690242863,
+      "grad_norm": 11.625,
+      "learning_rate": 2.47907326093047e-08,
+      "loss": 0.2202,
+      "step": 21986
+    },
+    {
+      "epoch": 1.873625905411163,
+      "grad_norm": 14.5,
+      "learning_rate": 2.4757443866220282e-08,
+      "loss": 0.425,
+      "step": 21987
+    },
+    {
+      "epoch": 1.873711120579463,
+      "grad_norm": 16.875,
+      "learning_rate": 2.4724177264500994e-08,
+      "loss": 0.7121,
+      "step": 21988
+    },
+    {
+      "epoch": 1.873796335747763,
+      "grad_norm": 28.25,
+      "learning_rate": 2.469093280474788e-08,
+      "loss": 0.3728,
+      "step": 21989
+    },
+    {
+      "epoch": 1.873881550916063,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.4657710487561854e-08,
+      "loss": 0.4249,
+      "step": 21990
+    },
+    {
+      "epoch": 1.8739667660843629,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.4624510313542986e-08,
+      "loss": 0.4398,
+      "step": 21991
+    },
+    {
+      "epoch": 1.8740519812526628,
+      "grad_norm": 21.625,
+      "learning_rate": 2.4591332283291498e-08,
+      "loss": 0.4363,
+      "step": 21992
+    },
+    {
+      "epoch": 1.8741371964209628,
+      "grad_norm": 21.625,
+      "learning_rate": 2.4558176397406902e-08,
+      "loss": 0.5688,
+      "step": 21993
+    },
+    {
+      "epoch": 1.8742224115892627,
+      "grad_norm": 20.625,
+      "learning_rate": 2.4525042656488306e-08,
+      "loss": 0.7042,
+      "step": 21994
+    },
+    {
+      "epoch": 1.8743076267575627,
+      "grad_norm": 17.75,
+      "learning_rate": 2.449193106113454e-08,
+      "loss": 0.7033,
+      "step": 21995
+    },
+    {
+      "epoch": 1.8743928419258626,
+      "grad_norm": 17.5,
+      "learning_rate": 2.4458841611943866e-08,
+      "loss": 1.0877,
+      "step": 21996
+    },
+    {
+      "epoch": 1.8744780570941626,
+      "grad_norm": 19.875,
+      "learning_rate": 2.442577430951443e-08,
+      "loss": 0.6551,
+      "step": 21997
+    },
+    {
+      "epoch": 1.8745632722624626,
+      "grad_norm": 15.5,
+      "learning_rate": 2.4392729154443516e-08,
+      "loss": 0.5409,
+      "step": 21998
+    },
+    {
+      "epoch": 1.8746484874307625,
+      "grad_norm": 19.875,
+      "learning_rate": 2.4359706147328576e-08,
+      "loss": 0.7905,
+      "step": 21999
+    },
+    {
+      "epoch": 1.8747337025990625,
+      "grad_norm": 18.375,
+      "learning_rate": 2.4326705288766074e-08,
+      "loss": 0.6612,
+      "step": 22000
+    },
+    {
+      "epoch": 1.8748189177673624,
+      "grad_norm": 14.25,
+      "learning_rate": 2.429372657935261e-08,
+      "loss": 0.7088,
+      "step": 22001
+    },
+    {
+      "epoch": 1.8749041329356624,
+      "grad_norm": 18.875,
+      "learning_rate": 2.4260770019683965e-08,
+      "loss": 0.6104,
+      "step": 22002
+    },
+    {
+      "epoch": 1.8749893481039623,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.422783561035591e-08,
+      "loss": 0.4768,
+      "step": 22003
+    },
+    {
+      "epoch": 1.8750745632722625,
+      "grad_norm": 15.5625,
+      "learning_rate": 2.4194923351963383e-08,
+      "loss": 0.8429,
+      "step": 22004
+    },
+    {
+      "epoch": 1.8751597784405625,
+      "grad_norm": 20.25,
+      "learning_rate": 2.416203324510133e-08,
+      "loss": 0.4881,
+      "step": 22005
+    },
+    {
+      "epoch": 1.8752449936088624,
+      "grad_norm": 14.25,
+      "learning_rate": 2.4129165290363854e-08,
+      "loss": 0.4009,
+      "step": 22006
+    },
+    {
+      "epoch": 1.8753302087771624,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.4096319488345204e-08,
+      "loss": 0.6384,
+      "step": 22007
+    },
+    {
+      "epoch": 1.8754154239454623,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.4063495839638655e-08,
+      "loss": 0.3404,
+      "step": 22008
+    },
+    {
+      "epoch": 1.8755006391137623,
+      "grad_norm": 11.5,
+      "learning_rate": 2.4030694344837625e-08,
+      "loss": 0.3316,
+      "step": 22009
+    },
+    {
+      "epoch": 1.8755858542820623,
+      "grad_norm": 17.5,
+      "learning_rate": 2.3997915004534687e-08,
+      "loss": 0.8472,
+      "step": 22010
+    },
+    {
+      "epoch": 1.8756710694503622,
+      "grad_norm": 25.625,
+      "learning_rate": 2.3965157819322433e-08,
+      "loss": 0.4261,
+      "step": 22011
+    },
+    {
+      "epoch": 1.8757562846186622,
+      "grad_norm": 10.625,
+      "learning_rate": 2.3932422789792463e-08,
+      "loss": 0.3023,
+      "step": 22012
+    },
+    {
+      "epoch": 1.8758414997869621,
+      "grad_norm": 18.5,
+      "learning_rate": 2.3899709916536675e-08,
+      "loss": 0.5454,
+      "step": 22013
+    },
+    {
+      "epoch": 1.875926714955262,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.3867019200145836e-08,
+      "loss": 0.52,
+      "step": 22014
+    },
+    {
+      "epoch": 1.876011930123562,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.3834350641211147e-08,
+      "loss": 0.2894,
+      "step": 22015
+    },
+    {
+      "epoch": 1.876097145291862,
+      "grad_norm": 23.0,
+      "learning_rate": 2.3801704240322688e-08,
+      "loss": 0.7118,
+      "step": 22016
+    },
+    {
+      "epoch": 1.876182360460162,
+      "grad_norm": 20.5,
+      "learning_rate": 2.3769079998070404e-08,
+      "loss": 0.6109,
+      "step": 22017
+    },
+    {
+      "epoch": 1.876267575628462,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3736477915043964e-08,
+      "loss": 0.4807,
+      "step": 22018
+    },
+    {
+      "epoch": 1.8763527907967619,
+      "grad_norm": 16.5,
+      "learning_rate": 2.370389799183248e-08,
+      "loss": 0.5337,
+      "step": 22019
+    },
+    {
+      "epoch": 1.8764380059650618,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.3671340229024646e-08,
+      "loss": 0.6018,
+      "step": 22020
+    },
+    {
+      "epoch": 1.8765232211333618,
+      "grad_norm": 17.875,
+      "learning_rate": 2.3638804627209022e-08,
+      "loss": 0.9148,
+      "step": 22021
+    },
+    {
+      "epoch": 1.8766084363016617,
+      "grad_norm": 19.25,
+      "learning_rate": 2.3606291186973328e-08,
+      "loss": 1.0585,
+      "step": 22022
+    },
+    {
+      "epoch": 1.8766936514699617,
+      "grad_norm": 16.375,
+      "learning_rate": 2.3573799908905013e-08,
+      "loss": 0.5612,
+      "step": 22023
+    },
+    {
+      "epoch": 1.8767788666382617,
+      "grad_norm": 15.0,
+      "learning_rate": 2.3541330793591665e-08,
+      "loss": 0.6944,
+      "step": 22024
+    },
+    {
+      "epoch": 1.8768640818065616,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.3508883841619617e-08,
+      "loss": 0.5887,
+      "step": 22025
+    },
+    {
+      "epoch": 1.8769492969748616,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3476459053575345e-08,
+      "loss": 0.8026,
+      "step": 22026
+    },
+    {
+      "epoch": 1.8770345121431615,
+      "grad_norm": 11.5,
+      "learning_rate": 2.344405643004477e-08,
+      "loss": 0.3071,
+      "step": 22027
+    },
+    {
+      "epoch": 1.8771197273114615,
+      "grad_norm": 15.375,
+      "learning_rate": 2.3411675971613673e-08,
+      "loss": 0.7051,
+      "step": 22028
+    },
+    {
+      "epoch": 1.8772049424797614,
+      "grad_norm": 11.0,
+      "learning_rate": 2.3379317678867002e-08,
+      "loss": 0.3305,
+      "step": 22029
+    },
+    {
+      "epoch": 1.8772901576480614,
+      "grad_norm": 12.0,
+      "learning_rate": 2.334698155238943e-08,
+      "loss": 0.3486,
+      "step": 22030
+    },
+    {
+      "epoch": 1.8773753728163614,
+      "grad_norm": 12.8125,
+      "learning_rate": 2.331466759276535e-08,
+      "loss": 0.5568,
+      "step": 22031
+    },
+    {
+      "epoch": 1.8774605879846613,
+      "grad_norm": 19.625,
+      "learning_rate": 2.3282375800578878e-08,
+      "loss": 0.6735,
+      "step": 22032
+    },
+    {
+      "epoch": 1.8775458031529613,
+      "grad_norm": 14.375,
+      "learning_rate": 2.3250106176413435e-08,
+      "loss": 0.4478,
+      "step": 22033
+    },
+    {
+      "epoch": 1.8776310183212612,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.3217858720852028e-08,
+      "loss": 0.6183,
+      "step": 22034
+    },
+    {
+      "epoch": 1.8777162334895612,
+      "grad_norm": 18.375,
+      "learning_rate": 2.318563343447766e-08,
+      "loss": 0.7437,
+      "step": 22035
+    },
+    {
+      "epoch": 1.8778014486578611,
+      "grad_norm": 17.875,
+      "learning_rate": 2.315343031787237e-08,
+      "loss": 0.9006,
+      "step": 22036
+    },
+    {
+      "epoch": 1.877886663826161,
+      "grad_norm": 13.625,
+      "learning_rate": 2.3121249371618747e-08,
+      "loss": 0.518,
+      "step": 22037
+    },
+    {
+      "epoch": 1.877971878994461,
+      "grad_norm": 16.0,
+      "learning_rate": 2.3089090596297435e-08,
+      "loss": 0.8933,
+      "step": 22038
+    },
+    {
+      "epoch": 1.878057094162761,
+      "grad_norm": 20.625,
+      "learning_rate": 2.3056953992490055e-08,
+      "loss": 1.0285,
+      "step": 22039
+    },
+    {
+      "epoch": 1.878142309331061,
+      "grad_norm": 17.5,
+      "learning_rate": 2.302483956077739e-08,
+      "loss": 0.4511,
+      "step": 22040
+    },
+    {
+      "epoch": 1.878227524499361,
+      "grad_norm": 19.75,
+      "learning_rate": 2.299274730173981e-08,
+      "loss": 0.5839,
+      "step": 22041
+    },
+    {
+      "epoch": 1.8783127396676609,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.2960677215956988e-08,
+      "loss": 0.5305,
+      "step": 22042
+    },
+    {
+      "epoch": 1.8783979548359608,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2928629304008602e-08,
+      "loss": 0.5578,
+      "step": 22043
+    },
+    {
+      "epoch": 1.8784831700042608,
+      "grad_norm": 21.0,
+      "learning_rate": 2.2896603566473773e-08,
+      "loss": 0.7825,
+      "step": 22044
+    },
+    {
+      "epoch": 1.8785683851725608,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.2864600003931343e-08,
+      "loss": 0.7924,
+      "step": 22045
+    },
+    {
+      "epoch": 1.8786536003408607,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.2832618616959596e-08,
+      "loss": 0.6439,
+      "step": 22046
+    },
+    {
+      "epoch": 1.8787388155091607,
+      "grad_norm": 18.125,
+      "learning_rate": 2.2800659406136404e-08,
+      "loss": 0.2738,
+      "step": 22047
+    },
+    {
+      "epoch": 1.8788240306774606,
+      "grad_norm": 17.875,
+      "learning_rate": 2.2768722372039364e-08,
+      "loss": 0.4208,
+      "step": 22048
+    },
+    {
+      "epoch": 1.8789092458457606,
+      "grad_norm": 17.875,
+      "learning_rate": 2.273680751524551e-08,
+      "loss": 0.7611,
+      "step": 22049
+    },
+    {
+      "epoch": 1.8789944610140605,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.2704914836331748e-08,
+      "loss": 0.4722,
+      "step": 22050
+    },
+    {
+      "epoch": 1.8790796761823605,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.2673044335874138e-08,
+      "loss": 0.8036,
+      "step": 22051
+    },
+    {
+      "epoch": 1.8791648913506604,
+      "grad_norm": 18.25,
+      "learning_rate": 2.264119601444889e-08,
+      "loss": 0.5202,
+      "step": 22052
+    },
+    {
+      "epoch": 1.8792501065189604,
+      "grad_norm": 12.625,
+      "learning_rate": 2.2609369872631652e-08,
+      "loss": 0.5062,
+      "step": 22053
+    },
+    {
+      "epoch": 1.8793353216872604,
+      "grad_norm": 17.5,
+      "learning_rate": 2.2577565910997106e-08,
+      "loss": 0.5519,
+      "step": 22054
+    },
+    {
+      "epoch": 1.8794205368555603,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.2545784130120347e-08,
+      "loss": 0.3928,
+      "step": 22055
+    },
+    {
+      "epoch": 1.8795057520238603,
+      "grad_norm": 17.0,
+      "learning_rate": 2.2514024530575357e-08,
+      "loss": 0.3066,
+      "step": 22056
+    },
+    {
+      "epoch": 1.8795909671921602,
+      "grad_norm": 13.75,
+      "learning_rate": 2.2482287112936403e-08,
+      "loss": 0.4484,
+      "step": 22057
+    },
+    {
+      "epoch": 1.8796761823604602,
+      "grad_norm": 12.625,
+      "learning_rate": 2.245057187777691e-08,
+      "loss": 0.4606,
+      "step": 22058
+    },
+    {
+      "epoch": 1.8797613975287601,
+      "grad_norm": 15.5,
+      "learning_rate": 2.2418878825669897e-08,
+      "loss": 0.7741,
+      "step": 22059
+    },
+    {
+      "epoch": 1.87984661269706,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.238720795718824e-08,
+      "loss": 0.5149,
+      "step": 22060
+    },
+    {
+      "epoch": 1.87993182786536,
+      "grad_norm": 11.9375,
+      "learning_rate": 2.2355559272904116e-08,
+      "loss": 0.434,
+      "step": 22061
+    },
+    {
+      "epoch": 1.88001704303366,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.2323932773389706e-08,
+      "loss": 0.5591,
+      "step": 22062
+    },
+    {
+      "epoch": 1.88010225820196,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.2292328459216362e-08,
+      "loss": 0.5445,
+      "step": 22063
+    },
+    {
+      "epoch": 1.88018747337026,
+      "grad_norm": 19.875,
+      "learning_rate": 2.2260746330955014e-08,
+      "loss": 0.9148,
+      "step": 22064
+    },
+    {
+      "epoch": 1.8802726885385599,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.2229186389176593e-08,
+      "loss": 0.5505,
+      "step": 22065
+    },
+    {
+      "epoch": 1.8803579037068598,
+      "grad_norm": 17.125,
+      "learning_rate": 2.219764863445134e-08,
+      "loss": 0.4022,
+      "step": 22066
+    },
+    {
+      "epoch": 1.8804431188751598,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.2166133067349354e-08,
+      "loss": 0.719,
+      "step": 22067
+    },
+    {
+      "epoch": 1.8805283340434598,
+      "grad_norm": 16.25,
+      "learning_rate": 2.213463968844004e-08,
+      "loss": 0.5788,
+      "step": 22068
+    },
+    {
+      "epoch": 1.8806135492117597,
+      "grad_norm": 21.0,
+      "learning_rate": 2.2103168498292387e-08,
+      "loss": 0.8303,
+      "step": 22069
+    },
+    {
+      "epoch": 1.8806987643800597,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.2071719497475384e-08,
+      "loss": 0.4182,
+      "step": 22070
+    },
+    {
+      "epoch": 1.8807839795483596,
+      "grad_norm": 16.5,
+      "learning_rate": 2.204029268655719e-08,
+      "loss": 0.741,
+      "step": 22071
+    },
+    {
+      "epoch": 1.8808691947166596,
+      "grad_norm": 16.625,
+      "learning_rate": 2.2008888066105544e-08,
+      "loss": 0.5421,
+      "step": 22072
+    },
+    {
+      "epoch": 1.8809544098849595,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1977505636688184e-08,
+      "loss": 0.5588,
+      "step": 22073
+    },
+    {
+      "epoch": 1.8810396250532595,
+      "grad_norm": 13.9375,
+      "learning_rate": 2.194614539887216e-08,
+      "loss": 0.4855,
+      "step": 22074
+    },
+    {
+      "epoch": 1.8811248402215595,
+      "grad_norm": 23.125,
+      "learning_rate": 2.1914807353224236e-08,
+      "loss": 0.6756,
+      "step": 22075
+    },
+    {
+      "epoch": 1.8812100553898594,
+      "grad_norm": 11.3125,
+      "learning_rate": 2.1883491500310634e-08,
+      "loss": 0.2515,
+      "step": 22076
+    },
+    {
+      "epoch": 1.8812952705581594,
+      "grad_norm": 18.375,
+      "learning_rate": 2.1852197840697282e-08,
+      "loss": 0.4925,
+      "step": 22077
+    },
+    {
+      "epoch": 1.8813804857264593,
+      "grad_norm": 19.25,
+      "learning_rate": 2.182092637494984e-08,
+      "loss": 0.636,
+      "step": 22078
+    },
+    {
+      "epoch": 1.8814657008947593,
+      "grad_norm": 15.75,
+      "learning_rate": 2.1789677103633278e-08,
+      "loss": 0.5575,
+      "step": 22079
+    },
+    {
+      "epoch": 1.8815509160630592,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.175845002731228e-08,
+      "loss": 0.6264,
+      "step": 22080
+    },
+    {
+      "epoch": 1.8816361312313592,
+      "grad_norm": 14.375,
+      "learning_rate": 2.172724514655111e-08,
+      "loss": 0.5365,
+      "step": 22081
+    },
+    {
+      "epoch": 1.8817213463996592,
+      "grad_norm": 13.5625,
+      "learning_rate": 2.1696062461913775e-08,
+      "loss": 0.3233,
+      "step": 22082
+    },
+    {
+      "epoch": 1.8818065615679591,
+      "grad_norm": 18.0,
+      "learning_rate": 2.1664901973963704e-08,
+      "loss": 0.5785,
+      "step": 22083
+    },
+    {
+      "epoch": 1.881891776736259,
+      "grad_norm": 14.875,
+      "learning_rate": 2.1633763683264197e-08,
+      "loss": 0.5567,
+      "step": 22084
+    },
+    {
+      "epoch": 1.881976991904559,
+      "grad_norm": 19.125,
+      "learning_rate": 2.1602647590377585e-08,
+      "loss": 0.6331,
+      "step": 22085
+    },
+    {
+      "epoch": 1.882062207072859,
+      "grad_norm": 16.75,
+      "learning_rate": 2.157155369586661e-08,
+      "loss": 0.5978,
+      "step": 22086
+    },
+    {
+      "epoch": 1.882147422241159,
+      "grad_norm": 19.0,
+      "learning_rate": 2.1540482000292767e-08,
+      "loss": 1.0022,
+      "step": 22087
+    },
+    {
+      "epoch": 1.882232637409459,
+      "grad_norm": 16.5,
+      "learning_rate": 2.1509432504217832e-08,
+      "loss": 0.8776,
+      "step": 22088
+    },
+    {
+      "epoch": 1.8823178525777589,
+      "grad_norm": 22.625,
+      "learning_rate": 2.1478405208202742e-08,
+      "loss": 0.9097,
+      "step": 22089
+    },
+    {
+      "epoch": 1.8824030677460588,
+      "grad_norm": 19.25,
+      "learning_rate": 2.1447400112808157e-08,
+      "loss": 0.9157,
+      "step": 22090
+    },
+    {
+      "epoch": 1.8824882829143588,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.141641721859461e-08,
+      "loss": 0.7165,
+      "step": 22091
+    },
+    {
+      "epoch": 1.8825734980826587,
+      "grad_norm": 13.0625,
+      "learning_rate": 2.1385456526121786e-08,
+      "loss": 0.5067,
+      "step": 22092
+    },
+    {
+      "epoch": 1.8826587132509587,
+      "grad_norm": 14.625,
+      "learning_rate": 2.1354518035949233e-08,
+      "loss": 0.5628,
+      "step": 22093
+    },
+    {
+      "epoch": 1.8827439284192586,
+      "grad_norm": 17.625,
+      "learning_rate": 2.1323601748636096e-08,
+      "loss": 0.9869,
+      "step": 22094
+    },
+    {
+      "epoch": 1.8828291435875586,
+      "grad_norm": 15.5,
+      "learning_rate": 2.1292707664740813e-08,
+      "loss": 0.5149,
+      "step": 22095
+    },
+    {
+      "epoch": 1.8829143587558586,
+      "grad_norm": 23.875,
+      "learning_rate": 2.1261835784821966e-08,
+      "loss": 0.6424,
+      "step": 22096
+    },
+    {
+      "epoch": 1.8829995739241585,
+      "grad_norm": 12.0625,
+      "learning_rate": 2.123098610943744e-08,
+      "loss": 0.3805,
+      "step": 22097
+    },
+    {
+      "epoch": 1.8830847890924585,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.120015863914443e-08,
+      "loss": 0.3814,
+      "step": 22098
+    },
+    {
+      "epoch": 1.8831700042607584,
+      "grad_norm": 15.75,
+      "learning_rate": 2.1169353374500407e-08,
+      "loss": 0.5576,
+      "step": 22099
+    },
+    {
+      "epoch": 1.8832552194290584,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.113857031606173e-08,
+      "loss": 0.7356,
+      "step": 22100
+    },
+    {
+      "epoch": 1.8833404345973583,
+      "grad_norm": 17.125,
+      "learning_rate": 2.110780946438476e-08,
+      "loss": 0.495,
+      "step": 22101
+    },
+    {
+      "epoch": 1.8834256497656583,
+      "grad_norm": 15.125,
+      "learning_rate": 2.107707082002558e-08,
+      "loss": 0.5506,
+      "step": 22102
+    },
+    {
+      "epoch": 1.8835108649339583,
+      "grad_norm": 11.375,
+      "learning_rate": 2.1046354383539442e-08,
+      "loss": 0.4423,
+      "step": 22103
+    },
+    {
+      "epoch": 1.8835960801022582,
+      "grad_norm": 21.625,
+      "learning_rate": 2.1015660155481456e-08,
+      "loss": 0.4508,
+      "step": 22104
+    },
+    {
+      "epoch": 1.8836812952705582,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0984988136406313e-08,
+      "loss": 0.413,
+      "step": 22105
+    },
+    {
+      "epoch": 1.8837665104388581,
+      "grad_norm": 20.75,
+      "learning_rate": 2.0954338326868303e-08,
+      "loss": 0.6871,
+      "step": 22106
+    },
+    {
+      "epoch": 1.883851725607158,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0923710727421277e-08,
+      "loss": 0.5738,
+      "step": 22107
+    },
+    {
+      "epoch": 1.883936940775458,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.0893105338618825e-08,
+      "loss": 0.4429,
+      "step": 22108
+    },
+    {
+      "epoch": 1.884022155943758,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.0862522161013975e-08,
+      "loss": 0.5472,
+      "step": 22109
+    },
+    {
+      "epoch": 1.884107371112058,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.0831961195159205e-08,
+      "loss": 0.6662,
+      "step": 22110
+    },
+    {
+      "epoch": 1.884192586280358,
+      "grad_norm": 17.875,
+      "learning_rate": 2.0801422441606987e-08,
+      "loss": 0.6349,
+      "step": 22111
+    },
+    {
+      "epoch": 1.8842778014486579,
+      "grad_norm": 16.25,
+      "learning_rate": 2.0770905900909104e-08,
+      "loss": 0.9843,
+      "step": 22112
+    },
+    {
+      "epoch": 1.8843630166169578,
+      "grad_norm": 13.4375,
+      "learning_rate": 2.074041157361692e-08,
+      "loss": 0.5592,
+      "step": 22113
+    },
+    {
+      "epoch": 1.8844482317852578,
+      "grad_norm": 14.875,
+      "learning_rate": 2.0709939460281798e-08,
+      "loss": 0.6278,
+      "step": 22114
+    },
+    {
+      "epoch": 1.8845334469535577,
+      "grad_norm": 17.0,
+      "learning_rate": 2.067948956145413e-08,
+      "loss": 0.591,
+      "step": 22115
+    },
+    {
+      "epoch": 1.8846186621218577,
+      "grad_norm": 20.125,
+      "learning_rate": 2.0649061877684173e-08,
+      "loss": 0.9596,
+      "step": 22116
+    },
+    {
+      "epoch": 1.8847038772901576,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.061865640952204e-08,
+      "loss": 0.2455,
+      "step": 22117
+    },
+    {
+      "epoch": 1.8847890924584576,
+      "grad_norm": 16.375,
+      "learning_rate": 2.058827315751702e-08,
+      "loss": 0.438,
+      "step": 22118
+    },
+    {
+      "epoch": 1.8848743076267576,
+      "grad_norm": 10.75,
+      "learning_rate": 2.055791212221811e-08,
+      "loss": 0.6009,
+      "step": 22119
+    },
+    {
+      "epoch": 1.8849595227950575,
+      "grad_norm": 12.5625,
+      "learning_rate": 2.052757330417404e-08,
+      "loss": 0.431,
+      "step": 22120
+    },
+    {
+      "epoch": 1.8850447379633575,
+      "grad_norm": 37.0,
+      "learning_rate": 2.0497256703932988e-08,
+      "loss": 1.1228,
+      "step": 22121
+    },
+    {
+      "epoch": 1.8851299531316574,
+      "grad_norm": 12.9375,
+      "learning_rate": 2.0466962322042984e-08,
+      "loss": 0.3412,
+      "step": 22122
+    },
+    {
+      "epoch": 1.8852151682999574,
+      "grad_norm": 15.625,
+      "learning_rate": 2.0436690159051225e-08,
+      "loss": 0.8763,
+      "step": 22123
+    },
+    {
+      "epoch": 1.8853003834682573,
+      "grad_norm": 16.875,
+      "learning_rate": 2.040644021550492e-08,
+      "loss": 0.5438,
+      "step": 22124
+    },
+    {
+      "epoch": 1.8853855986365573,
+      "grad_norm": 14.5,
+      "learning_rate": 2.037621249195071e-08,
+      "loss": 0.5155,
+      "step": 22125
+    },
+    {
+      "epoch": 1.8854708138048573,
+      "grad_norm": 19.25,
+      "learning_rate": 2.0346006988934962e-08,
+      "loss": 0.8729,
+      "step": 22126
+    },
+    {
+      "epoch": 1.8855560289731572,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.0315823707003358e-08,
+      "loss": 0.4281,
+      "step": 22127
+    },
+    {
+      "epoch": 1.8856412441414572,
+      "grad_norm": 16.375,
+      "learning_rate": 2.0285662646701286e-08,
+      "loss": 0.764,
+      "step": 22128
+    },
+    {
+      "epoch": 1.8857264593097571,
+      "grad_norm": 26.25,
+      "learning_rate": 2.0255523808573734e-08,
+      "loss": 0.5942,
+      "step": 22129
+    },
+    {
+      "epoch": 1.885811674478057,
+      "grad_norm": 38.75,
+      "learning_rate": 2.022540719316568e-08,
+      "loss": 0.916,
+      "step": 22130
+    },
+    {
+      "epoch": 1.885896889646357,
+      "grad_norm": 23.25,
+      "learning_rate": 2.0195312801020993e-08,
+      "loss": 0.9039,
+      "step": 22131
+    },
+    {
+      "epoch": 1.885982104814657,
+      "grad_norm": 12.75,
+      "learning_rate": 2.0165240632683825e-08,
+      "loss": 0.328,
+      "step": 22132
+    },
+    {
+      "epoch": 1.886067319982957,
+      "grad_norm": 16.625,
+      "learning_rate": 2.013519068869735e-08,
+      "loss": 0.7123,
+      "step": 22133
+    },
+    {
+      "epoch": 1.886152535151257,
+      "grad_norm": 16.375,
+      "learning_rate": 2.010516296960474e-08,
+      "loss": 0.6224,
+      "step": 22134
+    },
+    {
+      "epoch": 1.8862377503195569,
+      "grad_norm": 19.75,
+      "learning_rate": 2.0075157475948627e-08,
+      "loss": 0.8743,
+      "step": 22135
+    },
+    {
+      "epoch": 1.8863229654878568,
+      "grad_norm": 19.5,
+      "learning_rate": 2.0045174208271066e-08,
+      "loss": 0.7685,
+      "step": 22136
+    },
+    {
+      "epoch": 1.8864081806561568,
+      "grad_norm": 17.5,
+      "learning_rate": 2.0015213167114134e-08,
+      "loss": 0.1976,
+      "step": 22137
+    },
+    {
+      "epoch": 1.8864933958244567,
+      "grad_norm": 19.5,
+      "learning_rate": 1.9985274353019058e-08,
+      "loss": 0.9319,
+      "step": 22138
+    },
+    {
+      "epoch": 1.8865786109927567,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9955357766527072e-08,
+      "loss": 0.7175,
+      "step": 22139
+    },
+    {
+      "epoch": 1.8866638261610567,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.992546340817858e-08,
+      "loss": 0.6067,
+      "step": 22140
+    },
+    {
+      "epoch": 1.8867490413293566,
+      "grad_norm": 13.25,
+      "learning_rate": 1.9895591278513843e-08,
+      "loss": 0.4501,
+      "step": 22141
+    },
+    {
+      "epoch": 1.8868342564976566,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9865741378072846e-08,
+      "loss": 0.9939,
+      "step": 22142
+    },
+    {
+      "epoch": 1.8869194716659565,
+      "grad_norm": 25.625,
+      "learning_rate": 1.9835913707394876e-08,
+      "loss": 0.5444,
+      "step": 22143
+    },
+    {
+      "epoch": 1.8870046868342565,
+      "grad_norm": 20.875,
+      "learning_rate": 1.9806108267019093e-08,
+      "loss": 0.7615,
+      "step": 22144
+    },
+    {
+      "epoch": 1.8870899020025564,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9776325057483807e-08,
+      "loss": 0.6213,
+      "step": 22145
+    },
+    {
+      "epoch": 1.8871751171708564,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9746564079327486e-08,
+      "loss": 0.7745,
+      "step": 22146
+    },
+    {
+      "epoch": 1.8872603323391564,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9716825333087886e-08,
+      "loss": 0.7141,
+      "step": 22147
+    },
+    {
+      "epoch": 1.8873455475074563,
+      "grad_norm": 15.25,
+      "learning_rate": 1.968710881930236e-08,
+      "loss": 0.5947,
+      "step": 22148
+    },
+    {
+      "epoch": 1.8874307626757563,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9657414538507974e-08,
+      "loss": 0.5496,
+      "step": 22149
+    },
+    {
+      "epoch": 1.8875159778440562,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9627742491241247e-08,
+      "loss": 0.2495,
+      "step": 22150
+    },
+    {
+      "epoch": 1.8876011930123562,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.9598092678038556e-08,
+      "loss": 0.504,
+      "step": 22151
+    },
+    {
+      "epoch": 1.8876864081806561,
+      "grad_norm": 31.375,
+      "learning_rate": 1.956846509943558e-08,
+      "loss": 0.7275,
+      "step": 22152
+    },
+    {
+      "epoch": 1.887771623348956,
+      "grad_norm": 31.875,
+      "learning_rate": 1.9538859755967726e-08,
+      "loss": 0.6519,
+      "step": 22153
+    },
+    {
+      "epoch": 1.887856838517256,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.9509276648169846e-08,
+      "loss": 0.2972,
+      "step": 22154
+    },
+    {
+      "epoch": 1.887942053685556,
+      "grad_norm": 19.875,
+      "learning_rate": 1.9479715776576785e-08,
+      "loss": 0.6288,
+      "step": 22155
+    },
+    {
+      "epoch": 1.888027268853856,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9450177141722703e-08,
+      "loss": 0.3427,
+      "step": 22156
+    },
+    {
+      "epoch": 1.888112484022156,
+      "grad_norm": 16.5,
+      "learning_rate": 1.942066074414134e-08,
+      "loss": 0.6717,
+      "step": 22157
+    },
+    {
+      "epoch": 1.8881976991904559,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9391166584366017e-08,
+      "loss": 0.3025,
+      "step": 22158
+    },
+    {
+      "epoch": 1.8882829143587558,
+      "grad_norm": 27.375,
+      "learning_rate": 1.936169466292992e-08,
+      "loss": 0.7383,
+      "step": 22159
+    },
+    {
+      "epoch": 1.8883681295270558,
+      "grad_norm": 13.0,
+      "learning_rate": 1.93322449803654e-08,
+      "loss": 0.5046,
+      "step": 22160
+    },
+    {
+      "epoch": 1.8884533446953558,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9302817537204672e-08,
+      "loss": 0.4293,
+      "step": 22161
+    },
+    {
+      "epoch": 1.8885385598636557,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9273412333979667e-08,
+      "loss": 0.8356,
+      "step": 22162
+    },
+    {
+      "epoch": 1.8886237750319557,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9244029371221635e-08,
+      "loss": 0.5116,
+      "step": 22163
+    },
+    {
+      "epoch": 1.8887089902002556,
+      "grad_norm": 10.125,
+      "learning_rate": 1.9214668649461808e-08,
+      "loss": 0.4074,
+      "step": 22164
+    },
+    {
+      "epoch": 1.8887942053685556,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.9185330169230326e-08,
+      "loss": 0.3136,
+      "step": 22165
+    },
+    {
+      "epoch": 1.8888794205368555,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9156013931057733e-08,
+      "loss": 0.8767,
+      "step": 22166
+    },
+    {
+      "epoch": 1.8889646357051555,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.912671993547374e-08,
+      "loss": 0.6942,
+      "step": 22167
+    },
+    {
+      "epoch": 1.8890498508734554,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.9097448183007516e-08,
+      "loss": 0.4247,
+      "step": 22168
+    },
+    {
+      "epoch": 1.8891350660417554,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9068198674188214e-08,
+      "loss": 0.6581,
+      "step": 22169
+    },
+    {
+      "epoch": 1.8892202812100554,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9038971409544304e-08,
+      "loss": 0.7606,
+      "step": 22170
+    },
+    {
+      "epoch": 1.8893054963783553,
+      "grad_norm": 19.125,
+      "learning_rate": 1.900976638960411e-08,
+      "loss": 0.7506,
+      "step": 22171
+    },
+    {
+      "epoch": 1.8893907115466553,
+      "grad_norm": 16.625,
+      "learning_rate": 1.898058361489527e-08,
+      "loss": 0.7247,
+      "step": 22172
+    },
+    {
+      "epoch": 1.8894759267149552,
+      "grad_norm": 20.875,
+      "learning_rate": 1.8951423085945136e-08,
+      "loss": 0.7494,
+      "step": 22173
+    },
+    {
+      "epoch": 1.8895611418832552,
+      "grad_norm": 18.25,
+      "learning_rate": 1.8922284803280787e-08,
+      "loss": 0.4926,
+      "step": 22174
+    },
+    {
+      "epoch": 1.8896463570515551,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.889316876742861e-08,
+      "loss": 0.5829,
+      "step": 22175
+    },
+    {
+      "epoch": 1.889731572219855,
+      "grad_norm": 52.25,
+      "learning_rate": 1.886407497891485e-08,
+      "loss": 0.5139,
+      "step": 22176
+    },
+    {
+      "epoch": 1.889816787388155,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.8835003438265333e-08,
+      "loss": 0.5832,
+      "step": 22177
+    },
+    {
+      "epoch": 1.889902002556455,
+      "grad_norm": 18.75,
+      "learning_rate": 1.8805954146005334e-08,
+      "loss": 0.6383,
+      "step": 22178
+    },
+    {
+      "epoch": 1.889987217724755,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.8776927102659715e-08,
+      "loss": 0.6285,
+      "step": 22179
+    },
+    {
+      "epoch": 1.890072432893055,
+      "grad_norm": 20.25,
+      "learning_rate": 1.8747922308753053e-08,
+      "loss": 0.9649,
+      "step": 22180
+    },
+    {
+      "epoch": 1.890157648061355,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.871893976480965e-08,
+      "loss": 0.5265,
+      "step": 22181
+    },
+    {
+      "epoch": 1.8902428632296548,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.868997947135312e-08,
+      "loss": 0.385,
+      "step": 22182
+    },
+    {
+      "epoch": 1.8903280783979548,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.8661041428907068e-08,
+      "loss": 0.679,
+      "step": 22183
+    },
+    {
+      "epoch": 1.8904132935662548,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8632125637993853e-08,
+      "loss": 1.0204,
+      "step": 22184
+    },
+    {
+      "epoch": 1.8904985087345547,
+      "grad_norm": 14.0,
+      "learning_rate": 1.860323209913653e-08,
+      "loss": 0.4018,
+      "step": 22185
+    },
+    {
+      "epoch": 1.8905837239028547,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8574360812857183e-08,
+      "loss": 0.7517,
+      "step": 22186
+    },
+    {
+      "epoch": 1.8906689390711546,
+      "grad_norm": 17.0,
+      "learning_rate": 1.85455117796772e-08,
+      "loss": 0.8336,
+      "step": 22187
+    },
+    {
+      "epoch": 1.8907541542394546,
+      "grad_norm": 21.5,
+      "learning_rate": 1.8516685000118246e-08,
+      "loss": 0.6667,
+      "step": 22188
+    },
+    {
+      "epoch": 1.8908393694077545,
+      "grad_norm": 17.125,
+      "learning_rate": 1.8487880474701154e-08,
+      "loss": 0.524,
+      "step": 22189
+    },
+    {
+      "epoch": 1.8909245845760545,
+      "grad_norm": 13.5,
+      "learning_rate": 1.845909820394648e-08,
+      "loss": 0.434,
+      "step": 22190
+    },
+    {
+      "epoch": 1.8910097997443545,
+      "grad_norm": 16.375,
+      "learning_rate": 1.8430338188374224e-08,
+      "loss": 0.7357,
+      "step": 22191
+    },
+    {
+      "epoch": 1.8910950149126544,
+      "grad_norm": 15.0,
+      "learning_rate": 1.840160042850425e-08,
+      "loss": 0.7634,
+      "step": 22192
+    },
+    {
+      "epoch": 1.8911802300809544,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8372884924855867e-08,
+      "loss": 0.3374,
+      "step": 22193
+    },
+    {
+      "epoch": 1.8912654452492543,
+      "grad_norm": 17.75,
+      "learning_rate": 1.834419167794796e-08,
+      "loss": 0.965,
+      "step": 22194
+    },
+    {
+      "epoch": 1.8913506604175543,
+      "grad_norm": 18.5,
+      "learning_rate": 1.8315520688299005e-08,
+      "loss": 0.3337,
+      "step": 22195
+    },
+    {
+      "epoch": 1.8914358755858542,
+      "grad_norm": 19.75,
+      "learning_rate": 1.828687195642706e-08,
+      "loss": 0.9393,
+      "step": 22196
+    },
+    {
+      "epoch": 1.8915210907541542,
+      "grad_norm": 13.875,
+      "learning_rate": 1.8258245482849906e-08,
+      "loss": 0.4949,
+      "step": 22197
+    },
+    {
+      "epoch": 1.8916063059224542,
+      "grad_norm": 28.25,
+      "learning_rate": 1.8229641268085048e-08,
+      "loss": 0.7875,
+      "step": 22198
+    },
+    {
+      "epoch": 1.8916915210907541,
+      "grad_norm": 37.0,
+      "learning_rate": 1.820105931264901e-08,
+      "loss": 0.6778,
+      "step": 22199
+    },
+    {
+      "epoch": 1.891776736259054,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.8172499617058608e-08,
+      "loss": 0.428,
+      "step": 22200
+    },
+    {
+      "epoch": 1.891861951427354,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8143962181829815e-08,
+      "loss": 0.6965,
+      "step": 22201
+    },
+    {
+      "epoch": 1.891947166595654,
+      "grad_norm": 16.625,
+      "learning_rate": 1.8115447007478332e-08,
+      "loss": 0.5398,
+      "step": 22202
+    },
+    {
+      "epoch": 1.892032381763954,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.8086954094519576e-08,
+      "loss": 0.5666,
+      "step": 22203
+    },
+    {
+      "epoch": 1.892117596932254,
+      "grad_norm": 19.875,
+      "learning_rate": 1.805848344346828e-08,
+      "loss": 0.3861,
+      "step": 22204
+    },
+    {
+      "epoch": 1.8922028121005539,
+      "grad_norm": 12.75,
+      "learning_rate": 1.8030035054839022e-08,
+      "loss": 0.4156,
+      "step": 22205
+    },
+    {
+      "epoch": 1.8922880272688538,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.800160892914571e-08,
+      "loss": 0.2874,
+      "step": 22206
+    },
+    {
+      "epoch": 1.8923732424371538,
+      "grad_norm": 20.75,
+      "learning_rate": 1.7973205066902365e-08,
+      "loss": 1.0864,
+      "step": 22207
+    },
+    {
+      "epoch": 1.8924584576054537,
+      "grad_norm": 12.125,
+      "learning_rate": 1.7944823468622195e-08,
+      "loss": 0.5776,
+      "step": 22208
+    },
+    {
+      "epoch": 1.8925436727737537,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.7916464134817846e-08,
+      "loss": 0.5831,
+      "step": 22209
+    },
+    {
+      "epoch": 1.8926288879420536,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.788812706600196e-08,
+      "loss": 0.5283,
+      "step": 22210
+    },
+    {
+      "epoch": 1.8927141031103536,
+      "grad_norm": 14.375,
+      "learning_rate": 1.7859812262686628e-08,
+      "loss": 0.7142,
+      "step": 22211
+    },
+    {
+      "epoch": 1.8927993182786536,
+      "grad_norm": 15.875,
+      "learning_rate": 1.7831519725383385e-08,
+      "loss": 0.5516,
+      "step": 22212
+    },
+    {
+      "epoch": 1.8928845334469535,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.7803249454603765e-08,
+      "loss": 0.4874,
+      "step": 22213
+    },
+    {
+      "epoch": 1.8929697486152535,
+      "grad_norm": 20.875,
+      "learning_rate": 1.7775001450858474e-08,
+      "loss": 0.5609,
+      "step": 22214
+    },
+    {
+      "epoch": 1.8930549637835534,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.7746775714657938e-08,
+      "loss": 0.4931,
+      "step": 22215
+    },
+    {
+      "epoch": 1.8931401789518534,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.771857224651244e-08,
+      "loss": 0.6245,
+      "step": 22216
+    },
+    {
+      "epoch": 1.8932253941201533,
+      "grad_norm": 16.25,
+      "learning_rate": 1.7690391046931294e-08,
+      "loss": 0.591,
+      "step": 22217
+    },
+    {
+      "epoch": 1.8933106092884533,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7662232116424094e-08,
+      "loss": 0.7353,
+      "step": 22218
+    },
+    {
+      "epoch": 1.8933958244567533,
+      "grad_norm": 15.5,
+      "learning_rate": 1.7634095455499463e-08,
+      "loss": 0.4752,
+      "step": 22219
+    },
+    {
+      "epoch": 1.8934810396250532,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.7605981064666154e-08,
+      "loss": 0.3946,
+      "step": 22220
+    },
+    {
+      "epoch": 1.8935662547933532,
+      "grad_norm": 10.875,
+      "learning_rate": 1.7577888944431824e-08,
+      "loss": 0.3866,
+      "step": 22221
+    },
+    {
+      "epoch": 1.8936514699616531,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.75498190953044e-08,
+      "loss": 0.414,
+      "step": 22222
+    },
+    {
+      "epoch": 1.893736685129953,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.7521771517791247e-08,
+      "loss": 0.6509,
+      "step": 22223
+    },
+    {
+      "epoch": 1.893821900298253,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.749374621239891e-08,
+      "loss": 0.8166,
+      "step": 22224
+    },
+    {
+      "epoch": 1.893907115466553,
+      "grad_norm": 22.125,
+      "learning_rate": 1.7465743179633926e-08,
+      "loss": 0.5885,
+      "step": 22225
+    },
+    {
+      "epoch": 1.893992330634853,
+      "grad_norm": 16.375,
+      "learning_rate": 1.743776242000242e-08,
+      "loss": 0.4569,
+      "step": 22226
+    },
+    {
+      "epoch": 1.894077545803153,
+      "grad_norm": 13.5,
+      "learning_rate": 1.7409803934009816e-08,
+      "loss": 0.6827,
+      "step": 22227
+    },
+    {
+      "epoch": 1.8941627609714529,
+      "grad_norm": 14.25,
+      "learning_rate": 1.738186772216183e-08,
+      "loss": 0.5041,
+      "step": 22228
+    },
+    {
+      "epoch": 1.894247976139753,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.735395378496277e-08,
+      "loss": 0.5296,
+      "step": 22229
+    },
+    {
+      "epoch": 1.894333191308053,
+      "grad_norm": 18.875,
+      "learning_rate": 1.7326062122917238e-08,
+      "loss": 0.731,
+      "step": 22230
+    },
+    {
+      "epoch": 1.894418406476353,
+      "grad_norm": 15.75,
+      "learning_rate": 1.729819273652941e-08,
+      "loss": 0.5395,
+      "step": 22231
+    },
+    {
+      "epoch": 1.894503621644653,
+      "grad_norm": 18.125,
+      "learning_rate": 1.727034562630292e-08,
+      "loss": 0.4331,
+      "step": 22232
+    },
+    {
+      "epoch": 1.8945888368129529,
+      "grad_norm": 16.875,
+      "learning_rate": 1.724252079274083e-08,
+      "loss": 0.5287,
+      "step": 22233
+    },
+    {
+      "epoch": 1.8946740519812528,
+      "grad_norm": 26.0,
+      "learning_rate": 1.7214718236345934e-08,
+      "loss": 0.813,
+      "step": 22234
+    },
+    {
+      "epoch": 1.8947592671495528,
+      "grad_norm": 16.25,
+      "learning_rate": 1.718693795762075e-08,
+      "loss": 0.2972,
+      "step": 22235
+    },
+    {
+      "epoch": 1.8948444823178527,
+      "grad_norm": 18.25,
+      "learning_rate": 1.7159179957067228e-08,
+      "loss": 0.5384,
+      "step": 22236
+    },
+    {
+      "epoch": 1.8949296974861527,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.71314442351872e-08,
+      "loss": 0.434,
+      "step": 22237
+    },
+    {
+      "epoch": 1.8950149126544527,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.710373079248165e-08,
+      "loss": 0.4771,
+      "step": 22238
+    },
+    {
+      "epoch": 1.8951001278227526,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7076039629451424e-08,
+      "loss": 0.9232,
+      "step": 22239
+    },
+    {
+      "epoch": 1.8951853429910526,
+      "grad_norm": 19.875,
+      "learning_rate": 1.7048370746597098e-08,
+      "loss": 0.8364,
+      "step": 22240
+    },
+    {
+      "epoch": 1.8952705581593525,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.702072414441841e-08,
+      "loss": 0.4483,
+      "step": 22241
+    },
+    {
+      "epoch": 1.8953557733276525,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.6993099823415098e-08,
+      "loss": 0.5277,
+      "step": 22242
+    },
+    {
+      "epoch": 1.8954409884959524,
+      "grad_norm": 17.875,
+      "learning_rate": 1.696549778408635e-08,
+      "loss": 0.585,
+      "step": 22243
+    },
+    {
+      "epoch": 1.8955262036642524,
+      "grad_norm": 15.375,
+      "learning_rate": 1.6937918026931065e-08,
+      "loss": 0.7944,
+      "step": 22244
+    },
+    {
+      "epoch": 1.8956114188325524,
+      "grad_norm": 14.25,
+      "learning_rate": 1.6910360552447464e-08,
+      "loss": 0.5009,
+      "step": 22245
+    },
+    {
+      "epoch": 1.8956966340008523,
+      "grad_norm": 9.75,
+      "learning_rate": 1.6882825361133616e-08,
+      "loss": 0.2388,
+      "step": 22246
+    },
+    {
+      "epoch": 1.8957818491691523,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6855312453487177e-08,
+      "loss": 0.701,
+      "step": 22247
+    },
+    {
+      "epoch": 1.8958670643374522,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.6827821830005252e-08,
+      "loss": 0.4177,
+      "step": 22248
+    },
+    {
+      "epoch": 1.8959522795057522,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.6800353491184528e-08,
+      "loss": 0.4362,
+      "step": 22249
+    },
+    {
+      "epoch": 1.8960374946740521,
+      "grad_norm": 21.5,
+      "learning_rate": 1.677290743752169e-08,
+      "loss": 1.029,
+      "step": 22250
+    },
+    {
+      "epoch": 1.896122709842352,
+      "grad_norm": 12.625,
+      "learning_rate": 1.6745483669512312e-08,
+      "loss": 0.3774,
+      "step": 22251
+    },
+    {
+      "epoch": 1.896207925010652,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6718082187652252e-08,
+      "loss": 0.3903,
+      "step": 22252
+    },
+    {
+      "epoch": 1.896293140178952,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.6690702992436662e-08,
+      "loss": 0.5067,
+      "step": 22253
+    },
+    {
+      "epoch": 1.896378355347252,
+      "grad_norm": 20.875,
+      "learning_rate": 1.6663346084360153e-08,
+      "loss": 0.6823,
+      "step": 22254
+    },
+    {
+      "epoch": 1.896463570515552,
+      "grad_norm": 37.5,
+      "learning_rate": 1.663601146391719e-08,
+      "loss": 1.1558,
+      "step": 22255
+    },
+    {
+      "epoch": 1.8965487856838519,
+      "grad_norm": 30.375,
+      "learning_rate": 1.6608699131601818e-08,
+      "loss": 0.81,
+      "step": 22256
+    },
+    {
+      "epoch": 1.8966340008521518,
+      "grad_norm": 25.125,
+      "learning_rate": 1.6581409087907534e-08,
+      "loss": 0.609,
+      "step": 22257
+    },
+    {
+      "epoch": 1.8967192160204518,
+      "grad_norm": 15.75,
+      "learning_rate": 1.6554141333327554e-08,
+      "loss": 0.733,
+      "step": 22258
+    },
+    {
+      "epoch": 1.8968044311887517,
+      "grad_norm": 19.125,
+      "learning_rate": 1.6526895868354404e-08,
+      "loss": 0.3938,
+      "step": 22259
+    },
+    {
+      "epoch": 1.8968896463570517,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.649967269348074e-08,
+      "loss": 0.5497,
+      "step": 22260
+    },
+    {
+      "epoch": 1.8969748615253517,
+      "grad_norm": 17.75,
+      "learning_rate": 1.6472471809198253e-08,
+      "loss": 0.6176,
+      "step": 22261
+    },
+    {
+      "epoch": 1.8970600766936516,
+      "grad_norm": 14.125,
+      "learning_rate": 1.6445293215998636e-08,
+      "loss": 0.7186,
+      "step": 22262
+    },
+    {
+      "epoch": 1.8971452918619516,
+      "grad_norm": 19.125,
+      "learning_rate": 1.641813691437302e-08,
+      "loss": 0.6694,
+      "step": 22263
+    },
+    {
+      "epoch": 1.8972305070302515,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.639100290481213e-08,
+      "loss": 0.436,
+      "step": 22264
+    },
+    {
+      "epoch": 1.8973157221985515,
+      "grad_norm": 19.25,
+      "learning_rate": 1.6363891187806403e-08,
+      "loss": 0.7687,
+      "step": 22265
+    },
+    {
+      "epoch": 1.8974009373668514,
+      "grad_norm": 20.875,
+      "learning_rate": 1.6336801763845582e-08,
+      "loss": 0.9266,
+      "step": 22266
+    },
+    {
+      "epoch": 1.8974861525351514,
+      "grad_norm": 18.125,
+      "learning_rate": 1.6309734633419284e-08,
+      "loss": 0.4169,
+      "step": 22267
+    },
+    {
+      "epoch": 1.8975713677034514,
+      "grad_norm": 13.625,
+      "learning_rate": 1.6282689797016692e-08,
+      "loss": 0.5117,
+      "step": 22268
+    },
+    {
+      "epoch": 1.8976565828717513,
+      "grad_norm": 9.75,
+      "learning_rate": 1.625566725512645e-08,
+      "loss": 0.2369,
+      "step": 22269
+    },
+    {
+      "epoch": 1.8977417980400513,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6228667008237054e-08,
+      "loss": 0.5543,
+      "step": 22270
+    },
+    {
+      "epoch": 1.8978270132083512,
+      "grad_norm": 13.25,
+      "learning_rate": 1.6201689056836167e-08,
+      "loss": 0.713,
+      "step": 22271
+    },
+    {
+      "epoch": 1.8979122283766512,
+      "grad_norm": 18.375,
+      "learning_rate": 1.6174733401411457e-08,
+      "loss": 0.7061,
+      "step": 22272
+    },
+    {
+      "epoch": 1.8979974435449511,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6147800042450035e-08,
+      "loss": 0.5003,
+      "step": 22273
+    },
+    {
+      "epoch": 1.898082658713251,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.612088898043859e-08,
+      "loss": 0.373,
+      "step": 22274
+    },
+    {
+      "epoch": 1.898167873881551,
+      "grad_norm": 14.625,
+      "learning_rate": 1.609400021586355e-08,
+      "loss": 0.333,
+      "step": 22275
+    },
+    {
+      "epoch": 1.898253089049851,
+      "grad_norm": 11.75,
+      "learning_rate": 1.6067133749210486e-08,
+      "loss": 0.3897,
+      "step": 22276
+    },
+    {
+      "epoch": 1.898338304218151,
+      "grad_norm": 77.0,
+      "learning_rate": 1.6040289580965264e-08,
+      "loss": 0.6086,
+      "step": 22277
+    },
+    {
+      "epoch": 1.898423519386451,
+      "grad_norm": 15.5,
+      "learning_rate": 1.6013467711612773e-08,
+      "loss": 0.4753,
+      "step": 22278
+    },
+    {
+      "epoch": 1.8985087345547509,
+      "grad_norm": 14.75,
+      "learning_rate": 1.5986668141638046e-08,
+      "loss": 0.5092,
+      "step": 22279
+    },
+    {
+      "epoch": 1.8985939497230508,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.595989087152486e-08,
+      "loss": 0.512,
+      "step": 22280
+    },
+    {
+      "epoch": 1.8986791648913508,
+      "grad_norm": 15.625,
+      "learning_rate": 1.5933135901757547e-08,
+      "loss": 0.5688,
+      "step": 22281
+    },
+    {
+      "epoch": 1.8987643800596508,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.5906403232819195e-08,
+      "loss": 0.5521,
+      "step": 22282
+    },
+    {
+      "epoch": 1.8988495952279507,
+      "grad_norm": 30.875,
+      "learning_rate": 1.587969286519331e-08,
+      "loss": 0.7264,
+      "step": 22283
+    },
+    {
+      "epoch": 1.8989348103962507,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5853004799362275e-08,
+      "loss": 0.3812,
+      "step": 22284
+    },
+    {
+      "epoch": 1.8990200255645506,
+      "grad_norm": 20.75,
+      "learning_rate": 1.582633903580849e-08,
+      "loss": 0.5939,
+      "step": 22285
+    },
+    {
+      "epoch": 1.8991052407328506,
+      "grad_norm": 13.375,
+      "learning_rate": 1.579969557501393e-08,
+      "loss": 0.3817,
+      "step": 22286
+    },
+    {
+      "epoch": 1.8991904559011505,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5773074417459873e-08,
+      "loss": 0.6632,
+      "step": 22287
+    },
+    {
+      "epoch": 1.8992756710694505,
+      "grad_norm": 18.75,
+      "learning_rate": 1.5746475563627463e-08,
+      "loss": 1.0349,
+      "step": 22288
+    },
+    {
+      "epoch": 1.8993608862377505,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.5719899013997427e-08,
+      "loss": 0.5587,
+      "step": 22289
+    },
+    {
+      "epoch": 1.8994461014060504,
+      "grad_norm": 14.125,
+      "learning_rate": 1.5693344769049794e-08,
+      "loss": 0.6674,
+      "step": 22290
+    },
+    {
+      "epoch": 1.8995313165743504,
+      "grad_norm": 13.375,
+      "learning_rate": 1.5666812829264744e-08,
+      "loss": 0.656,
+      "step": 22291
+    },
+    {
+      "epoch": 1.8996165317426503,
+      "grad_norm": 22.5,
+      "learning_rate": 1.564030319512175e-08,
+      "loss": 0.9626,
+      "step": 22292
+    },
+    {
+      "epoch": 1.8997017469109503,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5613815867099453e-08,
+      "loss": 0.4243,
+      "step": 22293
+    },
+    {
+      "epoch": 1.8997869620792502,
+      "grad_norm": 20.875,
+      "learning_rate": 1.558735084567692e-08,
+      "loss": 0.6271,
+      "step": 22294
+    },
+    {
+      "epoch": 1.8998721772475502,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.5560908131332237e-08,
+      "loss": 0.7042,
+      "step": 22295
+    },
+    {
+      "epoch": 1.8999573924158502,
+      "grad_norm": 16.625,
+      "learning_rate": 1.5534487724543356e-08,
+      "loss": 0.5418,
+      "step": 22296
+    },
+    {
+      "epoch": 1.9000426075841501,
+      "grad_norm": 20.25,
+      "learning_rate": 1.5508089625787535e-08,
+      "loss": 0.576,
+      "step": 22297
+    },
+    {
+      "epoch": 1.90012782275245,
+      "grad_norm": 20.5,
+      "learning_rate": 1.5481713835541894e-08,
+      "loss": 0.883,
+      "step": 22298
+    },
+    {
+      "epoch": 1.90021303792075,
+      "grad_norm": 17.0,
+      "learning_rate": 1.5455360354283135e-08,
+      "loss": 0.6079,
+      "step": 22299
+    },
+    {
+      "epoch": 1.90029825308905,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5429029182487397e-08,
+      "loss": 0.6916,
+      "step": 22300
+    },
+    {
+      "epoch": 1.90038346825735,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.540272032063056e-08,
+      "loss": 0.4265,
+      "step": 22301
+    },
+    {
+      "epoch": 1.90046868342565,
+      "grad_norm": 25.5,
+      "learning_rate": 1.537643376918807e-08,
+      "loss": 0.9618,
+      "step": 22302
+    },
+    {
+      "epoch": 1.9005538985939499,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.5350169528634966e-08,
+      "loss": 0.4291,
+      "step": 22303
+    },
+    {
+      "epoch": 1.9006391137622498,
+      "grad_norm": 16.75,
+      "learning_rate": 1.532392759944587e-08,
+      "loss": 0.459,
+      "step": 22304
+    },
+    {
+      "epoch": 1.9007243289305498,
+      "grad_norm": 13.5,
+      "learning_rate": 1.5297707982094983e-08,
+      "loss": 0.3995,
+      "step": 22305
+    },
+    {
+      "epoch": 1.9008095440988497,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5271510677055956e-08,
+      "loss": 0.9546,
+      "step": 22306
+    },
+    {
+      "epoch": 1.9008947592671497,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.5245335684802437e-08,
+      "loss": 0.5611,
+      "step": 22307
+    },
+    {
+      "epoch": 1.9009799744354496,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.5219183005807242e-08,
+      "loss": 0.3231,
+      "step": 22308
+    },
+    {
+      "epoch": 1.9010651896037496,
+      "grad_norm": 17.375,
+      "learning_rate": 1.519305264054319e-08,
+      "loss": 0.7909,
+      "step": 22309
+    },
+    {
+      "epoch": 1.9011504047720496,
+      "grad_norm": 17.375,
+      "learning_rate": 1.5166944589482258e-08,
+      "loss": 0.9723,
+      "step": 22310
+    },
+    {
+      "epoch": 1.9012356199403495,
+      "grad_norm": 20.375,
+      "learning_rate": 1.5140858853096437e-08,
+      "loss": 0.5954,
+      "step": 22311
+    },
+    {
+      "epoch": 1.9013208351086495,
+      "grad_norm": 19.375,
+      "learning_rate": 1.511479543185701e-08,
+      "loss": 0.8528,
+      "step": 22312
+    },
+    {
+      "epoch": 1.9014060502769494,
+      "grad_norm": 27.125,
+      "learning_rate": 1.5088754326235274e-08,
+      "loss": 0.5171,
+      "step": 22313
+    },
+    {
+      "epoch": 1.9014912654452494,
+      "grad_norm": 24.875,
+      "learning_rate": 1.5062735536701263e-08,
+      "loss": 0.484,
+      "step": 22314
+    },
+    {
+      "epoch": 1.9015764806135493,
+      "grad_norm": 20.25,
+      "learning_rate": 1.5036739063725436e-08,
+      "loss": 0.9198,
+      "step": 22315
+    },
+    {
+      "epoch": 1.9016616957818493,
+      "grad_norm": 17.0,
+      "learning_rate": 1.501076490777756e-08,
+      "loss": 0.6133,
+      "step": 22316
+    },
+    {
+      "epoch": 1.9017469109501492,
+      "grad_norm": 10.1875,
+      "learning_rate": 1.4984813069327253e-08,
+      "loss": 0.2487,
+      "step": 22317
+    },
+    {
+      "epoch": 1.9018321261184492,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4958883548843033e-08,
+      "loss": 0.4993,
+      "step": 22318
+    },
+    {
+      "epoch": 1.9019173412867492,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4932976346793832e-08,
+      "loss": 0.5389,
+      "step": 22319
+    },
+    {
+      "epoch": 1.9020025564550491,
+      "grad_norm": 19.5,
+      "learning_rate": 1.4907091463647745e-08,
+      "loss": 0.781,
+      "step": 22320
+    },
+    {
+      "epoch": 1.902087771623349,
+      "grad_norm": 10.625,
+      "learning_rate": 1.4881228899872452e-08,
+      "loss": 0.2053,
+      "step": 22321
+    },
+    {
+      "epoch": 1.902172986791649,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.48553886559355e-08,
+      "loss": 0.3877,
+      "step": 22322
+    },
+    {
+      "epoch": 1.902258201959949,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4829570732303456e-08,
+      "loss": 0.743,
+      "step": 22323
+    },
+    {
+      "epoch": 1.902343417128249,
+      "grad_norm": 23.0,
+      "learning_rate": 1.480377512944331e-08,
+      "loss": 0.9175,
+      "step": 22324
+    },
+    {
+      "epoch": 1.902428632296549,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.4778001847820938e-08,
+      "loss": 0.5835,
+      "step": 22325
+    },
+    {
+      "epoch": 1.9025138474648489,
+      "grad_norm": 23.0,
+      "learning_rate": 1.475225088790222e-08,
+      "loss": 1.0265,
+      "step": 22326
+    },
+    {
+      "epoch": 1.9025990626331488,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4726522250152475e-08,
+      "loss": 0.5368,
+      "step": 22327
+    },
+    {
+      "epoch": 1.9026842778014488,
+      "grad_norm": 17.25,
+      "learning_rate": 1.4700815935036611e-08,
+      "loss": 0.717,
+      "step": 22328
+    },
+    {
+      "epoch": 1.9027694929697487,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.4675131943019116e-08,
+      "loss": 0.4175,
+      "step": 22329
+    },
+    {
+      "epoch": 1.9028547081380487,
+      "grad_norm": 19.875,
+      "learning_rate": 1.4649470274564481e-08,
+      "loss": 0.7567,
+      "step": 22330
+    },
+    {
+      "epoch": 1.9029399233063486,
+      "grad_norm": 21.125,
+      "learning_rate": 1.4623830930135946e-08,
+      "loss": 0.7178,
+      "step": 22331
+    },
+    {
+      "epoch": 1.9030251384746486,
+      "grad_norm": 13.25,
+      "learning_rate": 1.4598213910197029e-08,
+      "loss": 0.6074,
+      "step": 22332
+    },
+    {
+      "epoch": 1.9031103536429486,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4572619215210692e-08,
+      "loss": 0.7601,
+      "step": 22333
+    },
+    {
+      "epoch": 1.9031955688112485,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.4547046845639485e-08,
+      "loss": 0.3952,
+      "step": 22334
+    },
+    {
+      "epoch": 1.9032807839795485,
+      "grad_norm": 19.5,
+      "learning_rate": 1.4521496801945533e-08,
+      "loss": 0.9361,
+      "step": 22335
+    },
+    {
+      "epoch": 1.9033659991478484,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.4495969084590417e-08,
+      "loss": 0.7269,
+      "step": 22336
+    },
+    {
+      "epoch": 1.9034512143161484,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4470463694035431e-08,
+      "loss": 0.6829,
+      "step": 22337
+    },
+    {
+      "epoch": 1.9035364294844483,
+      "grad_norm": 17.375,
+      "learning_rate": 1.4444980630741873e-08,
+      "loss": 0.5051,
+      "step": 22338
+    },
+    {
+      "epoch": 1.9036216446527483,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.4419519895169792e-08,
+      "loss": 0.3317,
+      "step": 22339
+    },
+    {
+      "epoch": 1.9037068598210483,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4394081487779516e-08,
+      "loss": 0.5243,
+      "step": 22340
+    },
+    {
+      "epoch": 1.9037920749893482,
+      "grad_norm": 16.5,
+      "learning_rate": 1.4368665409030536e-08,
+      "loss": 0.5212,
+      "step": 22341
+    },
+    {
+      "epoch": 1.9038772901576482,
+      "grad_norm": 20.125,
+      "learning_rate": 1.4343271659382484e-08,
+      "loss": 1.0732,
+      "step": 22342
+    },
+    {
+      "epoch": 1.9039625053259481,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.4317900239294025e-08,
+      "loss": 0.3886,
+      "step": 22343
+    },
+    {
+      "epoch": 1.904047720494248,
+      "grad_norm": 12.25,
+      "learning_rate": 1.4292551149223538e-08,
+      "loss": 0.4434,
+      "step": 22344
+    },
+    {
+      "epoch": 1.904132935662548,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4267224389629408e-08,
+      "loss": 0.4923,
+      "step": 22345
+    },
+    {
+      "epoch": 1.904218150830848,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4241919960969185e-08,
+      "loss": 0.6493,
+      "step": 22346
+    },
+    {
+      "epoch": 1.904303365999148,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.4216637863700006e-08,
+      "loss": 0.3618,
+      "step": 22347
+    },
+    {
+      "epoch": 1.904388581167448,
+      "grad_norm": 14.75,
+      "learning_rate": 1.4191378098278863e-08,
+      "loss": 0.2869,
+      "step": 22348
+    },
+    {
+      "epoch": 1.9044737963357479,
+      "grad_norm": 15.375,
+      "learning_rate": 1.4166140665162198e-08,
+      "loss": 0.5116,
+      "step": 22349
+    },
+    {
+      "epoch": 1.9045590115040478,
+      "grad_norm": 20.0,
+      "learning_rate": 1.4140925564806173e-08,
+      "loss": 0.8809,
+      "step": 22350
+    },
+    {
+      "epoch": 1.9046442266723478,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4115732797666399e-08,
+      "loss": 0.525,
+      "step": 22351
+    },
+    {
+      "epoch": 1.9047294418406477,
+      "grad_norm": 12.75,
+      "learning_rate": 1.4090562364198063e-08,
+      "loss": 0.5143,
+      "step": 22352
+    },
+    {
+      "epoch": 1.9048146570089477,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.406541426485608e-08,
+      "loss": 0.3479,
+      "step": 22353
+    },
+    {
+      "epoch": 1.9048998721772477,
+      "grad_norm": 27.125,
+      "learning_rate": 1.404028850009495e-08,
+      "loss": 0.6341,
+      "step": 22354
+    },
+    {
+      "epoch": 1.9049850873455476,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4015185070368476e-08,
+      "loss": 0.442,
+      "step": 22355
+    },
+    {
+      "epoch": 1.9050703025138476,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3990103976130737e-08,
+      "loss": 0.649,
+      "step": 22356
+    },
+    {
+      "epoch": 1.9051555176821475,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.3965045217834427e-08,
+      "loss": 0.528,
+      "step": 22357
+    },
+    {
+      "epoch": 1.9052407328504475,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3940008795932935e-08,
+      "loss": 0.5905,
+      "step": 22358
+    },
+    {
+      "epoch": 1.9053259480187474,
+      "grad_norm": 18.5,
+      "learning_rate": 1.391499471087826e-08,
+      "loss": 0.5399,
+      "step": 22359
+    },
+    {
+      "epoch": 1.9054111631870474,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.3890002963122679e-08,
+      "loss": 0.5119,
+      "step": 22360
+    },
+    {
+      "epoch": 1.9054963783553474,
+      "grad_norm": 11.875,
+      "learning_rate": 1.3865033553117773e-08,
+      "loss": 0.4171,
+      "step": 22361
+    },
+    {
+      "epoch": 1.9055815935236473,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3840086481314852e-08,
+      "loss": 0.9304,
+      "step": 22362
+    },
+    {
+      "epoch": 1.9056668086919473,
+      "grad_norm": 17.375,
+      "learning_rate": 1.3815161748164385e-08,
+      "loss": 0.7465,
+      "step": 22363
+    },
+    {
+      "epoch": 1.9057520238602472,
+      "grad_norm": 13.375,
+      "learning_rate": 1.3790259354117264e-08,
+      "loss": 0.4436,
+      "step": 22364
+    },
+    {
+      "epoch": 1.9058372390285472,
+      "grad_norm": 22.0,
+      "learning_rate": 1.3765379299623128e-08,
+      "loss": 0.5752,
+      "step": 22365
+    },
+    {
+      "epoch": 1.9059224541968471,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.3740521585131617e-08,
+      "loss": 0.7235,
+      "step": 22366
+    },
+    {
+      "epoch": 1.906007669365147,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.3715686211092233e-08,
+      "loss": 0.4148,
+      "step": 22367
+    },
+    {
+      "epoch": 1.906092884533447,
+      "grad_norm": 17.125,
+      "learning_rate": 1.3690873177953507e-08,
+      "loss": 0.854,
+      "step": 22368
+    },
+    {
+      "epoch": 1.906178099701747,
+      "grad_norm": 17.0,
+      "learning_rate": 1.3666082486163829e-08,
+      "loss": 0.6283,
+      "step": 22369
+    },
+    {
+      "epoch": 1.906263314870047,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.3641314136171452e-08,
+      "loss": 0.3706,
+      "step": 22370
+    },
+    {
+      "epoch": 1.906348530038347,
+      "grad_norm": 21.125,
+      "learning_rate": 1.3616568128423658e-08,
+      "loss": 0.7561,
+      "step": 22371
+    },
+    {
+      "epoch": 1.9064337452066469,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.3591844463367726e-08,
+      "loss": 0.7676,
+      "step": 22372
+    },
+    {
+      "epoch": 1.9065189603749468,
+      "grad_norm": 15.625,
+      "learning_rate": 1.3567143141450523e-08,
+      "loss": 0.6314,
+      "step": 22373
+    },
+    {
+      "epoch": 1.9066041755432468,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.3542464163118358e-08,
+      "loss": 0.4293,
+      "step": 22374
+    },
+    {
+      "epoch": 1.9066893907115467,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.3517807528817123e-08,
+      "loss": 0.7607,
+      "step": 22375
+    },
+    {
+      "epoch": 1.9067746058798467,
+      "grad_norm": 16.875,
+      "learning_rate": 1.3493173238992574e-08,
+      "loss": 0.853,
+      "step": 22376
+    },
+    {
+      "epoch": 1.9068598210481467,
+      "grad_norm": 16.5,
+      "learning_rate": 1.346856129408991e-08,
+      "loss": 0.7179,
+      "step": 22377
+    },
+    {
+      "epoch": 1.9069450362164466,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.3443971694553636e-08,
+      "loss": 0.356,
+      "step": 22378
+    },
+    {
+      "epoch": 1.9070302513847466,
+      "grad_norm": 18.0,
+      "learning_rate": 1.341940444082826e-08,
+      "loss": 0.9654,
+      "step": 22379
+    },
+    {
+      "epoch": 1.9071154665530465,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.3394859533357729e-08,
+      "loss": 0.5506,
+      "step": 22380
+    },
+    {
+      "epoch": 1.9072006817213465,
+      "grad_norm": 16.875,
+      "learning_rate": 1.337033697258558e-08,
+      "loss": 0.5317,
+      "step": 22381
+    },
+    {
+      "epoch": 1.9072858968896464,
+      "grad_norm": 18.75,
+      "learning_rate": 1.3345836758955067e-08,
+      "loss": 0.7054,
+      "step": 22382
+    },
+    {
+      "epoch": 1.9073711120579464,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.3321358892908615e-08,
+      "loss": 0.2618,
+      "step": 22383
+    },
+    {
+      "epoch": 1.9074563272262464,
+      "grad_norm": 15.0,
+      "learning_rate": 1.3296903374888925e-08,
+      "loss": 0.5632,
+      "step": 22384
+    },
+    {
+      "epoch": 1.9075415423945463,
+      "grad_norm": 15.625,
+      "learning_rate": 1.3272470205337867e-08,
+      "loss": 0.6366,
+      "step": 22385
+    },
+    {
+      "epoch": 1.9076267575628463,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.3248059384696894e-08,
+      "loss": 0.5884,
+      "step": 22386
+    },
+    {
+      "epoch": 1.9077119727311462,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.3223670913407038e-08,
+      "loss": 0.4866,
+      "step": 22387
+    },
+    {
+      "epoch": 1.9077971878994462,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3199304791909063e-08,
+      "loss": 0.598,
+      "step": 22388
+    },
+    {
+      "epoch": 1.9078824030677461,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.3174961020643445e-08,
+      "loss": 0.7598,
+      "step": 22389
+    },
+    {
+      "epoch": 1.907967618236046,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3150639600050113e-08,
+      "loss": 0.7255,
+      "step": 22390
+    },
+    {
+      "epoch": 1.908052833404346,
+      "grad_norm": 22.75,
+      "learning_rate": 1.31263405305683e-08,
+      "loss": 0.5201,
+      "step": 22391
+    },
+    {
+      "epoch": 1.908138048572646,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3102063812637233e-08,
+      "loss": 0.7663,
+      "step": 22392
+    },
+    {
+      "epoch": 1.908223263740946,
+      "grad_norm": 12.75,
+      "learning_rate": 1.3077809446695871e-08,
+      "loss": 0.4035,
+      "step": 22393
+    },
+    {
+      "epoch": 1.908308478909246,
+      "grad_norm": 16.125,
+      "learning_rate": 1.3053577433182196e-08,
+      "loss": 0.5902,
+      "step": 22394
+    },
+    {
+      "epoch": 1.9083936940775459,
+      "grad_norm": 20.25,
+      "learning_rate": 1.302936777253433e-08,
+      "loss": 0.8054,
+      "step": 22395
+    },
+    {
+      "epoch": 1.9084789092458458,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3005180465189559e-08,
+      "loss": 0.8141,
+      "step": 22396
+    },
+    {
+      "epoch": 1.9085641244141458,
+      "grad_norm": 20.25,
+      "learning_rate": 1.2981015511585037e-08,
+      "loss": 0.8522,
+      "step": 22397
+    },
+    {
+      "epoch": 1.9086493395824458,
+      "grad_norm": 23.625,
+      "learning_rate": 1.2956872912157637e-08,
+      "loss": 0.6387,
+      "step": 22398
+    },
+    {
+      "epoch": 1.9087345547507457,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.2932752667343256e-08,
+      "loss": 0.6417,
+      "step": 22399
+    },
+    {
+      "epoch": 1.9088197699190457,
+      "grad_norm": 17.5,
+      "learning_rate": 1.2908654777578077e-08,
+      "loss": 0.6902,
+      "step": 22400
+    },
+    {
+      "epoch": 1.9089049850873456,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2884579243297585e-08,
+      "loss": 0.7657,
+      "step": 22401
+    },
+    {
+      "epoch": 1.9089902002556456,
+      "grad_norm": 15.25,
+      "learning_rate": 1.2860526064936707e-08,
+      "loss": 0.324,
+      "step": 22402
+    },
+    {
+      "epoch": 1.9090754154239455,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.2836495242930236e-08,
+      "loss": 0.5822,
+      "step": 22403
+    },
+    {
+      "epoch": 1.9091606305922455,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2812486777712408e-08,
+      "loss": 0.2957,
+      "step": 22404
+    },
+    {
+      "epoch": 1.9092458457605455,
+      "grad_norm": 18.125,
+      "learning_rate": 1.2788500669716902e-08,
+      "loss": 0.6466,
+      "step": 22405
+    },
+    {
+      "epoch": 1.9093310609288454,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2764536919377263e-08,
+      "loss": 0.699,
+      "step": 22406
+    },
+    {
+      "epoch": 1.9094162760971454,
+      "grad_norm": 18.875,
+      "learning_rate": 1.2740595527126615e-08,
+      "loss": 0.7627,
+      "step": 22407
+    },
+    {
+      "epoch": 1.9095014912654453,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2716676493397668e-08,
+      "loss": 0.4097,
+      "step": 22408
+    },
+    {
+      "epoch": 1.9095867064337453,
+      "grad_norm": 38.5,
+      "learning_rate": 1.2692779818622435e-08,
+      "loss": 1.0276,
+      "step": 22409
+    },
+    {
+      "epoch": 1.9096719216020452,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.2668905503232937e-08,
+      "loss": 0.7804,
+      "step": 22410
+    },
+    {
+      "epoch": 1.9097571367703452,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.2645053547660629e-08,
+      "loss": 0.4428,
+      "step": 22411
+    },
+    {
+      "epoch": 1.9098423519386452,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.262122395233642e-08,
+      "loss": 0.5864,
+      "step": 22412
+    },
+    {
+      "epoch": 1.9099275671069451,
+      "grad_norm": 14.0,
+      "learning_rate": 1.2597416717690797e-08,
+      "loss": 0.47,
+      "step": 22413
+    },
+    {
+      "epoch": 1.910012782275245,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2573631844154387e-08,
+      "loss": 0.5949,
+      "step": 22414
+    },
+    {
+      "epoch": 1.910097997443545,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.254986933215671e-08,
+      "loss": 0.69,
+      "step": 22415
+    },
+    {
+      "epoch": 1.910183212611845,
+      "grad_norm": 65.5,
+      "learning_rate": 1.2526129182127144e-08,
+      "loss": 0.7393,
+      "step": 22416
+    },
+    {
+      "epoch": 1.910268427780145,
+      "grad_norm": 11.0,
+      "learning_rate": 1.2502411394494928e-08,
+      "loss": 0.3549,
+      "step": 22417
+    },
+    {
+      "epoch": 1.910353642948445,
+      "grad_norm": 17.75,
+      "learning_rate": 1.2478715969688331e-08,
+      "loss": 0.7321,
+      "step": 22418
+    },
+    {
+      "epoch": 1.9104388581167449,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.245504290813604e-08,
+      "loss": 0.5578,
+      "step": 22419
+    },
+    {
+      "epoch": 1.9105240732850448,
+      "grad_norm": 11.5,
+      "learning_rate": 1.2431392210265487e-08,
+      "loss": 0.5024,
+      "step": 22420
+    },
+    {
+      "epoch": 1.9106092884533448,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2407763876504109e-08,
+      "loss": 0.629,
+      "step": 22421
+    },
+    {
+      "epoch": 1.9106945036216447,
+      "grad_norm": 19.625,
+      "learning_rate": 1.2384157907278926e-08,
+      "loss": 0.7235,
+      "step": 22422
+    },
+    {
+      "epoch": 1.9107797187899447,
+      "grad_norm": 14.0,
+      "learning_rate": 1.2360574303016543e-08,
+      "loss": 0.3473,
+      "step": 22423
+    },
+    {
+      "epoch": 1.9108649339582446,
+      "grad_norm": 24.875,
+      "learning_rate": 1.2337013064143144e-08,
+      "loss": 0.6643,
+      "step": 22424
+    },
+    {
+      "epoch": 1.9109501491265446,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.2313474191084363e-08,
+      "loss": 0.3494,
+      "step": 22425
+    },
+    {
+      "epoch": 1.9110353642948446,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.228995768426583e-08,
+      "loss": 0.6096,
+      "step": 22426
+    },
+    {
+      "epoch": 1.9111205794631445,
+      "grad_norm": 23.125,
+      "learning_rate": 1.2266463544112345e-08,
+      "loss": 0.8526,
+      "step": 22427
+    },
+    {
+      "epoch": 1.9112057946314445,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.2242991771048568e-08,
+      "loss": 0.5948,
+      "step": 22428
+    },
+    {
+      "epoch": 1.9112910097997444,
+      "grad_norm": 25.75,
+      "learning_rate": 1.2219542365498466e-08,
+      "loss": 0.6469,
+      "step": 22429
+    },
+    {
+      "epoch": 1.9113762249680444,
+      "grad_norm": 20.875,
+      "learning_rate": 1.2196115327885866e-08,
+      "loss": 0.943,
+      "step": 22430
+    },
+    {
+      "epoch": 1.9114614401363443,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.2172710658634318e-08,
+      "loss": 0.6619,
+      "step": 22431
+    },
+    {
+      "epoch": 1.9115466553046443,
+      "grad_norm": 17.375,
+      "learning_rate": 1.214932835816668e-08,
+      "loss": 0.7032,
+      "step": 22432
+    },
+    {
+      "epoch": 1.9116318704729443,
+      "grad_norm": 18.625,
+      "learning_rate": 1.212596842690525e-08,
+      "loss": 0.6572,
+      "step": 22433
+    },
+    {
+      "epoch": 1.9117170856412442,
+      "grad_norm": 16.625,
+      "learning_rate": 1.210263086527233e-08,
+      "loss": 0.588,
+      "step": 22434
+    },
+    {
+      "epoch": 1.9118023008095442,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2079315673689807e-08,
+      "loss": 0.8764,
+      "step": 22435
+    },
+    {
+      "epoch": 1.9118875159778441,
+      "grad_norm": 16.25,
+      "learning_rate": 1.205602285257887e-08,
+      "loss": 0.5839,
+      "step": 22436
+    },
+    {
+      "epoch": 1.911972731146144,
+      "grad_norm": 19.5,
+      "learning_rate": 1.203275240236043e-08,
+      "loss": 1.0433,
+      "step": 22437
+    },
+    {
+      "epoch": 1.912057946314444,
+      "grad_norm": 23.0,
+      "learning_rate": 1.200950432345499e-08,
+      "loss": 0.8281,
+      "step": 22438
+    },
+    {
+      "epoch": 1.912143161482744,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.1986278616282764e-08,
+      "loss": 0.4818,
+      "step": 22439
+    },
+    {
+      "epoch": 1.912228376651044,
+      "grad_norm": 17.375,
+      "learning_rate": 1.196307528126342e-08,
+      "loss": 0.6498,
+      "step": 22440
+    },
+    {
+      "epoch": 1.912313591819344,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.1939894318816065e-08,
+      "loss": 0.5491,
+      "step": 22441
+    },
+    {
+      "epoch": 1.9123988069876439,
+      "grad_norm": 23.125,
+      "learning_rate": 1.1916735729359946e-08,
+      "loss": 0.7516,
+      "step": 22442
+    },
+    {
+      "epoch": 1.9124840221559438,
+      "grad_norm": 11.3125,
+      "learning_rate": 1.1893599513313341e-08,
+      "loss": 0.397,
+      "step": 22443
+    },
+    {
+      "epoch": 1.9125692373242438,
+      "grad_norm": 20.25,
+      "learning_rate": 1.1870485671094528e-08,
+      "loss": 0.6726,
+      "step": 22444
+    },
+    {
+      "epoch": 1.9126544524925437,
+      "grad_norm": 13.125,
+      "learning_rate": 1.184739420312095e-08,
+      "loss": 0.4048,
+      "step": 22445
+    },
+    {
+      "epoch": 1.9127396676608437,
+      "grad_norm": 15.125,
+      "learning_rate": 1.1824325109810048e-08,
+      "loss": 0.799,
+      "step": 22446
+    },
+    {
+      "epoch": 1.9128248828291436,
+      "grad_norm": 14.125,
+      "learning_rate": 1.1801278391578574e-08,
+      "loss": 0.4129,
+      "step": 22447
+    },
+    {
+      "epoch": 1.9129100979974436,
+      "grad_norm": 15.875,
+      "learning_rate": 1.177825404884328e-08,
+      "loss": 0.8818,
+      "step": 22448
+    },
+    {
+      "epoch": 1.9129953131657436,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1755252082019942e-08,
+      "loss": 0.5926,
+      "step": 22449
+    },
+    {
+      "epoch": 1.9130805283340435,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.1732272491524338e-08,
+      "loss": 0.4745,
+      "step": 22450
+    },
+    {
+      "epoch": 1.9131657435023435,
+      "grad_norm": 22.5,
+      "learning_rate": 1.1709315277771972e-08,
+      "loss": 0.6819,
+      "step": 22451
+    },
+    {
+      "epoch": 1.9132509586706434,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1686380441177231e-08,
+      "loss": 1.0185,
+      "step": 22452
+    },
+    {
+      "epoch": 1.9133361738389434,
+      "grad_norm": 13.0,
+      "learning_rate": 1.1663467982154925e-08,
+      "loss": 0.4636,
+      "step": 22453
+    },
+    {
+      "epoch": 1.9134213890072433,
+      "grad_norm": 19.0,
+      "learning_rate": 1.1640577901118888e-08,
+      "loss": 0.533,
+      "step": 22454
+    },
+    {
+      "epoch": 1.9135066041755433,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.1617710198482956e-08,
+      "loss": 0.5736,
+      "step": 22455
+    },
+    {
+      "epoch": 1.9135918193438433,
+      "grad_norm": 22.375,
+      "learning_rate": 1.1594864874660272e-08,
+      "loss": 0.6739,
+      "step": 22456
+    },
+    {
+      "epoch": 1.9136770345121432,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1572041930063698e-08,
+      "loss": 0.629,
+      "step": 22457
+    },
+    {
+      "epoch": 1.9137622496804432,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.1549241365105684e-08,
+      "loss": 0.5069,
+      "step": 22458
+    },
+    {
+      "epoch": 1.9138474648487431,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1526463180198122e-08,
+      "loss": 0.5691,
+      "step": 22459
+    },
+    {
+      "epoch": 1.913932680017043,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1503707375752904e-08,
+      "loss": 0.4273,
+      "step": 22460
+    },
+    {
+      "epoch": 1.914017895185343,
+      "grad_norm": 17.375,
+      "learning_rate": 1.1480973952181091e-08,
+      "loss": 0.6816,
+      "step": 22461
+    },
+    {
+      "epoch": 1.914103110353643,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1458262909893464e-08,
+      "loss": 0.5591,
+      "step": 22462
+    },
+    {
+      "epoch": 1.914188325521943,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.143557424930053e-08,
+      "loss": 0.8277,
+      "step": 22463
+    },
+    {
+      "epoch": 1.914273540690243,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1412907970812098e-08,
+      "loss": 0.7017,
+      "step": 22464
+    },
+    {
+      "epoch": 1.9143587558585429,
+      "grad_norm": 18.125,
+      "learning_rate": 1.1390264074837981e-08,
+      "loss": 1.0599,
+      "step": 22465
+    },
+    {
+      "epoch": 1.9144439710268428,
+      "grad_norm": 16.5,
+      "learning_rate": 1.1367642561787295e-08,
+      "loss": 0.5307,
+      "step": 22466
+    },
+    {
+      "epoch": 1.9145291861951428,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.1345043432068881e-08,
+      "loss": 0.6101,
+      "step": 22467
+    },
+    {
+      "epoch": 1.9146144013634427,
+      "grad_norm": 13.5,
+      "learning_rate": 1.132246668609116e-08,
+      "loss": 0.5551,
+      "step": 22468
+    },
+    {
+      "epoch": 1.9146996165317427,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.1299912324262002e-08,
+      "loss": 0.6443,
+      "step": 22469
+    },
+    {
+      "epoch": 1.9147848317000427,
+      "grad_norm": 16.875,
+      "learning_rate": 1.1277380346988997e-08,
+      "loss": 0.9283,
+      "step": 22470
+    },
+    {
+      "epoch": 1.9148700468683426,
+      "grad_norm": 18.375,
+      "learning_rate": 1.1254870754679458e-08,
+      "loss": 0.6707,
+      "step": 22471
+    },
+    {
+      "epoch": 1.9149552620366426,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.1232383547739866e-08,
+      "loss": 0.3311,
+      "step": 22472
+    },
+    {
+      "epoch": 1.9150404772049425,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1209918726576979e-08,
+      "loss": 0.3583,
+      "step": 22473
+    },
+    {
+      "epoch": 1.9151256923732425,
+      "grad_norm": 37.75,
+      "learning_rate": 1.1187476291596445e-08,
+      "loss": 0.6923,
+      "step": 22474
+    },
+    {
+      "epoch": 1.9152109075415424,
+      "grad_norm": 14.5,
+      "learning_rate": 1.1165056243204048e-08,
+      "loss": 0.5156,
+      "step": 22475
+    },
+    {
+      "epoch": 1.9152961227098424,
+      "grad_norm": 13.875,
+      "learning_rate": 1.1142658581804744e-08,
+      "loss": 0.5436,
+      "step": 22476
+    },
+    {
+      "epoch": 1.9153813378781424,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1120283307803348e-08,
+      "loss": 0.5007,
+      "step": 22477
+    },
+    {
+      "epoch": 1.9154665530464423,
+      "grad_norm": 13.125,
+      "learning_rate": 1.1097930421604259e-08,
+      "loss": 0.4662,
+      "step": 22478
+    },
+    {
+      "epoch": 1.9155517682147423,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.1075599923611458e-08,
+      "loss": 0.5888,
+      "step": 22479
+    },
+    {
+      "epoch": 1.9156369833830422,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.1053291814228235e-08,
+      "loss": 0.5537,
+      "step": 22480
+    },
+    {
+      "epoch": 1.9157221985513422,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1031006093858015e-08,
+      "loss": 0.3978,
+      "step": 22481
+    },
+    {
+      "epoch": 1.9158074137196421,
+      "grad_norm": 36.0,
+      "learning_rate": 1.1008742762903258e-08,
+      "loss": 0.7182,
+      "step": 22482
+    },
+    {
+      "epoch": 1.915892628887942,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0986501821766554e-08,
+      "loss": 0.4365,
+      "step": 22483
+    },
+    {
+      "epoch": 1.915977844056242,
+      "grad_norm": 21.5,
+      "learning_rate": 1.0964283270849668e-08,
+      "loss": 0.6692,
+      "step": 22484
+    },
+    {
+      "epoch": 1.916063059224542,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.0942087110554084e-08,
+      "loss": 0.4515,
+      "step": 22485
+    },
+    {
+      "epoch": 1.916148274392842,
+      "grad_norm": 13.875,
+      "learning_rate": 1.0919913341281008e-08,
+      "loss": 0.5465,
+      "step": 22486
+    },
+    {
+      "epoch": 1.916233489561142,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0897761963431091e-08,
+      "loss": 0.6176,
+      "step": 22487
+    },
+    {
+      "epoch": 1.9163187047294419,
+      "grad_norm": 16.625,
+      "learning_rate": 1.087563297740457e-08,
+      "loss": 0.7088,
+      "step": 22488
+    },
+    {
+      "epoch": 1.9164039198977418,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.0853526383601404e-08,
+      "loss": 0.6015,
+      "step": 22489
+    },
+    {
+      "epoch": 1.9164891350660418,
+      "grad_norm": 15.625,
+      "learning_rate": 1.0831442182420992e-08,
+      "loss": 0.7489,
+      "step": 22490
+    },
+    {
+      "epoch": 1.9165743502343418,
+      "grad_norm": 16.375,
+      "learning_rate": 1.080938037426274e-08,
+      "loss": 0.6621,
+      "step": 22491
+    },
+    {
+      "epoch": 1.9166595654026417,
+      "grad_norm": 20.625,
+      "learning_rate": 1.078734095952494e-08,
+      "loss": 0.395,
+      "step": 22492
+    },
+    {
+      "epoch": 1.9167447805709417,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.0765323938606021e-08,
+      "loss": 0.4263,
+      "step": 22493
+    },
+    {
+      "epoch": 1.9168299957392416,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.074332931190386e-08,
+      "loss": 0.5519,
+      "step": 22494
+    },
+    {
+      "epoch": 1.9169152109075416,
+      "grad_norm": 14.875,
+      "learning_rate": 1.072135707981578e-08,
+      "loss": 0.4628,
+      "step": 22495
+    },
+    {
+      "epoch": 1.9170004260758415,
+      "grad_norm": 20.0,
+      "learning_rate": 1.069940724273924e-08,
+      "loss": 0.5568,
+      "step": 22496
+    },
+    {
+      "epoch": 1.9170856412441415,
+      "grad_norm": 16.75,
+      "learning_rate": 1.067747980107045e-08,
+      "loss": 0.3768,
+      "step": 22497
+    },
+    {
+      "epoch": 1.9171708564124414,
+      "grad_norm": 20.5,
+      "learning_rate": 1.065557475520576e-08,
+      "loss": 0.6064,
+      "step": 22498
+    },
+    {
+      "epoch": 1.9172560715807414,
+      "grad_norm": 14.125,
+      "learning_rate": 1.0633692105541244e-08,
+      "loss": 0.6848,
+      "step": 22499
+    },
+    {
+      "epoch": 1.9173412867490414,
+      "grad_norm": 13.75,
+      "learning_rate": 1.0611831852472138e-08,
+      "loss": 0.8272,
+      "step": 22500
+    },
+    {
+      "epoch": 1.9174265019173413,
+      "grad_norm": 17.875,
+      "learning_rate": 1.0589993996393544e-08,
+      "loss": 0.8754,
+      "step": 22501
+    },
+    {
+      "epoch": 1.9175117170856413,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.0568178537700008e-08,
+      "loss": 0.4858,
+      "step": 22502
+    },
+    {
+      "epoch": 1.9175969322539412,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0546385476785937e-08,
+      "loss": 0.5525,
+      "step": 22503
+    },
+    {
+      "epoch": 1.9176821474222412,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0524614814045042e-08,
+      "loss": 0.5528,
+      "step": 22504
+    },
+    {
+      "epoch": 1.9177673625905411,
+      "grad_norm": 20.75,
+      "learning_rate": 1.0502866549870622e-08,
+      "loss": 0.7828,
+      "step": 22505
+    },
+    {
+      "epoch": 1.917852577758841,
+      "grad_norm": 12.625,
+      "learning_rate": 1.0481140684655833e-08,
+      "loss": 0.4273,
+      "step": 22506
+    },
+    {
+      "epoch": 1.917937792927141,
+      "grad_norm": 16.375,
+      "learning_rate": 1.0459437218793279e-08,
+      "loss": 0.5491,
+      "step": 22507
+    },
+    {
+      "epoch": 1.918023008095441,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.0437756152675282e-08,
+      "loss": 0.8989,
+      "step": 22508
+    },
+    {
+      "epoch": 1.918108223263741,
+      "grad_norm": 18.375,
+      "learning_rate": 1.0416097486693476e-08,
+      "loss": 0.5715,
+      "step": 22509
+    },
+    {
+      "epoch": 1.918193438432041,
+      "grad_norm": 17.5,
+      "learning_rate": 1.0394461221239216e-08,
+      "loss": 0.7097,
+      "step": 22510
+    },
+    {
+      "epoch": 1.9182786536003409,
+      "grad_norm": 22.75,
+      "learning_rate": 1.0372847356703574e-08,
+      "loss": 0.8584,
+      "step": 22511
+    },
+    {
+      "epoch": 1.9183638687686408,
+      "grad_norm": 9.75,
+      "learning_rate": 1.0351255893477075e-08,
+      "loss": 0.235,
+      "step": 22512
+    },
+    {
+      "epoch": 1.9184490839369408,
+      "grad_norm": 23.375,
+      "learning_rate": 1.0329686831950098e-08,
+      "loss": 1.3183,
+      "step": 22513
+    },
+    {
+      "epoch": 1.9185342991052408,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.0308140172512054e-08,
+      "loss": 0.3388,
+      "step": 22514
+    },
+    {
+      "epoch": 1.9186195142735407,
+      "grad_norm": 24.375,
+      "learning_rate": 1.0286615915552633e-08,
+      "loss": 0.7583,
+      "step": 22515
+    },
+    {
+      "epoch": 1.9187047294418407,
+      "grad_norm": 16.375,
+      "learning_rate": 1.026511406146069e-08,
+      "loss": 0.6557,
+      "step": 22516
+    },
+    {
+      "epoch": 1.9187899446101406,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.0243634610624802e-08,
+      "loss": 0.3889,
+      "step": 22517
+    },
+    {
+      "epoch": 1.9188751597784406,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0222177563433134e-08,
+      "loss": 0.7879,
+      "step": 22518
+    },
+    {
+      "epoch": 1.9189603749467405,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.0200742920273288e-08,
+      "loss": 0.5139,
+      "step": 22519
+    },
+    {
+      "epoch": 1.9190455901150405,
+      "grad_norm": 14.5,
+      "learning_rate": 1.0179330681532873e-08,
+      "loss": 0.5789,
+      "step": 22520
+    },
+    {
+      "epoch": 1.9191308052833405,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0157940847598662e-08,
+      "loss": 0.6491,
+      "step": 22521
+    },
+    {
+      "epoch": 1.9192160204516404,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.0136573418857153e-08,
+      "loss": 0.4337,
+      "step": 22522
+    },
+    {
+      "epoch": 1.9193012356199404,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0115228395694565e-08,
+      "loss": 0.6612,
+      "step": 22523
+    },
+    {
+      "epoch": 1.9193864507882403,
+      "grad_norm": 15.25,
+      "learning_rate": 1.009390577849656e-08,
+      "loss": 0.9463,
+      "step": 22524
+    },
+    {
+      "epoch": 1.9194716659565403,
+      "grad_norm": 41.25,
+      "learning_rate": 1.0072605567648525e-08,
+      "loss": 1.2677,
+      "step": 22525
+    },
+    {
+      "epoch": 1.9195568811248402,
+      "grad_norm": 37.25,
+      "learning_rate": 1.0051327763535429e-08,
+      "loss": 0.7296,
+      "step": 22526
+    },
+    {
+      "epoch": 1.9196420962931402,
+      "grad_norm": 13.25,
+      "learning_rate": 1.0030072366541688e-08,
+      "loss": 0.405,
+      "step": 22527
+    },
+    {
+      "epoch": 1.9197273114614402,
+      "grad_norm": 13.625,
+      "learning_rate": 1.0008839377051299e-08,
+      "loss": 0.3942,
+      "step": 22528
+    },
+    {
+      "epoch": 1.9198125266297401,
+      "grad_norm": 18.25,
+      "learning_rate": 9.987628795448123e-09,
+      "loss": 0.7356,
+      "step": 22529
+    },
+    {
+      "epoch": 1.91989774179804,
+      "grad_norm": 16.125,
+      "learning_rate": 9.966440622115603e-09,
+      "loss": 0.3869,
+      "step": 22530
+    },
+    {
+      "epoch": 1.91998295696634,
+      "grad_norm": 16.25,
+      "learning_rate": 9.94527485743621e-09,
+      "loss": 0.6267,
+      "step": 22531
+    },
+    {
+      "epoch": 1.92006817213464,
+      "grad_norm": 17.125,
+      "learning_rate": 9.924131501792834e-09,
+      "loss": 0.7686,
+      "step": 22532
+    },
+    {
+      "epoch": 1.92015338730294,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.903010555567389e-09,
+      "loss": 0.2718,
+      "step": 22533
+    },
+    {
+      "epoch": 1.92023860247124,
+      "grad_norm": 14.375,
+      "learning_rate": 9.881912019141516e-09,
+      "loss": 0.6447,
+      "step": 22534
+    },
+    {
+      "epoch": 1.9203238176395399,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.860835892896575e-09,
+      "loss": 0.3742,
+      "step": 22535
+    },
+    {
+      "epoch": 1.9204090328078398,
+      "grad_norm": 14.375,
+      "learning_rate": 9.839782177213236e-09,
+      "loss": 0.6716,
+      "step": 22536
+    },
+    {
+      "epoch": 1.9204942479761398,
+      "grad_norm": 18.25,
+      "learning_rate": 9.818750872472166e-09,
+      "loss": 0.4224,
+      "step": 22537
+    },
+    {
+      "epoch": 1.9205794631444397,
+      "grad_norm": 12.1875,
+      "learning_rate": 9.797741979053477e-09,
+      "loss": 0.4535,
+      "step": 22538
+    },
+    {
+      "epoch": 1.9206646783127397,
+      "grad_norm": 10.5,
+      "learning_rate": 9.776755497336588e-09,
+      "loss": 0.3764,
+      "step": 22539
+    },
+    {
+      "epoch": 1.9207498934810396,
+      "grad_norm": 24.25,
+      "learning_rate": 9.755791427700917e-09,
+      "loss": 0.7252,
+      "step": 22540
+    },
+    {
+      "epoch": 1.9208351086493396,
+      "grad_norm": 32.75,
+      "learning_rate": 9.734849770525329e-09,
+      "loss": 0.7087,
+      "step": 22541
+    },
+    {
+      "epoch": 1.9209203238176396,
+      "grad_norm": 10.25,
+      "learning_rate": 9.71393052618813e-09,
+      "loss": 0.38,
+      "step": 22542
+    },
+    {
+      "epoch": 1.9210055389859395,
+      "grad_norm": 13.25,
+      "learning_rate": 9.693033695067494e-09,
+      "loss": 0.3871,
+      "step": 22543
+    },
+    {
+      "epoch": 1.9210907541542395,
+      "grad_norm": 11.8125,
+      "learning_rate": 9.672159277540754e-09,
+      "loss": 0.3912,
+      "step": 22544
+    },
+    {
+      "epoch": 1.9211759693225394,
+      "grad_norm": 11.875,
+      "learning_rate": 9.651307273985527e-09,
+      "loss": 0.2991,
+      "step": 22545
+    },
+    {
+      "epoch": 1.9212611844908394,
+      "grad_norm": 17.75,
+      "learning_rate": 9.630477684778316e-09,
+      "loss": 0.6307,
+      "step": 22546
+    },
+    {
+      "epoch": 1.9213463996591393,
+      "grad_norm": 16.25,
+      "learning_rate": 9.609670510295765e-09,
+      "loss": 0.7497,
+      "step": 22547
+    },
+    {
+      "epoch": 1.9214316148274393,
+      "grad_norm": 15.75,
+      "learning_rate": 9.588885750913685e-09,
+      "loss": 0.5276,
+      "step": 22548
+    },
+    {
+      "epoch": 1.9215168299957393,
+      "grad_norm": 18.625,
+      "learning_rate": 9.568123407007884e-09,
+      "loss": 0.5032,
+      "step": 22549
+    },
+    {
+      "epoch": 1.9216020451640392,
+      "grad_norm": 15.0,
+      "learning_rate": 9.547383478953343e-09,
+      "loss": 0.6313,
+      "step": 22550
+    },
+    {
+      "epoch": 1.9216872603323392,
+      "grad_norm": 22.0,
+      "learning_rate": 9.526665967125038e-09,
+      "loss": 0.8362,
+      "step": 22551
+    },
+    {
+      "epoch": 1.9217724755006391,
+      "grad_norm": 12.875,
+      "learning_rate": 9.505970871897252e-09,
+      "loss": 0.5725,
+      "step": 22552
+    },
+    {
+      "epoch": 1.921857690668939,
+      "grad_norm": 15.3125,
+      "learning_rate": 9.485298193643854e-09,
+      "loss": 0.5713,
+      "step": 22553
+    },
+    {
+      "epoch": 1.921942905837239,
+      "grad_norm": 23.5,
+      "learning_rate": 9.464647932738713e-09,
+      "loss": 1.1379,
+      "step": 22554
+    },
+    {
+      "epoch": 1.922028121005539,
+      "grad_norm": 16.0,
+      "learning_rate": 9.444020089554861e-09,
+      "loss": 0.6007,
+      "step": 22555
+    },
+    {
+      "epoch": 1.922113336173839,
+      "grad_norm": 20.125,
+      "learning_rate": 9.423414664464919e-09,
+      "loss": 0.6331,
+      "step": 22556
+    },
+    {
+      "epoch": 1.922198551342139,
+      "grad_norm": 16.25,
+      "learning_rate": 9.402831657841366e-09,
+      "loss": 0.9929,
+      "step": 22557
+    },
+    {
+      "epoch": 1.9222837665104389,
+      "grad_norm": 13.5625,
+      "learning_rate": 9.382271070056265e-09,
+      "loss": 0.391,
+      "step": 22558
+    },
+    {
+      "epoch": 1.9223689816787388,
+      "grad_norm": 10.625,
+      "learning_rate": 9.361732901480985e-09,
+      "loss": 0.2398,
+      "step": 22559
+    },
+    {
+      "epoch": 1.9224541968470388,
+      "grad_norm": 13.75,
+      "learning_rate": 9.34121715248676e-09,
+      "loss": 0.6736,
+      "step": 22560
+    },
+    {
+      "epoch": 1.9225394120153387,
+      "grad_norm": 11.0625,
+      "learning_rate": 9.3207238234444e-09,
+      "loss": 0.3752,
+      "step": 22561
+    },
+    {
+      "epoch": 1.9226246271836387,
+      "grad_norm": 17.375,
+      "learning_rate": 9.300252914724028e-09,
+      "loss": 0.4634,
+      "step": 22562
+    },
+    {
+      "epoch": 1.9227098423519386,
+      "grad_norm": 14.625,
+      "learning_rate": 9.279804426695765e-09,
+      "loss": 0.5634,
+      "step": 22563
+    },
+    {
+      "epoch": 1.9227950575202386,
+      "grad_norm": 13.5,
+      "learning_rate": 9.259378359729038e-09,
+      "loss": 0.4986,
+      "step": 22564
+    },
+    {
+      "epoch": 1.9228802726885386,
+      "grad_norm": 23.75,
+      "learning_rate": 9.238974714193133e-09,
+      "loss": 1.2789,
+      "step": 22565
+    },
+    {
+      "epoch": 1.9229654878568385,
+      "grad_norm": 19.75,
+      "learning_rate": 9.218593490456507e-09,
+      "loss": 0.7566,
+      "step": 22566
+    },
+    {
+      "epoch": 1.9230507030251385,
+      "grad_norm": 16.0,
+      "learning_rate": 9.198234688887614e-09,
+      "loss": 0.7058,
+      "step": 22567
+    },
+    {
+      "epoch": 1.9231359181934384,
+      "grad_norm": 20.75,
+      "learning_rate": 9.177898309854494e-09,
+      "loss": 0.8693,
+      "step": 22568
+    },
+    {
+      "epoch": 1.9232211333617384,
+      "grad_norm": 15.8125,
+      "learning_rate": 9.157584353724353e-09,
+      "loss": 0.8775,
+      "step": 22569
+    },
+    {
+      "epoch": 1.9233063485300383,
+      "grad_norm": 13.4375,
+      "learning_rate": 9.137292820864397e-09,
+      "loss": 0.5003,
+      "step": 22570
+    },
+    {
+      "epoch": 1.9233915636983383,
+      "grad_norm": 16.75,
+      "learning_rate": 9.117023711641415e-09,
+      "loss": 0.5035,
+      "step": 22571
+    },
+    {
+      "epoch": 1.9234767788666383,
+      "grad_norm": 12.4375,
+      "learning_rate": 9.096777026421643e-09,
+      "loss": 0.2213,
+      "step": 22572
+    },
+    {
+      "epoch": 1.9235619940349382,
+      "grad_norm": 12.3125,
+      "learning_rate": 9.0765527655709e-09,
+      "loss": 0.4109,
+      "step": 22573
+    },
+    {
+      "epoch": 1.9236472092032382,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.056350929454588e-09,
+      "loss": 0.5318,
+      "step": 22574
+    },
+    {
+      "epoch": 1.9237324243715381,
+      "grad_norm": 18.625,
+      "learning_rate": 9.036171518437969e-09,
+      "loss": 0.5912,
+      "step": 22575
+    },
+    {
+      "epoch": 1.923817639539838,
+      "grad_norm": 15.125,
+      "learning_rate": 9.016014532885752e-09,
+      "loss": 0.4754,
+      "step": 22576
+    },
+    {
+      "epoch": 1.923902854708138,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.995879973161814e-09,
+      "loss": 0.5072,
+      "step": 22577
+    },
+    {
+      "epoch": 1.923988069876438,
+      "grad_norm": 18.125,
+      "learning_rate": 8.975767839630444e-09,
+      "loss": 0.4691,
+      "step": 22578
+    },
+    {
+      "epoch": 1.924073285044738,
+      "grad_norm": 16.375,
+      "learning_rate": 8.955678132654688e-09,
+      "loss": 0.7053,
+      "step": 22579
+    },
+    {
+      "epoch": 1.924158500213038,
+      "grad_norm": 17.375,
+      "learning_rate": 8.935610852598003e-09,
+      "loss": 0.6823,
+      "step": 22580
+    },
+    {
+      "epoch": 1.9242437153813379,
+      "grad_norm": 12.125,
+      "learning_rate": 8.91556599982274e-09,
+      "loss": 0.3331,
+      "step": 22581
+    },
+    {
+      "epoch": 1.9243289305496378,
+      "grad_norm": 15.125,
+      "learning_rate": 8.895543574691245e-09,
+      "loss": 0.6116,
+      "step": 22582
+    },
+    {
+      "epoch": 1.9244141457179378,
+      "grad_norm": 14.5,
+      "learning_rate": 8.875543577565177e-09,
+      "loss": 0.5752,
+      "step": 22583
+    },
+    {
+      "epoch": 1.9244993608862377,
+      "grad_norm": 14.375,
+      "learning_rate": 8.855566008806189e-09,
+      "loss": 0.4723,
+      "step": 22584
+    },
+    {
+      "epoch": 1.9245845760545377,
+      "grad_norm": 24.5,
+      "learning_rate": 8.835610868775241e-09,
+      "loss": 0.6455,
+      "step": 22585
+    },
+    {
+      "epoch": 1.9246697912228377,
+      "grad_norm": 18.375,
+      "learning_rate": 8.81567815783274e-09,
+      "loss": 0.6921,
+      "step": 22586
+    },
+    {
+      "epoch": 1.9247550063911376,
+      "grad_norm": 15.25,
+      "learning_rate": 8.795767876339094e-09,
+      "loss": 0.5211,
+      "step": 22587
+    },
+    {
+      "epoch": 1.9248402215594376,
+      "grad_norm": 16.375,
+      "learning_rate": 8.775880024654155e-09,
+      "loss": 0.7015,
+      "step": 22588
+    },
+    {
+      "epoch": 1.9249254367277375,
+      "grad_norm": 12.0625,
+      "learning_rate": 8.756014603137075e-09,
+      "loss": 0.3585,
+      "step": 22589
+    },
+    {
+      "epoch": 1.9250106518960375,
+      "grad_norm": 12.75,
+      "learning_rate": 8.736171612147154e-09,
+      "loss": 0.6216,
+      "step": 22590
+    },
+    {
+      "epoch": 1.9250958670643374,
+      "grad_norm": 15.125,
+      "learning_rate": 8.716351052042577e-09,
+      "loss": 0.6348,
+      "step": 22591
+    },
+    {
+      "epoch": 1.9251810822326374,
+      "grad_norm": 31.625,
+      "learning_rate": 8.696552923181945e-09,
+      "loss": 0.9524,
+      "step": 22592
+    },
+    {
+      "epoch": 1.9252662974009374,
+      "grad_norm": 11.375,
+      "learning_rate": 8.676777225922751e-09,
+      "loss": 0.4501,
+      "step": 22593
+    },
+    {
+      "epoch": 1.9253515125692373,
+      "grad_norm": 14.1875,
+      "learning_rate": 8.657023960622345e-09,
+      "loss": 0.5972,
+      "step": 22594
+    },
+    {
+      "epoch": 1.9254367277375373,
+      "grad_norm": 16.625,
+      "learning_rate": 8.637293127637946e-09,
+      "loss": 0.4909,
+      "step": 22595
+    },
+    {
+      "epoch": 1.9255219429058372,
+      "grad_norm": 12.6875,
+      "learning_rate": 8.617584727325796e-09,
+      "loss": 0.3016,
+      "step": 22596
+    },
+    {
+      "epoch": 1.9256071580741372,
+      "grad_norm": 16.625,
+      "learning_rate": 8.597898760042272e-09,
+      "loss": 0.5652,
+      "step": 22597
+    },
+    {
+      "epoch": 1.9256923732424371,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.578235226143066e-09,
+      "loss": 0.5353,
+      "step": 22598
+    },
+    {
+      "epoch": 1.925777588410737,
+      "grad_norm": 12.625,
+      "learning_rate": 8.558594125983449e-09,
+      "loss": 0.3686,
+      "step": 22599
+    },
+    {
+      "epoch": 1.925862803579037,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.538975459918414e-09,
+      "loss": 0.623,
+      "step": 22600
+    },
+    {
+      "epoch": 1.925948018747337,
+      "grad_norm": 10.875,
+      "learning_rate": 8.519379228302398e-09,
+      "loss": 0.2783,
+      "step": 22601
+    },
+    {
+      "epoch": 1.926033233915637,
+      "grad_norm": 15.5,
+      "learning_rate": 8.499805431489844e-09,
+      "loss": 0.6176,
+      "step": 22602
+    },
+    {
+      "epoch": 1.926118449083937,
+      "grad_norm": 11.875,
+      "learning_rate": 8.480254069833938e-09,
+      "loss": 0.3249,
+      "step": 22603
+    },
+    {
+      "epoch": 1.9262036642522369,
+      "grad_norm": 19.375,
+      "learning_rate": 8.460725143688425e-09,
+      "loss": 0.7907,
+      "step": 22604
+    },
+    {
+      "epoch": 1.9262888794205368,
+      "grad_norm": 13.9375,
+      "learning_rate": 8.44121865340608e-09,
+      "loss": 0.4327,
+      "step": 22605
+    },
+    {
+      "epoch": 1.9263740945888368,
+      "grad_norm": 14.625,
+      "learning_rate": 8.421734599339537e-09,
+      "loss": 0.4961,
+      "step": 22606
+    },
+    {
+      "epoch": 1.9264593097571368,
+      "grad_norm": 15.375,
+      "learning_rate": 8.402272981840599e-09,
+      "loss": 0.6472,
+      "step": 22607
+    },
+    {
+      "epoch": 1.9265445249254367,
+      "grad_norm": 16.0,
+      "learning_rate": 8.382833801261065e-09,
+      "loss": 0.6273,
+      "step": 22608
+    },
+    {
+      "epoch": 1.9266297400937367,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.363417057952322e-09,
+      "loss": 0.4786,
+      "step": 22609
+    },
+    {
+      "epoch": 1.9267149552620366,
+      "grad_norm": 14.25,
+      "learning_rate": 8.344022752265201e-09,
+      "loss": 0.5719,
+      "step": 22610
+    },
+    {
+      "epoch": 1.9268001704303366,
+      "grad_norm": 13.875,
+      "learning_rate": 8.324650884550117e-09,
+      "loss": 0.408,
+      "step": 22611
+    },
+    {
+      "epoch": 1.9268853855986365,
+      "grad_norm": 12.875,
+      "learning_rate": 8.305301455157205e-09,
+      "loss": 0.5297,
+      "step": 22612
+    },
+    {
+      "epoch": 1.9269706007669365,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.285974464436186e-09,
+      "loss": 0.5254,
+      "step": 22613
+    },
+    {
+      "epoch": 1.9270558159352364,
+      "grad_norm": 16.5,
+      "learning_rate": 8.266669912736225e-09,
+      "loss": 0.7244,
+      "step": 22614
+    },
+    {
+      "epoch": 1.9271410311035364,
+      "grad_norm": 10.6875,
+      "learning_rate": 8.247387800406209e-09,
+      "loss": 0.2749,
+      "step": 22615
+    },
+    {
+      "epoch": 1.9272262462718364,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.228128127794472e-09,
+      "loss": 0.4528,
+      "step": 22616
+    },
+    {
+      "epoch": 1.9273114614401363,
+      "grad_norm": 14.25,
+      "learning_rate": 8.208890895249344e-09,
+      "loss": 0.5463,
+      "step": 22617
+    },
+    {
+      "epoch": 1.9273966766084363,
+      "grad_norm": 13.0625,
+      "learning_rate": 8.18967610311805e-09,
+      "loss": 0.4294,
+      "step": 22618
+    },
+    {
+      "epoch": 1.9274818917767362,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.170483751748226e-09,
+      "loss": 0.394,
+      "step": 22619
+    },
+    {
+      "epoch": 1.9275671069450362,
+      "grad_norm": 17.75,
+      "learning_rate": 8.151313841486402e-09,
+      "loss": 0.7623,
+      "step": 22620
+    },
+    {
+      "epoch": 1.9276523221133361,
+      "grad_norm": 20.125,
+      "learning_rate": 8.132166372679245e-09,
+      "loss": 0.9883,
+      "step": 22621
+    },
+    {
+      "epoch": 1.927737537281636,
+      "grad_norm": 14.125,
+      "learning_rate": 8.11304134567259e-09,
+      "loss": 0.4393,
+      "step": 22622
+    },
+    {
+      "epoch": 1.927822752449936,
+      "grad_norm": 17.5,
+      "learning_rate": 8.093938760811992e-09,
+      "loss": 0.6267,
+      "step": 22623
+    },
+    {
+      "epoch": 1.927907967618236,
+      "grad_norm": 19.25,
+      "learning_rate": 8.074858618442871e-09,
+      "loss": 1.0013,
+      "step": 22624
+    },
+    {
+      "epoch": 1.927993182786536,
+      "grad_norm": 14.0625,
+      "learning_rate": 8.055800918909813e-09,
+      "loss": 0.5435,
+      "step": 22625
+    },
+    {
+      "epoch": 1.928078397954836,
+      "grad_norm": 16.625,
+      "learning_rate": 8.036765662557262e-09,
+      "loss": 0.775,
+      "step": 22626
+    },
+    {
+      "epoch": 1.9281636131231359,
+      "grad_norm": 11.625,
+      "learning_rate": 8.017752849729387e-09,
+      "loss": 0.3296,
+      "step": 22627
+    },
+    {
+      "epoch": 1.9282488282914358,
+      "grad_norm": 15.0,
+      "learning_rate": 7.998762480769528e-09,
+      "loss": 0.4901,
+      "step": 22628
+    },
+    {
+      "epoch": 1.9283340434597358,
+      "grad_norm": 10.1875,
+      "learning_rate": 7.97979455602102e-09,
+      "loss": 0.2073,
+      "step": 22629
+    },
+    {
+      "epoch": 1.9284192586280358,
+      "grad_norm": 20.625,
+      "learning_rate": 7.960849075826643e-09,
+      "loss": 0.7853,
+      "step": 22630
+    },
+    {
+      "epoch": 1.9285044737963357,
+      "grad_norm": 15.125,
+      "learning_rate": 7.941926040528486e-09,
+      "loss": 0.4122,
+      "step": 22631
+    },
+    {
+      "epoch": 1.9285896889646357,
+      "grad_norm": 24.0,
+      "learning_rate": 7.923025450468914e-09,
+      "loss": 0.9443,
+      "step": 22632
+    },
+    {
+      "epoch": 1.9286749041329356,
+      "grad_norm": 14.5625,
+      "learning_rate": 7.90414730598918e-09,
+      "loss": 0.7375,
+      "step": 22633
+    },
+    {
+      "epoch": 1.9287601193012356,
+      "grad_norm": 14.0,
+      "learning_rate": 7.88529160743054e-09,
+      "loss": 0.364,
+      "step": 22634
+    },
+    {
+      "epoch": 1.9288453344695355,
+      "grad_norm": 14.75,
+      "learning_rate": 7.866458355133693e-09,
+      "loss": 0.4319,
+      "step": 22635
+    },
+    {
+      "epoch": 1.9289305496378355,
+      "grad_norm": 15.5,
+      "learning_rate": 7.847647549439064e-09,
+      "loss": 0.5604,
+      "step": 22636
+    },
+    {
+      "epoch": 1.9290157648061355,
+      "grad_norm": 20.5,
+      "learning_rate": 7.828859190686517e-09,
+      "loss": 1.0356,
+      "step": 22637
+    },
+    {
+      "epoch": 1.9291009799744354,
+      "grad_norm": 14.25,
+      "learning_rate": 7.810093279215502e-09,
+      "loss": 0.5609,
+      "step": 22638
+    },
+    {
+      "epoch": 1.9291861951427354,
+      "grad_norm": 22.25,
+      "learning_rate": 7.791349815365334e-09,
+      "loss": 0.7215,
+      "step": 22639
+    },
+    {
+      "epoch": 1.9292714103110353,
+      "grad_norm": 10.375,
+      "learning_rate": 7.77262879947463e-09,
+      "loss": 0.266,
+      "step": 22640
+    },
+    {
+      "epoch": 1.9293566254793353,
+      "grad_norm": 12.4375,
+      "learning_rate": 7.753930231881452e-09,
+      "loss": 0.433,
+      "step": 22641
+    },
+    {
+      "epoch": 1.9294418406476352,
+      "grad_norm": 13.0,
+      "learning_rate": 7.73525411292414e-09,
+      "loss": 0.4533,
+      "step": 22642
+    },
+    {
+      "epoch": 1.9295270558159352,
+      "grad_norm": 24.875,
+      "learning_rate": 7.716600442939926e-09,
+      "loss": 0.9608,
+      "step": 22643
+    },
+    {
+      "epoch": 1.9296122709842352,
+      "grad_norm": 16.875,
+      "learning_rate": 7.6979692222659e-09,
+      "loss": 0.5017,
+      "step": 22644
+    },
+    {
+      "epoch": 1.9296974861525351,
+      "grad_norm": 14.875,
+      "learning_rate": 7.679360451238877e-09,
+      "loss": 0.5903,
+      "step": 22645
+    },
+    {
+      "epoch": 1.929782701320835,
+      "grad_norm": 14.25,
+      "learning_rate": 7.660774130195114e-09,
+      "loss": 0.3448,
+      "step": 22646
+    },
+    {
+      "epoch": 1.929867916489135,
+      "grad_norm": 16.5,
+      "learning_rate": 7.642210259470318e-09,
+      "loss": 0.7383,
+      "step": 22647
+    },
+    {
+      "epoch": 1.929953131657435,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.62366883940005e-09,
+      "loss": 0.5643,
+      "step": 22648
+    },
+    {
+      "epoch": 1.930038346825735,
+      "grad_norm": 17.625,
+      "learning_rate": 7.605149870319462e-09,
+      "loss": 0.9396,
+      "step": 22649
+    },
+    {
+      "epoch": 1.930123561994035,
+      "grad_norm": 18.625,
+      "learning_rate": 7.586653352563145e-09,
+      "loss": 0.3825,
+      "step": 22650
+    },
+    {
+      "epoch": 1.9302087771623349,
+      "grad_norm": 35.0,
+      "learning_rate": 7.568179286465277e-09,
+      "loss": 1.088,
+      "step": 22651
+    },
+    {
+      "epoch": 1.9302939923306348,
+      "grad_norm": 17.875,
+      "learning_rate": 7.549727672359896e-09,
+      "loss": 0.9372,
+      "step": 22652
+    },
+    {
+      "epoch": 1.9303792074989348,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.53129851058021e-09,
+      "loss": 0.3839,
+      "step": 22653
+    },
+    {
+      "epoch": 1.9304644226672347,
+      "grad_norm": 16.875,
+      "learning_rate": 7.51289180145942e-09,
+      "loss": 0.8535,
+      "step": 22654
+    },
+    {
+      "epoch": 1.9305496378355347,
+      "grad_norm": 15.1875,
+      "learning_rate": 7.494507545330042e-09,
+      "loss": 0.44,
+      "step": 22655
+    },
+    {
+      "epoch": 1.9306348530038346,
+      "grad_norm": 16.25,
+      "learning_rate": 7.476145742524449e-09,
+      "loss": 0.7547,
+      "step": 22656
+    },
+    {
+      "epoch": 1.9307200681721346,
+      "grad_norm": 11.625,
+      "learning_rate": 7.457806393374317e-09,
+      "loss": 0.2323,
+      "step": 22657
+    },
+    {
+      "epoch": 1.9308052833404346,
+      "grad_norm": 15.75,
+      "learning_rate": 7.43948949821105e-09,
+      "loss": 0.5486,
+      "step": 22658
+    },
+    {
+      "epoch": 1.9308904985087345,
+      "grad_norm": 17.125,
+      "learning_rate": 7.421195057365632e-09,
+      "loss": 0.8541,
+      "step": 22659
+    },
+    {
+      "epoch": 1.9309757136770345,
+      "grad_norm": 14.75,
+      "learning_rate": 7.402923071168772e-09,
+      "loss": 0.6784,
+      "step": 22660
+    },
+    {
+      "epoch": 1.9310609288453344,
+      "grad_norm": 16.25,
+      "learning_rate": 7.3846735399506206e-09,
+      "loss": 0.6166,
+      "step": 22661
+    },
+    {
+      "epoch": 1.9311461440136344,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.366446464041055e-09,
+      "loss": 0.5266,
+      "step": 22662
+    },
+    {
+      "epoch": 1.9312313591819343,
+      "grad_norm": 16.625,
+      "learning_rate": 7.3482418437693936e-09,
+      "loss": 0.7341,
+      "step": 22663
+    },
+    {
+      "epoch": 1.9313165743502343,
+      "grad_norm": 12.75,
+      "learning_rate": 7.330059679464402e-09,
+      "loss": 0.4716,
+      "step": 22664
+    },
+    {
+      "epoch": 1.9314017895185343,
+      "grad_norm": 12.0,
+      "learning_rate": 7.311899971454983e-09,
+      "loss": 0.6512,
+      "step": 22665
+    },
+    {
+      "epoch": 1.9314870046868342,
+      "grad_norm": 15.25,
+      "learning_rate": 7.2937627200690695e-09,
+      "loss": 0.2825,
+      "step": 22666
+    },
+    {
+      "epoch": 1.9315722198551342,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.275647925634455e-09,
+      "loss": 0.6686,
+      "step": 22667
+    },
+    {
+      "epoch": 1.9316574350234341,
+      "grad_norm": 19.375,
+      "learning_rate": 7.257555588478654e-09,
+      "loss": 0.7956,
+      "step": 22668
+    },
+    {
+      "epoch": 1.931742650191734,
+      "grad_norm": 20.0,
+      "learning_rate": 7.2394857089284896e-09,
+      "loss": 0.6906,
+      "step": 22669
+    },
+    {
+      "epoch": 1.931827865360034,
+      "grad_norm": 15.25,
+      "learning_rate": 7.221438287310506e-09,
+      "loss": 0.7041,
+      "step": 22670
+    },
+    {
+      "epoch": 1.931913080528334,
+      "grad_norm": 17.125,
+      "learning_rate": 7.2034133239508304e-09,
+      "loss": 0.6499,
+      "step": 22671
+    },
+    {
+      "epoch": 1.931998295696634,
+      "grad_norm": 16.5,
+      "learning_rate": 7.185410819175176e-09,
+      "loss": 0.6882,
+      "step": 22672
+    },
+    {
+      "epoch": 1.932083510864934,
+      "grad_norm": 12.0625,
+      "learning_rate": 7.167430773308837e-09,
+      "loss": 0.3566,
+      "step": 22673
+    },
+    {
+      "epoch": 1.9321687260332339,
+      "grad_norm": 16.875,
+      "learning_rate": 7.1494731866769695e-09,
+      "loss": 0.2835,
+      "step": 22674
+    },
+    {
+      "epoch": 1.9322539412015338,
+      "grad_norm": 19.875,
+      "learning_rate": 7.131538059603759e-09,
+      "loss": 0.7884,
+      "step": 22675
+    },
+    {
+      "epoch": 1.9323391563698338,
+      "grad_norm": 11.9375,
+      "learning_rate": 7.113625392413393e-09,
+      "loss": 0.4734,
+      "step": 22676
+    },
+    {
+      "epoch": 1.9324243715381337,
+      "grad_norm": 14.125,
+      "learning_rate": 7.0957351854297754e-09,
+      "loss": 0.3725,
+      "step": 22677
+    },
+    {
+      "epoch": 1.9325095867064337,
+      "grad_norm": 15.5625,
+      "learning_rate": 7.0778674389759825e-09,
+      "loss": 0.3687,
+      "step": 22678
+    },
+    {
+      "epoch": 1.9325948018747336,
+      "grad_norm": 18.375,
+      "learning_rate": 7.06002215337509e-09,
+      "loss": 0.5184,
+      "step": 22679
+    },
+    {
+      "epoch": 1.9326800170430336,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.04219932894934e-09,
+      "loss": 0.5969,
+      "step": 22680
+    },
+    {
+      "epoch": 1.9327652322113336,
+      "grad_norm": 15.5,
+      "learning_rate": 7.0243989660209734e-09,
+      "loss": 0.7348,
+      "step": 22681
+    },
+    {
+      "epoch": 1.9328504473796335,
+      "grad_norm": 12.8125,
+      "learning_rate": 7.006621064911679e-09,
+      "loss": 0.5079,
+      "step": 22682
+    },
+    {
+      "epoch": 1.9329356625479335,
+      "grad_norm": 20.625,
+      "learning_rate": 6.988865625942726e-09,
+      "loss": 0.6684,
+      "step": 22683
+    },
+    {
+      "epoch": 1.9330208777162334,
+      "grad_norm": 20.375,
+      "learning_rate": 6.971132649434692e-09,
+      "loss": 1.1204,
+      "step": 22684
+    },
+    {
+      "epoch": 1.9331060928845334,
+      "grad_norm": 18.375,
+      "learning_rate": 6.953422135708432e-09,
+      "loss": 0.4149,
+      "step": 22685
+    },
+    {
+      "epoch": 1.9331913080528333,
+      "grad_norm": 18.375,
+      "learning_rate": 6.935734085083828e-09,
+      "loss": 0.6352,
+      "step": 22686
+    },
+    {
+      "epoch": 1.9332765232211333,
+      "grad_norm": 32.25,
+      "learning_rate": 6.918068497880487e-09,
+      "loss": 0.8614,
+      "step": 22687
+    },
+    {
+      "epoch": 1.9333617383894333,
+      "grad_norm": 20.25,
+      "learning_rate": 6.900425374417596e-09,
+      "loss": 0.8512,
+      "step": 22688
+    },
+    {
+      "epoch": 1.9334469535577332,
+      "grad_norm": 18.75,
+      "learning_rate": 6.8828047150140686e-09,
+      "loss": 0.8382,
+      "step": 22689
+    },
+    {
+      "epoch": 1.9335321687260332,
+      "grad_norm": 28.875,
+      "learning_rate": 6.86520651998826e-09,
+      "loss": 0.6198,
+      "step": 22690
+    },
+    {
+      "epoch": 1.9336173838943331,
+      "grad_norm": 14.125,
+      "learning_rate": 6.847630789658388e-09,
+      "loss": 0.686,
+      "step": 22691
+    },
+    {
+      "epoch": 1.933702599062633,
+      "grad_norm": 12.625,
+      "learning_rate": 6.830077524341838e-09,
+      "loss": 0.4367,
+      "step": 22692
+    },
+    {
+      "epoch": 1.933787814230933,
+      "grad_norm": 16.125,
+      "learning_rate": 6.8125467243558565e-09,
+      "loss": 0.6743,
+      "step": 22693
+    },
+    {
+      "epoch": 1.933873029399233,
+      "grad_norm": 15.5625,
+      "learning_rate": 6.795038390017272e-09,
+      "loss": 0.6767,
+      "step": 22694
+    },
+    {
+      "epoch": 1.933958244567533,
+      "grad_norm": 19.25,
+      "learning_rate": 6.7775525216425e-09,
+      "loss": 0.7401,
+      "step": 22695
+    },
+    {
+      "epoch": 1.934043459735833,
+      "grad_norm": 21.25,
+      "learning_rate": 6.760089119547536e-09,
+      "loss": 0.8233,
+      "step": 22696
+    },
+    {
+      "epoch": 1.9341286749041329,
+      "grad_norm": 12.0625,
+      "learning_rate": 6.742648184047962e-09,
+      "loss": 0.7685,
+      "step": 22697
+    },
+    {
+      "epoch": 1.9342138900724328,
+      "grad_norm": 20.875,
+      "learning_rate": 6.725229715458803e-09,
+      "loss": 0.8169,
+      "step": 22698
+    },
+    {
+      "epoch": 1.9342991052407328,
+      "grad_norm": 12.3125,
+      "learning_rate": 6.707833714095085e-09,
+      "loss": 0.5704,
+      "step": 22699
+    },
+    {
+      "epoch": 1.9343843204090327,
+      "grad_norm": 15.6875,
+      "learning_rate": 6.690460180271002e-09,
+      "loss": 0.5682,
+      "step": 22700
+    },
+    {
+      "epoch": 1.9344695355773327,
+      "grad_norm": 17.0,
+      "learning_rate": 6.673109114300469e-09,
+      "loss": 0.7426,
+      "step": 22701
+    },
+    {
+      "epoch": 1.9345547507456327,
+      "grad_norm": 13.375,
+      "learning_rate": 6.6557805164972625e-09,
+      "loss": 0.4975,
+      "step": 22702
+    },
+    {
+      "epoch": 1.9346399659139326,
+      "grad_norm": 21.0,
+      "learning_rate": 6.638474387174326e-09,
+      "loss": 0.8428,
+      "step": 22703
+    },
+    {
+      "epoch": 1.9347251810822326,
+      "grad_norm": 12.375,
+      "learning_rate": 6.621190726644467e-09,
+      "loss": 0.499,
+      "step": 22704
+    },
+    {
+      "epoch": 1.9348103962505325,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.6039295352199325e-09,
+      "loss": 0.6846,
+      "step": 22705
+    },
+    {
+      "epoch": 1.9348956114188325,
+      "grad_norm": 13.4375,
+      "learning_rate": 6.586690813212699e-09,
+      "loss": 0.5637,
+      "step": 22706
+    },
+    {
+      "epoch": 1.9349808265871324,
+      "grad_norm": 12.875,
+      "learning_rate": 6.569474560934319e-09,
+      "loss": 0.4794,
+      "step": 22707
+    },
+    {
+      "epoch": 1.9350660417554324,
+      "grad_norm": 14.25,
+      "learning_rate": 6.552280778696074e-09,
+      "loss": 0.6628,
+      "step": 22708
+    },
+    {
+      "epoch": 1.9351512569237324,
+      "grad_norm": 28.125,
+      "learning_rate": 6.535109466808271e-09,
+      "loss": 0.7128,
+      "step": 22709
+    },
+    {
+      "epoch": 1.9352364720920323,
+      "grad_norm": 16.75,
+      "learning_rate": 6.5179606255816315e-09,
+      "loss": 0.7488,
+      "step": 22710
+    },
+    {
+      "epoch": 1.9353216872603323,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.500834255325772e-09,
+      "loss": 0.6221,
+      "step": 22711
+    },
+    {
+      "epoch": 1.9354069024286322,
+      "grad_norm": 9.9375,
+      "learning_rate": 6.483730356350304e-09,
+      "loss": 0.3922,
+      "step": 22712
+    },
+    {
+      "epoch": 1.9354921175969322,
+      "grad_norm": 17.25,
+      "learning_rate": 6.4666489289642865e-09,
+      "loss": 0.5282,
+      "step": 22713
+    },
+    {
+      "epoch": 1.9355773327652321,
+      "grad_norm": 15.0625,
+      "learning_rate": 6.449589973476361e-09,
+      "loss": 0.458,
+      "step": 22714
+    },
+    {
+      "epoch": 1.935662547933532,
+      "grad_norm": 17.5,
+      "learning_rate": 6.432553490194754e-09,
+      "loss": 0.6653,
+      "step": 22715
+    },
+    {
+      "epoch": 1.935747763101832,
+      "grad_norm": 18.5,
+      "learning_rate": 6.4155394794275514e-09,
+      "loss": 0.7565,
+      "step": 22716
+    },
+    {
+      "epoch": 1.935832978270132,
+      "grad_norm": 13.3125,
+      "learning_rate": 6.398547941482008e-09,
+      "loss": 0.4604,
+      "step": 22717
+    },
+    {
+      "epoch": 1.935918193438432,
+      "grad_norm": 18.0,
+      "learning_rate": 6.38157887666524e-09,
+      "loss": 0.5182,
+      "step": 22718
+    },
+    {
+      "epoch": 1.936003408606732,
+      "grad_norm": 21.375,
+      "learning_rate": 6.364632285283806e-09,
+      "loss": 0.8154,
+      "step": 22719
+    },
+    {
+      "epoch": 1.9360886237750319,
+      "grad_norm": 14.5625,
+      "learning_rate": 6.3477081676441285e-09,
+      "loss": 0.6668,
+      "step": 22720
+    },
+    {
+      "epoch": 1.9361738389433318,
+      "grad_norm": 12.4375,
+      "learning_rate": 6.330806524052075e-09,
+      "loss": 0.4308,
+      "step": 22721
+    },
+    {
+      "epoch": 1.9362590541116318,
+      "grad_norm": 15.625,
+      "learning_rate": 6.313927354812816e-09,
+      "loss": 0.564,
+      "step": 22722
+    },
+    {
+      "epoch": 1.9363442692799318,
+      "grad_norm": 14.75,
+      "learning_rate": 6.297070660231525e-09,
+      "loss": 0.7012,
+      "step": 22723
+    },
+    {
+      "epoch": 1.9364294844482317,
+      "grad_norm": 19.125,
+      "learning_rate": 6.28023644061268e-09,
+      "loss": 0.5094,
+      "step": 22724
+    },
+    {
+      "epoch": 1.9365146996165317,
+      "grad_norm": 19.0,
+      "learning_rate": 6.2634246962609006e-09,
+      "loss": 0.5025,
+      "step": 22725
+    },
+    {
+      "epoch": 1.9365999147848316,
+      "grad_norm": 13.0,
+      "learning_rate": 6.246635427479414e-09,
+      "loss": 0.5157,
+      "step": 22726
+    },
+    {
+      "epoch": 1.9366851299531316,
+      "grad_norm": 21.0,
+      "learning_rate": 6.229868634572145e-09,
+      "loss": 0.7772,
+      "step": 22727
+    },
+    {
+      "epoch": 1.9367703451214315,
+      "grad_norm": 18.5,
+      "learning_rate": 6.213124317841768e-09,
+      "loss": 0.6077,
+      "step": 22728
+    },
+    {
+      "epoch": 1.9368555602897315,
+      "grad_norm": 25.375,
+      "learning_rate": 6.196402477590957e-09,
+      "loss": 0.7908,
+      "step": 22729
+    },
+    {
+      "epoch": 1.9369407754580314,
+      "grad_norm": 14.0,
+      "learning_rate": 6.179703114121971e-09,
+      "loss": 0.4777,
+      "step": 22730
+    },
+    {
+      "epoch": 1.9370259906263314,
+      "grad_norm": 20.875,
+      "learning_rate": 6.163026227736374e-09,
+      "loss": 0.6595,
+      "step": 22731
+    },
+    {
+      "epoch": 1.9371112057946314,
+      "grad_norm": 14.125,
+      "learning_rate": 6.1463718187357305e-09,
+      "loss": 0.4099,
+      "step": 22732
+    },
+    {
+      "epoch": 1.9371964209629313,
+      "grad_norm": 24.375,
+      "learning_rate": 6.129739887421049e-09,
+      "loss": 1.1084,
+      "step": 22733
+    },
+    {
+      "epoch": 1.9372816361312313,
+      "grad_norm": 10.8125,
+      "learning_rate": 6.113130434092645e-09,
+      "loss": 0.3347,
+      "step": 22734
+    },
+    {
+      "epoch": 1.9373668512995312,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.096543459050697e-09,
+      "loss": 0.8335,
+      "step": 22735
+    },
+    {
+      "epoch": 1.9374520664678312,
+      "grad_norm": 17.0,
+      "learning_rate": 6.07997896259524e-09,
+      "loss": 0.526,
+      "step": 22736
+    },
+    {
+      "epoch": 1.9375372816361311,
+      "grad_norm": 12.25,
+      "learning_rate": 6.063436945025341e-09,
+      "loss": 0.3248,
+      "step": 22737
+    },
+    {
+      "epoch": 1.937622496804431,
+      "grad_norm": 23.0,
+      "learning_rate": 6.046917406640068e-09,
+      "loss": 0.5618,
+      "step": 22738
+    },
+    {
+      "epoch": 1.937707711972731,
+      "grad_norm": 17.625,
+      "learning_rate": 6.030420347737792e-09,
+      "loss": 0.8767,
+      "step": 22739
+    },
+    {
+      "epoch": 1.937792927141031,
+      "grad_norm": 16.875,
+      "learning_rate": 6.013945768616885e-09,
+      "loss": 0.9016,
+      "step": 22740
+    },
+    {
+      "epoch": 1.937878142309331,
+      "grad_norm": 21.125,
+      "learning_rate": 5.9974936695747495e-09,
+      "loss": 1.1139,
+      "step": 22741
+    },
+    {
+      "epoch": 1.937963357477631,
+      "grad_norm": 15.625,
+      "learning_rate": 5.981064050908925e-09,
+      "loss": 0.4342,
+      "step": 22742
+    },
+    {
+      "epoch": 1.938048572645931,
+      "grad_norm": 14.875,
+      "learning_rate": 5.9646569129161165e-09,
+      "loss": 0.8037,
+      "step": 22743
+    },
+    {
+      "epoch": 1.9381337878142308,
+      "grad_norm": 15.1875,
+      "learning_rate": 5.948272255893173e-09,
+      "loss": 0.7155,
+      "step": 22744
+    },
+    {
+      "epoch": 1.9382190029825308,
+      "grad_norm": 23.875,
+      "learning_rate": 5.931910080135689e-09,
+      "loss": 0.8513,
+      "step": 22745
+    },
+    {
+      "epoch": 1.9383042181508308,
+      "grad_norm": 32.75,
+      "learning_rate": 5.9155703859398174e-09,
+      "loss": 0.8244,
+      "step": 22746
+    },
+    {
+      "epoch": 1.9383894333191307,
+      "grad_norm": 13.875,
+      "learning_rate": 5.899253173600461e-09,
+      "loss": 0.3749,
+      "step": 22747
+    },
+    {
+      "epoch": 1.9384746484874307,
+      "grad_norm": 18.375,
+      "learning_rate": 5.882958443412662e-09,
+      "loss": 0.7539,
+      "step": 22748
+    },
+    {
+      "epoch": 1.9385598636557306,
+      "grad_norm": 13.625,
+      "learning_rate": 5.866686195670768e-09,
+      "loss": 0.4786,
+      "step": 22749
+    },
+    {
+      "epoch": 1.9386450788240306,
+      "grad_norm": 17.5,
+      "learning_rate": 5.850436430669126e-09,
+      "loss": 0.6354,
+      "step": 22750
+    },
+    {
+      "epoch": 1.9387302939923305,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.834209148700976e-09,
+      "loss": 0.4538,
+      "step": 22751
+    },
+    {
+      "epoch": 1.9388155091606305,
+      "grad_norm": 19.375,
+      "learning_rate": 5.818004350059969e-09,
+      "loss": 0.7187,
+      "step": 22752
+    },
+    {
+      "epoch": 1.9389007243289305,
+      "grad_norm": 24.5,
+      "learning_rate": 5.801822035038651e-09,
+      "loss": 0.9284,
+      "step": 22753
+    },
+    {
+      "epoch": 1.9389859394972304,
+      "grad_norm": 20.375,
+      "learning_rate": 5.785662203929565e-09,
+      "loss": 0.7277,
+      "step": 22754
+    },
+    {
+      "epoch": 1.9390711546655304,
+      "grad_norm": 20.625,
+      "learning_rate": 5.7695248570246995e-09,
+      "loss": 0.7639,
+      "step": 22755
+    },
+    {
+      "epoch": 1.9391563698338303,
+      "grad_norm": 12.6875,
+      "learning_rate": 5.753409994615628e-09,
+      "loss": 0.5018,
+      "step": 22756
+    },
+    {
+      "epoch": 1.9392415850021303,
+      "grad_norm": 17.125,
+      "learning_rate": 5.737317616993782e-09,
+      "loss": 0.5999,
+      "step": 22757
+    },
+    {
+      "epoch": 1.9393268001704302,
+      "grad_norm": 23.0,
+      "learning_rate": 5.721247724449624e-09,
+      "loss": 1.0947,
+      "step": 22758
+    },
+    {
+      "epoch": 1.9394120153387302,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.705200317273896e-09,
+      "loss": 0.3761,
+      "step": 22759
+    },
+    {
+      "epoch": 1.9394972305070302,
+      "grad_norm": 13.9375,
+      "learning_rate": 5.6891753957565024e-09,
+      "loss": 0.5121,
+      "step": 22760
+    },
+    {
+      "epoch": 1.9395824456753301,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.673172960186796e-09,
+      "loss": 0.6826,
+      "step": 22761
+    },
+    {
+      "epoch": 1.93966766084363,
+      "grad_norm": 23.5,
+      "learning_rate": 5.6571930108541285e-09,
+      "loss": 1.0396,
+      "step": 22762
+    },
+    {
+      "epoch": 1.93975287601193,
+      "grad_norm": 21.5,
+      "learning_rate": 5.641235548047297e-09,
+      "loss": 1.1264,
+      "step": 22763
+    },
+    {
+      "epoch": 1.93983809118023,
+      "grad_norm": 12.25,
+      "learning_rate": 5.625300572054681e-09,
+      "loss": 0.35,
+      "step": 22764
+    },
+    {
+      "epoch": 1.93992330634853,
+      "grad_norm": 18.0,
+      "learning_rate": 5.609388083164108e-09,
+      "loss": 1.0244,
+      "step": 22765
+    },
+    {
+      "epoch": 1.94000852151683,
+      "grad_norm": 16.375,
+      "learning_rate": 5.593498081663262e-09,
+      "loss": 0.5349,
+      "step": 22766
+    },
+    {
+      "epoch": 1.9400937366851299,
+      "grad_norm": 23.125,
+      "learning_rate": 5.577630567839276e-09,
+      "loss": 0.874,
+      "step": 22767
+    },
+    {
+      "epoch": 1.9401789518534298,
+      "grad_norm": 14.8125,
+      "learning_rate": 5.561785541978865e-09,
+      "loss": 0.5889,
+      "step": 22768
+    },
+    {
+      "epoch": 1.9402641670217298,
+      "grad_norm": 15.25,
+      "learning_rate": 5.545963004368326e-09,
+      "loss": 0.3875,
+      "step": 22769
+    },
+    {
+      "epoch": 1.9403493821900297,
+      "grad_norm": 14.6875,
+      "learning_rate": 5.530162955293683e-09,
+      "loss": 0.5834,
+      "step": 22770
+    },
+    {
+      "epoch": 1.9404345973583297,
+      "grad_norm": 22.25,
+      "learning_rate": 5.514385395040401e-09,
+      "loss": 0.9507,
+      "step": 22771
+    },
+    {
+      "epoch": 1.9405198125266296,
+      "grad_norm": 11.125,
+      "learning_rate": 5.498630323893667e-09,
+      "loss": 0.2734,
+      "step": 22772
+    },
+    {
+      "epoch": 1.9406050276949296,
+      "grad_norm": 19.75,
+      "learning_rate": 5.482897742137977e-09,
+      "loss": 0.6552,
+      "step": 22773
+    },
+    {
+      "epoch": 1.9406902428632296,
+      "grad_norm": 14.625,
+      "learning_rate": 5.467187650057826e-09,
+      "loss": 0.5363,
+      "step": 22774
+    },
+    {
+      "epoch": 1.9407754580315295,
+      "grad_norm": 16.125,
+      "learning_rate": 5.451500047937153e-09,
+      "loss": 0.6275,
+      "step": 22775
+    },
+    {
+      "epoch": 1.9408606731998295,
+      "grad_norm": 18.75,
+      "learning_rate": 5.4358349360593425e-09,
+      "loss": 0.7074,
+      "step": 22776
+    },
+    {
+      "epoch": 1.9409458883681294,
+      "grad_norm": 14.1875,
+      "learning_rate": 5.420192314707501e-09,
+      "loss": 0.369,
+      "step": 22777
+    },
+    {
+      "epoch": 1.9410311035364294,
+      "grad_norm": 13.625,
+      "learning_rate": 5.404572184164181e-09,
+      "loss": 0.558,
+      "step": 22778
+    },
+    {
+      "epoch": 1.9411163187047293,
+      "grad_norm": 18.375,
+      "learning_rate": 5.388974544711933e-09,
+      "loss": 0.7846,
+      "step": 22779
+    },
+    {
+      "epoch": 1.9412015338730293,
+      "grad_norm": 24.125,
+      "learning_rate": 5.3733993966323396e-09,
+      "loss": 0.8802,
+      "step": 22780
+    },
+    {
+      "epoch": 1.9412867490413293,
+      "grad_norm": 11.8125,
+      "learning_rate": 5.357846740207118e-09,
+      "loss": 0.5686,
+      "step": 22781
+    },
+    {
+      "epoch": 1.9413719642096292,
+      "grad_norm": 15.25,
+      "learning_rate": 5.342316575717016e-09,
+      "loss": 0.6471,
+      "step": 22782
+    },
+    {
+      "epoch": 1.9414571793779292,
+      "grad_norm": 14.75,
+      "learning_rate": 5.326808903443059e-09,
+      "loss": 0.5192,
+      "step": 22783
+    },
+    {
+      "epoch": 1.9415423945462291,
+      "grad_norm": 13.5625,
+      "learning_rate": 5.311323723665163e-09,
+      "loss": 0.4873,
+      "step": 22784
+    },
+    {
+      "epoch": 1.941627609714529,
+      "grad_norm": 15.8125,
+      "learning_rate": 5.29586103666338e-09,
+      "loss": 0.6752,
+      "step": 22785
+    },
+    {
+      "epoch": 1.941712824882829,
+      "grad_norm": 14.375,
+      "learning_rate": 5.280420842716932e-09,
+      "loss": 0.6036,
+      "step": 22786
+    },
+    {
+      "epoch": 1.941798040051129,
+      "grad_norm": 14.5,
+      "learning_rate": 5.2650031421049015e-09,
+      "loss": 0.5152,
+      "step": 22787
+    },
+    {
+      "epoch": 1.941883255219429,
+      "grad_norm": 17.5,
+      "learning_rate": 5.2496079351059544e-09,
+      "loss": 0.691,
+      "step": 22788
+    },
+    {
+      "epoch": 1.941968470387729,
+      "grad_norm": 12.25,
+      "learning_rate": 5.234235221998479e-09,
+      "loss": 0.4721,
+      "step": 22789
+    },
+    {
+      "epoch": 1.9420536855560289,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.2188850030598925e-09,
+      "loss": 0.374,
+      "step": 22790
+    },
+    {
+      "epoch": 1.9421389007243288,
+      "grad_norm": 14.5625,
+      "learning_rate": 5.2035572785677505e-09,
+      "loss": 0.4807,
+      "step": 22791
+    },
+    {
+      "epoch": 1.9422241158926288,
+      "grad_norm": 25.0,
+      "learning_rate": 5.188252048799192e-09,
+      "loss": 0.554,
+      "step": 22792
+    },
+    {
+      "epoch": 1.9423093310609287,
+      "grad_norm": 13.1875,
+      "learning_rate": 5.172969314030663e-09,
+      "loss": 0.4729,
+      "step": 22793
+    },
+    {
+      "epoch": 1.9423945462292287,
+      "grad_norm": 20.125,
+      "learning_rate": 5.157709074538331e-09,
+      "loss": 0.4669,
+      "step": 22794
+    },
+    {
+      "epoch": 1.9424797613975286,
+      "grad_norm": 10.8125,
+      "learning_rate": 5.142471330597948e-09,
+      "loss": 0.5353,
+      "step": 22795
+    },
+    {
+      "epoch": 1.9425649765658286,
+      "grad_norm": 10.5625,
+      "learning_rate": 5.127256082484849e-09,
+      "loss": 0.3437,
+      "step": 22796
+    },
+    {
+      "epoch": 1.9426501917341286,
+      "grad_norm": 16.25,
+      "learning_rate": 5.112063330474232e-09,
+      "loss": 0.5875,
+      "step": 22797
+    },
+    {
+      "epoch": 1.9427354069024285,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.09689307484032e-09,
+      "loss": 0.5612,
+      "step": 22798
+    },
+    {
+      "epoch": 1.9428206220707285,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.08174531585734e-09,
+      "loss": 0.644,
+      "step": 22799
+    },
+    {
+      "epoch": 1.9429058372390284,
+      "grad_norm": 12.1875,
+      "learning_rate": 5.066620053799237e-09,
+      "loss": 0.5222,
+      "step": 22800
+    },
+    {
+      "epoch": 1.9429910524073284,
+      "grad_norm": 22.125,
+      "learning_rate": 5.051517288939128e-09,
+      "loss": 0.7805,
+      "step": 22801
+    },
+    {
+      "epoch": 1.9430762675756283,
+      "grad_norm": 15.375,
+      "learning_rate": 5.036437021549989e-09,
+      "loss": 0.4743,
+      "step": 22802
+    },
+    {
+      "epoch": 1.9431614827439283,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.021379251904379e-09,
+      "loss": 0.508,
+      "step": 22803
+    },
+    {
+      "epoch": 1.9432466979122283,
+      "grad_norm": 15.6875,
+      "learning_rate": 5.006343980274303e-09,
+      "loss": 0.6215,
+      "step": 22804
+    },
+    {
+      "epoch": 1.9433319130805282,
+      "grad_norm": 11.3125,
+      "learning_rate": 4.991331206931488e-09,
+      "loss": 0.392,
+      "step": 22805
+    },
+    {
+      "epoch": 1.9434171282488282,
+      "grad_norm": 19.125,
+      "learning_rate": 4.976340932147383e-09,
+      "loss": 0.6619,
+      "step": 22806
+    },
+    {
+      "epoch": 1.9435023434171281,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.961373156192745e-09,
+      "loss": 0.6131,
+      "step": 22807
+    },
+    {
+      "epoch": 1.943587558585428,
+      "grad_norm": 25.125,
+      "learning_rate": 4.946427879338051e-09,
+      "loss": 0.8166,
+      "step": 22808
+    },
+    {
+      "epoch": 1.943672773753728,
+      "grad_norm": 11.625,
+      "learning_rate": 4.931505101853501e-09,
+      "loss": 0.415,
+      "step": 22809
+    },
+    {
+      "epoch": 1.943757988922028,
+      "grad_norm": 16.5,
+      "learning_rate": 4.916604824008603e-09,
+      "loss": 0.7277,
+      "step": 22810
+    },
+    {
+      "epoch": 1.943843204090328,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.901727046072585e-09,
+      "loss": 0.822,
+      "step": 22811
+    },
+    {
+      "epoch": 1.943928419258628,
+      "grad_norm": 13.625,
+      "learning_rate": 4.8868717683144e-09,
+      "loss": 0.3708,
+      "step": 22812
+    },
+    {
+      "epoch": 1.9440136344269279,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.872038991002581e-09,
+      "loss": 0.4393,
+      "step": 22813
+    },
+    {
+      "epoch": 1.9440988495952278,
+      "grad_norm": 19.75,
+      "learning_rate": 4.85722871440511e-09,
+      "loss": 0.7644,
+      "step": 22814
+    },
+    {
+      "epoch": 1.9441840647635278,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.84244093878955e-09,
+      "loss": 0.5745,
+      "step": 22815
+    },
+    {
+      "epoch": 1.9442692799318277,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.827675664423187e-09,
+      "loss": 0.3719,
+      "step": 22816
+    },
+    {
+      "epoch": 1.9443544951001277,
+      "grad_norm": 18.625,
+      "learning_rate": 4.8129328915728915e-09,
+      "loss": 0.6716,
+      "step": 22817
+    },
+    {
+      "epoch": 1.9444397102684277,
+      "grad_norm": 24.0,
+      "learning_rate": 4.798212620505116e-09,
+      "loss": 0.6655,
+      "step": 22818
+    },
+    {
+      "epoch": 1.9445249254367276,
+      "grad_norm": 18.0,
+      "learning_rate": 4.783514851485621e-09,
+      "loss": 0.3232,
+      "step": 22819
+    },
+    {
+      "epoch": 1.9446101406050276,
+      "grad_norm": 16.0,
+      "learning_rate": 4.768839584780305e-09,
+      "loss": 0.7177,
+      "step": 22820
+    },
+    {
+      "epoch": 1.9446953557733275,
+      "grad_norm": 13.875,
+      "learning_rate": 4.754186820654094e-09,
+      "loss": 0.6823,
+      "step": 22821
+    },
+    {
+      "epoch": 1.9447805709416275,
+      "grad_norm": 21.125,
+      "learning_rate": 4.7395565593719165e-09,
+      "loss": 0.948,
+      "step": 22822
+    },
+    {
+      "epoch": 1.9448657861099274,
+      "grad_norm": 21.75,
+      "learning_rate": 4.7249488011982815e-09,
+      "loss": 0.479,
+      "step": 22823
+    },
+    {
+      "epoch": 1.9449510012782274,
+      "grad_norm": 24.375,
+      "learning_rate": 4.710363546397006e-09,
+      "loss": 0.4929,
+      "step": 22824
+    },
+    {
+      "epoch": 1.9450362164465274,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.695800795231631e-09,
+      "loss": 0.414,
+      "step": 22825
+    },
+    {
+      "epoch": 1.9451214316148273,
+      "grad_norm": 26.0,
+      "learning_rate": 4.681260547965416e-09,
+      "loss": 0.9006,
+      "step": 22826
+    },
+    {
+      "epoch": 1.9452066467831273,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.666742804861208e-09,
+      "loss": 0.3901,
+      "step": 22827
+    },
+    {
+      "epoch": 1.9452918619514272,
+      "grad_norm": 14.375,
+      "learning_rate": 4.652247566181156e-09,
+      "loss": 0.4674,
+      "step": 22828
+    },
+    {
+      "epoch": 1.9453770771197272,
+      "grad_norm": 19.625,
+      "learning_rate": 4.637774832187136e-09,
+      "loss": 0.4595,
+      "step": 22829
+    },
+    {
+      "epoch": 1.9454622922880271,
+      "grad_norm": 13.8125,
+      "learning_rate": 4.623324603141022e-09,
+      "loss": 0.5765,
+      "step": 22830
+    },
+    {
+      "epoch": 1.945547507456327,
+      "grad_norm": 14.0,
+      "learning_rate": 4.608896879303715e-09,
+      "loss": 0.4858,
+      "step": 22831
+    },
+    {
+      "epoch": 1.945632722624627,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.594491660935979e-09,
+      "loss": 0.7352,
+      "step": 22832
+    },
+    {
+      "epoch": 1.945717937792927,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.580108948298162e-09,
+      "loss": 0.4173,
+      "step": 22833
+    },
+    {
+      "epoch": 1.945803152961227,
+      "grad_norm": 10.8125,
+      "learning_rate": 4.565748741650055e-09,
+      "loss": 0.3293,
+      "step": 22834
+    },
+    {
+      "epoch": 1.945888368129527,
+      "grad_norm": 23.375,
+      "learning_rate": 4.5514110412511746e-09,
+      "loss": 0.5904,
+      "step": 22835
+    },
+    {
+      "epoch": 1.9459735832978269,
+      "grad_norm": 12.5,
+      "learning_rate": 4.537095847360895e-09,
+      "loss": 0.4584,
+      "step": 22836
+    },
+    {
+      "epoch": 1.9460587984661268,
+      "grad_norm": 17.25,
+      "learning_rate": 4.522803160237621e-09,
+      "loss": 0.627,
+      "step": 22837
+    },
+    {
+      "epoch": 1.9461440136344268,
+      "grad_norm": 16.125,
+      "learning_rate": 4.5085329801396186e-09,
+      "loss": 0.7121,
+      "step": 22838
+    },
+    {
+      "epoch": 1.9462292288027268,
+      "grad_norm": 14.375,
+      "learning_rate": 4.494285307325014e-09,
+      "loss": 0.5242,
+      "step": 22839
+    },
+    {
+      "epoch": 1.9463144439710267,
+      "grad_norm": 12.75,
+      "learning_rate": 4.480060142051102e-09,
+      "loss": 0.5207,
+      "step": 22840
+    },
+    {
+      "epoch": 1.9463996591393267,
+      "grad_norm": 17.125,
+      "learning_rate": 4.465857484574898e-09,
+      "loss": 0.6064,
+      "step": 22841
+    },
+    {
+      "epoch": 1.9464848743076266,
+      "grad_norm": 14.5,
+      "learning_rate": 4.4516773351531425e-09,
+      "loss": 0.5086,
+      "step": 22842
+    },
+    {
+      "epoch": 1.9465700894759266,
+      "grad_norm": 20.5,
+      "learning_rate": 4.4375196940421585e-09,
+      "loss": 0.4294,
+      "step": 22843
+    },
+    {
+      "epoch": 1.9466553046442265,
+      "grad_norm": 15.375,
+      "learning_rate": 4.423384561497573e-09,
+      "loss": 0.5493,
+      "step": 22844
+    },
+    {
+      "epoch": 1.9467405198125265,
+      "grad_norm": 16.125,
+      "learning_rate": 4.409271937775017e-09,
+      "loss": 0.3998,
+      "step": 22845
+    },
+    {
+      "epoch": 1.9468257349808265,
+      "grad_norm": 20.0,
+      "learning_rate": 4.395181823129424e-09,
+      "loss": 0.9498,
+      "step": 22846
+    },
+    {
+      "epoch": 1.9469109501491264,
+      "grad_norm": 13.9375,
+      "learning_rate": 4.381114217815452e-09,
+      "loss": 0.633,
+      "step": 22847
+    },
+    {
+      "epoch": 1.9469961653174264,
+      "grad_norm": 16.5,
+      "learning_rate": 4.367069122087481e-09,
+      "loss": 0.4816,
+      "step": 22848
+    },
+    {
+      "epoch": 1.9470813804857263,
+      "grad_norm": 23.375,
+      "learning_rate": 4.353046536199057e-09,
+      "loss": 0.871,
+      "step": 22849
+    },
+    {
+      "epoch": 1.9471665956540263,
+      "grad_norm": 17.25,
+      "learning_rate": 4.3390464604035895e-09,
+      "loss": 0.7952,
+      "step": 22850
+    },
+    {
+      "epoch": 1.9472518108223262,
+      "grad_norm": 12.5625,
+      "learning_rate": 4.325068894954209e-09,
+      "loss": 0.4503,
+      "step": 22851
+    },
+    {
+      "epoch": 1.9473370259906262,
+      "grad_norm": 13.75,
+      "learning_rate": 4.3111138401033535e-09,
+      "loss": 0.7788,
+      "step": 22852
+    },
+    {
+      "epoch": 1.9474222411589261,
+      "grad_norm": 18.125,
+      "learning_rate": 4.2971812961034584e-09,
+      "loss": 0.6312,
+      "step": 22853
+    },
+    {
+      "epoch": 1.947507456327226,
+      "grad_norm": 17.125,
+      "learning_rate": 4.283271263206129e-09,
+      "loss": 0.4665,
+      "step": 22854
+    },
+    {
+      "epoch": 1.947592671495526,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.269383741662692e-09,
+      "loss": 0.4937,
+      "step": 22855
+    },
+    {
+      "epoch": 1.947677886663826,
+      "grad_norm": 13.3125,
+      "learning_rate": 4.255518731724195e-09,
+      "loss": 0.4926,
+      "step": 22856
+    },
+    {
+      "epoch": 1.947763101832126,
+      "grad_norm": 14.875,
+      "learning_rate": 4.241676233641273e-09,
+      "loss": 0.5833,
+      "step": 22857
+    },
+    {
+      "epoch": 1.947848317000426,
+      "grad_norm": 17.75,
+      "learning_rate": 4.227856247663864e-09,
+      "loss": 0.7203,
+      "step": 22858
+    },
+    {
+      "epoch": 1.947933532168726,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.214058774041768e-09,
+      "loss": 0.5708,
+      "step": 22859
+    },
+    {
+      "epoch": 1.9480187473370258,
+      "grad_norm": 11.5,
+      "learning_rate": 4.20028381302437e-09,
+      "loss": 0.6724,
+      "step": 22860
+    },
+    {
+      "epoch": 1.9481039625053258,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.1865313648606356e-09,
+      "loss": 0.5599,
+      "step": 22861
+    },
+    {
+      "epoch": 1.9481891776736258,
+      "grad_norm": 13.5,
+      "learning_rate": 4.172801429799117e-09,
+      "loss": 0.586,
+      "step": 22862
+    },
+    {
+      "epoch": 1.9482743928419257,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.159094008087672e-09,
+      "loss": 0.5967,
+      "step": 22863
+    },
+    {
+      "epoch": 1.9483596080102257,
+      "grad_norm": 14.875,
+      "learning_rate": 4.145409099974296e-09,
+      "loss": 0.4505,
+      "step": 22864
+    },
+    {
+      "epoch": 1.9484448231785256,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.131746705706291e-09,
+      "loss": 0.5463,
+      "step": 22865
+    },
+    {
+      "epoch": 1.9485300383468256,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.1181068255302656e-09,
+      "loss": 0.317,
+      "step": 22866
+    },
+    {
+      "epoch": 1.9486152535151255,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.104489459692967e-09,
+      "loss": 0.7845,
+      "step": 22867
+    },
+    {
+      "epoch": 1.9487004686834255,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.09089460844031e-09,
+      "loss": 0.7015,
+      "step": 22868
+    },
+    {
+      "epoch": 1.9487856838517255,
+      "grad_norm": 17.0,
+      "learning_rate": 4.077322272018208e-09,
+      "loss": 0.5473,
+      "step": 22869
+    },
+    {
+      "epoch": 1.9488708990200254,
+      "grad_norm": 14.75,
+      "learning_rate": 4.063772450671605e-09,
+      "loss": 0.373,
+      "step": 22870
+    },
+    {
+      "epoch": 1.9489561141883254,
+      "grad_norm": 15.1875,
+      "learning_rate": 4.0502451446455816e-09,
+      "loss": 0.4812,
+      "step": 22871
+    },
+    {
+      "epoch": 1.9490413293566253,
+      "grad_norm": 15.6875,
+      "learning_rate": 4.036740354184526e-09,
+      "loss": 0.6789,
+      "step": 22872
+    },
+    {
+      "epoch": 1.9491265445249253,
+      "grad_norm": 17.625,
+      "learning_rate": 4.02325807953241e-09,
+      "loss": 0.7824,
+      "step": 22873
+    },
+    {
+      "epoch": 1.9492117596932252,
+      "grad_norm": 14.75,
+      "learning_rate": 4.009798320933067e-09,
+      "loss": 0.504,
+      "step": 22874
+    },
+    {
+      "epoch": 1.9492969748615252,
+      "grad_norm": 15.6875,
+      "learning_rate": 3.996361078629496e-09,
+      "loss": 0.5327,
+      "step": 22875
+    },
+    {
+      "epoch": 1.9493821900298252,
+      "grad_norm": 13.125,
+      "learning_rate": 3.982946352864697e-09,
+      "loss": 0.3957,
+      "step": 22876
+    },
+    {
+      "epoch": 1.9494674051981251,
+      "grad_norm": 15.625,
+      "learning_rate": 3.969554143881116e-09,
+      "loss": 0.5723,
+      "step": 22877
+    },
+    {
+      "epoch": 1.949552620366425,
+      "grad_norm": 15.625,
+      "learning_rate": 3.9561844519205025e-09,
+      "loss": 0.9371,
+      "step": 22878
+    },
+    {
+      "epoch": 1.949637835534725,
+      "grad_norm": 17.75,
+      "learning_rate": 3.942837277224748e-09,
+      "loss": 0.7208,
+      "step": 22879
+    },
+    {
+      "epoch": 1.949723050703025,
+      "grad_norm": 13.75,
+      "learning_rate": 3.929512620034909e-09,
+      "loss": 0.4366,
+      "step": 22880
+    },
+    {
+      "epoch": 1.949808265871325,
+      "grad_norm": 20.125,
+      "learning_rate": 3.916210480591765e-09,
+      "loss": 0.5498,
+      "step": 22881
+    },
+    {
+      "epoch": 1.949893481039625,
+      "grad_norm": 17.125,
+      "learning_rate": 3.9029308591356786e-09,
+      "loss": 0.82,
+      "step": 22882
+    },
+    {
+      "epoch": 1.9499786962079249,
+      "grad_norm": 14.125,
+      "learning_rate": 3.889673755906875e-09,
+      "loss": 0.6988,
+      "step": 22883
+    },
+    {
+      "epoch": 1.9500639113762248,
+      "grad_norm": 21.5,
+      "learning_rate": 3.876439171144469e-09,
+      "loss": 0.8951,
+      "step": 22884
+    },
+    {
+      "epoch": 1.9501491265445248,
+      "grad_norm": 19.625,
+      "learning_rate": 3.863227105087991e-09,
+      "loss": 0.7829,
+      "step": 22885
+    },
+    {
+      "epoch": 1.9502343417128247,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.850037557976139e-09,
+      "loss": 0.499,
+      "step": 22886
+    },
+    {
+      "epoch": 1.9503195568811247,
+      "grad_norm": 16.25,
+      "learning_rate": 3.8368705300471934e-09,
+      "loss": 0.6692,
+      "step": 22887
+    },
+    {
+      "epoch": 1.9504047720494246,
+      "grad_norm": 13.875,
+      "learning_rate": 3.82372602153902e-09,
+      "loss": 0.5806,
+      "step": 22888
+    },
+    {
+      "epoch": 1.9504899872177246,
+      "grad_norm": 15.3125,
+      "learning_rate": 3.810604032689347e-09,
+      "loss": 0.7795,
+      "step": 22889
+    },
+    {
+      "epoch": 1.9505752023860246,
+      "grad_norm": 18.0,
+      "learning_rate": 3.797504563735205e-09,
+      "loss": 0.6544,
+      "step": 22890
+    },
+    {
+      "epoch": 1.9506604175543245,
+      "grad_norm": 22.75,
+      "learning_rate": 3.784427614913349e-09,
+      "loss": 0.8243,
+      "step": 22891
+    },
+    {
+      "epoch": 1.9507456327226245,
+      "grad_norm": 24.0,
+      "learning_rate": 3.771373186459981e-09,
+      "loss": 0.8327,
+      "step": 22892
+    },
+    {
+      "epoch": 1.9508308478909244,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.758341278611022e-09,
+      "loss": 0.8261,
+      "step": 22893
+    },
+    {
+      "epoch": 1.9509160630592244,
+      "grad_norm": 16.5,
+      "learning_rate": 3.745331891602116e-09,
+      "loss": 0.6354,
+      "step": 22894
+    },
+    {
+      "epoch": 1.9510012782275243,
+      "grad_norm": 10.5625,
+      "learning_rate": 3.732345025668216e-09,
+      "loss": 0.3424,
+      "step": 22895
+    },
+    {
+      "epoch": 1.9510864933958243,
+      "grad_norm": 16.375,
+      "learning_rate": 3.7193806810439935e-09,
+      "loss": 0.5841,
+      "step": 22896
+    },
+    {
+      "epoch": 1.9511717085641243,
+      "grad_norm": 11.75,
+      "learning_rate": 3.7064388579639844e-09,
+      "loss": 0.3738,
+      "step": 22897
+    },
+    {
+      "epoch": 1.9512569237324242,
+      "grad_norm": 33.0,
+      "learning_rate": 3.693519556661751e-09,
+      "loss": 0.5823,
+      "step": 22898
+    },
+    {
+      "epoch": 1.9513421389007242,
+      "grad_norm": 19.375,
+      "learning_rate": 3.680622777370857e-09,
+      "loss": 0.67,
+      "step": 22899
+    },
+    {
+      "epoch": 1.9514273540690241,
+      "grad_norm": 12.3125,
+      "learning_rate": 3.667748520324449e-09,
+      "loss": 0.3177,
+      "step": 22900
+    },
+    {
+      "epoch": 1.951512569237324,
+      "grad_norm": 16.875,
+      "learning_rate": 3.6548967857551186e-09,
+      "loss": 0.8116,
+      "step": 22901
+    },
+    {
+      "epoch": 1.951597784405624,
+      "grad_norm": 15.0,
+      "learning_rate": 3.642067573895042e-09,
+      "loss": 0.8315,
+      "step": 22902
+    },
+    {
+      "epoch": 1.951682999573924,
+      "grad_norm": 10.1875,
+      "learning_rate": 3.6292608849761156e-09,
+      "loss": 0.2401,
+      "step": 22903
+    },
+    {
+      "epoch": 1.951768214742224,
+      "grad_norm": 19.75,
+      "learning_rate": 3.6164767192298222e-09,
+      "loss": 0.7365,
+      "step": 22904
+    },
+    {
+      "epoch": 1.951853429910524,
+      "grad_norm": 17.0,
+      "learning_rate": 3.603715076886949e-09,
+      "loss": 0.5623,
+      "step": 22905
+    },
+    {
+      "epoch": 1.951938645078824,
+      "grad_norm": 13.0,
+      "learning_rate": 3.590975958178561e-09,
+      "loss": 0.5289,
+      "step": 22906
+    },
+    {
+      "epoch": 1.952023860247124,
+      "grad_norm": 14.25,
+      "learning_rate": 3.5782593633344743e-09,
+      "loss": 0.344,
+      "step": 22907
+    },
+    {
+      "epoch": 1.952109075415424,
+      "grad_norm": 11.75,
+      "learning_rate": 3.565565292584644e-09,
+      "loss": 0.4612,
+      "step": 22908
+    },
+    {
+      "epoch": 1.952194290583724,
+      "grad_norm": 18.625,
+      "learning_rate": 3.5528937461583314e-09,
+      "loss": 0.6618,
+      "step": 22909
+    },
+    {
+      "epoch": 1.952279505752024,
+      "grad_norm": 24.75,
+      "learning_rate": 3.5402447242846584e-09,
+      "loss": 0.3983,
+      "step": 22910
+    },
+    {
+      "epoch": 1.9523647209203239,
+      "grad_norm": 12.375,
+      "learning_rate": 3.527618227192192e-09,
+      "loss": 0.5091,
+      "step": 22911
+    },
+    {
+      "epoch": 1.9524499360886238,
+      "grad_norm": 11.0625,
+      "learning_rate": 3.5150142551092225e-09,
+      "loss": 0.3522,
+      "step": 22912
+    },
+    {
+      "epoch": 1.9525351512569238,
+      "grad_norm": 15.75,
+      "learning_rate": 3.5024328082632063e-09,
+      "loss": 0.8253,
+      "step": 22913
+    },
+    {
+      "epoch": 1.9526203664252237,
+      "grad_norm": 16.25,
+      "learning_rate": 3.4898738868817394e-09,
+      "loss": 0.5758,
+      "step": 22914
+    },
+    {
+      "epoch": 1.9527055815935237,
+      "grad_norm": 16.125,
+      "learning_rate": 3.4773374911917233e-09,
+      "loss": 0.6468,
+      "step": 22915
+    },
+    {
+      "epoch": 1.9527907967618237,
+      "grad_norm": 13.5625,
+      "learning_rate": 3.4648236214197828e-09,
+      "loss": 0.4465,
+      "step": 22916
+    },
+    {
+      "epoch": 1.9528760119301236,
+      "grad_norm": 11.4375,
+      "learning_rate": 3.4523322777918476e-09,
+      "loss": 0.4035,
+      "step": 22917
+    },
+    {
+      "epoch": 1.9529612270984236,
+      "grad_norm": 17.125,
+      "learning_rate": 3.43986346053371e-09,
+      "loss": 0.7947,
+      "step": 22918
+    },
+    {
+      "epoch": 1.9530464422667235,
+      "grad_norm": 14.6875,
+      "learning_rate": 3.4274171698708835e-09,
+      "loss": 0.5772,
+      "step": 22919
+    },
+    {
+      "epoch": 1.9531316574350235,
+      "grad_norm": 17.125,
+      "learning_rate": 3.4149934060281886e-09,
+      "loss": 0.7187,
+      "step": 22920
+    },
+    {
+      "epoch": 1.9532168726033234,
+      "grad_norm": 20.375,
+      "learning_rate": 3.402592169230029e-09,
+      "loss": 0.6684,
+      "step": 22921
+    },
+    {
+      "epoch": 1.9533020877716234,
+      "grad_norm": 19.375,
+      "learning_rate": 3.390213459700531e-09,
+      "loss": 0.6214,
+      "step": 22922
+    },
+    {
+      "epoch": 1.9533873029399234,
+      "grad_norm": 21.25,
+      "learning_rate": 3.377857277663543e-09,
+      "loss": 1.0383,
+      "step": 22923
+    },
+    {
+      "epoch": 1.9534725181082233,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.365523623342082e-09,
+      "loss": 0.6702,
+      "step": 22924
+    },
+    {
+      "epoch": 1.9535577332765233,
+      "grad_norm": 13.25,
+      "learning_rate": 3.3532124969593017e-09,
+      "loss": 0.4363,
+      "step": 22925
+    },
+    {
+      "epoch": 1.9536429484448232,
+      "grad_norm": 27.5,
+      "learning_rate": 3.3409238987376637e-09,
+      "loss": 1.3392,
+      "step": 22926
+    },
+    {
+      "epoch": 1.9537281636131232,
+      "grad_norm": 32.0,
+      "learning_rate": 3.3286578288989345e-09,
+      "loss": 1.1013,
+      "step": 22927
+    },
+    {
+      "epoch": 1.9538133787814231,
+      "grad_norm": 15.625,
+      "learning_rate": 3.3164142876651595e-09,
+      "loss": 0.6894,
+      "step": 22928
+    },
+    {
+      "epoch": 1.953898593949723,
+      "grad_norm": 13.75,
+      "learning_rate": 3.3041932752574113e-09,
+      "loss": 0.3678,
+      "step": 22929
+    },
+    {
+      "epoch": 1.953983809118023,
+      "grad_norm": 13.5,
+      "learning_rate": 3.2919947918964856e-09,
+      "loss": 0.5912,
+      "step": 22930
+    },
+    {
+      "epoch": 1.954069024286323,
+      "grad_norm": 11.625,
+      "learning_rate": 3.2798188378029005e-09,
+      "loss": 0.3738,
+      "step": 22931
+    },
+    {
+      "epoch": 1.954154239454623,
+      "grad_norm": 15.0,
+      "learning_rate": 3.267665413196619e-09,
+      "loss": 0.5846,
+      "step": 22932
+    },
+    {
+      "epoch": 1.954239454622923,
+      "grad_norm": 12.875,
+      "learning_rate": 3.2555345182973266e-09,
+      "loss": 0.4001,
+      "step": 22933
+    },
+    {
+      "epoch": 1.9543246697912229,
+      "grad_norm": 13.5,
+      "learning_rate": 3.2434261533242915e-09,
+      "loss": 0.4544,
+      "step": 22934
+    },
+    {
+      "epoch": 1.9544098849595228,
+      "grad_norm": 13.875,
+      "learning_rate": 3.23134031849609e-09,
+      "loss": 0.4502,
+      "step": 22935
+    },
+    {
+      "epoch": 1.9544951001278228,
+      "grad_norm": 21.25,
+      "learning_rate": 3.2192770140314357e-09,
+      "loss": 1.0115,
+      "step": 22936
+    },
+    {
+      "epoch": 1.9545803152961228,
+      "grad_norm": 10.875,
+      "learning_rate": 3.2072362401482092e-09,
+      "loss": 0.3436,
+      "step": 22937
+    },
+    {
+      "epoch": 1.9546655304644227,
+      "grad_norm": 11.375,
+      "learning_rate": 3.1952179970638763e-09,
+      "loss": 0.292,
+      "step": 22938
+    },
+    {
+      "epoch": 1.9547507456327227,
+      "grad_norm": 17.5,
+      "learning_rate": 3.1832222849957626e-09,
+      "loss": 0.6814,
+      "step": 22939
+    },
+    {
+      "epoch": 1.9548359608010226,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.171249104160501e-09,
+      "loss": 0.7186,
+      "step": 22940
+    },
+    {
+      "epoch": 1.9549211759693226,
+      "grad_norm": 13.875,
+      "learning_rate": 3.159298454774584e-09,
+      "loss": 0.3969,
+      "step": 22941
+    },
+    {
+      "epoch": 1.9550063911376225,
+      "grad_norm": 19.5,
+      "learning_rate": 3.1473703370540896e-09,
+      "loss": 0.5557,
+      "step": 22942
+    },
+    {
+      "epoch": 1.9550916063059225,
+      "grad_norm": 15.9375,
+      "learning_rate": 3.1354647512142624e-09,
+      "loss": 0.7334,
+      "step": 22943
+    },
+    {
+      "epoch": 1.9551768214742224,
+      "grad_norm": 37.0,
+      "learning_rate": 3.123581697470346e-09,
+      "loss": 1.1417,
+      "step": 22944
+    },
+    {
+      "epoch": 1.9552620366425224,
+      "grad_norm": 17.625,
+      "learning_rate": 3.1117211760373077e-09,
+      "loss": 0.6734,
+      "step": 22945
+    },
+    {
+      "epoch": 1.9553472518108224,
+      "grad_norm": 13.375,
+      "learning_rate": 3.0998831871292823e-09,
+      "loss": 0.5857,
+      "step": 22946
+    },
+    {
+      "epoch": 1.9554324669791223,
+      "grad_norm": 17.0,
+      "learning_rate": 3.0880677309602646e-09,
+      "loss": 0.5576,
+      "step": 22947
+    },
+    {
+      "epoch": 1.9555176821474223,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.0762748077435566e-09,
+      "loss": 0.7198,
+      "step": 22948
+    },
+    {
+      "epoch": 1.9556028973157222,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.0645044176925985e-09,
+      "loss": 0.5303,
+      "step": 22949
+    },
+    {
+      "epoch": 1.9556881124840222,
+      "grad_norm": 14.5,
+      "learning_rate": 3.0527565610198596e-09,
+      "loss": 0.5773,
+      "step": 22950
+    },
+    {
+      "epoch": 1.9557733276523221,
+      "grad_norm": 12.0625,
+      "learning_rate": 3.0410312379376695e-09,
+      "loss": 0.3299,
+      "step": 22951
+    },
+    {
+      "epoch": 1.955858542820622,
+      "grad_norm": 14.125,
+      "learning_rate": 3.029328448658081e-09,
+      "loss": 0.576,
+      "step": 22952
+    },
+    {
+      "epoch": 1.955943757988922,
+      "grad_norm": 16.25,
+      "learning_rate": 3.0176481933923142e-09,
+      "loss": 0.8553,
+      "step": 22953
+    },
+    {
+      "epoch": 1.956028973157222,
+      "grad_norm": 17.25,
+      "learning_rate": 3.0059904723517275e-09,
+      "loss": 0.698,
+      "step": 22954
+    },
+    {
+      "epoch": 1.956114188325522,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.9943552857468473e-09,
+      "loss": 0.6108,
+      "step": 22955
+    },
+    {
+      "epoch": 1.956199403493822,
+      "grad_norm": 15.75,
+      "learning_rate": 2.9827426337877828e-09,
+      "loss": 0.5805,
+      "step": 22956
+    },
+    {
+      "epoch": 1.9562846186621219,
+      "grad_norm": 14.375,
+      "learning_rate": 2.9711525166846445e-09,
+      "loss": 0.6658,
+      "step": 22957
+    },
+    {
+      "epoch": 1.9563698338304218,
+      "grad_norm": 17.125,
+      "learning_rate": 2.9595849346467087e-09,
+      "loss": 0.4852,
+      "step": 22958
+    },
+    {
+      "epoch": 1.9564550489987218,
+      "grad_norm": 17.625,
+      "learning_rate": 2.948039887883114e-09,
+      "loss": 0.7131,
+      "step": 22959
+    },
+    {
+      "epoch": 1.9565402641670218,
+      "grad_norm": 19.25,
+      "learning_rate": 2.9365173766023048e-09,
+      "loss": 0.8092,
+      "step": 22960
+    },
+    {
+      "epoch": 1.9566254793353217,
+      "grad_norm": 21.375,
+      "learning_rate": 2.9250174010127254e-09,
+      "loss": 0.8703,
+      "step": 22961
+    },
+    {
+      "epoch": 1.9567106945036217,
+      "grad_norm": 13.5,
+      "learning_rate": 2.913539961322126e-09,
+      "loss": 0.596,
+      "step": 22962
+    },
+    {
+      "epoch": 1.9567959096719216,
+      "grad_norm": 14.1875,
+      "learning_rate": 2.90208505773798e-09,
+      "loss": 0.465,
+      "step": 22963
+    },
+    {
+      "epoch": 1.9568811248402216,
+      "grad_norm": 14.75,
+      "learning_rate": 2.890652690467205e-09,
+      "loss": 0.5611,
+      "step": 22964
+    },
+    {
+      "epoch": 1.9569663400085215,
+      "grad_norm": 17.625,
+      "learning_rate": 2.8792428597163023e-09,
+      "loss": 0.7974,
+      "step": 22965
+    },
+    {
+      "epoch": 1.9570515551768215,
+      "grad_norm": 14.875,
+      "learning_rate": 2.867855565691635e-09,
+      "loss": 0.7436,
+      "step": 22966
+    },
+    {
+      "epoch": 1.9571367703451215,
+      "grad_norm": 13.375,
+      "learning_rate": 2.856490808598872e-09,
+      "loss": 0.4579,
+      "step": 22967
+    },
+    {
+      "epoch": 1.9572219855134214,
+      "grad_norm": 19.375,
+      "learning_rate": 2.845148588643404e-09,
+      "loss": 0.9404,
+      "step": 22968
+    },
+    {
+      "epoch": 1.9573072006817214,
+      "grad_norm": 10.5,
+      "learning_rate": 2.833828906030206e-09,
+      "loss": 0.3313,
+      "step": 22969
+    },
+    {
+      "epoch": 1.9573924158500213,
+      "grad_norm": 14.25,
+      "learning_rate": 2.822531760963837e-09,
+      "loss": 0.578,
+      "step": 22970
+    },
+    {
+      "epoch": 1.9574776310183213,
+      "grad_norm": 14.375,
+      "learning_rate": 2.811257153648578e-09,
+      "loss": 0.5709,
+      "step": 22971
+    },
+    {
+      "epoch": 1.9575628461866212,
+      "grad_norm": 16.625,
+      "learning_rate": 2.8000050842880155e-09,
+      "loss": 0.639,
+      "step": 22972
+    },
+    {
+      "epoch": 1.9576480613549212,
+      "grad_norm": 11.75,
+      "learning_rate": 2.788775553085321e-09,
+      "loss": 0.3511,
+      "step": 22973
+    },
+    {
+      "epoch": 1.9577332765232212,
+      "grad_norm": 19.0,
+      "learning_rate": 2.777568560243804e-09,
+      "loss": 1.2777,
+      "step": 22974
+    },
+    {
+      "epoch": 1.9578184916915211,
+      "grad_norm": 15.25,
+      "learning_rate": 2.7663841059658025e-09,
+      "loss": 0.5507,
+      "step": 22975
+    },
+    {
+      "epoch": 1.957903706859821,
+      "grad_norm": 18.0,
+      "learning_rate": 2.7552221904533772e-09,
+      "loss": 0.5241,
+      "step": 22976
+    },
+    {
+      "epoch": 1.957988922028121,
+      "grad_norm": 10.0,
+      "learning_rate": 2.7440828139083107e-09,
+      "loss": 0.24,
+      "step": 22977
+    },
+    {
+      "epoch": 1.958074137196421,
+      "grad_norm": 17.375,
+      "learning_rate": 2.7329659765318317e-09,
+      "loss": 0.5549,
+      "step": 22978
+    },
+    {
+      "epoch": 1.958159352364721,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.72187167852489e-09,
+      "loss": 0.4907,
+      "step": 22979
+    },
+    {
+      "epoch": 1.958244567533021,
+      "grad_norm": 15.8125,
+      "learning_rate": 2.7107999200880197e-09,
+      "loss": 0.561,
+      "step": 22980
+    },
+    {
+      "epoch": 1.9583297827013209,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.6997507014212e-09,
+      "loss": 0.7476,
+      "step": 22981
+    },
+    {
+      "epoch": 1.9584149978696208,
+      "grad_norm": 15.5,
+      "learning_rate": 2.688724022724132e-09,
+      "loss": 0.6087,
+      "step": 22982
+    },
+    {
+      "epoch": 1.9585002130379208,
+      "grad_norm": 19.875,
+      "learning_rate": 2.6777198841961003e-09,
+      "loss": 0.774,
+      "step": 22983
+    },
+    {
+      "epoch": 1.9585854282062207,
+      "grad_norm": 13.25,
+      "learning_rate": 2.6667382860359746e-09,
+      "loss": 0.582,
+      "step": 22984
+    },
+    {
+      "epoch": 1.9586706433745207,
+      "grad_norm": 16.25,
+      "learning_rate": 2.655779228442068e-09,
+      "loss": 0.5083,
+      "step": 22985
+    },
+    {
+      "epoch": 1.9587558585428206,
+      "grad_norm": 13.75,
+      "learning_rate": 2.644842711612694e-09,
+      "loss": 0.5607,
+      "step": 22986
+    },
+    {
+      "epoch": 1.9588410737111206,
+      "grad_norm": 14.0625,
+      "learning_rate": 2.6339287357451947e-09,
+      "loss": 0.4884,
+      "step": 22987
+    },
+    {
+      "epoch": 1.9589262888794206,
+      "grad_norm": 15.3125,
+      "learning_rate": 2.6230373010370513e-09,
+      "loss": 0.3584,
+      "step": 22988
+    },
+    {
+      "epoch": 1.9590115040477205,
+      "grad_norm": 26.25,
+      "learning_rate": 2.612168407684912e-09,
+      "loss": 0.6188,
+      "step": 22989
+    },
+    {
+      "epoch": 1.9590967192160205,
+      "grad_norm": 16.125,
+      "learning_rate": 2.601322055885147e-09,
+      "loss": 0.633,
+      "step": 22990
+    },
+    {
+      "epoch": 1.9591819343843204,
+      "grad_norm": 14.75,
+      "learning_rate": 2.5904982458339888e-09,
+      "loss": 0.604,
+      "step": 22991
+    },
+    {
+      "epoch": 1.9592671495526204,
+      "grad_norm": 22.625,
+      "learning_rate": 2.579696977726698e-09,
+      "loss": 0.7801,
+      "step": 22992
+    },
+    {
+      "epoch": 1.9593523647209203,
+      "grad_norm": 15.875,
+      "learning_rate": 2.568918251758812e-09,
+      "loss": 0.5949,
+      "step": 22993
+    },
+    {
+      "epoch": 1.9594375798892203,
+      "grad_norm": 26.375,
+      "learning_rate": 2.5581620681250364e-09,
+      "loss": 0.7303,
+      "step": 22994
+    },
+    {
+      "epoch": 1.9595227950575203,
+      "grad_norm": 13.875,
+      "learning_rate": 2.5474284270195214e-09,
+      "loss": 0.4028,
+      "step": 22995
+    },
+    {
+      "epoch": 1.9596080102258202,
+      "grad_norm": 18.75,
+      "learning_rate": 2.5367173286364178e-09,
+      "loss": 1.0238,
+      "step": 22996
+    },
+    {
+      "epoch": 1.9596932253941202,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.52602877316932e-09,
+      "loss": 0.2802,
+      "step": 22997
+    },
+    {
+      "epoch": 1.9597784405624201,
+      "grad_norm": 14.6875,
+      "learning_rate": 2.5153627608112683e-09,
+      "loss": 0.6173,
+      "step": 22998
+    },
+    {
+      "epoch": 1.95986365573072,
+      "grad_norm": 19.25,
+      "learning_rate": 2.5047192917550257e-09,
+      "loss": 0.6868,
+      "step": 22999
+    },
+    {
+      "epoch": 1.95994887089902,
+      "grad_norm": 13.3125,
+      "learning_rate": 2.4940983661929375e-09,
+      "loss": 0.4554,
+      "step": 23000
+    },
+    {
+      "epoch": 1.96003408606732,
+      "grad_norm": 20.25,
+      "learning_rate": 2.4834999843170724e-09,
+      "loss": 0.5386,
+      "step": 23001
+    },
+    {
+      "epoch": 1.96011930123562,
+      "grad_norm": 12.375,
+      "learning_rate": 2.472924146318806e-09,
+      "loss": 0.3945,
+      "step": 23002
+    },
+    {
+      "epoch": 1.96020451640392,
+      "grad_norm": 15.1875,
+      "learning_rate": 2.462370852389234e-09,
+      "loss": 0.6153,
+      "step": 23003
+    },
+    {
+      "epoch": 1.9602897315722199,
+      "grad_norm": 25.25,
+      "learning_rate": 2.4518401027190387e-09,
+      "loss": 0.8293,
+      "step": 23004
+    },
+    {
+      "epoch": 1.9603749467405198,
+      "grad_norm": 37.25,
+      "learning_rate": 2.441331897498761e-09,
+      "loss": 0.7087,
+      "step": 23005
+    },
+    {
+      "epoch": 1.9604601619088198,
+      "grad_norm": 16.125,
+      "learning_rate": 2.4308462369181107e-09,
+      "loss": 0.7311,
+      "step": 23006
+    },
+    {
+      "epoch": 1.9605453770771197,
+      "grad_norm": 20.375,
+      "learning_rate": 2.4203831211665196e-09,
+      "loss": 0.7735,
+      "step": 23007
+    },
+    {
+      "epoch": 1.9606305922454197,
+      "grad_norm": 11.8125,
+      "learning_rate": 2.409942550433142e-09,
+      "loss": 0.3098,
+      "step": 23008
+    },
+    {
+      "epoch": 1.9607158074137196,
+      "grad_norm": 15.9375,
+      "learning_rate": 2.3995245249067157e-09,
+      "loss": 0.4936,
+      "step": 23009
+    },
+    {
+      "epoch": 1.9608010225820196,
+      "grad_norm": 27.375,
+      "learning_rate": 2.3891290447754236e-09,
+      "loss": 0.7621,
+      "step": 23010
+    },
+    {
+      "epoch": 1.9608862377503196,
+      "grad_norm": 19.125,
+      "learning_rate": 2.3787561102270317e-09,
+      "loss": 0.8545,
+      "step": 23011
+    },
+    {
+      "epoch": 1.9609714529186195,
+      "grad_norm": 27.625,
+      "learning_rate": 2.3684057214493073e-09,
+      "loss": 0.895,
+      "step": 23012
+    },
+    {
+      "epoch": 1.9610566680869195,
+      "grad_norm": 13.0,
+      "learning_rate": 2.3580778786289062e-09,
+      "loss": 0.5552,
+      "step": 23013
+    },
+    {
+      "epoch": 1.9611418832552194,
+      "grad_norm": 17.125,
+      "learning_rate": 2.347772581952762e-09,
+      "loss": 0.8315,
+      "step": 23014
+    },
+    {
+      "epoch": 1.9612270984235194,
+      "grad_norm": 16.75,
+      "learning_rate": 2.3374898316068375e-09,
+      "loss": 0.6713,
+      "step": 23015
+    },
+    {
+      "epoch": 1.9613123135918193,
+      "grad_norm": 19.375,
+      "learning_rate": 2.327229627777233e-09,
+      "loss": 0.727,
+      "step": 23016
+    },
+    {
+      "epoch": 1.9613975287601193,
+      "grad_norm": 17.0,
+      "learning_rate": 2.3169919706490797e-09,
+      "loss": 0.6789,
+      "step": 23017
+    },
+    {
+      "epoch": 1.9614827439284193,
+      "grad_norm": 17.625,
+      "learning_rate": 2.3067768604076448e-09,
+      "loss": 0.5146,
+      "step": 23018
+    },
+    {
+      "epoch": 1.9615679590967192,
+      "grad_norm": 14.125,
+      "learning_rate": 2.296584297237364e-09,
+      "loss": 0.4511,
+      "step": 23019
+    },
+    {
+      "epoch": 1.9616531742650192,
+      "grad_norm": 15.75,
+      "learning_rate": 2.2864142813223965e-09,
+      "loss": 0.7398,
+      "step": 23020
+    },
+    {
+      "epoch": 1.9617383894333191,
+      "grad_norm": 14.125,
+      "learning_rate": 2.276266812846761e-09,
+      "loss": 0.517,
+      "step": 23021
+    },
+    {
+      "epoch": 1.961823604601619,
+      "grad_norm": 17.0,
+      "learning_rate": 2.2661418919935062e-09,
+      "loss": 0.6405,
+      "step": 23022
+    },
+    {
+      "epoch": 1.961908819769919,
+      "grad_norm": 17.875,
+      "learning_rate": 2.2560395189458182e-09,
+      "loss": 0.6748,
+      "step": 23023
+    },
+    {
+      "epoch": 1.961994034938219,
+      "grad_norm": 13.875,
+      "learning_rate": 2.2459596938861905e-09,
+      "loss": 0.416,
+      "step": 23024
+    },
+    {
+      "epoch": 1.962079250106519,
+      "grad_norm": 21.375,
+      "learning_rate": 2.2359024169968378e-09,
+      "loss": 0.6156,
+      "step": 23025
+    },
+    {
+      "epoch": 1.962164465274819,
+      "grad_norm": 15.875,
+      "learning_rate": 2.225867688459421e-09,
+      "loss": 1.0962,
+      "step": 23026
+    },
+    {
+      "epoch": 1.9622496804431189,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.2158555084553224e-09,
+      "loss": 0.417,
+      "step": 23027
+    },
+    {
+      "epoch": 1.9623348956114188,
+      "grad_norm": 14.375,
+      "learning_rate": 2.2058658771653697e-09,
+      "loss": 0.4612,
+      "step": 23028
+    },
+    {
+      "epoch": 1.9624201107797188,
+      "grad_norm": 31.375,
+      "learning_rate": 2.1958987947702514e-09,
+      "loss": 0.7578,
+      "step": 23029
+    },
+    {
+      "epoch": 1.9625053259480187,
+      "grad_norm": 30.0,
+      "learning_rate": 2.1859542614501016e-09,
+      "loss": 0.9946,
+      "step": 23030
+    },
+    {
+      "epoch": 1.9625905411163187,
+      "grad_norm": 14.9375,
+      "learning_rate": 2.17603227738436e-09,
+      "loss": 0.5833,
+      "step": 23031
+    },
+    {
+      "epoch": 1.9626757562846187,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.1661328427526053e-09,
+      "loss": 0.7809,
+      "step": 23032
+    },
+    {
+      "epoch": 1.9627609714529186,
+      "grad_norm": 13.5,
+      "learning_rate": 2.156255957733583e-09,
+      "loss": 0.3277,
+      "step": 23033
+    },
+    {
+      "epoch": 1.9628461866212186,
+      "grad_norm": 17.125,
+      "learning_rate": 2.1464016225057614e-09,
+      "loss": 0.5945,
+      "step": 23034
+    },
+    {
+      "epoch": 1.9629314017895185,
+      "grad_norm": 12.25,
+      "learning_rate": 2.136569837247332e-09,
+      "loss": 0.5264,
+      "step": 23035
+    },
+    {
+      "epoch": 1.9630166169578185,
+      "grad_norm": 15.75,
+      "learning_rate": 2.12676060213593e-09,
+      "loss": 0.6614,
+      "step": 23036
+    },
+    {
+      "epoch": 1.9631018321261184,
+      "grad_norm": 14.0,
+      "learning_rate": 2.1169739173487745e-09,
+      "loss": 0.5344,
+      "step": 23037
+    },
+    {
+      "epoch": 1.9631870472944184,
+      "grad_norm": 16.625,
+      "learning_rate": 2.107209783062808e-09,
+      "loss": 0.7302,
+      "step": 23038
+    },
+    {
+      "epoch": 1.9632722624627184,
+      "grad_norm": 28.25,
+      "learning_rate": 2.0974681994544165e-09,
+      "loss": 0.5176,
+      "step": 23039
+    },
+    {
+      "epoch": 1.9633574776310183,
+      "grad_norm": 14.875,
+      "learning_rate": 2.087749166699571e-09,
+      "loss": 0.5706,
+      "step": 23040
+    },
+    {
+      "epoch": 1.9634426927993183,
+      "grad_norm": 13.25,
+      "learning_rate": 2.0780526849739634e-09,
+      "loss": 0.4429,
+      "step": 23041
+    },
+    {
+      "epoch": 1.9635279079676182,
+      "grad_norm": 17.0,
+      "learning_rate": 2.068378754452732e-09,
+      "loss": 0.7627,
+      "step": 23042
+    },
+    {
+      "epoch": 1.9636131231359182,
+      "grad_norm": 21.875,
+      "learning_rate": 2.058727375311015e-09,
+      "loss": 1.0042,
+      "step": 23043
+    },
+    {
+      "epoch": 1.9636983383042181,
+      "grad_norm": 18.0,
+      "learning_rate": 2.0490985477228386e-09,
+      "loss": 0.6375,
+      "step": 23044
+    },
+    {
+      "epoch": 1.963783553472518,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.0394922718623698e-09,
+      "loss": 0.2024,
+      "step": 23045
+    },
+    {
+      "epoch": 1.963868768640818,
+      "grad_norm": 13.0,
+      "learning_rate": 2.02990854790322e-09,
+      "loss": 0.5834,
+      "step": 23046
+    },
+    {
+      "epoch": 1.963953983809118,
+      "grad_norm": 47.5,
+      "learning_rate": 2.020347376018583e-09,
+      "loss": 0.8603,
+      "step": 23047
+    },
+    {
+      "epoch": 1.964039198977418,
+      "grad_norm": 11.0,
+      "learning_rate": 2.0108087563812385e-09,
+      "loss": 0.2898,
+      "step": 23048
+    },
+    {
+      "epoch": 1.964124414145718,
+      "grad_norm": 15.25,
+      "learning_rate": 2.0012926891635475e-09,
+      "loss": 0.4462,
+      "step": 23049
+    },
+    {
+      "epoch": 1.9642096293140179,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.9917991745373176e-09,
+      "loss": 0.3421,
+      "step": 23050
+    },
+    {
+      "epoch": 1.9642948444823178,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9823282126743558e-09,
+      "loss": 0.5738,
+      "step": 23051
+    },
+    {
+      "epoch": 1.9643800596506178,
+      "grad_norm": 25.875,
+      "learning_rate": 1.9728798037456366e-09,
+      "loss": 0.8628,
+      "step": 23052
+    },
+    {
+      "epoch": 1.9644652748189178,
+      "grad_norm": 20.375,
+      "learning_rate": 1.963453947921995e-09,
+      "loss": 0.9686,
+      "step": 23053
+    },
+    {
+      "epoch": 1.9645504899872177,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9540506453737117e-09,
+      "loss": 0.77,
+      "step": 23054
+    },
+    {
+      "epoch": 1.9646357051555177,
+      "grad_norm": 33.0,
+      "learning_rate": 1.94466989627079e-09,
+      "loss": 0.5928,
+      "step": 23055
+    },
+    {
+      "epoch": 1.9647209203238176,
+      "grad_norm": 10.625,
+      "learning_rate": 1.935311700782677e-09,
+      "loss": 0.425,
+      "step": 23056
+    },
+    {
+      "epoch": 1.9648061354921176,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9259760590785438e-09,
+      "loss": 0.4426,
+      "step": 23057
+    },
+    {
+      "epoch": 1.9648913506604175,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9166629713270046e-09,
+      "loss": 0.4274,
+      "step": 23058
+    },
+    {
+      "epoch": 1.9649765658287175,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9073724376963974e-09,
+      "loss": 0.2492,
+      "step": 23059
+    },
+    {
+      "epoch": 1.9650617809970174,
+      "grad_norm": 18.625,
+      "learning_rate": 1.8981044583547823e-09,
+      "loss": 0.6669,
+      "step": 23060
+    },
+    {
+      "epoch": 1.9651469961653174,
+      "grad_norm": 38.25,
+      "learning_rate": 1.8888590334693866e-09,
+      "loss": 0.8172,
+      "step": 23061
+    },
+    {
+      "epoch": 1.9652322113336174,
+      "grad_norm": 12.0,
+      "learning_rate": 1.879636163207299e-09,
+      "loss": 0.3306,
+      "step": 23062
+    },
+    {
+      "epoch": 1.9653174265019173,
+      "grad_norm": 17.375,
+      "learning_rate": 1.87043584773533e-09,
+      "loss": 0.7051,
+      "step": 23063
+    },
+    {
+      "epoch": 1.9654026416702173,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8612580872195974e-09,
+      "loss": 0.5711,
+      "step": 23064
+    },
+    {
+      "epoch": 1.9654878568385172,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8521028818262178e-09,
+      "loss": 0.5496,
+      "step": 23065
+    },
+    {
+      "epoch": 1.9655730720068172,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.842970231720198e-09,
+      "loss": 0.3839,
+      "step": 23066
+    },
+    {
+      "epoch": 1.9656582871751171,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8338601370669618e-09,
+      "loss": 0.9758,
+      "step": 23067
+    },
+    {
+      "epoch": 1.965743502343417,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8247725980310993e-09,
+      "loss": 0.643,
+      "step": 23068
+    },
+    {
+      "epoch": 1.965828717511717,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.815707614776646e-09,
+      "loss": 0.4918,
+      "step": 23069
+    },
+    {
+      "epoch": 1.965913932680017,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8066651874674989e-09,
+      "loss": 0.4265,
+      "step": 23070
+    },
+    {
+      "epoch": 1.965999147848317,
+      "grad_norm": 23.0,
+      "learning_rate": 1.7976453162669994e-09,
+      "loss": 0.9298,
+      "step": 23071
+    },
+    {
+      "epoch": 1.966084363016617,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.7886480013383501e-09,
+      "loss": 0.4361,
+      "step": 23072
+    },
+    {
+      "epoch": 1.9661695781849169,
+      "grad_norm": 15.0,
+      "learning_rate": 1.7796732428439213e-09,
+      "loss": 0.7417,
+      "step": 23073
+    },
+    {
+      "epoch": 1.9662547933532168,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.770721040945944e-09,
+      "loss": 0.5085,
+      "step": 23074
+    },
+    {
+      "epoch": 1.9663400085215168,
+      "grad_norm": 17.875,
+      "learning_rate": 1.7617913958062338e-09,
+      "loss": 0.8252,
+      "step": 23075
+    },
+    {
+      "epoch": 1.9664252236898168,
+      "grad_norm": 15.5,
+      "learning_rate": 1.752884307586189e-09,
+      "loss": 0.716,
+      "step": 23076
+    },
+    {
+      "epoch": 1.9665104388581167,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.743999776446792e-09,
+      "loss": 0.5298,
+      "step": 23077
+    },
+    {
+      "epoch": 1.9665956540264167,
+      "grad_norm": 13.75,
+      "learning_rate": 1.73513780254847e-09,
+      "loss": 0.625,
+      "step": 23078
+    },
+    {
+      "epoch": 1.9666808691947166,
+      "grad_norm": 11.375,
+      "learning_rate": 1.7262983860515114e-09,
+      "loss": 0.237,
+      "step": 23079
+    },
+    {
+      "epoch": 1.9667660843630166,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.717481527115511e-09,
+      "loss": 0.3264,
+      "step": 23080
+    },
+    {
+      "epoch": 1.9668512995313165,
+      "grad_norm": 19.625,
+      "learning_rate": 1.7086872258997855e-09,
+      "loss": 0.7508,
+      "step": 23081
+    },
+    {
+      "epoch": 1.9669365146996165,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.6999154825635134e-09,
+      "loss": 0.2095,
+      "step": 23082
+    },
+    {
+      "epoch": 1.9670217298679165,
+      "grad_norm": 23.125,
+      "learning_rate": 1.6911662972649012e-09,
+      "loss": 0.5202,
+      "step": 23083
+    },
+    {
+      "epoch": 1.9671069450362164,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.682439670162156e-09,
+      "loss": 0.5985,
+      "step": 23084
+    },
+    {
+      "epoch": 1.9671921602045164,
+      "grad_norm": 18.0,
+      "learning_rate": 1.673735601413068e-09,
+      "loss": 0.5777,
+      "step": 23085
+    },
+    {
+      "epoch": 1.9672773753728163,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6650540911750113e-09,
+      "loss": 0.6757,
+      "step": 23086
+    },
+    {
+      "epoch": 1.9673625905411163,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6563951396045275e-09,
+      "loss": 1.2129,
+      "step": 23087
+    },
+    {
+      "epoch": 1.9674478057094162,
+      "grad_norm": 21.375,
+      "learning_rate": 1.6477587468582968e-09,
+      "loss": 0.7593,
+      "step": 23088
+    },
+    {
+      "epoch": 1.9675330208777162,
+      "grad_norm": 19.5,
+      "learning_rate": 1.6391449130924442e-09,
+      "loss": 0.7467,
+      "step": 23089
+    },
+    {
+      "epoch": 1.9676182360460162,
+      "grad_norm": 16.5,
+      "learning_rate": 1.6305536384625399e-09,
+      "loss": 0.6727,
+      "step": 23090
+    },
+    {
+      "epoch": 1.9677034512143161,
+      "grad_norm": 17.875,
+      "learning_rate": 1.621984923123876e-09,
+      "loss": 0.7189,
+      "step": 23091
+    },
+    {
+      "epoch": 1.967788666382616,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.613438767231329e-09,
+      "loss": 0.7864,
+      "step": 23092
+    },
+    {
+      "epoch": 1.967873881550916,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.6049151709392196e-09,
+      "loss": 0.602,
+      "step": 23093
+    },
+    {
+      "epoch": 1.967959096719216,
+      "grad_norm": 26.75,
+      "learning_rate": 1.5964141344017302e-09,
+      "loss": 0.7621,
+      "step": 23094
+    },
+    {
+      "epoch": 1.968044311887516,
+      "grad_norm": 18.0,
+      "learning_rate": 1.5879356577723494e-09,
+      "loss": 1.0914,
+      "step": 23095
+    },
+    {
+      "epoch": 1.968129527055816,
+      "grad_norm": 23.625,
+      "learning_rate": 1.5794797412042872e-09,
+      "loss": 0.8505,
+      "step": 23096
+    },
+    {
+      "epoch": 1.9682147422241159,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.5710463848504776e-09,
+      "loss": 0.5152,
+      "step": 23097
+    },
+    {
+      "epoch": 1.9682999573924158,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.5626355888632982e-09,
+      "loss": 1.0984,
+      "step": 23098
+    },
+    {
+      "epoch": 1.9683851725607158,
+      "grad_norm": 18.5,
+      "learning_rate": 1.554247353394711e-09,
+      "loss": 0.7624,
+      "step": 23099
+    },
+    {
+      "epoch": 1.9684703877290157,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.5458816785961228e-09,
+      "loss": 0.6482,
+      "step": 23100
+    },
+    {
+      "epoch": 1.9685556028973157,
+      "grad_norm": 51.5,
+      "learning_rate": 1.5375385646190787e-09,
+      "loss": 0.8037,
+      "step": 23101
+    },
+    {
+      "epoch": 1.9686408180656156,
+      "grad_norm": 24.0,
+      "learning_rate": 1.5292180116140143e-09,
+      "loss": 1.2655,
+      "step": 23102
+    },
+    {
+      "epoch": 1.9687260332339156,
+      "grad_norm": 13.25,
+      "learning_rate": 1.5209200197313645e-09,
+      "loss": 0.4831,
+      "step": 23103
+    },
+    {
+      "epoch": 1.9688112484022156,
+      "grad_norm": 14.75,
+      "learning_rate": 1.5126445891212871e-09,
+      "loss": 0.6919,
+      "step": 23104
+    },
+    {
+      "epoch": 1.9688964635705155,
+      "grad_norm": 19.25,
+      "learning_rate": 1.5043917199329682e-09,
+      "loss": 0.7322,
+      "step": 23105
+    },
+    {
+      "epoch": 1.9689816787388155,
+      "grad_norm": 19.0,
+      "learning_rate": 1.4961614123158718e-09,
+      "loss": 0.6141,
+      "step": 23106
+    },
+    {
+      "epoch": 1.9690668939071154,
+      "grad_norm": 14.625,
+      "learning_rate": 1.487953666418629e-09,
+      "loss": 0.4339,
+      "step": 23107
+    },
+    {
+      "epoch": 1.9691521090754154,
+      "grad_norm": 24.75,
+      "learning_rate": 1.4797684823894543e-09,
+      "loss": 0.3565,
+      "step": 23108
+    },
+    {
+      "epoch": 1.9692373242437153,
+      "grad_norm": 11.0,
+      "learning_rate": 1.4716058603764239e-09,
+      "loss": 0.3067,
+      "step": 23109
+    },
+    {
+      "epoch": 1.9693225394120153,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.4634658005270586e-09,
+      "loss": 0.8576,
+      "step": 23110
+    },
+    {
+      "epoch": 1.9694077545803153,
+      "grad_norm": 14.5,
+      "learning_rate": 1.4553483029881854e-09,
+      "loss": 0.6379,
+      "step": 23111
+    },
+    {
+      "epoch": 1.9694929697486152,
+      "grad_norm": 19.0,
+      "learning_rate": 1.44725336790677e-09,
+      "loss": 0.8173,
+      "step": 23112
+    },
+    {
+      "epoch": 1.9695781849169152,
+      "grad_norm": 16.875,
+      "learning_rate": 1.439180995428946e-09,
+      "loss": 1.0586,
+      "step": 23113
+    },
+    {
+      "epoch": 1.9696634000852151,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4311311857007072e-09,
+      "loss": 0.7584,
+      "step": 23114
+    },
+    {
+      "epoch": 1.969748615253515,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.423103938867354e-09,
+      "loss": 0.7925,
+      "step": 23115
+    },
+    {
+      "epoch": 1.969833830421815,
+      "grad_norm": 14.625,
+      "learning_rate": 1.415099255074187e-09,
+      "loss": 0.592,
+      "step": 23116
+    },
+    {
+      "epoch": 1.969919045590115,
+      "grad_norm": 19.75,
+      "learning_rate": 1.4071171344656742e-09,
+      "loss": 0.5554,
+      "step": 23117
+    },
+    {
+      "epoch": 1.970004260758415,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.3991575771861443e-09,
+      "loss": 0.3557,
+      "step": 23118
+    },
+    {
+      "epoch": 1.970089475926715,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.3912205833793713e-09,
+      "loss": 0.3957,
+      "step": 23119
+    },
+    {
+      "epoch": 1.9701746910950149,
+      "grad_norm": 16.5,
+      "learning_rate": 1.3833061531888514e-09,
+      "loss": 0.5071,
+      "step": 23120
+    },
+    {
+      "epoch": 1.9702599062633148,
+      "grad_norm": 18.125,
+      "learning_rate": 1.375414286757526e-09,
+      "loss": 0.7778,
+      "step": 23121
+    },
+    {
+      "epoch": 1.9703451214316148,
+      "grad_norm": 17.875,
+      "learning_rate": 1.3675449842280585e-09,
+      "loss": 0.8993,
+      "step": 23122
+    },
+    {
+      "epoch": 1.9704303365999147,
+      "grad_norm": 37.25,
+      "learning_rate": 1.3596982457425579e-09,
+      "loss": 0.5876,
+      "step": 23123
+    },
+    {
+      "epoch": 1.9705155517682147,
+      "grad_norm": 14.0,
+      "learning_rate": 1.3518740714431321e-09,
+      "loss": 0.3453,
+      "step": 23124
+    },
+    {
+      "epoch": 1.9706007669365146,
+      "grad_norm": 15.375,
+      "learning_rate": 1.3440724614707802e-09,
+      "loss": 0.4762,
+      "step": 23125
+    },
+    {
+      "epoch": 1.9706859821048146,
+      "grad_norm": 17.75,
+      "learning_rate": 1.3362934159666385e-09,
+      "loss": 0.5011,
+      "step": 23126
+    },
+    {
+      "epoch": 1.9707711972731146,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.3285369350714284e-09,
+      "loss": 0.776,
+      "step": 23127
+    },
+    {
+      "epoch": 1.9708564124414145,
+      "grad_norm": 18.875,
+      "learning_rate": 1.3208030189251764e-09,
+      "loss": 0.7474,
+      "step": 23128
+    },
+    {
+      "epoch": 1.9709416276097145,
+      "grad_norm": 13.75,
+      "learning_rate": 1.313091667667632e-09,
+      "loss": 0.5405,
+      "step": 23129
+    },
+    {
+      "epoch": 1.9710268427780144,
+      "grad_norm": 27.75,
+      "learning_rate": 1.305402881438128e-09,
+      "loss": 0.4619,
+      "step": 23130
+    },
+    {
+      "epoch": 1.9711120579463144,
+      "grad_norm": 12.625,
+      "learning_rate": 1.2977366603757202e-09,
+      "loss": 0.4542,
+      "step": 23131
+    },
+    {
+      "epoch": 1.9711972731146146,
+      "grad_norm": 16.25,
+      "learning_rate": 1.29009300461877e-09,
+      "loss": 0.6752,
+      "step": 23132
+    },
+    {
+      "epoch": 1.9712824882829145,
+      "grad_norm": 18.375,
+      "learning_rate": 1.2824719143056385e-09,
+      "loss": 0.5236,
+      "step": 23133
+    },
+    {
+      "epoch": 1.9713677034512145,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2748733895738551e-09,
+      "loss": 0.4807,
+      "step": 23134
+    },
+    {
+      "epoch": 1.9714529186195144,
+      "grad_norm": 13.5,
+      "learning_rate": 1.26729743056081e-09,
+      "loss": 0.5434,
+      "step": 23135
+    },
+    {
+      "epoch": 1.9715381337878144,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2597440374034764e-09,
+      "loss": 0.3572,
+      "step": 23136
+    },
+    {
+      "epoch": 1.9716233489561144,
+      "grad_norm": 17.125,
+      "learning_rate": 1.2522132102381345e-09,
+      "loss": 0.725,
+      "step": 23137
+    },
+    {
+      "epoch": 1.9717085641244143,
+      "grad_norm": 19.5,
+      "learning_rate": 1.2447049492010643e-09,
+      "loss": 0.529,
+      "step": 23138
+    },
+    {
+      "epoch": 1.9717937792927143,
+      "grad_norm": 13.375,
+      "learning_rate": 1.2372192544279904e-09,
+      "loss": 0.5407,
+      "step": 23139
+    },
+    {
+      "epoch": 1.9718789944610142,
+      "grad_norm": 42.25,
+      "learning_rate": 1.2297561260539436e-09,
+      "loss": 0.7963,
+      "step": 23140
+    },
+    {
+      "epoch": 1.9719642096293142,
+      "grad_norm": 10.5,
+      "learning_rate": 1.2223155642140938e-09,
+      "loss": 0.3503,
+      "step": 23141
+    },
+    {
+      "epoch": 1.9720494247976141,
+      "grad_norm": 17.25,
+      "learning_rate": 1.214897569042639e-09,
+      "loss": 0.4181,
+      "step": 23142
+    },
+    {
+      "epoch": 1.972134639965914,
+      "grad_norm": 19.875,
+      "learning_rate": 1.2075021406737775e-09,
+      "loss": 0.6367,
+      "step": 23143
+    },
+    {
+      "epoch": 1.972219855134214,
+      "grad_norm": 18.875,
+      "learning_rate": 1.2001292792410136e-09,
+      "loss": 0.8345,
+      "step": 23144
+    },
+    {
+      "epoch": 1.972305070302514,
+      "grad_norm": 18.5,
+      "learning_rate": 1.1927789848777127e-09,
+      "loss": 0.535,
+      "step": 23145
+    },
+    {
+      "epoch": 1.972390285470814,
+      "grad_norm": 18.625,
+      "learning_rate": 1.1854512577168242e-09,
+      "loss": 0.6726,
+      "step": 23146
+    },
+    {
+      "epoch": 1.972475500639114,
+      "grad_norm": 18.875,
+      "learning_rate": 1.1781460978904646e-09,
+      "loss": 0.8215,
+      "step": 23147
+    },
+    {
+      "epoch": 1.9725607158074139,
+      "grad_norm": 14.25,
+      "learning_rate": 1.1708635055307505e-09,
+      "loss": 0.6118,
+      "step": 23148
+    },
+    {
+      "epoch": 1.9726459309757138,
+      "grad_norm": 17.0,
+      "learning_rate": 1.1636034807695207e-09,
+      "loss": 0.3633,
+      "step": 23149
+    },
+    {
+      "epoch": 1.9727311461440138,
+      "grad_norm": 11.375,
+      "learning_rate": 1.1563660237376428e-09,
+      "loss": 0.3144,
+      "step": 23150
+    },
+    {
+      "epoch": 1.9728163613123137,
+      "grad_norm": 12.5,
+      "learning_rate": 1.1491511345659844e-09,
+      "loss": 0.4339,
+      "step": 23151
+    },
+    {
+      "epoch": 1.9729015764806137,
+      "grad_norm": 12.875,
+      "learning_rate": 1.1419588133851357e-09,
+      "loss": 0.4516,
+      "step": 23152
+    },
+    {
+      "epoch": 1.9729867916489137,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.1347890603247146e-09,
+      "loss": 0.4728,
+      "step": 23153
+    },
+    {
+      "epoch": 1.9730720068172136,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.1276418755147566e-09,
+      "loss": 0.5482,
+      "step": 23154
+    },
+    {
+      "epoch": 1.9731572219855136,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.1205172590839086e-09,
+      "loss": 0.6259,
+      "step": 23155
+    },
+    {
+      "epoch": 1.9732424371538135,
+      "grad_norm": 13.875,
+      "learning_rate": 1.113415211161234e-09,
+      "loss": 0.5267,
+      "step": 23156
+    },
+    {
+      "epoch": 1.9733276523221135,
+      "grad_norm": 14.75,
+      "learning_rate": 1.1063357318749634e-09,
+      "loss": 0.4483,
+      "step": 23157
+    },
+    {
+      "epoch": 1.9734128674904134,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.0992788213531892e-09,
+      "loss": 0.5168,
+      "step": 23158
+    },
+    {
+      "epoch": 1.9734980826587134,
+      "grad_norm": 20.875,
+      "learning_rate": 1.092244479723309e-09,
+      "loss": 0.6854,
+      "step": 23159
+    },
+    {
+      "epoch": 1.9735832978270134,
+      "grad_norm": 20.625,
+      "learning_rate": 1.0852327071123047e-09,
+      "loss": 1.0474,
+      "step": 23160
+    },
+    {
+      "epoch": 1.9736685129953133,
+      "grad_norm": 25.625,
+      "learning_rate": 1.0782435036471583e-09,
+      "loss": 0.395,
+      "step": 23161
+    },
+    {
+      "epoch": 1.9737537281636133,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0712768694540188e-09,
+      "loss": 0.588,
+      "step": 23162
+    },
+    {
+      "epoch": 1.9738389433319132,
+      "grad_norm": 12.125,
+      "learning_rate": 1.0643328046586188e-09,
+      "loss": 0.2899,
+      "step": 23163
+    },
+    {
+      "epoch": 1.9739241585002132,
+      "grad_norm": 16.125,
+      "learning_rate": 1.0574113093868298e-09,
+      "loss": 0.6833,
+      "step": 23164
+    },
+    {
+      "epoch": 1.9740093736685131,
+      "grad_norm": 12.75,
+      "learning_rate": 1.0505123837634135e-09,
+      "loss": 0.4787,
+      "step": 23165
+    },
+    {
+      "epoch": 1.974094588836813,
+      "grad_norm": 17.125,
+      "learning_rate": 1.0436360279132696e-09,
+      "loss": 0.6312,
+      "step": 23166
+    },
+    {
+      "epoch": 1.974179804005113,
+      "grad_norm": 17.5,
+      "learning_rate": 1.0367822419604657e-09,
+      "loss": 0.553,
+      "step": 23167
+    },
+    {
+      "epoch": 1.974265019173413,
+      "grad_norm": 16.25,
+      "learning_rate": 1.0299510260289302e-09,
+      "loss": 0.8832,
+      "step": 23168
+    },
+    {
+      "epoch": 1.974350234341713,
+      "grad_norm": 17.0,
+      "learning_rate": 1.0231423802421758e-09,
+      "loss": 0.8022,
+      "step": 23169
+    },
+    {
+      "epoch": 1.974435449510013,
+      "grad_norm": 26.0,
+      "learning_rate": 1.0163563047231596e-09,
+      "loss": 0.5269,
+      "step": 23170
+    },
+    {
+      "epoch": 1.9745206646783129,
+      "grad_norm": 19.625,
+      "learning_rate": 1.0095927995945608e-09,
+      "loss": 0.7514,
+      "step": 23171
+    },
+    {
+      "epoch": 1.9746058798466128,
+      "grad_norm": 11.75,
+      "learning_rate": 1.0028518649785047e-09,
+      "loss": 0.3114,
+      "step": 23172
+    },
+    {
+      "epoch": 1.9746910950149128,
+      "grad_norm": 15.4375,
+      "learning_rate": 9.961335009968377e-10,
+      "loss": 0.9102,
+      "step": 23173
+    },
+    {
+      "epoch": 1.9747763101832128,
+      "grad_norm": 14.4375,
+      "learning_rate": 9.894377077711291e-10,
+      "loss": 0.5267,
+      "step": 23174
+    },
+    {
+      "epoch": 1.9748615253515127,
+      "grad_norm": 19.0,
+      "learning_rate": 9.82764485422255e-10,
+      "loss": 0.8224,
+      "step": 23175
+    },
+    {
+      "epoch": 1.9749467405198127,
+      "grad_norm": 43.75,
+      "learning_rate": 9.761138340708132e-10,
+      "loss": 0.6936,
+      "step": 23176
+    },
+    {
+      "epoch": 1.9750319556881126,
+      "grad_norm": 13.5,
+      "learning_rate": 9.694857538368463e-10,
+      "loss": 0.5346,
+      "step": 23177
+    },
+    {
+      "epoch": 1.9751171708564126,
+      "grad_norm": 12.8125,
+      "learning_rate": 9.628802448403974e-10,
+      "loss": 0.4519,
+      "step": 23178
+    },
+    {
+      "epoch": 1.9752023860247125,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.562973072006766e-10,
+      "loss": 0.9345,
+      "step": 23179
+    },
+    {
+      "epoch": 1.9752876011930125,
+      "grad_norm": 14.625,
+      "learning_rate": 9.497369410366164e-10,
+      "loss": 0.4527,
+      "step": 23180
+    },
+    {
+      "epoch": 1.9753728163613125,
+      "grad_norm": 13.125,
+      "learning_rate": 9.43199146466872e-10,
+      "loss": 0.4334,
+      "step": 23181
+    },
+    {
+      "epoch": 1.9754580315296124,
+      "grad_norm": 12.5,
+      "learning_rate": 9.366839236096826e-10,
+      "loss": 0.3288,
+      "step": 23182
+    },
+    {
+      "epoch": 1.9755432466979124,
+      "grad_norm": 11.5625,
+      "learning_rate": 9.301912725824536e-10,
+      "loss": 0.7561,
+      "step": 23183
+    },
+    {
+      "epoch": 1.9756284618662123,
+      "grad_norm": 16.25,
+      "learning_rate": 9.237211935027301e-10,
+      "loss": 0.6324,
+      "step": 23184
+    },
+    {
+      "epoch": 1.9757136770345123,
+      "grad_norm": 15.6875,
+      "learning_rate": 9.17273686487502e-10,
+      "loss": 0.663,
+      "step": 23185
+    },
+    {
+      "epoch": 1.9757988922028122,
+      "grad_norm": 30.625,
+      "learning_rate": 9.108487516530651e-10,
+      "loss": 0.5023,
+      "step": 23186
+    },
+    {
+      "epoch": 1.9758841073711122,
+      "grad_norm": 15.5625,
+      "learning_rate": 9.044463891157151e-10,
+      "loss": 0.742,
+      "step": 23187
+    },
+    {
+      "epoch": 1.9759693225394122,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.980665989910542e-10,
+      "loss": 0.5283,
+      "step": 23188
+    },
+    {
+      "epoch": 1.976054537707712,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.917093813945454e-10,
+      "loss": 0.5337,
+      "step": 23189
+    },
+    {
+      "epoch": 1.976139752876012,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.853747364408194e-10,
+      "loss": 0.4634,
+      "step": 23190
+    },
+    {
+      "epoch": 1.976224968044312,
+      "grad_norm": 16.375,
+      "learning_rate": 8.790626642443678e-10,
+      "loss": 0.6058,
+      "step": 23191
+    },
+    {
+      "epoch": 1.976310183212612,
+      "grad_norm": 16.875,
+      "learning_rate": 8.727731649194049e-10,
+      "loss": 0.3757,
+      "step": 23192
+    },
+    {
+      "epoch": 1.976395398380912,
+      "grad_norm": 22.0,
+      "learning_rate": 8.665062385795897e-10,
+      "loss": 0.5928,
+      "step": 23193
+    },
+    {
+      "epoch": 1.976480613549212,
+      "grad_norm": 21.125,
+      "learning_rate": 8.602618853380262e-10,
+      "loss": 0.7695,
+      "step": 23194
+    },
+    {
+      "epoch": 1.9765658287175119,
+      "grad_norm": 19.25,
+      "learning_rate": 8.540401053076797e-10,
+      "loss": 0.5815,
+      "step": 23195
+    },
+    {
+      "epoch": 1.9766510438858118,
+      "grad_norm": 17.625,
+      "learning_rate": 8.478408986008213e-10,
+      "loss": 0.4575,
+      "step": 23196
+    },
+    {
+      "epoch": 1.9767362590541118,
+      "grad_norm": 22.75,
+      "learning_rate": 8.416642653297224e-10,
+      "loss": 1.0323,
+      "step": 23197
+    },
+    {
+      "epoch": 1.9768214742224117,
+      "grad_norm": 12.5,
+      "learning_rate": 8.355102056058217e-10,
+      "loss": 0.3818,
+      "step": 23198
+    },
+    {
+      "epoch": 1.9769066893907117,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.29378719540419e-10,
+      "loss": 0.3655,
+      "step": 23199
+    },
+    {
+      "epoch": 1.9769919045590116,
+      "grad_norm": 15.0625,
+      "learning_rate": 8.232698072442591e-10,
+      "loss": 0.4592,
+      "step": 23200
+    },
+    {
+      "epoch": 1.9770771197273116,
+      "grad_norm": 14.6875,
+      "learning_rate": 8.171834688278091e-10,
+      "loss": 0.4865,
+      "step": 23201
+    },
+    {
+      "epoch": 1.9771623348956116,
+      "grad_norm": 17.75,
+      "learning_rate": 8.11119704400981e-10,
+      "loss": 0.6239,
+      "step": 23202
+    },
+    {
+      "epoch": 1.9772475500639115,
+      "grad_norm": 15.5625,
+      "learning_rate": 8.050785140734096e-10,
+      "loss": 0.653,
+      "step": 23203
+    },
+    {
+      "epoch": 1.9773327652322115,
+      "grad_norm": 16.75,
+      "learning_rate": 7.990598979541742e-10,
+      "loss": 0.4711,
+      "step": 23204
+    },
+    {
+      "epoch": 1.9774179804005114,
+      "grad_norm": 30.125,
+      "learning_rate": 7.930638561520764e-10,
+      "loss": 1.1897,
+      "step": 23205
+    },
+    {
+      "epoch": 1.9775031955688114,
+      "grad_norm": 14.5,
+      "learning_rate": 7.870903887755021e-10,
+      "loss": 0.6074,
+      "step": 23206
+    },
+    {
+      "epoch": 1.9775884107371113,
+      "grad_norm": 14.375,
+      "learning_rate": 7.811394959324203e-10,
+      "loss": 0.4635,
+      "step": 23207
+    },
+    {
+      "epoch": 1.9776736259054113,
+      "grad_norm": 16.0,
+      "learning_rate": 7.752111777303839e-10,
+      "loss": 0.3942,
+      "step": 23208
+    },
+    {
+      "epoch": 1.9777588410737112,
+      "grad_norm": 18.625,
+      "learning_rate": 7.693054342763906e-10,
+      "loss": 0.6744,
+      "step": 23209
+    },
+    {
+      "epoch": 1.9778440562420112,
+      "grad_norm": 19.875,
+      "learning_rate": 7.634222656772994e-10,
+      "loss": 0.8584,
+      "step": 23210
+    },
+    {
+      "epoch": 1.9779292714103112,
+      "grad_norm": 13.9375,
+      "learning_rate": 7.575616720392754e-10,
+      "loss": 0.335,
+      "step": 23211
+    },
+    {
+      "epoch": 1.9780144865786111,
+      "grad_norm": 10.625,
+      "learning_rate": 7.517236534683447e-10,
+      "loss": 0.3457,
+      "step": 23212
+    },
+    {
+      "epoch": 1.978099701746911,
+      "grad_norm": 21.75,
+      "learning_rate": 7.459082100701176e-10,
+      "loss": 0.7888,
+      "step": 23213
+    },
+    {
+      "epoch": 1.978184916915211,
+      "grad_norm": 18.75,
+      "learning_rate": 7.401153419493712e-10,
+      "loss": 0.6673,
+      "step": 23214
+    },
+    {
+      "epoch": 1.978270132083511,
+      "grad_norm": 21.75,
+      "learning_rate": 7.343450492111604e-10,
+      "loss": 1.1788,
+      "step": 23215
+    },
+    {
+      "epoch": 1.978355347251811,
+      "grad_norm": 15.4375,
+      "learning_rate": 7.285973319594297e-10,
+      "loss": 0.596,
+      "step": 23216
+    },
+    {
+      "epoch": 1.978440562420111,
+      "grad_norm": 17.75,
+      "learning_rate": 7.228721902982628e-10,
+      "loss": 0.688,
+      "step": 23217
+    },
+    {
+      "epoch": 1.9785257775884109,
+      "grad_norm": 18.375,
+      "learning_rate": 7.1716962433091e-10,
+      "loss": 0.5774,
+      "step": 23218
+    },
+    {
+      "epoch": 1.9786109927567108,
+      "grad_norm": 12.625,
+      "learning_rate": 7.114896341606225e-10,
+      "loss": 0.4801,
+      "step": 23219
+    },
+    {
+      "epoch": 1.9786962079250108,
+      "grad_norm": 15.5,
+      "learning_rate": 7.058322198900957e-10,
+      "loss": 0.3919,
+      "step": 23220
+    },
+    {
+      "epoch": 1.9787814230933107,
+      "grad_norm": 17.0,
+      "learning_rate": 7.001973816213314e-10,
+      "loss": 0.7137,
+      "step": 23221
+    },
+    {
+      "epoch": 1.9788666382616107,
+      "grad_norm": 19.0,
+      "learning_rate": 6.945851194561926e-10,
+      "loss": 0.5701,
+      "step": 23222
+    },
+    {
+      "epoch": 1.9789518534299106,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.889954334962646e-10,
+      "loss": 0.4946,
+      "step": 23223
+    },
+    {
+      "epoch": 1.9790370685982106,
+      "grad_norm": 20.0,
+      "learning_rate": 6.834283238424389e-10,
+      "loss": 0.8554,
+      "step": 23224
+    },
+    {
+      "epoch": 1.9791222837665106,
+      "grad_norm": 21.875,
+      "learning_rate": 6.778837905954683e-10,
+      "loss": 0.7566,
+      "step": 23225
+    },
+    {
+      "epoch": 1.9792074989348105,
+      "grad_norm": 19.25,
+      "learning_rate": 6.723618338552729e-10,
+      "loss": 0.4918,
+      "step": 23226
+    },
+    {
+      "epoch": 1.9792927141031105,
+      "grad_norm": 23.125,
+      "learning_rate": 6.668624537219115e-10,
+      "loss": 0.6919,
+      "step": 23227
+    },
+    {
+      "epoch": 1.9793779292714104,
+      "grad_norm": 15.0,
+      "learning_rate": 6.613856502944715e-10,
+      "loss": 0.1969,
+      "step": 23228
+    },
+    {
+      "epoch": 1.9794631444397104,
+      "grad_norm": 18.25,
+      "learning_rate": 6.55931423672318e-10,
+      "loss": 0.7214,
+      "step": 23229
+    },
+    {
+      "epoch": 1.9795483596080103,
+      "grad_norm": 18.75,
+      "learning_rate": 6.504997739537056e-10,
+      "loss": 0.5708,
+      "step": 23230
+    },
+    {
+      "epoch": 1.9796335747763103,
+      "grad_norm": 12.0625,
+      "learning_rate": 6.45090701236889e-10,
+      "loss": 0.4487,
+      "step": 23231
+    },
+    {
+      "epoch": 1.9797187899446103,
+      "grad_norm": 12.625,
+      "learning_rate": 6.397042056195679e-10,
+      "loss": 0.2648,
+      "step": 23232
+    },
+    {
+      "epoch": 1.9798040051129102,
+      "grad_norm": 28.625,
+      "learning_rate": 6.343402871991644e-10,
+      "loss": 1.1565,
+      "step": 23233
+    },
+    {
+      "epoch": 1.9798892202812102,
+      "grad_norm": 25.375,
+      "learning_rate": 6.289989460724066e-10,
+      "loss": 0.6484,
+      "step": 23234
+    },
+    {
+      "epoch": 1.9799744354495101,
+      "grad_norm": 12.75,
+      "learning_rate": 6.236801823361616e-10,
+      "loss": 0.5695,
+      "step": 23235
+    },
+    {
+      "epoch": 1.98005965061781,
+      "grad_norm": 14.9375,
+      "learning_rate": 6.183839960863247e-10,
+      "loss": 0.4795,
+      "step": 23236
+    },
+    {
+      "epoch": 1.98014486578611,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.131103874186528e-10,
+      "loss": 0.4457,
+      "step": 23237
+    },
+    {
+      "epoch": 1.98023008095441,
+      "grad_norm": 12.1875,
+      "learning_rate": 6.078593564283475e-10,
+      "loss": 0.6531,
+      "step": 23238
+    },
+    {
+      "epoch": 1.98031529612271,
+      "grad_norm": 15.8125,
+      "learning_rate": 6.026309032104716e-10,
+      "loss": 0.4453,
+      "step": 23239
+    },
+    {
+      "epoch": 1.98040051129101,
+      "grad_norm": 66.0,
+      "learning_rate": 5.974250278593941e-10,
+      "loss": 0.9572,
+      "step": 23240
+    },
+    {
+      "epoch": 1.9804857264593099,
+      "grad_norm": 21.5,
+      "learning_rate": 5.922417304692063e-10,
+      "loss": 0.6631,
+      "step": 23241
+    },
+    {
+      "epoch": 1.9805709416276098,
+      "grad_norm": 15.75,
+      "learning_rate": 5.870810111335834e-10,
+      "loss": 0.6596,
+      "step": 23242
+    },
+    {
+      "epoch": 1.9806561567959098,
+      "grad_norm": 18.375,
+      "learning_rate": 5.819428699459229e-10,
+      "loss": 0.4828,
+      "step": 23243
+    },
+    {
+      "epoch": 1.9807413719642097,
+      "grad_norm": 20.375,
+      "learning_rate": 5.768273069987896e-10,
+      "loss": 0.6896,
+      "step": 23244
+    },
+    {
+      "epoch": 1.9808265871325097,
+      "grad_norm": 18.75,
+      "learning_rate": 5.717343223847482e-10,
+      "loss": 0.5104,
+      "step": 23245
+    },
+    {
+      "epoch": 1.9809118023008097,
+      "grad_norm": 12.8125,
+      "learning_rate": 5.666639161959475e-10,
+      "loss": 0.6057,
+      "step": 23246
+    },
+    {
+      "epoch": 1.9809970174691096,
+      "grad_norm": 21.375,
+      "learning_rate": 5.616160885239808e-10,
+      "loss": 0.5526,
+      "step": 23247
+    },
+    {
+      "epoch": 1.9810822326374096,
+      "grad_norm": 16.5,
+      "learning_rate": 5.565908394598863e-10,
+      "loss": 0.5616,
+      "step": 23248
+    },
+    {
+      "epoch": 1.9811674478057095,
+      "grad_norm": 15.25,
+      "learning_rate": 5.515881690947023e-10,
+      "loss": 0.3627,
+      "step": 23249
+    },
+    {
+      "epoch": 1.9812526629740095,
+      "grad_norm": 17.375,
+      "learning_rate": 5.466080775186344e-10,
+      "loss": 0.9317,
+      "step": 23250
+    },
+    {
+      "epoch": 1.9813378781423094,
+      "grad_norm": 13.8125,
+      "learning_rate": 5.416505648218884e-10,
+      "loss": 0.5098,
+      "step": 23251
+    },
+    {
+      "epoch": 1.9814230933106094,
+      "grad_norm": 19.875,
+      "learning_rate": 5.367156310939759e-10,
+      "loss": 0.7808,
+      "step": 23252
+    },
+    {
+      "epoch": 1.9815083084789094,
+      "grad_norm": 11.375,
+      "learning_rate": 5.318032764238535e-10,
+      "loss": 0.3538,
+      "step": 23253
+    },
+    {
+      "epoch": 1.9815935236472093,
+      "grad_norm": 17.875,
+      "learning_rate": 5.269135009006165e-10,
+      "loss": 0.7669,
+      "step": 23254
+    },
+    {
+      "epoch": 1.9816787388155093,
+      "grad_norm": 18.375,
+      "learning_rate": 5.220463046125279e-10,
+      "loss": 0.4303,
+      "step": 23255
+    },
+    {
+      "epoch": 1.9817639539838092,
+      "grad_norm": 14.75,
+      "learning_rate": 5.172016876474339e-10,
+      "loss": 0.6543,
+      "step": 23256
+    },
+    {
+      "epoch": 1.9818491691521092,
+      "grad_norm": 20.625,
+      "learning_rate": 5.123796500930422e-10,
+      "loss": 0.707,
+      "step": 23257
+    },
+    {
+      "epoch": 1.9819343843204091,
+      "grad_norm": 14.5,
+      "learning_rate": 5.075801920363665e-10,
+      "loss": 0.5756,
+      "step": 23258
+    },
+    {
+      "epoch": 1.982019599488709,
+      "grad_norm": 16.5,
+      "learning_rate": 5.02803313564143e-10,
+      "loss": 1.2815,
+      "step": 23259
+    },
+    {
+      "epoch": 1.982104814657009,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.980490147628303e-10,
+      "loss": 0.3331,
+      "step": 23260
+    },
+    {
+      "epoch": 1.982190029825309,
+      "grad_norm": 17.125,
+      "learning_rate": 4.933172957180543e-10,
+      "loss": 0.5887,
+      "step": 23261
+    },
+    {
+      "epoch": 1.982275244993609,
+      "grad_norm": 18.5,
+      "learning_rate": 4.886081565157186e-10,
+      "loss": 0.5691,
+      "step": 23262
+    },
+    {
+      "epoch": 1.982360460161909,
+      "grad_norm": 16.125,
+      "learning_rate": 4.839215972406164e-10,
+      "loss": 0.6155,
+      "step": 23263
+    },
+    {
+      "epoch": 1.9824456753302089,
+      "grad_norm": 20.875,
+      "learning_rate": 4.79257617977541e-10,
+      "loss": 0.928,
+      "step": 23264
+    },
+    {
+      "epoch": 1.9825308904985088,
+      "grad_norm": 18.5,
+      "learning_rate": 4.746162188107306e-10,
+      "loss": 1.0592,
+      "step": 23265
+    },
+    {
+      "epoch": 1.9826161056668088,
+      "grad_norm": 16.125,
+      "learning_rate": 4.699973998241459e-10,
+      "loss": 0.561,
+      "step": 23266
+    },
+    {
+      "epoch": 1.9827013208351087,
+      "grad_norm": 15.5625,
+      "learning_rate": 4.6540116110119217e-10,
+      "loss": 0.5419,
+      "step": 23267
+    },
+    {
+      "epoch": 1.9827865360034087,
+      "grad_norm": 33.0,
+      "learning_rate": 4.6082750272499756e-10,
+      "loss": 0.8275,
+      "step": 23268
+    },
+    {
+      "epoch": 1.9828717511717087,
+      "grad_norm": 11.625,
+      "learning_rate": 4.5627642477813485e-10,
+      "loss": 0.345,
+      "step": 23269
+    },
+    {
+      "epoch": 1.9829569663400086,
+      "grad_norm": 14.3125,
+      "learning_rate": 4.517479273427605e-10,
+      "loss": 0.3628,
+      "step": 23270
+    },
+    {
+      "epoch": 1.9830421815083086,
+      "grad_norm": 15.0625,
+      "learning_rate": 4.4724201050089235e-10,
+      "loss": 0.7604,
+      "step": 23271
+    },
+    {
+      "epoch": 1.9831273966766085,
+      "grad_norm": 12.875,
+      "learning_rate": 4.4275867433399286e-10,
+      "loss": 0.3542,
+      "step": 23272
+    },
+    {
+      "epoch": 1.9832126118449085,
+      "grad_norm": 13.5625,
+      "learning_rate": 4.382979189228309e-10,
+      "loss": 0.5714,
+      "step": 23273
+    },
+    {
+      "epoch": 1.9832978270132084,
+      "grad_norm": 15.125,
+      "learning_rate": 4.338597443481751e-10,
+      "loss": 0.655,
+      "step": 23274
+    },
+    {
+      "epoch": 1.9833830421815084,
+      "grad_norm": 18.375,
+      "learning_rate": 4.2944415069023915e-10,
+      "loss": 0.4823,
+      "step": 23275
+    },
+    {
+      "epoch": 1.9834682573498084,
+      "grad_norm": 17.25,
+      "learning_rate": 4.2505113802882023e-10,
+      "loss": 0.8536,
+      "step": 23276
+    },
+    {
+      "epoch": 1.9835534725181083,
+      "grad_norm": 26.125,
+      "learning_rate": 4.206807064432994e-10,
+      "loss": 0.8074,
+      "step": 23277
+    },
+    {
+      "epoch": 1.9836386876864083,
+      "grad_norm": 13.4375,
+      "learning_rate": 4.163328560125024e-10,
+      "loss": 0.5872,
+      "step": 23278
+    },
+    {
+      "epoch": 1.9837239028547082,
+      "grad_norm": 15.3125,
+      "learning_rate": 4.120075868152551e-10,
+      "loss": 0.6299,
+      "step": 23279
+    },
+    {
+      "epoch": 1.9838091180230082,
+      "grad_norm": 17.0,
+      "learning_rate": 4.077048989295507e-10,
+      "loss": 0.6026,
+      "step": 23280
+    },
+    {
+      "epoch": 1.9838943331913081,
+      "grad_norm": 18.5,
+      "learning_rate": 4.0342479243310475e-10,
+      "loss": 0.3762,
+      "step": 23281
+    },
+    {
+      "epoch": 1.983979548359608,
+      "grad_norm": 19.25,
+      "learning_rate": 3.991672674033553e-10,
+      "loss": 0.5901,
+      "step": 23282
+    },
+    {
+      "epoch": 1.984064763527908,
+      "grad_norm": 15.375,
+      "learning_rate": 3.9493232391732415e-10,
+      "loss": 0.6387,
+      "step": 23283
+    },
+    {
+      "epoch": 1.984149978696208,
+      "grad_norm": 15.0,
+      "learning_rate": 3.907199620513391e-10,
+      "loss": 0.5621,
+      "step": 23284
+    },
+    {
+      "epoch": 1.984235193864508,
+      "grad_norm": 14.125,
+      "learning_rate": 3.8653018188158917e-10,
+      "loss": 0.3461,
+      "step": 23285
+    },
+    {
+      "epoch": 1.984320409032808,
+      "grad_norm": 14.625,
+      "learning_rate": 3.8236298348398594e-10,
+      "loss": 0.5143,
+      "step": 23286
+    },
+    {
+      "epoch": 1.9844056242011079,
+      "grad_norm": 14.1875,
+      "learning_rate": 3.7821836693346935e-10,
+      "loss": 0.5395,
+      "step": 23287
+    },
+    {
+      "epoch": 1.9844908393694078,
+      "grad_norm": 13.8125,
+      "learning_rate": 3.740963323052571e-10,
+      "loss": 0.4294,
+      "step": 23288
+    },
+    {
+      "epoch": 1.9845760545377078,
+      "grad_norm": 15.0,
+      "learning_rate": 3.699968796737341e-10,
+      "loss": 0.6105,
+      "step": 23289
+    },
+    {
+      "epoch": 1.9846612697060078,
+      "grad_norm": 16.875,
+      "learning_rate": 3.659200091130077e-10,
+      "loss": 0.6301,
+      "step": 23290
+    },
+    {
+      "epoch": 1.9847464848743077,
+      "grad_norm": 13.3125,
+      "learning_rate": 3.618657206966303e-10,
+      "loss": 0.3805,
+      "step": 23291
+    },
+    {
+      "epoch": 1.9848317000426077,
+      "grad_norm": 13.75,
+      "learning_rate": 3.5783401449801525e-10,
+      "loss": 0.7186,
+      "step": 23292
+    },
+    {
+      "epoch": 1.9849169152109076,
+      "grad_norm": 17.5,
+      "learning_rate": 3.5382489058988224e-10,
+      "loss": 0.4876,
+      "step": 23293
+    },
+    {
+      "epoch": 1.9850021303792076,
+      "grad_norm": 13.1875,
+      "learning_rate": 3.498383490446733e-10,
+      "loss": 0.5962,
+      "step": 23294
+    },
+    {
+      "epoch": 1.9850873455475075,
+      "grad_norm": 11.875,
+      "learning_rate": 3.458743899346917e-10,
+      "loss": 0.3941,
+      "step": 23295
+    },
+    {
+      "epoch": 1.9851725607158075,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.4193301333126927e-10,
+      "loss": 0.6987,
+      "step": 23296
+    },
+    {
+      "epoch": 1.9852577758841075,
+      "grad_norm": 16.5,
+      "learning_rate": 3.3801421930573784e-10,
+      "loss": 0.536,
+      "step": 23297
+    },
+    {
+      "epoch": 1.9853429910524074,
+      "grad_norm": 26.875,
+      "learning_rate": 3.3411800792901293e-10,
+      "loss": 0.9491,
+      "step": 23298
+    },
+    {
+      "epoch": 1.9854282062207074,
+      "grad_norm": 14.9375,
+      "learning_rate": 3.3024437927131614e-10,
+      "loss": 0.5384,
+      "step": 23299
+    },
+    {
+      "epoch": 1.9855134213890073,
+      "grad_norm": 12.4375,
+      "learning_rate": 3.263933334027303e-10,
+      "loss": 0.4388,
+      "step": 23300
+    },
+    {
+      "epoch": 1.9855986365573073,
+      "grad_norm": 19.875,
+      "learning_rate": 3.2256487039292186e-10,
+      "loss": 0.8038,
+      "step": 23301
+    },
+    {
+      "epoch": 1.9856838517256072,
+      "grad_norm": 16.875,
+      "learning_rate": 3.187589903110022e-10,
+      "loss": 0.7514,
+      "step": 23302
+    },
+    {
+      "epoch": 1.9857690668939072,
+      "grad_norm": 13.375,
+      "learning_rate": 3.1497569322580526e-10,
+      "loss": 0.5579,
+      "step": 23303
+    },
+    {
+      "epoch": 1.9858542820622072,
+      "grad_norm": 24.25,
+      "learning_rate": 3.112149792056096e-10,
+      "loss": 0.8181,
+      "step": 23304
+    },
+    {
+      "epoch": 1.9859394972305071,
+      "grad_norm": 10.9375,
+      "learning_rate": 3.074768483184165e-10,
+      "loss": 0.4085,
+      "step": 23305
+    },
+    {
+      "epoch": 1.986024712398807,
+      "grad_norm": 17.75,
+      "learning_rate": 3.0376130063181073e-10,
+      "loss": 0.5633,
+      "step": 23306
+    },
+    {
+      "epoch": 1.986109927567107,
+      "grad_norm": 23.75,
+      "learning_rate": 3.0006833621282204e-10,
+      "loss": 0.5607,
+      "step": 23307
+    },
+    {
+      "epoch": 1.986195142735407,
+      "grad_norm": 29.375,
+      "learning_rate": 2.9639795512834133e-10,
+      "loss": 1.0904,
+      "step": 23308
+    },
+    {
+      "epoch": 1.986280357903707,
+      "grad_norm": 17.125,
+      "learning_rate": 2.927501574445657e-10,
+      "loss": 1.0816,
+      "step": 23309
+    },
+    {
+      "epoch": 1.986365573072007,
+      "grad_norm": 16.375,
+      "learning_rate": 2.8912494322755335e-10,
+      "loss": 0.431,
+      "step": 23310
+    },
+    {
+      "epoch": 1.9864507882403069,
+      "grad_norm": 19.125,
+      "learning_rate": 2.855223125426687e-10,
+      "loss": 0.6658,
+      "step": 23311
+    },
+    {
+      "epoch": 1.9865360034086068,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.819422654551374e-10,
+      "loss": 0.6147,
+      "step": 23312
+    },
+    {
+      "epoch": 1.9866212185769068,
+      "grad_norm": 21.375,
+      "learning_rate": 2.7838480202949105e-10,
+      "loss": 0.7043,
+      "step": 23313
+    },
+    {
+      "epoch": 1.9867064337452067,
+      "grad_norm": 21.125,
+      "learning_rate": 2.748499223302614e-10,
+      "loss": 0.7865,
+      "step": 23314
+    },
+    {
+      "epoch": 1.9867916489135067,
+      "grad_norm": 12.1875,
+      "learning_rate": 2.7133762642114756e-10,
+      "loss": 0.7474,
+      "step": 23315
+    },
+    {
+      "epoch": 1.9868768640818066,
+      "grad_norm": 15.4375,
+      "learning_rate": 2.67847914365571e-10,
+      "loss": 0.6573,
+      "step": 23316
+    },
+    {
+      "epoch": 1.9869620792501066,
+      "grad_norm": 14.875,
+      "learning_rate": 2.643807862268144e-10,
+      "loss": 0.5092,
+      "step": 23317
+    },
+    {
+      "epoch": 1.9870472944184066,
+      "grad_norm": 17.5,
+      "learning_rate": 2.609362420673278e-10,
+      "loss": 0.5382,
+      "step": 23318
+    },
+    {
+      "epoch": 1.9871325095867065,
+      "grad_norm": 16.375,
+      "learning_rate": 2.575142819494225e-10,
+      "loss": 0.5776,
+      "step": 23319
+    },
+    {
+      "epoch": 1.9872177247550065,
+      "grad_norm": 20.5,
+      "learning_rate": 2.541149059349934e-10,
+      "loss": 0.8098,
+      "step": 23320
+    },
+    {
+      "epoch": 1.9873029399233064,
+      "grad_norm": 16.75,
+      "learning_rate": 2.507381140853804e-10,
+      "loss": 0.7249,
+      "step": 23321
+    },
+    {
+      "epoch": 1.9873881550916064,
+      "grad_norm": 18.25,
+      "learning_rate": 2.4738390646164567e-10,
+      "loss": 0.6222,
+      "step": 23322
+    },
+    {
+      "epoch": 1.9874733702599063,
+      "grad_norm": 16.375,
+      "learning_rate": 2.4405228312443516e-10,
+      "loss": 0.4627,
+      "step": 23323
+    },
+    {
+      "epoch": 1.9875585854282063,
+      "grad_norm": 16.75,
+      "learning_rate": 2.407432441339785e-10,
+      "loss": 0.4503,
+      "step": 23324
+    },
+    {
+      "epoch": 1.9876438005965062,
+      "grad_norm": 19.125,
+      "learning_rate": 2.374567895499502e-10,
+      "loss": 0.6984,
+      "step": 23325
+    },
+    {
+      "epoch": 1.9877290157648062,
+      "grad_norm": 19.75,
+      "learning_rate": 2.3419291943174716e-10,
+      "loss": 0.8107,
+      "step": 23326
+    },
+    {
+      "epoch": 1.9878142309331062,
+      "grad_norm": 15.6875,
+      "learning_rate": 2.3095163383848873e-10,
+      "loss": 1.0178,
+      "step": 23327
+    },
+    {
+      "epoch": 1.9878994461014061,
+      "grad_norm": 13.75,
+      "learning_rate": 2.277329328287392e-10,
+      "loss": 0.4053,
+      "step": 23328
+    },
+    {
+      "epoch": 1.987984661269706,
+      "grad_norm": 13.625,
+      "learning_rate": 2.2453681646050774e-10,
+      "loss": 0.5675,
+      "step": 23329
+    },
+    {
+      "epoch": 1.988069876438006,
+      "grad_norm": 18.0,
+      "learning_rate": 2.213632847916647e-10,
+      "loss": 0.7549,
+      "step": 23330
+    },
+    {
+      "epoch": 1.988155091606306,
+      "grad_norm": 12.4375,
+      "learning_rate": 2.1821233787966412e-10,
+      "loss": 0.5089,
+      "step": 23331
+    },
+    {
+      "epoch": 1.988240306774606,
+      "grad_norm": 17.5,
+      "learning_rate": 2.1508397578112738e-10,
+      "loss": 0.7466,
+      "step": 23332
+    },
+    {
+      "epoch": 1.988325521942906,
+      "grad_norm": 13.875,
+      "learning_rate": 2.1197819855295342e-10,
+      "loss": 0.5205,
+      "step": 23333
+    },
+    {
+      "epoch": 1.9884107371112059,
+      "grad_norm": 18.25,
+      "learning_rate": 2.0889500625093095e-10,
+      "loss": 0.8504,
+      "step": 23334
+    },
+    {
+      "epoch": 1.9884959522795058,
+      "grad_norm": 15.0,
+      "learning_rate": 2.0583439893098746e-10,
+      "loss": 0.6026,
+      "step": 23335
+    },
+    {
+      "epoch": 1.9885811674478058,
+      "grad_norm": 14.3125,
+      "learning_rate": 2.027963766483565e-10,
+      "loss": 0.5881,
+      "step": 23336
+    },
+    {
+      "epoch": 1.9886663826161057,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9978093945799415e-10,
+      "loss": 0.6151,
+      "step": 23337
+    },
+    {
+      "epoch": 1.9887515977844057,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9678808741444012e-10,
+      "loss": 0.5574,
+      "step": 23338
+    },
+    {
+      "epoch": 1.9888368129527056,
+      "grad_norm": 24.125,
+      "learning_rate": 1.9381782057167896e-10,
+      "loss": 0.7941,
+      "step": 23339
+    },
+    {
+      "epoch": 1.9889220281210056,
+      "grad_norm": 15.25,
+      "learning_rate": 1.90870138983279e-10,
+      "loss": 0.6016,
+      "step": 23340
+    },
+    {
+      "epoch": 1.9890072432893056,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8794504270266967e-10,
+      "loss": 0.6921,
+      "step": 23341
+    },
+    {
+      "epoch": 1.9890924584576055,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8504253178272535e-10,
+      "loss": 0.4144,
+      "step": 23342
+    },
+    {
+      "epoch": 1.9891776736259055,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.821626062759041e-10,
+      "loss": 0.4146,
+      "step": 23343
+    },
+    {
+      "epoch": 1.9892628887942054,
+      "grad_norm": 16.75,
+      "learning_rate": 1.7930526623410883e-10,
+      "loss": 0.6635,
+      "step": 23344
+    },
+    {
+      "epoch": 1.9893481039625054,
+      "grad_norm": 12.875,
+      "learning_rate": 1.7647051170910369e-10,
+      "loss": 0.3788,
+      "step": 23345
+    },
+    {
+      "epoch": 1.9894333191308053,
+      "grad_norm": 21.375,
+      "learning_rate": 1.736583427520977e-10,
+      "loss": 1.052,
+      "step": 23346
+    },
+    {
+      "epoch": 1.9895185342991053,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7086875941374482e-10,
+      "loss": 0.7536,
+      "step": 23347
+    },
+    {
+      "epoch": 1.9896037494674053,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.6810176174469894e-10,
+      "loss": 0.4115,
+      "step": 23348
+    },
+    {
+      "epoch": 1.9896889646357052,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.6535734979492013e-10,
+      "loss": 0.3688,
+      "step": 23349
+    },
+    {
+      "epoch": 1.9897741798040052,
+      "grad_norm": 15.5,
+      "learning_rate": 1.626355236138133e-10,
+      "loss": 0.5693,
+      "step": 23350
+    },
+    {
+      "epoch": 1.9898593949723051,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.5993628325078338e-10,
+      "loss": 0.7412,
+      "step": 23351
+    },
+    {
+      "epoch": 1.989944610140605,
+      "grad_norm": 23.0,
+      "learning_rate": 1.572596287544026e-10,
+      "loss": 1.0815,
+      "step": 23352
+    },
+    {
+      "epoch": 1.990029825308905,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5460556017324325e-10,
+      "loss": 0.5194,
+      "step": 23353
+    },
+    {
+      "epoch": 1.990115040477205,
+      "grad_norm": 19.5,
+      "learning_rate": 1.5197407755518369e-10,
+      "loss": 0.5725,
+      "step": 23354
+    },
+    {
+      "epoch": 1.990200255645505,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.493651809478247e-10,
+      "loss": 0.2269,
+      "step": 23355
+    },
+    {
+      "epoch": 1.990285470813805,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4677887039807327e-10,
+      "loss": 0.248,
+      "step": 23356
+    },
+    {
+      "epoch": 1.9903706859821049,
+      "grad_norm": 26.625,
+      "learning_rate": 1.4421514595297504e-10,
+      "loss": 0.7722,
+      "step": 23357
+    },
+    {
+      "epoch": 1.9904559011504048,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4167400765874307e-10,
+      "loss": 0.5324,
+      "step": 23358
+    },
+    {
+      "epoch": 1.9905411163187048,
+      "grad_norm": 13.875,
+      "learning_rate": 1.3915545556131283e-10,
+      "loss": 0.699,
+      "step": 23359
+    },
+    {
+      "epoch": 1.9906263314870047,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.3665948970606468e-10,
+      "loss": 0.5638,
+      "step": 23360
+    },
+    {
+      "epoch": 1.9907115466553047,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3418611013837901e-10,
+      "loss": 0.5347,
+      "step": 23361
+    },
+    {
+      "epoch": 1.9907967618236047,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3173531690266473e-10,
+      "loss": 0.7225,
+      "step": 23362
+    },
+    {
+      "epoch": 1.9908819769919046,
+      "grad_norm": 15.125,
+      "learning_rate": 1.2930711004333073e-10,
+      "loss": 0.6781,
+      "step": 23363
+    },
+    {
+      "epoch": 1.9909671921602046,
+      "grad_norm": 18.0,
+      "learning_rate": 1.2690148960436966e-10,
+      "loss": 0.536,
+      "step": 23364
+    },
+    {
+      "epoch": 1.9910524073285045,
+      "grad_norm": 16.625,
+      "learning_rate": 1.2451845562894137e-10,
+      "loss": 0.4712,
+      "step": 23365
+    },
+    {
+      "epoch": 1.9911376224968045,
+      "grad_norm": 21.625,
+      "learning_rate": 1.2215800816048339e-10,
+      "loss": 0.7013,
+      "step": 23366
+    },
+    {
+      "epoch": 1.9912228376651044,
+      "grad_norm": 16.75,
+      "learning_rate": 1.1982014724146174e-10,
+      "loss": 0.8034,
+      "step": 23367
+    },
+    {
+      "epoch": 1.9913080528334044,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1750487291406488e-10,
+      "loss": 0.5762,
+      "step": 23368
+    },
+    {
+      "epoch": 1.9913932680017044,
+      "grad_norm": 18.5,
+      "learning_rate": 1.152121852203425e-10,
+      "loss": 0.6992,
+      "step": 23369
+    },
+    {
+      "epoch": 1.9914784831700043,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1294208420151165e-10,
+      "loss": 0.6394,
+      "step": 23370
+    },
+    {
+      "epoch": 1.9915636983383043,
+      "grad_norm": 15.75,
+      "learning_rate": 1.1069456989865057e-10,
+      "loss": 0.4918,
+      "step": 23371
+    },
+    {
+      "epoch": 1.9916489135066042,
+      "grad_norm": 18.5,
+      "learning_rate": 1.0846964235242119e-10,
+      "loss": 0.9906,
+      "step": 23372
+    },
+    {
+      "epoch": 1.9917341286749042,
+      "grad_norm": 15.625,
+      "learning_rate": 1.062673016029303e-10,
+      "loss": 0.8268,
+      "step": 23373
+    },
+    {
+      "epoch": 1.9918193438432041,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.0408754769014595e-10,
+      "loss": 0.4655,
+      "step": 23374
+    },
+    {
+      "epoch": 1.991904559011504,
+      "grad_norm": 18.5,
+      "learning_rate": 1.0193038065334226e-10,
+      "loss": 0.4301,
+      "step": 23375
+    },
+    {
+      "epoch": 1.991989774179804,
+      "grad_norm": 12.6875,
+      "learning_rate": 9.979580053151582e-11,
+      "loss": 0.4808,
+      "step": 23376
+    },
+    {
+      "epoch": 1.992074989348104,
+      "grad_norm": 15.9375,
+      "learning_rate": 9.768380736338568e-11,
+      "loss": 0.8434,
+      "step": 23377
+    },
+    {
+      "epoch": 1.992160204516404,
+      "grad_norm": 14.0,
+      "learning_rate": 9.559440118683816e-11,
+      "loss": 0.3944,
+      "step": 23378
+    },
+    {
+      "epoch": 1.992245419684704,
+      "grad_norm": 12.5625,
+      "learning_rate": 9.352758203975965e-11,
+      "loss": 0.2396,
+      "step": 23379
+    },
+    {
+      "epoch": 1.9923306348530039,
+      "grad_norm": 18.125,
+      "learning_rate": 9.148334995962016e-11,
+      "loss": 0.6497,
+      "step": 23380
+    },
+    {
+      "epoch": 1.9924158500213038,
+      "grad_norm": 12.5625,
+      "learning_rate": 8.946170498319584e-11,
+      "loss": 0.5476,
+      "step": 23381
+    },
+    {
+      "epoch": 1.9925010651896038,
+      "grad_norm": 16.125,
+      "learning_rate": 8.746264714698527e-11,
+      "loss": 0.8115,
+      "step": 23382
+    },
+    {
+      "epoch": 1.9925862803579037,
+      "grad_norm": 16.25,
+      "learning_rate": 8.548617648734825e-11,
+      "loss": 0.6526,
+      "step": 23383
+    },
+    {
+      "epoch": 1.9926714955262037,
+      "grad_norm": 14.5625,
+      "learning_rate": 8.353229303981191e-11,
+      "loss": 0.6659,
+      "step": 23384
+    },
+    {
+      "epoch": 1.9927567106945037,
+      "grad_norm": 12.375,
+      "learning_rate": 8.160099683976463e-11,
+      "loss": 0.6611,
+      "step": 23385
+    },
+    {
+      "epoch": 1.9928419258628036,
+      "grad_norm": 19.125,
+      "learning_rate": 7.969228792203965e-11,
+      "loss": 0.8846,
+      "step": 23386
+    },
+    {
+      "epoch": 1.9929271410311036,
+      "grad_norm": 14.9375,
+      "learning_rate": 7.780616632119264e-11,
+      "loss": 0.4177,
+      "step": 23387
+    },
+    {
+      "epoch": 1.9930123561994035,
+      "grad_norm": 12.75,
+      "learning_rate": 7.594263207122421e-11,
+      "loss": 0.4071,
+      "step": 23388
+    },
+    {
+      "epoch": 1.9930975713677035,
+      "grad_norm": 14.3125,
+      "learning_rate": 7.410168520599614e-11,
+      "loss": 0.5367,
+      "step": 23389
+    },
+    {
+      "epoch": 1.9931827865360034,
+      "grad_norm": 14.25,
+      "learning_rate": 7.228332575853758e-11,
+      "loss": 0.528,
+      "step": 23390
+    },
+    {
+      "epoch": 1.9932680017043034,
+      "grad_norm": 13.0,
+      "learning_rate": 7.048755376187766e-11,
+      "loss": 0.4956,
+      "step": 23391
+    },
+    {
+      "epoch": 1.9933532168726034,
+      "grad_norm": 17.0,
+      "learning_rate": 6.871436924835162e-11,
+      "loss": 0.7185,
+      "step": 23392
+    },
+    {
+      "epoch": 1.9934384320409033,
+      "grad_norm": 19.0,
+      "learning_rate": 6.696377225015594e-11,
+      "loss": 0.9757,
+      "step": 23393
+    },
+    {
+      "epoch": 1.9935236472092033,
+      "grad_norm": 24.0,
+      "learning_rate": 6.52357627986544e-11,
+      "loss": 0.6906,
+      "step": 23394
+    },
+    {
+      "epoch": 1.9936088623775032,
+      "grad_norm": 18.25,
+      "learning_rate": 6.353034092548838e-11,
+      "loss": 0.5773,
+      "step": 23395
+    },
+    {
+      "epoch": 1.9936940775458032,
+      "grad_norm": 11.5,
+      "learning_rate": 6.184750666105022e-11,
+      "loss": 0.2112,
+      "step": 23396
+    },
+    {
+      "epoch": 1.9937792927141031,
+      "grad_norm": 18.375,
+      "learning_rate": 6.018726003600983e-11,
+      "loss": 0.6498,
+      "step": 23397
+    },
+    {
+      "epoch": 1.993864507882403,
+      "grad_norm": 15.3125,
+      "learning_rate": 5.854960108020447e-11,
+      "loss": 0.5303,
+      "step": 23398
+    },
+    {
+      "epoch": 1.993949723050703,
+      "grad_norm": 14.875,
+      "learning_rate": 5.693452982347136e-11,
+      "loss": 0.7258,
+      "step": 23399
+    },
+    {
+      "epoch": 1.994034938219003,
+      "grad_norm": 18.125,
+      "learning_rate": 5.53420462948151e-11,
+      "loss": 0.5227,
+      "step": 23400
+    },
+    {
+      "epoch": 1.994120153387303,
+      "grad_norm": 19.0,
+      "learning_rate": 5.37721505229627e-11,
+      "loss": 0.9858,
+      "step": 23401
+    },
+    {
+      "epoch": 1.994205368555603,
+      "grad_norm": 12.375,
+      "learning_rate": 5.222484253636362e-11,
+      "loss": 0.3322,
+      "step": 23402
+    },
+    {
+      "epoch": 1.9942905837239029,
+      "grad_norm": 14.9375,
+      "learning_rate": 5.0700122363050995e-11,
+      "loss": 0.6295,
+      "step": 23403
+    },
+    {
+      "epoch": 1.9943757988922028,
+      "grad_norm": 19.875,
+      "learning_rate": 4.919799003050285e-11,
+      "loss": 0.7383,
+      "step": 23404
+    },
+    {
+      "epoch": 1.9944610140605028,
+      "grad_norm": 9.625,
+      "learning_rate": 4.771844556578087e-11,
+      "loss": 0.2615,
+      "step": 23405
+    },
+    {
+      "epoch": 1.9945462292288028,
+      "grad_norm": 13.125,
+      "learning_rate": 4.6261488995807956e-11,
+      "loss": 0.3842,
+      "step": 23406
+    },
+    {
+      "epoch": 1.9946314443971027,
+      "grad_norm": 14.0,
+      "learning_rate": 4.4827120346813134e-11,
+      "loss": 0.435,
+      "step": 23407
+    },
+    {
+      "epoch": 1.9947166595654027,
+      "grad_norm": 18.625,
+      "learning_rate": 4.341533964474787e-11,
+      "loss": 0.3524,
+      "step": 23408
+    },
+    {
+      "epoch": 1.9948018747337026,
+      "grad_norm": 17.5,
+      "learning_rate": 4.2026146915008505e-11,
+      "loss": 0.5361,
+      "step": 23409
+    },
+    {
+      "epoch": 1.9948870899020026,
+      "grad_norm": 15.4375,
+      "learning_rate": 4.0659542182852616e-11,
+      "loss": 0.6589,
+      "step": 23410
+    },
+    {
+      "epoch": 1.9949723050703025,
+      "grad_norm": 15.75,
+      "learning_rate": 3.93155254728439e-11,
+      "loss": 0.6018,
+      "step": 23411
+    },
+    {
+      "epoch": 1.9950575202386025,
+      "grad_norm": 16.5,
+      "learning_rate": 3.799409680940724e-11,
+      "loss": 0.4634,
+      "step": 23412
+    },
+    {
+      "epoch": 1.9951427354069025,
+      "grad_norm": 20.125,
+      "learning_rate": 3.6695256216273675e-11,
+      "loss": 0.5748,
+      "step": 23413
+    },
+    {
+      "epoch": 1.9952279505752024,
+      "grad_norm": 14.875,
+      "learning_rate": 3.541900371703544e-11,
+      "loss": 0.6229,
+      "step": 23414
+    },
+    {
+      "epoch": 1.9953131657435024,
+      "grad_norm": 15.5625,
+      "learning_rate": 3.4165339334729654e-11,
+      "loss": 0.3247,
+      "step": 23415
+    },
+    {
+      "epoch": 1.9953983809118023,
+      "grad_norm": 30.75,
+      "learning_rate": 3.2934263091977115e-11,
+      "loss": 0.4226,
+      "step": 23416
+    },
+    {
+      "epoch": 1.9954835960801023,
+      "grad_norm": 33.5,
+      "learning_rate": 3.1725775010982284e-11,
+      "loss": 1.1166,
+      "step": 23417
+    },
+    {
+      "epoch": 1.9955688112484022,
+      "grad_norm": 14.3125,
+      "learning_rate": 3.053987511367207e-11,
+      "loss": 0.4443,
+      "step": 23418
+    },
+    {
+      "epoch": 1.9956540264167022,
+      "grad_norm": 13.375,
+      "learning_rate": 2.937656342155704e-11,
+      "loss": 0.6238,
+      "step": 23419
+    },
+    {
+      "epoch": 1.9957392415850022,
+      "grad_norm": 10.5,
+      "learning_rate": 2.823583995545387e-11,
+      "loss": 0.2475,
+      "step": 23420
+    },
+    {
+      "epoch": 1.9958244567533021,
+      "grad_norm": 19.125,
+      "learning_rate": 2.7117704736040475e-11,
+      "loss": 0.5291,
+      "step": 23421
+    },
+    {
+      "epoch": 1.995909671921602,
+      "grad_norm": 25.0,
+      "learning_rate": 2.6022157783578416e-11,
+      "loss": 0.5864,
+      "step": 23422
+    },
+    {
+      "epoch": 1.995994887089902,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.4949199117912937e-11,
+      "loss": 0.2874,
+      "step": 23423
+    },
+    {
+      "epoch": 1.996080102258202,
+      "grad_norm": 25.625,
+      "learning_rate": 2.389882875833416e-11,
+      "loss": 0.5458,
+      "step": 23424
+    },
+    {
+      "epoch": 1.996165317426502,
+      "grad_norm": 13.25,
+      "learning_rate": 2.2871046723854652e-11,
+      "loss": 0.3514,
+      "step": 23425
+    },
+    {
+      "epoch": 1.996250532594802,
+      "grad_norm": 21.75,
+      "learning_rate": 2.186585303293187e-11,
+      "loss": 0.9411,
+      "step": 23426
+    },
+    {
+      "epoch": 1.9963357477631019,
+      "grad_norm": 14.8125,
+      "learning_rate": 2.088324770402328e-11,
+      "loss": 0.694,
+      "step": 23427
+    },
+    {
+      "epoch": 1.9964209629314018,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9923230754614886e-11,
+      "loss": 0.5039,
+      "step": 23428
+    },
+    {
+      "epoch": 1.9965061780997018,
+      "grad_norm": 12.875,
+      "learning_rate": 1.8985802202053926e-11,
+      "loss": 0.5889,
+      "step": 23429
+    },
+    {
+      "epoch": 1.9965913932680017,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8070962063548856e-11,
+      "loss": 0.6589,
+      "step": 23430
+    },
+    {
+      "epoch": 1.9966766084363017,
+      "grad_norm": 13.375,
+      "learning_rate": 1.717871035533669e-11,
+      "loss": 0.3435,
+      "step": 23431
+    },
+    {
+      "epoch": 1.9967618236046016,
+      "grad_norm": 13.5,
+      "learning_rate": 1.630904709365444e-11,
+      "loss": 0.7092,
+      "step": 23432
+    },
+    {
+      "epoch": 1.9968470387729016,
+      "grad_norm": 22.125,
+      "learning_rate": 1.546197229432278e-11,
+      "loss": 0.9153,
+      "step": 23433
+    },
+    {
+      "epoch": 1.9969322539412016,
+      "grad_norm": 18.875,
+      "learning_rate": 1.4637485972468502e-11,
+      "loss": 0.6093,
+      "step": 23434
+    },
+    {
+      "epoch": 1.9970174691095015,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.3835588143079615e-11,
+      "loss": 0.3844,
+      "step": 23435
+    },
+    {
+      "epoch": 1.9971026842778015,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.3056278820727797e-11,
+      "loss": 0.2483,
+      "step": 23436
+    },
+    {
+      "epoch": 1.9971878994461014,
+      "grad_norm": 15.625,
+      "learning_rate": 1.2299558019290836e-11,
+      "loss": 0.6519,
+      "step": 23437
+    },
+    {
+      "epoch": 1.9972731146144014,
+      "grad_norm": 19.125,
+      "learning_rate": 1.156542575264652e-11,
+      "loss": 0.5929,
+      "step": 23438
+    },
+    {
+      "epoch": 1.9973583297827013,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.0853882033839969e-11,
+      "loss": 0.5593,
+      "step": 23439
+    },
+    {
+      "epoch": 1.9974435449510013,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.0164926876055082e-11,
+      "loss": 0.4518,
+      "step": 23440
+    },
+    {
+      "epoch": 1.9975287601193012,
+      "grad_norm": 16.625,
+      "learning_rate": 9.498560291365532e-12,
+      "loss": 0.6395,
+      "step": 23441
+    },
+    {
+      "epoch": 1.9976139752876012,
+      "grad_norm": 17.0,
+      "learning_rate": 8.854782292122554e-12,
+      "loss": 0.8704,
+      "step": 23442
+    },
+    {
+      "epoch": 1.9976991904559012,
+      "grad_norm": 16.375,
+      "learning_rate": 8.23359288984471e-12,
+      "loss": 0.7305,
+      "step": 23443
+    },
+    {
+      "epoch": 1.9977844056242011,
+      "grad_norm": 12.0,
+      "learning_rate": 7.634992095634231e-12,
+      "loss": 0.3576,
+      "step": 23444
+    },
+    {
+      "epoch": 1.997869620792501,
+      "grad_norm": 15.75,
+      "learning_rate": 7.058979920593345e-12,
+      "loss": 0.4708,
+      "step": 23445
+    },
+    {
+      "epoch": 1.997954835960801,
+      "grad_norm": 16.125,
+      "learning_rate": 6.505556374852839e-12,
+      "loss": 0.7115,
+      "step": 23446
+    },
+    {
+      "epoch": 1.998040051129101,
+      "grad_norm": 16.375,
+      "learning_rate": 5.974721468543498e-12,
+      "loss": 0.4002,
+      "step": 23447
+    },
+    {
+      "epoch": 1.998125266297401,
+      "grad_norm": 19.5,
+      "learning_rate": 5.466475211240996e-12,
+      "loss": 0.958,
+      "step": 23448
+    },
+    {
+      "epoch": 1.998210481465701,
+      "grad_norm": 16.875,
+      "learning_rate": 4.9808176121046715e-12,
+      "loss": 0.9991,
+      "step": 23449
+    },
+    {
+      "epoch": 1.9982956966340009,
+      "grad_norm": 15.75,
+      "learning_rate": 4.517748679877531e-12,
+      "loss": 0.4989,
+      "step": 23450
+    },
+    {
+      "epoch": 1.9983809118023008,
+      "grad_norm": 13.0625,
+      "learning_rate": 4.077268423025027e-12,
+      "loss": 0.5058,
+      "step": 23451
+    },
+    {
+      "epoch": 1.9984661269706008,
+      "grad_norm": 15.1875,
+      "learning_rate": 3.659376849596274e-12,
+      "loss": 0.4856,
+      "step": 23452
+    },
+    {
+      "epoch": 1.9985513421389007,
+      "grad_norm": 16.625,
+      "learning_rate": 3.2640739668077237e-12,
+      "loss": 0.6734,
+      "step": 23453
+    },
+    {
+      "epoch": 1.9986365573072007,
+      "grad_norm": 20.75,
+      "learning_rate": 2.89135978215338e-12,
+      "loss": 0.6762,
+      "step": 23454
+    },
+    {
+      "epoch": 1.9987217724755006,
+      "grad_norm": 17.0,
+      "learning_rate": 2.541234302155804e-12,
+      "loss": 0.6329,
+      "step": 23455
+    },
+    {
+      "epoch": 1.9988069876438006,
+      "grad_norm": 16.375,
+      "learning_rate": 2.2136975331987775e-12,
+      "loss": 0.4668,
+      "step": 23456
+    },
+    {
+      "epoch": 1.9988922028121006,
+      "grad_norm": 18.125,
+      "learning_rate": 1.90874948124975e-12,
+      "loss": 0.5856,
+      "step": 23457
+    },
+    {
+      "epoch": 1.9989774179804005,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6263901517210579e-12,
+      "loss": 0.9124,
+      "step": 23458
+    },
+    {
+      "epoch": 1.9990626331487005,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.366619549886261e-12,
+      "loss": 0.7548,
+      "step": 23459
+    },
+    {
+      "epoch": 1.9991478483170004,
+      "grad_norm": 16.125,
+      "learning_rate": 1.1294376801862517e-12,
+      "loss": 0.7113,
+      "step": 23460
+    },
+    {
+      "epoch": 1.9992330634853004,
+      "grad_norm": 12.125,
+      "learning_rate": 9.148445472006994e-13,
+      "loss": 0.7011,
+      "step": 23461
+    },
+    {
+      "epoch": 1.9993182786536003,
+      "grad_norm": 14.875,
+      "learning_rate": 7.228401545378294e-13,
+      "loss": 0.5103,
+      "step": 23462
+    },
+    {
+      "epoch": 1.9994034938219003,
+      "grad_norm": 13.625,
+      "learning_rate": 5.534245058058663e-13,
+      "loss": 0.5976,
+      "step": 23463
+    },
+    {
+      "epoch": 1.9994887089902003,
+      "grad_norm": 15.625,
+      "learning_rate": 4.0659760405792335e-13,
+      "loss": 0.7625,
+      "step": 23464
+    },
+    {
+      "epoch": 1.9995739241585002,
+      "grad_norm": 14.5625,
+      "learning_rate": 2.823594519307804e-13,
+      "loss": 0.6131,
+      "step": 23465
+    },
+    {
+      "epoch": 1.9996591393268002,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8071005164488343e-13,
+      "loss": 0.6435,
+      "step": 23466
+    },
+    {
+      "epoch": 1.9997443544951001,
+      "grad_norm": 13.9375,
+      "learning_rate": 1.0164940514312272e-13,
+      "loss": 0.5344,
+      "step": 23467
+    },
+    {
+      "epoch": 1.9998295696634,
+      "grad_norm": 14.8125,
+      "learning_rate": 4.517751367449919e-14,
+      "loss": 0.3828,
+      "step": 23468
+    },
+    {
+      "epoch": 1.9999147848317,
+      "grad_norm": 22.875,
+      "learning_rate": 1.1294378488013736e-14,
+      "loss": 0.5472,
+      "step": 23469
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 16.75,
+      "learning_rate": 0.0,
+      "loss": 0.7522,
+      "step": 23470
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 23470,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 5868,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3209448924525363e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}