diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,42081 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5266174974149115,
+  "eval_steps": 1000,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.776958290248525e-05,
+      "grad_norm": 121.5,
+      "learning_rate": 3e-06,
+      "loss": 7.9287,
+      "step": 1
+    },
+    {
+      "epoch": 0.0001755391658049705,
+      "grad_norm": 112.0,
+      "learning_rate": 6e-06,
+      "loss": 7.9365,
+      "step": 2
+    },
+    {
+      "epoch": 0.00026330874870745575,
+      "grad_norm": 73.0,
+      "learning_rate": 9e-06,
+      "loss": 7.9922,
+      "step": 3
+    },
+    {
+      "epoch": 0.000351078331609941,
+      "grad_norm": 67.0,
+      "learning_rate": 1.2e-05,
+      "loss": 8.0557,
+      "step": 4
+    },
+    {
+      "epoch": 0.00043884791451242626,
+      "grad_norm": 82.0,
+      "learning_rate": 1.5e-05,
+      "loss": 7.8389,
+      "step": 5
+    },
+    {
+      "epoch": 0.0005266174974149115,
+      "grad_norm": 66.5,
+      "learning_rate": 1.8e-05,
+      "loss": 7.8574,
+      "step": 6
+    },
+    {
+      "epoch": 0.0006143870803173967,
+      "grad_norm": 158.0,
+      "learning_rate": 2.1000000000000002e-05,
+      "loss": 7.708,
+      "step": 7
+    },
+    {
+      "epoch": 0.000702156663219882,
+      "grad_norm": 76.5,
+      "learning_rate": 2.4e-05,
+      "loss": 7.4805,
+      "step": 8
+    },
+    {
+      "epoch": 0.0007899262461223672,
+      "grad_norm": 76.5,
+      "learning_rate": 2.7e-05,
+      "loss": 7.667,
+      "step": 9
+    },
+    {
+      "epoch": 0.0008776958290248525,
+      "grad_norm": 59.25,
+      "learning_rate": 3e-05,
+      "loss": 7.4551,
+      "step": 10
+    },
+    {
+      "epoch": 0.0009654654119273377,
+      "grad_norm": 60.0,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 7.4072,
+      "step": 11
+    },
+    {
+      "epoch": 0.001053234994829823,
+      "grad_norm": 169.0,
+      "learning_rate": 3.6e-05,
+      "loss": 7.2051,
+      "step": 12
+    },
+    {
+      "epoch": 0.0011410045777323083,
+      "grad_norm": 198.0,
+      "learning_rate": 3.9e-05,
+      "loss": 6.9873,
+      "step": 13
+    },
+    {
+      "epoch": 0.0012287741606347934,
+      "grad_norm": 52.75,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 6.8613,
+      "step": 14
+    },
+    {
+      "epoch": 0.0013165437435372787,
+      "grad_norm": 84.5,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 6.665,
+      "step": 15
+    },
+    {
+      "epoch": 0.001404313326439764,
+      "grad_norm": 42.75,
+      "learning_rate": 4.8e-05,
+      "loss": 6.6465,
+      "step": 16
+    },
+    {
+      "epoch": 0.0014920829093422493,
+      "grad_norm": 139.0,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 6.5166,
+      "step": 17
+    },
+    {
+      "epoch": 0.0015798524922447344,
+      "grad_norm": 41.75,
+      "learning_rate": 5.4e-05,
+      "loss": 6.4814,
+      "step": 18
+    },
+    {
+      "epoch": 0.0016676220751472197,
+      "grad_norm": 39.5,
+      "learning_rate": 5.7e-05,
+      "loss": 6.2256,
+      "step": 19
+    },
+    {
+      "epoch": 0.001755391658049705,
+      "grad_norm": 17.125,
+      "learning_rate": 6e-05,
+      "loss": 6.0605,
+      "step": 20
+    },
+    {
+      "epoch": 0.0018431612409521903,
+      "grad_norm": 16.75,
+      "learning_rate": 6.3e-05,
+      "loss": 6.0781,
+      "step": 21
+    },
+    {
+      "epoch": 0.0019309308238546754,
+      "grad_norm": 83.5,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 6.0078,
+      "step": 22
+    },
+    {
+      "epoch": 0.002018700406757161,
+      "grad_norm": 44.75,
+      "learning_rate": 6.9e-05,
+      "loss": 5.9941,
+      "step": 23
+    },
+    {
+      "epoch": 0.002106469989659646,
+      "grad_norm": 25.625,
+      "learning_rate": 7.2e-05,
+      "loss": 5.917,
+      "step": 24
+    },
+    {
+      "epoch": 0.002194239572562131,
+      "grad_norm": 65.5,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 5.7607,
+      "step": 25
+    },
+    {
+      "epoch": 0.0022820091554646166,
+      "grad_norm": 16.875,
+      "learning_rate": 7.8e-05,
+      "loss": 5.7617,
+      "step": 26
+    },
+    {
+      "epoch": 0.0023697787383671017,
+      "grad_norm": 22.625,
+      "learning_rate": 8.1e-05,
+      "loss": 5.8018,
+      "step": 27
+    },
+    {
+      "epoch": 0.002457548321269587,
+      "grad_norm": 27.375,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 5.6592,
+      "step": 28
+    },
+    {
+      "epoch": 0.0025453179041720723,
+      "grad_norm": 28.875,
+      "learning_rate": 8.7e-05,
+      "loss": 5.5781,
+      "step": 29
+    },
+    {
+      "epoch": 0.0026330874870745574,
+      "grad_norm": 25.875,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 5.4229,
+      "step": 30
+    },
+    {
+      "epoch": 0.002720857069977043,
+      "grad_norm": 256.0,
+      "learning_rate": 9.3e-05,
+      "loss": 5.373,
+      "step": 31
+    },
+    {
+      "epoch": 0.002808626652879528,
+      "grad_norm": 21.5,
+      "learning_rate": 9.6e-05,
+      "loss": 5.374,
+      "step": 32
+    },
+    {
+      "epoch": 0.002896396235782013,
+      "grad_norm": 15.625,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 5.3916,
+      "step": 33
+    },
+    {
+      "epoch": 0.0029841658186844987,
+      "grad_norm": 17.625,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 5.2695,
+      "step": 34
+    },
+    {
+      "epoch": 0.0030719354015869837,
+      "grad_norm": 26.75,
+      "learning_rate": 0.00010500000000000002,
+      "loss": 5.1943,
+      "step": 35
+    },
+    {
+      "epoch": 0.003159704984489469,
+      "grad_norm": 18.375,
+      "learning_rate": 0.000108,
+      "loss": 5.2988,
+      "step": 36
+    },
+    {
+      "epoch": 0.0032474745673919544,
+      "grad_norm": 15.375,
+      "learning_rate": 0.000111,
+      "loss": 5.1738,
+      "step": 37
+    },
+    {
+      "epoch": 0.0033352441502944394,
+      "grad_norm": 3.1875,
+      "learning_rate": 0.000114,
+      "loss": 5.0938,
+      "step": 38
+    },
+    {
+      "epoch": 0.003423013733196925,
+      "grad_norm": 13.875,
+      "learning_rate": 0.000117,
+      "loss": 5.0488,
+      "step": 39
+    },
+    {
+      "epoch": 0.00351078331609941,
+      "grad_norm": 18.375,
+      "learning_rate": 0.00012,
+      "loss": 5.0342,
+      "step": 40
+    },
+    {
+      "epoch": 0.003598552899001895,
+      "grad_norm": 13.25,
+      "learning_rate": 0.000123,
+      "loss": 4.9072,
+      "step": 41
+    },
+    {
+      "epoch": 0.0036863224819043807,
+      "grad_norm": 14.375,
+      "learning_rate": 0.000126,
+      "loss": 5.0068,
+      "step": 42
+    },
+    {
+      "epoch": 0.0037740920648068658,
+      "grad_norm": 8.5625,
+      "learning_rate": 0.000129,
+      "loss": 4.9346,
+      "step": 43
+    },
+    {
+      "epoch": 0.003861861647709351,
+      "grad_norm": 4.53125,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 4.9004,
+      "step": 44
+    },
+    {
+      "epoch": 0.003949631230611836,
+      "grad_norm": 44.5,
+      "learning_rate": 0.000135,
+      "loss": 4.96,
+      "step": 45
+    },
+    {
+      "epoch": 0.004037400813514322,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.000138,
+      "loss": 4.8945,
+      "step": 46
+    },
+    {
+      "epoch": 0.0041251703964168066,
+      "grad_norm": 5.3125,
+      "learning_rate": 0.000141,
+      "loss": 4.9229,
+      "step": 47
+    },
+    {
+      "epoch": 0.004212939979319292,
+      "grad_norm": 4.8125,
+      "learning_rate": 0.000144,
+      "loss": 4.8545,
+      "step": 48
+    },
+    {
+      "epoch": 0.004300709562221778,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.000147,
+      "loss": 4.8369,
+      "step": 49
+    },
+    {
+      "epoch": 0.004388479145124262,
+      "grad_norm": 6.8125,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 4.9775,
+      "step": 50
+    },
+    {
+      "epoch": 0.004476248728026748,
+      "grad_norm": 5.0,
+      "learning_rate": 0.000153,
+      "loss": 4.9795,
+      "step": 51
+    },
+    {
+      "epoch": 0.004564018310929233,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.000156,
+      "loss": 4.7822,
+      "step": 52
+    },
+    {
+      "epoch": 0.004651787893831718,
+      "grad_norm": 7.40625,
+      "learning_rate": 0.000159,
+      "loss": 4.8193,
+      "step": 53
+    },
+    {
+      "epoch": 0.0047395574767342035,
+      "grad_norm": 5.65625,
+      "learning_rate": 0.000162,
+      "loss": 4.8438,
+      "step": 54
+    },
+    {
+      "epoch": 0.004827327059636689,
+      "grad_norm": 6.0,
+      "learning_rate": 0.000165,
+      "loss": 4.8115,
+      "step": 55
+    },
+    {
+      "epoch": 0.004915096642539174,
+      "grad_norm": 6.75,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 4.7568,
+      "step": 56
+    },
+    {
+      "epoch": 0.005002866225441659,
+      "grad_norm": 8.125,
+      "learning_rate": 0.000171,
+      "loss": 4.8145,
+      "step": 57
+    },
+    {
+      "epoch": 0.005090635808344145,
+      "grad_norm": 12.125,
+      "learning_rate": 0.000174,
+      "loss": 4.8076,
+      "step": 58
+    },
+    {
+      "epoch": 0.005178405391246629,
+      "grad_norm": 11.875,
+      "learning_rate": 0.000177,
+      "loss": 4.7861,
+      "step": 59
+    },
+    {
+      "epoch": 0.005266174974149115,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.7412,
+      "step": 60
+    },
+    {
+      "epoch": 0.0053539445570516,
+      "grad_norm": 5.875,
+      "learning_rate": 0.000183,
+      "loss": 4.7051,
+      "step": 61
+    },
+    {
+      "epoch": 0.005441714139954086,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.000186,
+      "loss": 4.7061,
+      "step": 62
+    },
+    {
+      "epoch": 0.005529483722856571,
+      "grad_norm": 5.90625,
+      "learning_rate": 0.000189,
+      "loss": 4.7012,
+      "step": 63
+    },
+    {
+      "epoch": 0.005617253305759056,
+      "grad_norm": 4.375,
+      "learning_rate": 0.000192,
+      "loss": 4.7412,
+      "step": 64
+    },
+    {
+      "epoch": 0.005705022888661542,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00019500000000000002,
+      "loss": 4.708,
+      "step": 65
+    },
+    {
+      "epoch": 0.005792792471564026,
+      "grad_norm": 5.5,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 4.7158,
+      "step": 66
+    },
+    {
+      "epoch": 0.005880562054466512,
+      "grad_norm": 3.375,
+      "learning_rate": 0.000201,
+      "loss": 4.6943,
+      "step": 67
+    },
+    {
+      "epoch": 0.005968331637368997,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 4.6689,
+      "step": 68
+    },
+    {
+      "epoch": 0.006056101220271482,
+      "grad_norm": 5.625,
+      "learning_rate": 0.00020700000000000002,
+      "loss": 4.7139,
+      "step": 69
+    },
+    {
+      "epoch": 0.0061438708031739675,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00021000000000000004,
+      "loss": 4.748,
+      "step": 70
+    },
+    {
+      "epoch": 0.006231640386076453,
+      "grad_norm": 6.40625,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 4.7041,
+      "step": 71
+    },
+    {
+      "epoch": 0.006319409968978938,
+      "grad_norm": 2.25,
+      "learning_rate": 0.000216,
+      "loss": 4.709,
+      "step": 72
+    },
+    {
+      "epoch": 0.006407179551881423,
+      "grad_norm": 13.25,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 4.8428,
+      "step": 73
+    },
+    {
+      "epoch": 0.006494949134783909,
+      "grad_norm": 14.5,
+      "learning_rate": 0.000222,
+      "loss": 4.7979,
+      "step": 74
+    },
+    {
+      "epoch": 0.006582718717686393,
+      "grad_norm": 7.4375,
+      "learning_rate": 0.000225,
+      "loss": 4.6299,
+      "step": 75
+    },
+    {
+      "epoch": 0.006670488300588879,
+      "grad_norm": 13.75,
+      "learning_rate": 0.000228,
+      "loss": 4.7627,
+      "step": 76
+    },
+    {
+      "epoch": 0.006758257883491364,
+      "grad_norm": 15.1875,
+      "learning_rate": 0.000231,
+      "loss": 4.8945,
+      "step": 77
+    },
+    {
+      "epoch": 0.00684602746639385,
+      "grad_norm": 12.5,
+      "learning_rate": 0.000234,
+      "loss": 4.7734,
+      "step": 78
+    },
+    {
+      "epoch": 0.006933797049296335,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00023700000000000001,
+      "loss": 4.6152,
+      "step": 79
+    },
+    {
+      "epoch": 0.00702156663219882,
+      "grad_norm": 11.125,
+      "learning_rate": 0.00024,
+      "loss": 4.7637,
+      "step": 80
+    },
+    {
+      "epoch": 0.007109336215101306,
+      "grad_norm": 11.5,
+      "learning_rate": 0.00024300000000000002,
+      "loss": 4.7783,
+      "step": 81
+    },
+    {
+      "epoch": 0.00719710579800379,
+      "grad_norm": 4.4375,
+      "learning_rate": 0.000246,
+      "loss": 4.6436,
+      "step": 82
+    },
+    {
+      "epoch": 0.007284875380906276,
+      "grad_norm": 12.6875,
+      "learning_rate": 0.00024900000000000004,
+      "loss": 4.8057,
+      "step": 83
+    },
+    {
+      "epoch": 0.007372644963808761,
+      "grad_norm": 14.0625,
+      "learning_rate": 0.000252,
+      "loss": 4.8232,
+      "step": 84
+    },
+    {
+      "epoch": 0.007460414546711246,
+      "grad_norm": 10.875,
+      "learning_rate": 0.000255,
+      "loss": 4.6621,
+      "step": 85
+    },
+    {
+      "epoch": 0.0075481841296137315,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.000258,
+      "loss": 4.6309,
+      "step": 86
+    },
+    {
+      "epoch": 0.007635953712516217,
+      "grad_norm": 6.5,
+      "learning_rate": 0.000261,
+      "loss": 4.6016,
+      "step": 87
+    },
+    {
+      "epoch": 0.007723723295418702,
+      "grad_norm": 5.5625,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 4.6582,
+      "step": 88
+    },
+    {
+      "epoch": 0.007811492878321187,
+      "grad_norm": 3.453125,
+      "learning_rate": 0.000267,
+      "loss": 4.6748,
+      "step": 89
+    },
+    {
+      "epoch": 0.007899262461223673,
+      "grad_norm": 4.03125,
+      "learning_rate": 0.00027,
+      "loss": 4.6484,
+      "step": 90
+    },
+    {
+      "epoch": 0.007987032044126157,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.000273,
+      "loss": 4.6318,
+      "step": 91
+    },
+    {
+      "epoch": 0.008074801627028644,
+      "grad_norm": 6.25,
+      "learning_rate": 0.000276,
+      "loss": 4.6748,
+      "step": 92
+    },
+    {
+      "epoch": 0.008162571209931128,
+      "grad_norm": 6.03125,
+      "learning_rate": 0.000279,
+      "loss": 4.6309,
+      "step": 93
+    },
+    {
+      "epoch": 0.008250340792833613,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.000282,
+      "loss": 4.6621,
+      "step": 94
+    },
+    {
+      "epoch": 0.0083381103757361,
+      "grad_norm": 5.03125,
+      "learning_rate": 0.000285,
+      "loss": 4.6953,
+      "step": 95
+    },
+    {
+      "epoch": 0.008425879958638584,
+      "grad_norm": 3.984375,
+      "learning_rate": 0.000288,
+      "loss": 4.6104,
+      "step": 96
+    },
+    {
+      "epoch": 0.008513649541541069,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.000291,
+      "loss": 4.6055,
+      "step": 97
+    },
+    {
+      "epoch": 0.008601419124443555,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.000294,
+      "loss": 4.6123,
+      "step": 98
+    },
+    {
+      "epoch": 0.00868918870734604,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.000297,
+      "loss": 4.5986,
+      "step": 99
+    },
+    {
+      "epoch": 0.008776958290248525,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 4.5312,
+      "step": 100
+    },
+    {
+      "epoch": 0.008864727873151011,
+      "grad_norm": 3.296875,
+      "learning_rate": 0.00030300000000000005,
+      "loss": 4.6201,
+      "step": 101
+    },
+    {
+      "epoch": 0.008952497456053496,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.000306,
+      "loss": 4.626,
+      "step": 102
+    },
+    {
+      "epoch": 0.00904026703895598,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.000309,
+      "loss": 4.624,
+      "step": 103
+    },
+    {
+      "epoch": 0.009128036621858467,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.000312,
+      "loss": 4.5771,
+      "step": 104
+    },
+    {
+      "epoch": 0.009215806204760951,
+      "grad_norm": 2.9375,
+      "learning_rate": 0.000315,
+      "loss": 4.6221,
+      "step": 105
+    },
+    {
+      "epoch": 0.009303575787663436,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.000318,
+      "loss": 4.5703,
+      "step": 106
+    },
+    {
+      "epoch": 0.009391345370565922,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.000321,
+      "loss": 4.5283,
+      "step": 107
+    },
+    {
+      "epoch": 0.009479114953468407,
+      "grad_norm": 1.9765625,
+      "learning_rate": 0.000324,
+      "loss": 4.6152,
+      "step": 108
+    },
+    {
+      "epoch": 0.009566884536370892,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.000327,
+      "loss": 4.5557,
+      "step": 109
+    },
+    {
+      "epoch": 0.009654654119273378,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00033,
+      "loss": 4.6279,
+      "step": 110
+    },
+    {
+      "epoch": 0.009742423702175863,
+      "grad_norm": 3.046875,
+      "learning_rate": 0.000333,
+      "loss": 4.542,
+      "step": 111
+    },
+    {
+      "epoch": 0.009830193285078347,
+      "grad_norm": 2.71875,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 4.5586,
+      "step": 112
+    },
+    {
+      "epoch": 0.009917962867980834,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.000339,
+      "loss": 4.5127,
+      "step": 113
+    },
+    {
+      "epoch": 0.010005732450883318,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.000342,
+      "loss": 4.5742,
+      "step": 114
+    },
+    {
+      "epoch": 0.010093502033785803,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00034500000000000004,
+      "loss": 4.5088,
+      "step": 115
+    },
+    {
+      "epoch": 0.01018127161668829,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.000348,
+      "loss": 4.543,
+      "step": 116
+    },
+    {
+      "epoch": 0.010269041199590774,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.000351,
+      "loss": 4.4609,
+      "step": 117
+    },
+    {
+      "epoch": 0.010356810782493259,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.000354,
+      "loss": 4.5303,
+      "step": 118
+    },
+    {
+      "epoch": 0.010444580365395745,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.000357,
+      "loss": 4.4971,
+      "step": 119
+    },
+    {
+      "epoch": 0.01053234994829823,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 4.5342,
+      "step": 120
+    },
+    {
+      "epoch": 0.010620119531200716,
+      "grad_norm": 2.640625,
+      "learning_rate": 0.000363,
+      "loss": 4.4941,
+      "step": 121
+    },
+    {
+      "epoch": 0.0107078891141032,
+      "grad_norm": 1.9453125,
+      "learning_rate": 0.000366,
+      "loss": 4.4658,
+      "step": 122
+    },
+    {
+      "epoch": 0.010795658697005685,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.000369,
+      "loss": 4.4941,
+      "step": 123
+    },
+    {
+      "epoch": 0.010883428279908172,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.000372,
+      "loss": 4.4248,
+      "step": 124
+    },
+    {
+      "epoch": 0.010971197862810656,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.000375,
+      "loss": 4.4961,
+      "step": 125
+    },
+    {
+      "epoch": 0.011058967445713141,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.000378,
+      "loss": 4.4292,
+      "step": 126
+    },
+    {
+      "epoch": 0.011146737028615628,
+      "grad_norm": 2.96875,
+      "learning_rate": 0.000381,
+      "loss": 4.5234,
+      "step": 127
+    },
+    {
+      "epoch": 0.011234506611518112,
+      "grad_norm": 2.640625,
+      "learning_rate": 0.000384,
+      "loss": 4.5049,
+      "step": 128
+    },
+    {
+      "epoch": 0.011322276194420597,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00038700000000000003,
+      "loss": 4.4512,
+      "step": 129
+    },
+    {
+      "epoch": 0.011410045777323083,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 4.4355,
+      "step": 130
+    },
+    {
+      "epoch": 0.011497815360225568,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.000393,
+      "loss": 4.4238,
+      "step": 131
+    },
+    {
+      "epoch": 0.011585584943128053,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 4.5137,
+      "step": 132
+    },
+    {
+      "epoch": 0.011673354526030539,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00039900000000000005,
+      "loss": 4.4004,
+      "step": 133
+    },
+    {
+      "epoch": 0.011761124108933024,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.000402,
+      "loss": 4.4141,
+      "step": 134
+    },
+    {
+      "epoch": 0.011848893691835508,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00040500000000000003,
+      "loss": 4.4688,
+      "step": 135
+    },
+    {
+      "epoch": 0.011936663274737995,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00040800000000000005,
+      "loss": 4.4326,
+      "step": 136
+    },
+    {
+      "epoch": 0.01202443285764048,
+      "grad_norm": 2.90625,
+      "learning_rate": 0.000411,
+      "loss": 4.5049,
+      "step": 137
+    },
+    {
+      "epoch": 0.012112202440542964,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00041400000000000003,
+      "loss": 4.5088,
+      "step": 138
+    },
+    {
+      "epoch": 0.01219997202344545,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00041700000000000005,
+      "loss": 4.4717,
+      "step": 139
+    },
+    {
+      "epoch": 0.012287741606347935,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00042000000000000007,
+      "loss": 4.4717,
+      "step": 140
+    },
+    {
+      "epoch": 0.01237551118925042,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.000423,
+      "loss": 4.3867,
+      "step": 141
+    },
+    {
+      "epoch": 0.012463280772152906,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00042599999999999995,
+      "loss": 4.3896,
+      "step": 142
+    },
+    {
+      "epoch": 0.01255105035505539,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00042899999999999997,
+      "loss": 4.3428,
+      "step": 143
+    },
+    {
+      "epoch": 0.012638819937957875,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.000432,
+      "loss": 4.4297,
+      "step": 144
+    },
+    {
+      "epoch": 0.012726589520860362,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.000435,
+      "loss": 4.418,
+      "step": 145
+    },
+    {
+      "epoch": 0.012814359103762846,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00043799999999999997,
+      "loss": 4.3877,
+      "step": 146
+    },
+    {
+      "epoch": 0.012902128686665331,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.000441,
+      "loss": 4.415,
+      "step": 147
+    },
+    {
+      "epoch": 0.012989898269567817,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.000444,
+      "loss": 4.4492,
+      "step": 148
+    },
+    {
+      "epoch": 0.013077667852470302,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00044699999999999997,
+      "loss": 4.4111,
+      "step": 149
+    },
+    {
+      "epoch": 0.013165437435372787,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00045,
+      "loss": 4.3945,
+      "step": 150
+    },
+    {
+      "epoch": 0.013253207018275273,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.000453,
+      "loss": 4.3818,
+      "step": 151
+    },
+    {
+      "epoch": 0.013340976601177758,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.000456,
+      "loss": 4.3496,
+      "step": 152
+    },
+    {
+      "epoch": 0.013428746184080244,
+      "grad_norm": 3.734375,
+      "learning_rate": 0.000459,
+      "loss": 4.418,
+      "step": 153
+    },
+    {
+      "epoch": 0.013516515766982729,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.000462,
+      "loss": 4.4609,
+      "step": 154
+    },
+    {
+      "epoch": 0.013604285349885213,
+      "grad_norm": 6.5,
+      "learning_rate": 0.000465,
+      "loss": 4.4658,
+      "step": 155
+    },
+    {
+      "epoch": 0.0136920549327877,
+      "grad_norm": 5.8125,
+      "learning_rate": 0.000468,
+      "loss": 4.4727,
+      "step": 156
+    },
+    {
+      "epoch": 0.013779824515690185,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.000471,
+      "loss": 4.3848,
+      "step": 157
+    },
+    {
+      "epoch": 0.01386759409859267,
+      "grad_norm": 3.390625,
+      "learning_rate": 0.00047400000000000003,
+      "loss": 4.3906,
+      "step": 158
+    },
+    {
+      "epoch": 0.013955363681495156,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.000477,
+      "loss": 4.373,
+      "step": 159
+    },
+    {
+      "epoch": 0.01404313326439764,
+      "grad_norm": 3.671875,
+      "learning_rate": 0.00048,
+      "loss": 4.3984,
+      "step": 160
+    },
+    {
+      "epoch": 0.014130902847300125,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00048300000000000003,
+      "loss": 4.4072,
+      "step": 161
+    },
+    {
+      "epoch": 0.014218672430202611,
+      "grad_norm": 3.703125,
+      "learning_rate": 0.00048600000000000005,
+      "loss": 4.418,
+      "step": 162
+    },
+    {
+      "epoch": 0.014306442013105096,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0004890000000000001,
+      "loss": 4.4209,
+      "step": 163
+    },
+    {
+      "epoch": 0.01439421159600758,
+      "grad_norm": 4.53125,
+      "learning_rate": 0.000492,
+      "loss": 4.374,
+      "step": 164
+    },
+    {
+      "epoch": 0.014481981178910067,
+      "grad_norm": 3.890625,
+      "learning_rate": 0.000495,
+      "loss": 4.4346,
+      "step": 165
+    },
+    {
+      "epoch": 0.014569750761812552,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.0004980000000000001,
+      "loss": 4.3232,
+      "step": 166
+    },
+    {
+      "epoch": 0.014657520344715036,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.000501,
+      "loss": 4.332,
+      "step": 167
+    },
+    {
+      "epoch": 0.014745289927617523,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.000504,
+      "loss": 4.4102,
+      "step": 168
+    },
+    {
+      "epoch": 0.014833059510520007,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0005070000000000001,
+      "loss": 4.3809,
+      "step": 169
+    },
+    {
+      "epoch": 0.014920829093422492,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00051,
+      "loss": 4.3984,
+      "step": 170
+    },
+    {
+      "epoch": 0.015008598676324978,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.000513,
+      "loss": 4.3887,
+      "step": 171
+    },
+    {
+      "epoch": 0.015096368259227463,
+      "grad_norm": 3.09375,
+      "learning_rate": 0.000516,
+      "loss": 4.3398,
+      "step": 172
+    },
+    {
+      "epoch": 0.015184137842129948,
+      "grad_norm": 2.625,
+      "learning_rate": 0.0005189999999999999,
+      "loss": 4.418,
+      "step": 173
+    },
+    {
+      "epoch": 0.015271907425032434,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.000522,
+      "loss": 4.3408,
+      "step": 174
+    },
+    {
+      "epoch": 0.015359677007934919,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.000525,
+      "loss": 4.4092,
+      "step": 175
+    },
+    {
+      "epoch": 0.015447446590837403,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.0005279999999999999,
+      "loss": 4.3594,
+      "step": 176
+    },
+    {
+      "epoch": 0.01553521617373989,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.000531,
+      "loss": 4.3291,
+      "step": 177
+    },
+    {
+      "epoch": 0.015622985756642374,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.000534,
+      "loss": 4.3228,
+      "step": 178
+    },
+    {
+      "epoch": 0.01571075533954486,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.000537,
+      "loss": 4.3418,
+      "step": 179
+    },
+    {
+      "epoch": 0.015798524922447345,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00054,
+      "loss": 4.3486,
+      "step": 180
+    },
+    {
+      "epoch": 0.01588629450534983,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.000543,
+      "loss": 4.3457,
+      "step": 181
+    },
+    {
+      "epoch": 0.015974064088252315,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.000546,
+      "loss": 4.3867,
+      "step": 182
+    },
+    {
+      "epoch": 0.0160618336711548,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.000549,
+      "loss": 4.3555,
+      "step": 183
+    },
+    {
+      "epoch": 0.016149603254057288,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.000552,
+      "loss": 4.334,
+      "step": 184
+    },
+    {
+      "epoch": 0.016237372836959772,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.000555,
+      "loss": 4.292,
+      "step": 185
+    },
+    {
+      "epoch": 0.016325142419862257,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.000558,
+      "loss": 4.3467,
+      "step": 186
+    },
+    {
+      "epoch": 0.01641291200276474,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.000561,
+      "loss": 4.3145,
+      "step": 187
+    },
+    {
+      "epoch": 0.016500681585667226,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.000564,
+      "loss": 4.2754,
+      "step": 188
+    },
+    {
+      "epoch": 0.01658845116856971,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.000567,
+      "loss": 4.2783,
+      "step": 189
+    },
+    {
+      "epoch": 0.0166762207514722,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00057,
+      "loss": 4.269,
+      "step": 190
+    },
+    {
+      "epoch": 0.016763990334374684,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.000573,
+      "loss": 4.332,
+      "step": 191
+    },
+    {
+      "epoch": 0.01685175991727717,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.000576,
+      "loss": 4.3018,
+      "step": 192
+    },
+    {
+      "epoch": 0.016939529500179653,
+      "grad_norm": 0.625,
+      "learning_rate": 0.000579,
+      "loss": 4.2607,
+      "step": 193
+    },
+    {
+      "epoch": 0.017027299083082138,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.000582,
+      "loss": 4.3398,
+      "step": 194
+    },
+    {
+      "epoch": 0.017115068665984622,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.000585,
+      "loss": 4.2788,
+      "step": 195
+    },
+    {
+      "epoch": 0.01720283824888711,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.000588,
+      "loss": 4.2266,
+      "step": 196
+    },
+    {
+      "epoch": 0.017290607831789595,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.000591,
+      "loss": 4.3252,
+      "step": 197
+    },
+    {
+      "epoch": 0.01737837741469208,
+      "grad_norm": 2.375,
+      "learning_rate": 0.000594,
+      "loss": 4.2686,
+      "step": 198
+    },
+    {
+      "epoch": 0.017466146997594564,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0005970000000000001,
+      "loss": 4.2939,
+      "step": 199
+    },
+    {
+      "epoch": 0.01755391658049705,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.0006000000000000001,
+      "loss": 4.2803,
+      "step": 200
+    },
+    {
+      "epoch": 0.017641686163399534,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.000603,
+      "loss": 4.2744,
+      "step": 201
+    },
+    {
+      "epoch": 0.017729455746302022,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0006060000000000001,
+      "loss": 4.292,
+      "step": 202
+    },
+    {
+      "epoch": 0.017817225329204506,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0006090000000000001,
+      "loss": 4.2046,
+      "step": 203
+    },
+    {
+      "epoch": 0.01790499491210699,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.000612,
+      "loss": 4.2178,
+      "step": 204
+    },
+    {
+      "epoch": 0.017992764495009476,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.000615,
+      "loss": 4.2183,
+      "step": 205
+    },
+    {
+      "epoch": 0.01808053407791196,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.000618,
+      "loss": 4.2139,
+      "step": 206
+    },
+    {
+      "epoch": 0.01816830366081445,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.000621,
+      "loss": 4.2363,
+      "step": 207
+    },
+    {
+      "epoch": 0.018256073243716933,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.000624,
+      "loss": 4.2891,
+      "step": 208
+    },
+    {
+      "epoch": 0.018343842826619418,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.000627,
+      "loss": 4.2212,
+      "step": 209
+    },
+    {
+      "epoch": 0.018431612409521902,
+      "grad_norm": 3.390625,
+      "learning_rate": 0.00063,
+      "loss": 4.3203,
+      "step": 210
+    },
+    {
+      "epoch": 0.018519381992424387,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.000633,
+      "loss": 4.3159,
+      "step": 211
+    },
+    {
+      "epoch": 0.018607151575326872,
+      "grad_norm": 3.734375,
+      "learning_rate": 0.000636,
+      "loss": 4.272,
+      "step": 212
+    },
+    {
+      "epoch": 0.01869492115822936,
+      "grad_norm": 3.0625,
+      "learning_rate": 0.000639,
+      "loss": 4.2368,
+      "step": 213
+    },
+    {
+      "epoch": 0.018782690741131845,
+      "grad_norm": 2.875,
+      "learning_rate": 0.000642,
+      "loss": 4.2749,
+      "step": 214
+    },
+    {
+      "epoch": 0.01887046032403433,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.000645,
+      "loss": 4.1772,
+      "step": 215
+    },
+    {
+      "epoch": 0.018958229906936814,
+      "grad_norm": 2.125,
+      "learning_rate": 0.000648,
+      "loss": 4.3271,
+      "step": 216
+    },
+    {
+      "epoch": 0.0190459994898393,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.000651,
+      "loss": 4.2402,
+      "step": 217
+    },
+    {
+      "epoch": 0.019133769072741783,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.000654,
+      "loss": 4.2407,
+      "step": 218
+    },
+    {
+      "epoch": 0.01922153865564427,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.000657,
+      "loss": 4.2031,
+      "step": 219
+    },
+    {
+      "epoch": 0.019309308238546756,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00066,
+      "loss": 4.2534,
+      "step": 220
+    },
+    {
+      "epoch": 0.01939707782144924,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0006630000000000001,
+      "loss": 4.2451,
+      "step": 221
+    },
+    {
+      "epoch": 0.019484847404351725,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.000666,
+      "loss": 4.3438,
+      "step": 222
+    },
+    {
+      "epoch": 0.01957261698725421,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.000669,
+      "loss": 4.2666,
+      "step": 223
+    },
+    {
+      "epoch": 0.019660386570156695,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 4.2119,
+      "step": 224
+    },
+    {
+      "epoch": 0.019748156153059183,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.000675,
+      "loss": 4.1357,
+      "step": 225
+    },
+    {
+      "epoch": 0.019835925735961667,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.000678,
+      "loss": 4.2373,
+      "step": 226
+    },
+    {
+      "epoch": 0.019923695318864152,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0006810000000000001,
+      "loss": 4.2134,
+      "step": 227
+    },
+    {
+      "epoch": 0.020011464901766637,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.000684,
+      "loss": 4.1768,
+      "step": 228
+    },
+    {
+      "epoch": 0.02009923448466912,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.000687,
+      "loss": 4.165,
+      "step": 229
+    },
+    {
+      "epoch": 0.020187004067571606,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.0006900000000000001,
+      "loss": 4.1826,
+      "step": 230
+    },
+    {
+      "epoch": 0.020274773650474094,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.000693,
+      "loss": 4.1729,
+      "step": 231
+    },
+    {
+      "epoch": 0.02036254323337658,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.000696,
+      "loss": 4.1919,
+      "step": 232
+    },
+    {
+      "epoch": 0.020450312816279063,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0006990000000000001,
+      "loss": 4.2305,
+      "step": 233
+    },
+    {
+      "epoch": 0.020538082399181548,
+      "grad_norm": 2.71875,
+      "learning_rate": 0.000702,
+      "loss": 4.2744,
+      "step": 234
+    },
+    {
+      "epoch": 0.020625851982084033,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.000705,
+      "loss": 4.1924,
+      "step": 235
+    },
+    {
+      "epoch": 0.020713621564986517,
+      "grad_norm": 5.4375,
+      "learning_rate": 0.000708,
+      "loss": 4.2466,
+      "step": 236
+    },
+    {
+      "epoch": 0.020801391147889006,
+      "grad_norm": 4.6875,
+      "learning_rate": 0.0007109999999999999,
+      "loss": 4.312,
+      "step": 237
+    },
+    {
+      "epoch": 0.02088916073079149,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.000714,
+      "loss": 4.1802,
+      "step": 238
+    },
+    {
+      "epoch": 0.020976930313693975,
+      "grad_norm": 2.75,
+      "learning_rate": 0.000717,
+      "loss": 4.3037,
+      "step": 239
+    },
+    {
+      "epoch": 0.02106469989659646,
+      "grad_norm": 1.9140625,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 4.2271,
+      "step": 240
+    },
+    {
+      "epoch": 0.021152469479498944,
+      "grad_norm": 2.890625,
+      "learning_rate": 0.000723,
+      "loss": 4.2114,
+      "step": 241
+    },
+    {
+      "epoch": 0.021240239062401432,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.000726,
+      "loss": 4.2104,
+      "step": 242
+    },
+    {
+      "epoch": 0.021328008645303917,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.000729,
+      "loss": 4.1665,
+      "step": 243
+    },
+    {
+      "epoch": 0.0214157782282064,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.000732,
+      "loss": 4.1475,
+      "step": 244
+    },
+    {
+      "epoch": 0.021503547811108886,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.000735,
+      "loss": 4.2339,
+      "step": 245
+    },
+    {
+      "epoch": 0.02159131739401137,
+      "grad_norm": 1.125,
+      "learning_rate": 0.000738,
+      "loss": 4.2329,
+      "step": 246
+    },
+    {
+      "epoch": 0.021679086976913856,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.000741,
+      "loss": 4.21,
+      "step": 247
+    },
+    {
+      "epoch": 0.021766856559816344,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.000744,
+      "loss": 4.2168,
+      "step": 248
+    },
+    {
+      "epoch": 0.02185462614271883,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.000747,
+      "loss": 4.1685,
+      "step": 249
+    },
+    {
+      "epoch": 0.021942395725621313,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00075,
+      "loss": 4.1772,
+      "step": 250
+    },
+    {
+      "epoch": 0.022030165308523798,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.000753,
+      "loss": 4.1802,
+      "step": 251
+    },
+    {
+      "epoch": 0.022117934891426282,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.000756,
+      "loss": 4.1489,
+      "step": 252
+    },
+    {
+      "epoch": 0.022205704474328767,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.000759,
+      "loss": 4.1973,
+      "step": 253
+    },
+    {
+      "epoch": 0.022293474057231255,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.000762,
+      "loss": 4.1558,
+      "step": 254
+    },
+    {
+      "epoch": 0.02238124364013374,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0007650000000000001,
+      "loss": 4.2026,
+      "step": 255
+    },
+    {
+      "epoch": 0.022469013223036224,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.000768,
+      "loss": 4.123,
+      "step": 256
+    },
+    {
+      "epoch": 0.02255678280593871,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.000771,
+      "loss": 4.1147,
+      "step": 257
+    },
+    {
+      "epoch": 0.022644552388841194,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 4.1553,
+      "step": 258
+    },
+    {
+      "epoch": 0.02273232197174368,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.000777,
+      "loss": 4.21,
+      "step": 259
+    },
+    {
+      "epoch": 0.022820091554646166,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 4.1685,
+      "step": 260
+    },
+    {
+      "epoch": 0.02290786113754865,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0007830000000000001,
+      "loss": 4.0781,
+      "step": 261
+    },
+    {
+      "epoch": 0.022995630720451136,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.000786,
+      "loss": 4.1138,
+      "step": 262
+    },
+    {
+      "epoch": 0.02308340030335362,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0007890000000000001,
+      "loss": 4.1313,
+      "step": 263
+    },
+    {
+      "epoch": 0.023171169886256105,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 4.1675,
+      "step": 264
+    },
+    {
+      "epoch": 0.02325893946915859,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.000795,
+      "loss": 4.228,
+      "step": 265
+    },
+    {
+      "epoch": 0.023346709052061078,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 4.1475,
+      "step": 266
+    },
+    {
+      "epoch": 0.023434478634963563,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0008010000000000001,
+      "loss": 4.1196,
+      "step": 267
+    },
+    {
+      "epoch": 0.023522248217866047,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.000804,
+      "loss": 4.1523,
+      "step": 268
+    },
+    {
+      "epoch": 0.023610017800768532,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0008070000000000001,
+      "loss": 4.1338,
+      "step": 269
+    },
+    {
+      "epoch": 0.023697787383671017,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 4.0962,
+      "step": 270
+    },
+    {
+      "epoch": 0.023785556966573505,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.000813,
+      "loss": 4.1567,
+      "step": 271
+    },
+    {
+      "epoch": 0.02387332654947599,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.0008160000000000001,
+      "loss": 4.1865,
+      "step": 272
+    },
+    {
+      "epoch": 0.023961096132378474,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0008190000000000001,
+      "loss": 4.061,
+      "step": 273
+    },
+    {
+      "epoch": 0.02404886571528096,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.000822,
+      "loss": 4.0957,
+      "step": 274
+    },
+    {
+      "epoch": 0.024136635298183443,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0008250000000000001,
+      "loss": 4.1304,
+      "step": 275
+    },
+    {
+      "epoch": 0.024224404881085928,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0008280000000000001,
+      "loss": 4.1309,
+      "step": 276
+    },
+    {
+      "epoch": 0.024312174463988416,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0008310000000000001,
+      "loss": 4.1338,
+      "step": 277
+    },
+    {
+      "epoch": 0.0243999440468909,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0008340000000000001,
+      "loss": 4.0967,
+      "step": 278
+    },
+    {
+      "epoch": 0.024487713629793385,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0008370000000000001,
+      "loss": 4.0981,
+      "step": 279
+    },
+    {
+      "epoch": 0.02457548321269587,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0008400000000000001,
+      "loss": 4.0859,
+      "step": 280
+    },
+    {
+      "epoch": 0.024663252795598355,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0008430000000000001,
+      "loss": 4.126,
+      "step": 281
+    },
+    {
+      "epoch": 0.02475102237850084,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.000846,
+      "loss": 4.2065,
+      "step": 282
+    },
+    {
+      "epoch": 0.024838791961403327,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0008489999999999999,
+      "loss": 4.0684,
+      "step": 283
+    },
+    {
+      "epoch": 0.024926561544305812,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0008519999999999999,
+      "loss": 4.0898,
+      "step": 284
+    },
+    {
+      "epoch": 0.025014331127208297,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.000855,
+      "loss": 4.1406,
+      "step": 285
+    },
+    {
+      "epoch": 0.02510210071011078,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0008579999999999999,
+      "loss": 4.1655,
+      "step": 286
+    },
+    {
+      "epoch": 0.025189870293013266,
+      "grad_norm": 5.3125,
+      "learning_rate": 0.000861,
+      "loss": 4.1816,
+      "step": 287
+    },
+    {
+      "epoch": 0.02527763987591575,
+      "grad_norm": 4.75,
+      "learning_rate": 0.000864,
+      "loss": 4.252,
+      "step": 288
+    },
+    {
+      "epoch": 0.02536540945881824,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0008669999999999999,
+      "loss": 4.146,
+      "step": 289
+    },
+    {
+      "epoch": 0.025453179041720723,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.00087,
+      "loss": 4.1196,
+      "step": 290
+    },
+    {
+      "epoch": 0.025540948624623208,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.000873,
+      "loss": 4.123,
+      "step": 291
+    },
+    {
+      "epoch": 0.025628718207525693,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.0008759999999999999,
+      "loss": 4.1416,
+      "step": 292
+    },
+    {
+      "epoch": 0.025716487790428177,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.000879,
+      "loss": 4.1191,
+      "step": 293
+    },
+    {
+      "epoch": 0.025804257373330662,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.000882,
+      "loss": 4.1421,
+      "step": 294
+    },
+    {
+      "epoch": 0.02589202695623315,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0008849999999999999,
+      "loss": 4.123,
+      "step": 295
+    },
+    {
+      "epoch": 0.025979796539135635,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.000888,
+      "loss": 4.1802,
+      "step": 296
+    },
+    {
+      "epoch": 0.02606756612203812,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.000891,
+      "loss": 4.0566,
+      "step": 297
+    },
+    {
+      "epoch": 0.026155335704940604,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0008939999999999999,
+      "loss": 4.1284,
+      "step": 298
+    },
+    {
+      "epoch": 0.02624310528784309,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.000897,
+      "loss": 4.0513,
+      "step": 299
+    },
+    {
+      "epoch": 0.026330874870745574,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0009,
+      "loss": 4.1318,
+      "step": 300
+    },
+    {
+      "epoch": 0.02641864445364806,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0009029999999999999,
+      "loss": 4.0923,
+      "step": 301
+    },
+    {
+      "epoch": 0.026506414036550546,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.000906,
+      "loss": 4.063,
+      "step": 302
+    },
+    {
+      "epoch": 0.02659418361945303,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.000909,
+      "loss": 4.0669,
+      "step": 303
+    },
+    {
+      "epoch": 0.026681953202355516,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.000912,
+      "loss": 4.1074,
+      "step": 304
+    },
+    {
+      "epoch": 0.026769722785258,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.000915,
+      "loss": 4.1069,
+      "step": 305
+    },
+    {
+      "epoch": 0.02685749236816049,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.000918,
+      "loss": 4.0347,
+      "step": 306
+    },
+    {
+      "epoch": 0.026945261951062973,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.000921,
+      "loss": 4.0503,
+      "step": 307
+    },
+    {
+      "epoch": 0.027033031533965458,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.000924,
+      "loss": 4.0518,
+      "step": 308
+    },
+    {
+      "epoch": 0.027120801116867942,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.000927,
+      "loss": 4.0605,
+      "step": 309
+    },
+    {
+      "epoch": 0.027208570699770427,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.00093,
+      "loss": 4.1006,
+      "step": 310
+    },
+    {
+      "epoch": 0.02729634028267291,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.000933,
+      "loss": 4.0596,
+      "step": 311
+    },
+    {
+      "epoch": 0.0273841098655754,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.000936,
+      "loss": 4.1206,
+      "step": 312
+    },
+    {
+      "epoch": 0.027471879448477884,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.0009390000000000001,
+      "loss": 4.0425,
+      "step": 313
+    },
+    {
+      "epoch": 0.02755964903138037,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.000942,
+      "loss": 4.1162,
+      "step": 314
+    },
+    {
+      "epoch": 0.027647418614282854,
+      "grad_norm": 3.515625,
+      "learning_rate": 0.000945,
+      "loss": 4.0737,
+      "step": 315
+    },
+    {
+      "epoch": 0.02773518819718534,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0009480000000000001,
+      "loss": 4.0366,
+      "step": 316
+    },
+    {
+      "epoch": 0.027822957780087823,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.000951,
+      "loss": 4.021,
+      "step": 317
+    },
+    {
+      "epoch": 0.02791072736299031,
+      "grad_norm": 1.625,
+      "learning_rate": 0.000954,
+      "loss": 4.0576,
+      "step": 318
+    },
+    {
+      "epoch": 0.027998496945892796,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0009570000000000001,
+      "loss": 4.0444,
+      "step": 319
+    },
+    {
+      "epoch": 0.02808626652879528,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00096,
+      "loss": 4.0332,
+      "step": 320
+    },
+    {
+      "epoch": 0.028174036111697765,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.000963,
+      "loss": 4.1543,
+      "step": 321
+    },
+    {
+      "epoch": 0.02826180569460025,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0009660000000000001,
+      "loss": 4.0342,
+      "step": 322
+    },
+    {
+      "epoch": 0.028349575277502734,
+      "grad_norm": 3.53125,
+      "learning_rate": 0.000969,
+      "loss": 4.064,
+      "step": 323
+    },
+    {
+      "epoch": 0.028437344860405223,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.0009720000000000001,
+      "loss": 4.0474,
+      "step": 324
+    },
+    {
+      "epoch": 0.028525114443307707,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0009750000000000001,
+      "loss": 4.1118,
+      "step": 325
+    },
+    {
+      "epoch": 0.028612884026210192,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0009780000000000001,
+      "loss": 4.0039,
+      "step": 326
+    },
+    {
+      "epoch": 0.028700653609112677,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.000981,
+      "loss": 4.061,
+      "step": 327
+    },
+    {
+      "epoch": 0.02878842319201516,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.000984,
+      "loss": 4.0532,
+      "step": 328
+    },
+    {
+      "epoch": 0.028876192774917646,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.000987,
+      "loss": 4.0806,
+      "step": 329
+    },
+    {
+      "epoch": 0.028963962357820134,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00099,
+      "loss": 4.022,
+      "step": 330
+    },
+    {
+      "epoch": 0.02905173194072262,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.0009930000000000002,
+      "loss": 4.0552,
+      "step": 331
+    },
+    {
+      "epoch": 0.029139501523625103,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0009960000000000001,
+      "loss": 3.9961,
+      "step": 332
+    },
+    {
+      "epoch": 0.029227271106527588,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.000999,
+      "loss": 4.0605,
+      "step": 333
+    },
+    {
+      "epoch": 0.029315040689430073,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.001002,
+      "loss": 3.9985,
+      "step": 334
+    },
+    {
+      "epoch": 0.029402810272332557,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.001005,
+      "loss": 4.0396,
+      "step": 335
+    },
+    {
+      "epoch": 0.029490579855235045,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.001008,
+      "loss": 4.0532,
+      "step": 336
+    },
+    {
+      "epoch": 0.02957834943813753,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0010110000000000002,
+      "loss": 4.0146,
+      "step": 337
+    },
+    {
+      "epoch": 0.029666119021040015,
+      "grad_norm": 2.0,
+      "learning_rate": 0.0010140000000000001,
+      "loss": 4.0566,
+      "step": 338
+    },
+    {
+      "epoch": 0.0297538886039425,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0010170000000000001,
+      "loss": 4.0024,
+      "step": 339
+    },
+    {
+      "epoch": 0.029841658186844984,
+      "grad_norm": 3.4375,
+      "learning_rate": 0.00102,
+      "loss": 4.0923,
+      "step": 340
+    },
+    {
+      "epoch": 0.029929427769747472,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.001023,
+      "loss": 4.0166,
+      "step": 341
+    },
+    {
+      "epoch": 0.030017197352649957,
+      "grad_norm": 3.578125,
+      "learning_rate": 0.001026,
+      "loss": 4.1646,
+      "step": 342
+    },
+    {
+      "epoch": 0.03010496693555244,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.0010290000000000002,
+      "loss": 4.0503,
+      "step": 343
+    },
+    {
+      "epoch": 0.030192736518454926,
+      "grad_norm": 4.125,
+      "learning_rate": 0.001032,
+      "loss": 4.1504,
+      "step": 344
+    },
+    {
+      "epoch": 0.03028050610135741,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.001035,
+      "loss": 4.0747,
+      "step": 345
+    },
+    {
+      "epoch": 0.030368275684259895,
+      "grad_norm": 4.71875,
+      "learning_rate": 0.0010379999999999999,
+      "loss": 4.1255,
+      "step": 346
+    },
+    {
+      "epoch": 0.030456045267162384,
+      "grad_norm": 7.4375,
+      "learning_rate": 0.001041,
+      "loss": 4.1182,
+      "step": 347
+    },
+    {
+      "epoch": 0.030543814850064868,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.001044,
+      "loss": 4.0532,
+      "step": 348
+    },
+    {
+      "epoch": 0.030631584432967353,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.001047,
+      "loss": 4.0298,
+      "step": 349
+    },
+    {
+      "epoch": 0.030719354015869837,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00105,
+      "loss": 4.0298,
+      "step": 350
+    },
+    {
+      "epoch": 0.030807123598772322,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.001053,
+      "loss": 4.0601,
+      "step": 351
+    },
+    {
+      "epoch": 0.030894893181674807,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0010559999999999999,
+      "loss": 3.9873,
+      "step": 352
+    },
+    {
+      "epoch": 0.030982662764577295,
+      "grad_norm": 2.375,
+      "learning_rate": 0.001059,
+      "loss": 4.0327,
+      "step": 353
+    },
+    {
+      "epoch": 0.03107043234747978,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.001062,
+      "loss": 4.0474,
+      "step": 354
+    },
+    {
+      "epoch": 0.031158201930382264,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.001065,
+      "loss": 4.085,
+      "step": 355
+    },
+    {
+      "epoch": 0.03124597151328475,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.001068,
+      "loss": 4.0176,
+      "step": 356
+    },
+    {
+      "epoch": 0.03133374109618724,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.001071,
+      "loss": 3.9917,
+      "step": 357
+    },
+    {
+      "epoch": 0.03142151067908972,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.001074,
+      "loss": 4.0342,
+      "step": 358
+    },
+    {
+      "epoch": 0.031509280261992206,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.001077,
+      "loss": 4.0479,
+      "step": 359
+    },
+    {
+      "epoch": 0.03159704984489469,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00108,
+      "loss": 4.1104,
+      "step": 360
+    },
+    {
+      "epoch": 0.031684819427797176,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.001083,
+      "loss": 4.0,
+      "step": 361
+    },
+    {
+      "epoch": 0.03177258901069966,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.001086,
+      "loss": 4.1157,
+      "step": 362
+    },
+    {
+      "epoch": 0.031860358593602145,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.001089,
+      "loss": 4.0947,
+      "step": 363
+    },
+    {
+      "epoch": 0.03194812817650463,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.001092,
+      "loss": 3.9888,
+      "step": 364
+    },
+    {
+      "epoch": 0.032035897759407114,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.001095,
+      "loss": 4.0068,
+      "step": 365
+    },
+    {
+      "epoch": 0.0321236673423096,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.001098,
+      "loss": 4.0063,
+      "step": 366
+    },
+    {
+      "epoch": 0.032211436925212084,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001101,
+      "loss": 4.0361,
+      "step": 367
+    },
+    {
+      "epoch": 0.032299206508114575,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001104,
+      "loss": 4.042,
+      "step": 368
+    },
+    {
+      "epoch": 0.03238697609101706,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.001107,
+      "loss": 3.9937,
+      "step": 369
+    },
+    {
+      "epoch": 0.032474745673919544,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00111,
+      "loss": 4.0044,
+      "step": 370
+    },
+    {
+      "epoch": 0.03256251525682203,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.001113,
+      "loss": 3.9927,
+      "step": 371
+    },
+    {
+      "epoch": 0.032650284839724514,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.001116,
+      "loss": 3.9639,
+      "step": 372
+    },
+    {
+      "epoch": 0.032738054422627,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.001119,
+      "loss": 3.9951,
+      "step": 373
+    },
+    {
+      "epoch": 0.03282582400552948,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001122,
+      "loss": 4.0352,
+      "step": 374
+    },
+    {
+      "epoch": 0.03291359358843197,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0011250000000000001,
+      "loss": 3.9482,
+      "step": 375
+    },
+    {
+      "epoch": 0.03300136317133445,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.001128,
+      "loss": 3.9829,
+      "step": 376
+    },
+    {
+      "epoch": 0.03308913275423694,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001131,
+      "loss": 4.0322,
+      "step": 377
+    },
+    {
+      "epoch": 0.03317690233713942,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.001134,
+      "loss": 3.9609,
+      "step": 378
+    },
+    {
+      "epoch": 0.03326467192004191,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001137,
+      "loss": 4.0239,
+      "step": 379
+    },
+    {
+      "epoch": 0.0333524415029444,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00114,
+      "loss": 3.9888,
+      "step": 380
+    },
+    {
+      "epoch": 0.03344021108584688,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0011430000000000001,
+      "loss": 3.9292,
+      "step": 381
+    },
+    {
+      "epoch": 0.03352798066874937,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001146,
+      "loss": 3.958,
+      "step": 382
+    },
+    {
+      "epoch": 0.03361575025165185,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001149,
+      "loss": 3.9438,
+      "step": 383
+    },
+    {
+      "epoch": 0.03370351983455434,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001152,
+      "loss": 3.9805,
+      "step": 384
+    },
+    {
+      "epoch": 0.03379128941745682,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001155,
+      "loss": 4.0659,
+      "step": 385
+    },
+    {
+      "epoch": 0.033879059000359306,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.001158,
+      "loss": 3.9604,
+      "step": 386
+    },
+    {
+      "epoch": 0.03396682858326179,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.0011610000000000001,
+      "loss": 3.9927,
+      "step": 387
+    },
+    {
+      "epoch": 0.034054598166164275,
+      "grad_norm": 1.0,
+      "learning_rate": 0.001164,
+      "loss": 4.0088,
+      "step": 388
+    },
+    {
+      "epoch": 0.03414236774906676,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.001167,
+      "loss": 3.9712,
+      "step": 389
+    },
+    {
+      "epoch": 0.034230137331969245,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00117,
+      "loss": 3.9395,
+      "step": 390
+    },
+    {
+      "epoch": 0.034317906914871736,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.001173,
+      "loss": 3.9497,
+      "step": 391
+    },
+    {
+      "epoch": 0.03440567649777422,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.001176,
+      "loss": 3.998,
+      "step": 392
+    },
+    {
+      "epoch": 0.034493446080676705,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0011790000000000001,
+      "loss": 3.9795,
+      "step": 393
+    },
+    {
+      "epoch": 0.03458121566357919,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001182,
+      "loss": 3.9468,
+      "step": 394
+    },
+    {
+      "epoch": 0.034668985246481675,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001185,
+      "loss": 3.9731,
+      "step": 395
+    },
+    {
+      "epoch": 0.03475675482938416,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.001188,
+      "loss": 3.9263,
+      "step": 396
+    },
+    {
+      "epoch": 0.034844524412286644,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.001191,
+      "loss": 3.9502,
+      "step": 397
+    },
+    {
+      "epoch": 0.03493229399518913,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0011940000000000002,
+      "loss": 3.9434,
+      "step": 398
+    },
+    {
+      "epoch": 0.03502006357809161,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.0011970000000000001,
+      "loss": 4.0049,
+      "step": 399
+    },
+    {
+      "epoch": 0.0351078331609941,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0012000000000000001,
+      "loss": 3.9722,
+      "step": 400
+    },
+    {
+      "epoch": 0.03519560274389658,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.001203,
+      "loss": 3.9683,
+      "step": 401
+    },
+    {
+      "epoch": 0.03528337232679907,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.001206,
+      "loss": 3.9512,
+      "step": 402
+    },
+    {
+      "epoch": 0.03537114190970156,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.001209,
+      "loss": 4.0049,
+      "step": 403
+    },
+    {
+      "epoch": 0.035458911492604044,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0012120000000000002,
+      "loss": 4.0151,
+      "step": 404
+    },
+    {
+      "epoch": 0.03554668107550653,
+      "grad_norm": 3.0,
+      "learning_rate": 0.0012150000000000002,
+      "loss": 4.0083,
+      "step": 405
+    },
+    {
+      "epoch": 0.03563445065840901,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.0012180000000000001,
+      "loss": 4.0024,
+      "step": 406
+    },
+    {
+      "epoch": 0.0357222202413115,
+      "grad_norm": 3.171875,
+      "learning_rate": 0.0012209999999999999,
+      "loss": 4.0562,
+      "step": 407
+    },
+    {
+      "epoch": 0.03580998982421398,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.001224,
+      "loss": 3.9976,
+      "step": 408
+    },
+    {
+      "epoch": 0.03589775940711647,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.001227,
+      "loss": 4.021,
+      "step": 409
+    },
+    {
+      "epoch": 0.03598552899001895,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00123,
+      "loss": 4.002,
+      "step": 410
+    },
+    {
+      "epoch": 0.036073298572921436,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.001233,
+      "loss": 4.0156,
+      "step": 411
+    },
+    {
+      "epoch": 0.03616106815582392,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.001236,
+      "loss": 4.0059,
+      "step": 412
+    },
+    {
+      "epoch": 0.036248837738726405,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.0012389999999999999,
+      "loss": 3.959,
+      "step": 413
+    },
+    {
+      "epoch": 0.0363366073216289,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.001242,
+      "loss": 3.9736,
+      "step": 414
+    },
+    {
+      "epoch": 0.03642437690453138,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.001245,
+      "loss": 4.0049,
+      "step": 415
+    },
+    {
+      "epoch": 0.036512146487433866,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.001248,
+      "loss": 4.002,
+      "step": 416
+    },
+    {
+      "epoch": 0.03659991607033635,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.001251,
+      "loss": 3.9365,
+      "step": 417
+    },
+    {
+      "epoch": 0.036687685653238836,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.001254,
+      "loss": 3.9277,
+      "step": 418
+    },
+    {
+      "epoch": 0.03677545523614132,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0012569999999999999,
+      "loss": 3.9312,
+      "step": 419
+    },
+    {
+      "epoch": 0.036863224819043805,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00126,
+      "loss": 3.98,
+      "step": 420
+    },
+    {
+      "epoch": 0.03695099440194629,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.001263,
+      "loss": 3.9614,
+      "step": 421
+    },
+    {
+      "epoch": 0.037038763984848774,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001266,
+      "loss": 3.9854,
+      "step": 422
+    },
+    {
+      "epoch": 0.03712653356775126,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001269,
+      "loss": 3.9487,
+      "step": 423
+    },
+    {
+      "epoch": 0.037214303150653744,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.001272,
+      "loss": 3.9883,
+      "step": 424
+    },
+    {
+      "epoch": 0.03730207273355623,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001275,
+      "loss": 3.9438,
+      "step": 425
+    },
+    {
+      "epoch": 0.03738984231645872,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001278,
+      "loss": 3.9678,
+      "step": 426
+    },
+    {
+      "epoch": 0.037477611899361205,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001281,
+      "loss": 3.9448,
+      "step": 427
+    },
+    {
+      "epoch": 0.03756538148226369,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001284,
+      "loss": 3.9224,
+      "step": 428
+    },
+    {
+      "epoch": 0.037653151065166174,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001287,
+      "loss": 3.874,
+      "step": 429
+    },
+    {
+      "epoch": 0.03774092064806866,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00129,
+      "loss": 3.9141,
+      "step": 430
+    },
+    {
+      "epoch": 0.03782869023097114,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001293,
+      "loss": 3.9487,
+      "step": 431
+    },
+    {
+      "epoch": 0.03791645981387363,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001296,
+      "loss": 3.9058,
+      "step": 432
+    },
+    {
+      "epoch": 0.03800422939677611,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001299,
+      "loss": 4.0063,
+      "step": 433
+    },
+    {
+      "epoch": 0.0380919989796786,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001302,
+      "loss": 3.9141,
+      "step": 434
+    },
+    {
+      "epoch": 0.03817976856258108,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001305,
+      "loss": 3.8501,
+      "step": 435
+    },
+    {
+      "epoch": 0.038267538145483566,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001308,
+      "loss": 3.8936,
+      "step": 436
+    },
+    {
+      "epoch": 0.03835530772838605,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001311,
+      "loss": 3.9009,
+      "step": 437
+    },
+    {
+      "epoch": 0.03844307731128854,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.001314,
+      "loss": 3.8857,
+      "step": 438
+    },
+    {
+      "epoch": 0.03853084689419103,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001317,
+      "loss": 3.8975,
+      "step": 439
+    },
+    {
+      "epoch": 0.03861861647709351,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00132,
+      "loss": 3.9072,
+      "step": 440
+    },
+    {
+      "epoch": 0.038706386059996,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.001323,
+      "loss": 3.9507,
+      "step": 441
+    },
+    {
+      "epoch": 0.03879415564289848,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0013260000000000001,
+      "loss": 3.9082,
+      "step": 442
+    },
+    {
+      "epoch": 0.038881925225800966,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.001329,
+      "loss": 3.8965,
+      "step": 443
+    },
+    {
+      "epoch": 0.03896969480870345,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.001332,
+      "loss": 3.9629,
+      "step": 444
+    },
+    {
+      "epoch": 0.039057464391605935,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001335,
+      "loss": 4.0127,
+      "step": 445
+    },
+    {
+      "epoch": 0.03914523397450842,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.001338,
+      "loss": 3.9023,
+      "step": 446
+    },
+    {
+      "epoch": 0.039233003557410905,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.001341,
+      "loss": 3.9448,
+      "step": 447
+    },
+    {
+      "epoch": 0.03932077314031339,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0013440000000000001,
+      "loss": 3.9077,
+      "step": 448
+    },
+    {
+      "epoch": 0.03940854272321588,
+      "grad_norm": 2.890625,
+      "learning_rate": 0.001347,
+      "loss": 4.0171,
+      "step": 449
+    },
+    {
+      "epoch": 0.039496312306118365,
+      "grad_norm": 1.9765625,
+      "learning_rate": 0.00135,
+      "loss": 3.9956,
+      "step": 450
+    },
+    {
+      "epoch": 0.03958408188902085,
+      "grad_norm": 2.625,
+      "learning_rate": 0.001353,
+      "loss": 3.8921,
+      "step": 451
+    },
+    {
+      "epoch": 0.039671851471923335,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.001356,
+      "loss": 3.9917,
+      "step": 452
+    },
+    {
+      "epoch": 0.03975962105482582,
+      "grad_norm": 1.875,
+      "learning_rate": 0.001359,
+      "loss": 3.9468,
+      "step": 453
+    },
+    {
+      "epoch": 0.039847390637728304,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0013620000000000001,
+      "loss": 3.8755,
+      "step": 454
+    },
+    {
+      "epoch": 0.03993516022063079,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.0013650000000000001,
+      "loss": 3.9326,
+      "step": 455
+    },
+    {
+      "epoch": 0.04002292980353327,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.001368,
+      "loss": 3.918,
+      "step": 456
+    },
+    {
+      "epoch": 0.04011069938643576,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.001371,
+      "loss": 3.8735,
+      "step": 457
+    },
+    {
+      "epoch": 0.04019846896933824,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.001374,
+      "loss": 3.9316,
+      "step": 458
+    },
+    {
+      "epoch": 0.04028623855224073,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0013770000000000002,
+      "loss": 3.9751,
+      "step": 459
+    },
+    {
+      "epoch": 0.04037400813514321,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0013800000000000002,
+      "loss": 3.9214,
+      "step": 460
+    },
+    {
+      "epoch": 0.040461777718045704,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0013830000000000001,
+      "loss": 3.9629,
+      "step": 461
+    },
+    {
+      "epoch": 0.04054954730094819,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.001386,
+      "loss": 3.9033,
+      "step": 462
+    },
+    {
+      "epoch": 0.04063731688385067,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.001389,
+      "loss": 3.9839,
+      "step": 463
+    },
+    {
+      "epoch": 0.04072508646675316,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.001392,
+      "loss": 3.937,
+      "step": 464
+    },
+    {
+      "epoch": 0.04081285604965564,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0013950000000000002,
+      "loss": 3.9165,
+      "step": 465
+    },
+    {
+      "epoch": 0.04090062563255813,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0013980000000000002,
+      "loss": 3.9062,
+      "step": 466
+    },
+    {
+      "epoch": 0.04098839521546061,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0014010000000000001,
+      "loss": 3.9053,
+      "step": 467
+    },
+    {
+      "epoch": 0.041076164798363096,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001404,
+      "loss": 3.9014,
+      "step": 468
+    },
+    {
+      "epoch": 0.04116393438126558,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001407,
+      "loss": 3.9634,
+      "step": 469
+    },
+    {
+      "epoch": 0.041251703964168066,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00141,
+      "loss": 3.9297,
+      "step": 470
+    },
+    {
+      "epoch": 0.04133947354707055,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001413,
+      "loss": 3.9546,
+      "step": 471
+    },
+    {
+      "epoch": 0.041427243129973035,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001416,
+      "loss": 3.9058,
+      "step": 472
+    },
+    {
+      "epoch": 0.041515012712875526,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001419,
+      "loss": 3.8677,
+      "step": 473
+    },
+    {
+      "epoch": 0.04160278229577801,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0014219999999999999,
+      "loss": 3.9424,
+      "step": 474
+    },
+    {
+      "epoch": 0.041690551878680496,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001425,
+      "loss": 3.9219,
+      "step": 475
+    },
+    {
+      "epoch": 0.04177832146158298,
+      "grad_norm": 1.125,
+      "learning_rate": 0.001428,
+      "loss": 3.8311,
+      "step": 476
+    },
+    {
+      "epoch": 0.041866091044485465,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001431,
+      "loss": 3.8345,
+      "step": 477
+    },
+    {
+      "epoch": 0.04195386062738795,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001434,
+      "loss": 3.8394,
+      "step": 478
+    },
+    {
+      "epoch": 0.042041630210290434,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001437,
+      "loss": 3.9087,
+      "step": 479
+    },
+    {
+      "epoch": 0.04212939979319292,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0014399999999999999,
+      "loss": 3.8911,
+      "step": 480
+    },
+    {
+      "epoch": 0.042217169376095404,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.001443,
+      "loss": 3.9209,
+      "step": 481
+    },
+    {
+      "epoch": 0.04230493895899789,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.001446,
+      "loss": 3.9258,
+      "step": 482
+    },
+    {
+      "epoch": 0.04239270854190037,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001449,
+      "loss": 3.9434,
+      "step": 483
+    },
+    {
+      "epoch": 0.042480478124802865,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001452,
+      "loss": 3.9185,
+      "step": 484
+    },
+    {
+      "epoch": 0.04256824770770535,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.001455,
+      "loss": 3.957,
+      "step": 485
+    },
+    {
+      "epoch": 0.042656017290607834,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.001458,
+      "loss": 3.9399,
+      "step": 486
+    },
+    {
+      "epoch": 0.04274378687351032,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.001461,
+      "loss": 3.9497,
+      "step": 487
+    },
+    {
+      "epoch": 0.0428315564564128,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.001464,
+      "loss": 3.8267,
+      "step": 488
+    },
+    {
+      "epoch": 0.04291932603931529,
+      "grad_norm": 2.25,
+      "learning_rate": 0.001467,
+      "loss": 3.918,
+      "step": 489
+    },
+    {
+      "epoch": 0.04300709562221777,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00147,
+      "loss": 3.8647,
+      "step": 490
+    },
+    {
+      "epoch": 0.04309486520512026,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.001473,
+      "loss": 3.9663,
+      "step": 491
+    },
+    {
+      "epoch": 0.04318263478802274,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.001476,
+      "loss": 3.9067,
+      "step": 492
+    },
+    {
+      "epoch": 0.043270404370925226,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.001479,
+      "loss": 3.8442,
+      "step": 493
+    },
+    {
+      "epoch": 0.04335817395382771,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.001482,
+      "loss": 3.9346,
+      "step": 494
+    },
+    {
+      "epoch": 0.043445943536730196,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.001485,
+      "loss": 3.9224,
+      "step": 495
+    },
+    {
+      "epoch": 0.04353371311963269,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001488,
+      "loss": 3.9116,
+      "step": 496
+    },
+    {
+      "epoch": 0.04362148270253517,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.001491,
+      "loss": 3.8931,
+      "step": 497
+    },
+    {
+      "epoch": 0.04370925228543766,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.001494,
+      "loss": 3.9458,
+      "step": 498
+    },
+    {
+      "epoch": 0.04379702186834014,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.001497,
+      "loss": 3.8657,
+      "step": 499
+    },
+    {
+      "epoch": 0.043884791451242626,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0015,
+      "loss": 3.937,
+      "step": 500
+    },
+    {
+      "epoch": 0.04397256103414511,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.001503,
+      "loss": 3.8296,
+      "step": 501
+    },
+    {
+      "epoch": 0.044060330617047595,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001506,
+      "loss": 3.8706,
+      "step": 502
+    },
+    {
+      "epoch": 0.04414810019995008,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.0015090000000000001,
+      "loss": 3.8838,
+      "step": 503
+    },
+    {
+      "epoch": 0.044235869782852565,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.001512,
+      "loss": 3.8638,
+      "step": 504
+    },
+    {
+      "epoch": 0.04432363936575505,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.001515,
+      "loss": 3.936,
+      "step": 505
+    },
+    {
+      "epoch": 0.044411408948657534,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.001518,
+      "loss": 3.8638,
+      "step": 506
+    },
+    {
+      "epoch": 0.044499178531560026,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.001521,
+      "loss": 3.9365,
+      "step": 507
+    },
+    {
+      "epoch": 0.04458694811446251,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001524,
+      "loss": 3.8721,
+      "step": 508
+    },
+    {
+      "epoch": 0.044674717697364995,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0015270000000000001,
+      "loss": 3.8794,
+      "step": 509
+    },
+    {
+      "epoch": 0.04476248728026748,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0015300000000000001,
+      "loss": 3.8472,
+      "step": 510
+    },
+    {
+      "epoch": 0.044850256863169964,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001533,
+      "loss": 3.8096,
+      "step": 511
+    },
+    {
+      "epoch": 0.04493802644607245,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.001536,
+      "loss": 3.877,
+      "step": 512
+    },
+    {
+      "epoch": 0.04502579602897493,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001539,
+      "loss": 3.8677,
+      "step": 513
+    },
+    {
+      "epoch": 0.04511356561187742,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001542,
+      "loss": 3.8823,
+      "step": 514
+    },
+    {
+      "epoch": 0.0452013351947799,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0015450000000000001,
+      "loss": 3.8525,
+      "step": 515
+    },
+    {
+      "epoch": 0.04528910477768239,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0015480000000000001,
+      "loss": 3.957,
+      "step": 516
+    },
+    {
+      "epoch": 0.04537687436058487,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.001551,
+      "loss": 3.8887,
+      "step": 517
+    },
+    {
+      "epoch": 0.04546464394348736,
+      "grad_norm": 4.5625,
+      "learning_rate": 0.001554,
+      "loss": 3.8979,
+      "step": 518
+    },
+    {
+      "epoch": 0.04555241352638985,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.001557,
+      "loss": 3.8882,
+      "step": 519
+    },
+    {
+      "epoch": 0.04564018310929233,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0015600000000000002,
+      "loss": 3.8604,
+      "step": 520
+    },
+    {
+      "epoch": 0.04572795269219482,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0015630000000000002,
+      "loss": 3.8828,
+      "step": 521
+    },
+    {
+      "epoch": 0.0458157222750973,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0015660000000000001,
+      "loss": 3.9438,
+      "step": 522
+    },
+    {
+      "epoch": 0.04590349185799979,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.001569,
+      "loss": 3.873,
+      "step": 523
+    },
+    {
+      "epoch": 0.04599126144090227,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.001572,
+      "loss": 3.834,
+      "step": 524
+    },
+    {
+      "epoch": 0.046079031023804756,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001575,
+      "loss": 3.9067,
+      "step": 525
+    },
+    {
+      "epoch": 0.04616680060670724,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0015780000000000002,
+      "loss": 3.8608,
+      "step": 526
+    },
+    {
+      "epoch": 0.046254570189609726,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0015810000000000002,
+      "loss": 3.8252,
+      "step": 527
+    },
+    {
+      "epoch": 0.04634233977251221,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0015840000000000001,
+      "loss": 3.8516,
+      "step": 528
+    },
+    {
+      "epoch": 0.046430109355414695,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001587,
+      "loss": 3.8423,
+      "step": 529
+    },
+    {
+      "epoch": 0.04651787893831718,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00159,
+      "loss": 3.9004,
+      "step": 530
+    },
+    {
+      "epoch": 0.04660564852121967,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.001593,
+      "loss": 3.9033,
+      "step": 531
+    },
+    {
+      "epoch": 0.046693418104122156,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0015960000000000002,
+      "loss": 3.8877,
+      "step": 532
+    },
+    {
+      "epoch": 0.04678118768702464,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0015990000000000002,
+      "loss": 3.833,
+      "step": 533
+    },
+    {
+      "epoch": 0.046868957269927125,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0016020000000000001,
+      "loss": 3.8491,
+      "step": 534
+    },
+    {
+      "epoch": 0.04695672685282961,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001605,
+      "loss": 3.7886,
+      "step": 535
+    },
+    {
+      "epoch": 0.047044496435732094,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001608,
+      "loss": 3.8159,
+      "step": 536
+    },
+    {
+      "epoch": 0.04713226601863458,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0016110000000000002,
+      "loss": 3.8486,
+      "step": 537
+    },
+    {
+      "epoch": 0.047220035601537064,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0016140000000000002,
+      "loss": 3.8882,
+      "step": 538
+    },
+    {
+      "epoch": 0.04730780518443955,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0016170000000000002,
+      "loss": 3.8584,
+      "step": 539
+    },
+    {
+      "epoch": 0.04739557476734203,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.0016200000000000001,
+      "loss": 3.8384,
+      "step": 540
+    },
+    {
+      "epoch": 0.04748334435024452,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.001623,
+      "loss": 3.9512,
+      "step": 541
+    },
+    {
+      "epoch": 0.04757111393314701,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.001626,
+      "loss": 3.9414,
+      "step": 542
+    },
+    {
+      "epoch": 0.047658883516049494,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0016290000000000002,
+      "loss": 3.8887,
+      "step": 543
+    },
+    {
+      "epoch": 0.04774665309895198,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0016320000000000002,
+      "loss": 3.9502,
+      "step": 544
+    },
+    {
+      "epoch": 0.04783442268185446,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.0016350000000000002,
+      "loss": 3.8711,
+      "step": 545
+    },
+    {
+      "epoch": 0.04792219226475695,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0016380000000000001,
+      "loss": 3.9072,
+      "step": 546
+    },
+    {
+      "epoch": 0.04800996184765943,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.001641,
+      "loss": 3.9033,
+      "step": 547
+    },
+    {
+      "epoch": 0.04809773143056192,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.001644,
+      "loss": 3.8638,
+      "step": 548
+    },
+    {
+      "epoch": 0.0481855010134644,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0016470000000000002,
+      "loss": 3.8281,
+      "step": 549
+    },
+    {
+      "epoch": 0.048273270596366886,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0016500000000000002,
+      "loss": 3.8472,
+      "step": 550
+    },
+    {
+      "epoch": 0.04836104017926937,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.0016530000000000002,
+      "loss": 3.8496,
+      "step": 551
+    },
+    {
+      "epoch": 0.048448809762171856,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0016560000000000001,
+      "loss": 3.9116,
+      "step": 552
+    },
+    {
+      "epoch": 0.04853657934507434,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001659,
+      "loss": 3.8252,
+      "step": 553
+    },
+    {
+      "epoch": 0.04862434892797683,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0016620000000000003,
+      "loss": 3.79,
+      "step": 554
+    },
+    {
+      "epoch": 0.04871211851087932,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0016650000000000002,
+      "loss": 3.8506,
+      "step": 555
+    },
+    {
+      "epoch": 0.0487998880937818,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0016680000000000002,
+      "loss": 3.8159,
+      "step": 556
+    },
+    {
+      "epoch": 0.048887657676684286,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0016710000000000002,
+      "loss": 3.811,
+      "step": 557
+    },
+    {
+      "epoch": 0.04897542725958677,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0016740000000000001,
+      "loss": 3.877,
+      "step": 558
+    },
+    {
+      "epoch": 0.049063196842489255,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001677,
+      "loss": 3.873,
+      "step": 559
+    },
+    {
+      "epoch": 0.04915096642539174,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0016800000000000003,
+      "loss": 3.8652,
+      "step": 560
+    },
+    {
+      "epoch": 0.049238736008294225,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0016830000000000003,
+      "loss": 3.8296,
+      "step": 561
+    },
+    {
+      "epoch": 0.04932650559119671,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0016860000000000002,
+      "loss": 3.9292,
+      "step": 562
+    },
+    {
+      "epoch": 0.049414275174099194,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001689,
+      "loss": 3.7979,
+      "step": 563
+    },
+    {
+      "epoch": 0.04950204475700168,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.001692,
+      "loss": 3.8418,
+      "step": 564
+    },
+    {
+      "epoch": 0.04958981433990416,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.001695,
+      "loss": 3.8462,
+      "step": 565
+    },
+    {
+      "epoch": 0.049677583922806655,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.0016979999999999999,
+      "loss": 3.8521,
+      "step": 566
+    },
+    {
+      "epoch": 0.04976535350570914,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.0017009999999999998,
+      "loss": 3.8589,
+      "step": 567
+    },
+    {
+      "epoch": 0.049853123088611624,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0017039999999999998,
+      "loss": 3.7671,
+      "step": 568
+    },
+    {
+      "epoch": 0.04994089267151411,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.001707,
+      "loss": 3.8135,
+      "step": 569
+    },
+    {
+      "epoch": 0.050028662254416593,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00171,
+      "loss": 3.8589,
+      "step": 570
+    },
+    {
+      "epoch": 0.05011643183731908,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001713,
+      "loss": 3.8003,
+      "step": 571
+    },
+    {
+      "epoch": 0.05020420142022156,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0017159999999999999,
+      "loss": 3.8389,
+      "step": 572
+    },
+    {
+      "epoch": 0.05029197100312405,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0017189999999999998,
+      "loss": 3.7534,
+      "step": 573
+    },
+    {
+      "epoch": 0.05037974058602653,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001722,
+      "loss": 3.7896,
+      "step": 574
+    },
+    {
+      "epoch": 0.05046751016892902,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.001725,
+      "loss": 3.8091,
+      "step": 575
+    },
+    {
+      "epoch": 0.0505552797518315,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.001728,
+      "loss": 3.8789,
+      "step": 576
+    },
+    {
+      "epoch": 0.05064304933473399,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001731,
+      "loss": 3.7671,
+      "step": 577
+    },
+    {
+      "epoch": 0.05073081891763648,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0017339999999999999,
+      "loss": 3.8643,
+      "step": 578
+    },
+    {
+      "epoch": 0.05081858850053896,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.0017369999999999998,
+      "loss": 3.8315,
+      "step": 579
+    },
+    {
+      "epoch": 0.05090635808344145,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00174,
+      "loss": 3.7935,
+      "step": 580
+    },
+    {
+      "epoch": 0.05099412766634393,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001743,
+      "loss": 3.7778,
+      "step": 581
+    },
+    {
+      "epoch": 0.051081897249246416,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001746,
+      "loss": 3.8335,
+      "step": 582
+    },
+    {
+      "epoch": 0.0511696668321489,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001749,
+      "loss": 3.7993,
+      "step": 583
+    },
+    {
+      "epoch": 0.051257436415051386,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0017519999999999999,
+      "loss": 3.8672,
+      "step": 584
+    },
+    {
+      "epoch": 0.05134520599795387,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.0017549999999999998,
+      "loss": 3.8003,
+      "step": 585
+    },
+    {
+      "epoch": 0.051432975580856355,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001758,
+      "loss": 3.7607,
+      "step": 586
+    },
+    {
+      "epoch": 0.05152074516375884,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.001761,
+      "loss": 3.8931,
+      "step": 587
+    },
+    {
+      "epoch": 0.051608514746661324,
+      "grad_norm": 2.65625,
+      "learning_rate": 0.001764,
+      "loss": 3.7676,
+      "step": 588
+    },
+    {
+      "epoch": 0.051696284329563816,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.001767,
+      "loss": 3.8403,
+      "step": 589
+    },
+    {
+      "epoch": 0.0517840539124663,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0017699999999999999,
+      "loss": 3.8208,
+      "step": 590
+    },
+    {
+      "epoch": 0.051871823495368785,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.001773,
+      "loss": 3.8799,
+      "step": 591
+    },
+    {
+      "epoch": 0.05195959307827127,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.001776,
+      "loss": 3.8438,
+      "step": 592
+    },
+    {
+      "epoch": 0.052047362661173754,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.001779,
+      "loss": 3.8594,
+      "step": 593
+    },
+    {
+      "epoch": 0.05213513224407624,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.001782,
+      "loss": 3.8071,
+      "step": 594
+    },
+    {
+      "epoch": 0.052222901826978724,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.001785,
+      "loss": 3.8213,
+      "step": 595
+    },
+    {
+      "epoch": 0.05231067140988121,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.0017879999999999999,
+      "loss": 3.812,
+      "step": 596
+    },
+    {
+      "epoch": 0.05239844099278369,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001791,
+      "loss": 3.8872,
+      "step": 597
+    },
+    {
+      "epoch": 0.05248621057568618,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.001794,
+      "loss": 3.894,
+      "step": 598
+    },
+    {
+      "epoch": 0.05257398015858866,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001797,
+      "loss": 3.8413,
+      "step": 599
+    },
+    {
+      "epoch": 0.05266174974149115,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0018,
+      "loss": 3.9023,
+      "step": 600
+    },
+    {
+      "epoch": 0.05274951932439364,
+      "grad_norm": 6.1875,
+      "learning_rate": 0.001803,
+      "loss": 3.8579,
+      "step": 601
+    },
+    {
+      "epoch": 0.05283728890729612,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0018059999999999999,
+      "loss": 3.7905,
+      "step": 602
+    },
+    {
+      "epoch": 0.05292505849019861,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.001809,
+      "loss": 3.8398,
+      "step": 603
+    },
+    {
+      "epoch": 0.05301282807310109,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.001812,
+      "loss": 3.7539,
+      "step": 604
+    },
+    {
+      "epoch": 0.05310059765600358,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001815,
+      "loss": 3.8086,
+      "step": 605
+    },
+    {
+      "epoch": 0.05318836723890606,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001818,
+      "loss": 3.7642,
+      "step": 606
+    },
+    {
+      "epoch": 0.05327613682180855,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001821,
+      "loss": 3.7954,
+      "step": 607
+    },
+    {
+      "epoch": 0.05336390640471103,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001824,
+      "loss": 3.7739,
+      "step": 608
+    },
+    {
+      "epoch": 0.053451675987613516,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001827,
+      "loss": 3.8013,
+      "step": 609
+    },
+    {
+      "epoch": 0.053539445570516,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00183,
+      "loss": 3.7886,
+      "step": 610
+    },
+    {
+      "epoch": 0.053627215153418485,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.001833,
+      "loss": 3.7124,
+      "step": 611
+    },
+    {
+      "epoch": 0.05371498473632098,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001836,
+      "loss": 3.8032,
+      "step": 612
+    },
+    {
+      "epoch": 0.05380275431922346,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.001839,
+      "loss": 3.7935,
+      "step": 613
+    },
+    {
+      "epoch": 0.053890523902125946,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.001842,
+      "loss": 3.73,
+      "step": 614
+    },
+    {
+      "epoch": 0.05397829348502843,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.001845,
+      "loss": 3.8667,
+      "step": 615
+    },
+    {
+      "epoch": 0.054066063067930915,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001848,
+      "loss": 3.8242,
+      "step": 616
+    },
+    {
+      "epoch": 0.0541538326508334,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001851,
+      "loss": 3.8115,
+      "step": 617
+    },
+    {
+      "epoch": 0.054241602233735885,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001854,
+      "loss": 3.7793,
+      "step": 618
+    },
+    {
+      "epoch": 0.05432937181663837,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.001857,
+      "loss": 3.7954,
+      "step": 619
+    },
+    {
+      "epoch": 0.054417141399540854,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00186,
+      "loss": 3.8774,
+      "step": 620
+    },
+    {
+      "epoch": 0.05450491098244334,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.001863,
+      "loss": 3.8413,
+      "step": 621
+    },
+    {
+      "epoch": 0.05459268056534582,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001866,
+      "loss": 3.8218,
+      "step": 622
+    },
+    {
+      "epoch": 0.05468045014824831,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001869,
+      "loss": 3.7842,
+      "step": 623
+    },
+    {
+      "epoch": 0.0547682197311508,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001872,
+      "loss": 3.8037,
+      "step": 624
+    },
+    {
+      "epoch": 0.054855989314053284,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.001875,
+      "loss": 3.7056,
+      "step": 625
+    },
+    {
+      "epoch": 0.05494375889695577,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0018780000000000001,
+      "loss": 3.8032,
+      "step": 626
+    },
+    {
+      "epoch": 0.055031528479858254,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001881,
+      "loss": 3.8687,
+      "step": 627
+    },
+    {
+      "epoch": 0.05511929806276074,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.001884,
+      "loss": 3.7104,
+      "step": 628
+    },
+    {
+      "epoch": 0.05520706764566322,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.001887,
+      "loss": 3.7793,
+      "step": 629
+    },
+    {
+      "epoch": 0.05529483722856571,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00189,
+      "loss": 3.8047,
+      "step": 630
+    },
+    {
+      "epoch": 0.05538260681146819,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0018930000000000002,
+      "loss": 3.791,
+      "step": 631
+    },
+    {
+      "epoch": 0.05547037639437068,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0018960000000000001,
+      "loss": 3.7681,
+      "step": 632
+    },
+    {
+      "epoch": 0.05555814597727316,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.001899,
+      "loss": 3.7451,
+      "step": 633
+    },
+    {
+      "epoch": 0.055645915560175646,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.001902,
+      "loss": 3.772,
+      "step": 634
+    },
+    {
+      "epoch": 0.05573368514307813,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.001905,
+      "loss": 3.8345,
+      "step": 635
+    },
+    {
+      "epoch": 0.05582145472598062,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001908,
+      "loss": 3.77,
+      "step": 636
+    },
+    {
+      "epoch": 0.05590922430888311,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0019110000000000002,
+      "loss": 3.8013,
+      "step": 637
+    },
+    {
+      "epoch": 0.05599699389178559,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0019140000000000001,
+      "loss": 3.7749,
+      "step": 638
+    },
+    {
+      "epoch": 0.056084763474688076,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001917,
+      "loss": 3.8413,
+      "step": 639
+    },
+    {
+      "epoch": 0.05617253305759056,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00192,
+      "loss": 3.7783,
+      "step": 640
+    },
+    {
+      "epoch": 0.056260302640493046,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001923,
+      "loss": 3.7515,
+      "step": 641
+    },
+    {
+      "epoch": 0.05634807222339553,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001926,
+      "loss": 3.8101,
+      "step": 642
+    },
+    {
+      "epoch": 0.056435841806298015,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0019290000000000002,
+      "loss": 3.8691,
+      "step": 643
+    },
+    {
+      "epoch": 0.0565236113892005,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0019320000000000001,
+      "loss": 3.8071,
+      "step": 644
+    },
+    {
+      "epoch": 0.056611380972102984,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.001935,
+      "loss": 3.8032,
+      "step": 645
+    },
+    {
+      "epoch": 0.05669915055500547,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.001938,
+      "loss": 3.7886,
+      "step": 646
+    },
+    {
+      "epoch": 0.05678692013790796,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.001941,
+      "loss": 3.7485,
+      "step": 647
+    },
+    {
+      "epoch": 0.056874689720810445,
+      "grad_norm": 0.75,
+      "learning_rate": 0.0019440000000000002,
+      "loss": 3.7437,
+      "step": 648
+    },
+    {
+      "epoch": 0.05696245930371293,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0019470000000000002,
+      "loss": 3.8135,
+      "step": 649
+    },
+    {
+      "epoch": 0.057050228886615414,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0019500000000000001,
+      "loss": 3.8784,
+      "step": 650
+    },
+    {
+      "epoch": 0.0571379984695179,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001953,
+      "loss": 3.7544,
+      "step": 651
+    },
+    {
+      "epoch": 0.057225768052420384,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0019560000000000003,
+      "loss": 3.7856,
+      "step": 652
+    },
+    {
+      "epoch": 0.05731353763532287,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0019590000000000002,
+      "loss": 3.7559,
+      "step": 653
+    },
+    {
+      "epoch": 0.05740130721822535,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001962,
+      "loss": 3.6909,
+      "step": 654
+    },
+    {
+      "epoch": 0.05748907680112784,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001965,
+      "loss": 3.8457,
+      "step": 655
+    },
+    {
+      "epoch": 0.05757684638403032,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.001968,
+      "loss": 3.7324,
+      "step": 656
+    },
+    {
+      "epoch": 0.05766461596693281,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.001971,
+      "loss": 3.7222,
+      "step": 657
+    },
+    {
+      "epoch": 0.05775238554983529,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.001974,
+      "loss": 3.7612,
+      "step": 658
+    },
+    {
+      "epoch": 0.05784015513273778,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.001977,
+      "loss": 3.6855,
+      "step": 659
+    },
+    {
+      "epoch": 0.05792792471564027,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00198,
+      "loss": 3.7402,
+      "step": 660
+    },
+    {
+      "epoch": 0.05801569429854275,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.001983,
+      "loss": 3.7339,
+      "step": 661
+    },
+    {
+      "epoch": 0.05810346388144524,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0019860000000000004,
+      "loss": 3.7827,
+      "step": 662
+    },
+    {
+      "epoch": 0.05819123346434772,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0019890000000000003,
+      "loss": 3.7568,
+      "step": 663
+    },
+    {
+      "epoch": 0.05827900304725021,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0019920000000000003,
+      "loss": 3.8013,
+      "step": 664
+    },
+    {
+      "epoch": 0.05836677263015269,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0019950000000000002,
+      "loss": 3.7363,
+      "step": 665
+    },
+    {
+      "epoch": 0.058454542213055176,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001998,
+      "loss": 3.7886,
+      "step": 666
+    },
+    {
+      "epoch": 0.05854231179595766,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.002001,
+      "loss": 3.7139,
+      "step": 667
+    },
+    {
+      "epoch": 0.058630081378860145,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.002004,
+      "loss": 3.689,
+      "step": 668
+    },
+    {
+      "epoch": 0.05871785096176263,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.002007,
+      "loss": 3.6919,
+      "step": 669
+    },
+    {
+      "epoch": 0.058805620544665115,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00201,
+      "loss": 3.7578,
+      "step": 670
+    },
+    {
+      "epoch": 0.058893390127567606,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.002013,
+      "loss": 3.813,
+      "step": 671
+    },
+    {
+      "epoch": 0.05898115971047009,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.002016,
+      "loss": 3.814,
+      "step": 672
+    },
+    {
+      "epoch": 0.059068929293372575,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.002019,
+      "loss": 3.8003,
+      "step": 673
+    },
+    {
+      "epoch": 0.05915669887627506,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0020220000000000004,
+      "loss": 3.73,
+      "step": 674
+    },
+    {
+      "epoch": 0.059244468459177545,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0020250000000000003,
+      "loss": 3.7837,
+      "step": 675
+    },
+    {
+      "epoch": 0.05933223804208003,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.0020280000000000003,
+      "loss": 3.7627,
+      "step": 676
+    },
+    {
+      "epoch": 0.059420007624982514,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0020310000000000003,
+      "loss": 3.7266,
+      "step": 677
+    },
+    {
+      "epoch": 0.059507777207885,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0020340000000000002,
+      "loss": 3.7349,
+      "step": 678
+    },
+    {
+      "epoch": 0.05959554679078748,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.002037,
+      "loss": 3.832,
+      "step": 679
+    },
+    {
+      "epoch": 0.05968331637368997,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00204,
+      "loss": 3.8428,
+      "step": 680
+    },
+    {
+      "epoch": 0.05977108595659245,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.002043,
+      "loss": 3.8057,
+      "step": 681
+    },
+    {
+      "epoch": 0.059858855539494944,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.002046,
+      "loss": 3.7397,
+      "step": 682
+    },
+    {
+      "epoch": 0.05994662512239743,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.002049,
+      "loss": 3.769,
+      "step": 683
+    },
+    {
+      "epoch": 0.060034394705299914,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.002052,
+      "loss": 3.7515,
+      "step": 684
+    },
+    {
+      "epoch": 0.0601221642882024,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0020550000000000004,
+      "loss": 3.7881,
+      "step": 685
+    },
+    {
+      "epoch": 0.06020993387110488,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0020580000000000004,
+      "loss": 3.7822,
+      "step": 686
+    },
+    {
+      "epoch": 0.06029770345400737,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0020610000000000003,
+      "loss": 3.7397,
+      "step": 687
+    },
+    {
+      "epoch": 0.06038547303690985,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002064,
+      "loss": 3.7529,
+      "step": 688
+    },
+    {
+      "epoch": 0.06047324261981234,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002067,
+      "loss": 3.6914,
+      "step": 689
+    },
+    {
+      "epoch": 0.06056101220271482,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00207,
+      "loss": 3.7534,
+      "step": 690
+    },
+    {
+      "epoch": 0.060648781785617306,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0020729999999999998,
+      "loss": 3.731,
+      "step": 691
+    },
+    {
+      "epoch": 0.06073655136851979,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0020759999999999997,
+      "loss": 3.7261,
+      "step": 692
+    },
+    {
+      "epoch": 0.060824320951422275,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0020789999999999997,
+      "loss": 3.7568,
+      "step": 693
+    },
+    {
+      "epoch": 0.06091209053432477,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.002082,
+      "loss": 3.7568,
+      "step": 694
+    },
+    {
+      "epoch": 0.06099986011722725,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.002085,
+      "loss": 3.8052,
+      "step": 695
+    },
+    {
+      "epoch": 0.061087629700129736,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.002088,
+      "loss": 3.7393,
+      "step": 696
+    },
+    {
+      "epoch": 0.06117539928303222,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.002091,
+      "loss": 3.7939,
+      "step": 697
+    },
+    {
+      "epoch": 0.061263168865934706,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.002094,
+      "loss": 3.7563,
+      "step": 698
+    },
+    {
+      "epoch": 0.06135093844883719,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.002097,
+      "loss": 3.7168,
+      "step": 699
+    },
+    {
+      "epoch": 0.061438708031739675,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0021,
+      "loss": 3.7871,
+      "step": 700
+    },
+    {
+      "epoch": 0.06152647761464216,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.002103,
+      "loss": 3.7847,
+      "step": 701
+    },
+    {
+      "epoch": 0.061614247197544644,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.002106,
+      "loss": 3.7852,
+      "step": 702
+    },
+    {
+      "epoch": 0.06170201678044713,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0021089999999999998,
+      "loss": 3.7314,
+      "step": 703
+    },
+    {
+      "epoch": 0.061789786363349614,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0021119999999999997,
+      "loss": 3.7412,
+      "step": 704
+    },
+    {
+      "epoch": 0.0618775559462521,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.002115,
+      "loss": 3.7026,
+      "step": 705
+    },
+    {
+      "epoch": 0.06196532552915459,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.002118,
+      "loss": 3.7529,
+      "step": 706
+    },
+    {
+      "epoch": 0.062053095112057075,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002121,
+      "loss": 3.7627,
+      "step": 707
+    },
+    {
+      "epoch": 0.06214086469495956,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.002124,
+      "loss": 3.7388,
+      "step": 708
+    },
+    {
+      "epoch": 0.062228634277862044,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.002127,
+      "loss": 3.7339,
+      "step": 709
+    },
+    {
+      "epoch": 0.06231640386076453,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00213,
+      "loss": 3.7705,
+      "step": 710
+    },
+    {
+      "epoch": 0.06240417344366701,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.002133,
+      "loss": 3.7134,
+      "step": 711
+    },
+    {
+      "epoch": 0.0624919430265695,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002136,
+      "loss": 3.7803,
+      "step": 712
+    },
+    {
+      "epoch": 0.06257971260947198,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.002139,
+      "loss": 3.7539,
+      "step": 713
+    },
+    {
+      "epoch": 0.06266748219237447,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002142,
+      "loss": 3.8052,
+      "step": 714
+    },
+    {
+      "epoch": 0.06275525177527695,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0021449999999999998,
+      "loss": 3.6963,
+      "step": 715
+    },
+    {
+      "epoch": 0.06284302135817944,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.002148,
+      "loss": 3.7109,
+      "step": 716
+    },
+    {
+      "epoch": 0.06293079094108192,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.002151,
+      "loss": 3.665,
+      "step": 717
+    },
+    {
+      "epoch": 0.06301856052398441,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.002154,
+      "loss": 3.71,
+      "step": 718
+    },
+    {
+      "epoch": 0.06310633010688689,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.002157,
+      "loss": 3.7148,
+      "step": 719
+    },
+    {
+      "epoch": 0.06319409968978938,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00216,
+      "loss": 3.6938,
+      "step": 720
+    },
+    {
+      "epoch": 0.06328186927269186,
+      "grad_norm": 6.9375,
+      "learning_rate": 0.002163,
+      "loss": 3.8232,
+      "step": 721
+    },
+    {
+      "epoch": 0.06336963885559435,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.002166,
+      "loss": 3.8701,
+      "step": 722
+    },
+    {
+      "epoch": 0.06345740843849684,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.002169,
+      "loss": 3.7744,
+      "step": 723
+    },
+    {
+      "epoch": 0.06354517802139932,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.002172,
+      "loss": 3.7949,
+      "step": 724
+    },
+    {
+      "epoch": 0.06363294760430181,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.002175,
+      "loss": 3.7656,
+      "step": 725
+    },
+    {
+      "epoch": 0.06372071718720429,
+      "grad_norm": 0.5,
+      "learning_rate": 0.002178,
+      "loss": 3.7471,
+      "step": 726
+    },
+    {
+      "epoch": 0.06380848677010678,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0021809999999999998,
+      "loss": 3.686,
+      "step": 727
+    },
+    {
+      "epoch": 0.06389625635300926,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.002184,
+      "loss": 3.7002,
+      "step": 728
+    },
+    {
+      "epoch": 0.06398402593591175,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.002187,
+      "loss": 3.6968,
+      "step": 729
+    },
+    {
+      "epoch": 0.06407179551881423,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00219,
+      "loss": 3.7095,
+      "step": 730
+    },
+    {
+      "epoch": 0.06415956510171672,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.002193,
+      "loss": 3.7622,
+      "step": 731
+    },
+    {
+      "epoch": 0.0642473346846192,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.002196,
+      "loss": 3.8252,
+      "step": 732
+    },
+    {
+      "epoch": 0.06433510426752169,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002199,
+      "loss": 3.7144,
+      "step": 733
+    },
+    {
+      "epoch": 0.06442287385042417,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002202,
+      "loss": 3.7085,
+      "step": 734
+    },
+    {
+      "epoch": 0.06451064343332666,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002205,
+      "loss": 3.7637,
+      "step": 735
+    },
+    {
+      "epoch": 0.06459841301622915,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.002208,
+      "loss": 3.7861,
+      "step": 736
+    },
+    {
+      "epoch": 0.06468618259913163,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002211,
+      "loss": 3.7471,
+      "step": 737
+    },
+    {
+      "epoch": 0.06477395218203412,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002214,
+      "loss": 3.748,
+      "step": 738
+    },
+    {
+      "epoch": 0.0648617217649366,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0022170000000000002,
+      "loss": 3.7354,
+      "step": 739
+    },
+    {
+      "epoch": 0.06494949134783909,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00222,
+      "loss": 3.6177,
+      "step": 740
+    },
+    {
+      "epoch": 0.06503726093074157,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.002223,
+      "loss": 3.7783,
+      "step": 741
+    },
+    {
+      "epoch": 0.06512503051364406,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.002226,
+      "loss": 3.7266,
+      "step": 742
+    },
+    {
+      "epoch": 0.06521280009654654,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.002229,
+      "loss": 3.7812,
+      "step": 743
+    },
+    {
+      "epoch": 0.06530056967944903,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.002232,
+      "loss": 3.6904,
+      "step": 744
+    },
+    {
+      "epoch": 0.0653883392623515,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002235,
+      "loss": 3.7261,
+      "step": 745
+    },
+    {
+      "epoch": 0.065476108845254,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.002238,
+      "loss": 3.7495,
+      "step": 746
+    },
+    {
+      "epoch": 0.06556387842815649,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.002241,
+      "loss": 3.7397,
+      "step": 747
+    },
+    {
+      "epoch": 0.06565164801105897,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002244,
+      "loss": 3.7744,
+      "step": 748
+    },
+    {
+      "epoch": 0.06573941759396146,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.002247,
+      "loss": 3.7305,
+      "step": 749
+    },
+    {
+      "epoch": 0.06582718717686394,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0022500000000000003,
+      "loss": 3.6714,
+      "step": 750
+    },
+    {
+      "epoch": 0.06591495675976643,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0022530000000000002,
+      "loss": 3.7026,
+      "step": 751
+    },
+    {
+      "epoch": 0.0660027263426689,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.002256,
+      "loss": 3.7227,
+      "step": 752
+    },
+    {
+      "epoch": 0.0660904959255714,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.002259,
+      "loss": 3.6636,
+      "step": 753
+    },
+    {
+      "epoch": 0.06617826550847387,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.002262,
+      "loss": 3.7505,
+      "step": 754
+    },
+    {
+      "epoch": 0.06626603509137637,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.002265,
+      "loss": 3.6899,
+      "step": 755
+    },
+    {
+      "epoch": 0.06635380467427884,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.002268,
+      "loss": 3.7573,
+      "step": 756
+    },
+    {
+      "epoch": 0.06644157425718134,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002271,
+      "loss": 3.7549,
+      "step": 757
+    },
+    {
+      "epoch": 0.06652934384008383,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002274,
+      "loss": 3.7002,
+      "step": 758
+    },
+    {
+      "epoch": 0.0666171134229863,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.002277,
+      "loss": 3.7773,
+      "step": 759
+    },
+    {
+      "epoch": 0.0667048830058888,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00228,
+      "loss": 3.769,
+      "step": 760
+    },
+    {
+      "epoch": 0.06679265258879127,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002283,
+      "loss": 3.6855,
+      "step": 761
+    },
+    {
+      "epoch": 0.06688042217169377,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0022860000000000003,
+      "loss": 3.7285,
+      "step": 762
+    },
+    {
+      "epoch": 0.06696819175459624,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0022890000000000002,
+      "loss": 3.7085,
+      "step": 763
+    },
+    {
+      "epoch": 0.06705596133749873,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002292,
+      "loss": 3.7393,
+      "step": 764
+    },
+    {
+      "epoch": 0.06714373092040121,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.002295,
+      "loss": 3.7393,
+      "step": 765
+    },
+    {
+      "epoch": 0.0672315005033037,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.002298,
+      "loss": 3.7456,
+      "step": 766
+    },
+    {
+      "epoch": 0.06731927008620618,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.002301,
+      "loss": 3.686,
+      "step": 767
+    },
+    {
+      "epoch": 0.06740703966910867,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.002304,
+      "loss": 3.7363,
+      "step": 768
+    },
+    {
+      "epoch": 0.06749480925201115,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.002307,
+      "loss": 3.686,
+      "step": 769
+    },
+    {
+      "epoch": 0.06758257883491364,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00231,
+      "loss": 3.7188,
+      "step": 770
+    },
+    {
+      "epoch": 0.06767034841781613,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.002313,
+      "loss": 3.6997,
+      "step": 771
+    },
+    {
+      "epoch": 0.06775811800071861,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.002316,
+      "loss": 3.7358,
+      "step": 772
+    },
+    {
+      "epoch": 0.0678458875836211,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0023190000000000003,
+      "loss": 3.7314,
+      "step": 773
+    },
+    {
+      "epoch": 0.06793365716652358,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.0023220000000000003,
+      "loss": 3.7417,
+      "step": 774
+    },
+    {
+      "epoch": 0.06802142674942607,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0023250000000000002,
+      "loss": 3.7671,
+      "step": 775
+    },
+    {
+      "epoch": 0.06810919633232855,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.002328,
+      "loss": 3.7241,
+      "step": 776
+    },
+    {
+      "epoch": 0.06819696591523104,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002331,
+      "loss": 3.748,
+      "step": 777
+    },
+    {
+      "epoch": 0.06828473549813352,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.002334,
+      "loss": 3.7295,
+      "step": 778
+    },
+    {
+      "epoch": 0.06837250508103601,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.002337,
+      "loss": 3.7632,
+      "step": 779
+    },
+    {
+      "epoch": 0.06846027466393849,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00234,
+      "loss": 3.708,
+      "step": 780
+    },
+    {
+      "epoch": 0.06854804424684098,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.002343,
+      "loss": 3.7207,
+      "step": 781
+    },
+    {
+      "epoch": 0.06863581382974347,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002346,
+      "loss": 3.6948,
+      "step": 782
+    },
+    {
+      "epoch": 0.06872358341264595,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002349,
+      "loss": 3.7104,
+      "step": 783
+    },
+    {
+      "epoch": 0.06881135299554844,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002352,
+      "loss": 3.6846,
+      "step": 784
+    },
+    {
+      "epoch": 0.06889912257845092,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0023550000000000003,
+      "loss": 3.7046,
+      "step": 785
+    },
+    {
+      "epoch": 0.06898689216135341,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0023580000000000003,
+      "loss": 3.6807,
+      "step": 786
+    },
+    {
+      "epoch": 0.06907466174425589,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0023610000000000003,
+      "loss": 3.7573,
+      "step": 787
+    },
+    {
+      "epoch": 0.06916243132715838,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002364,
+      "loss": 3.6899,
+      "step": 788
+    },
+    {
+      "epoch": 0.06925020091006086,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002367,
+      "loss": 3.7749,
+      "step": 789
+    },
+    {
+      "epoch": 0.06933797049296335,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.00237,
+      "loss": 3.6626,
+      "step": 790
+    },
+    {
+      "epoch": 0.06942574007586583,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.002373,
+      "loss": 3.7949,
+      "step": 791
+    },
+    {
+      "epoch": 0.06951350965876832,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.002376,
+      "loss": 3.6807,
+      "step": 792
+    },
+    {
+      "epoch": 0.06960127924167081,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.002379,
+      "loss": 3.6987,
+      "step": 793
+    },
+    {
+      "epoch": 0.06968904882457329,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.002382,
+      "loss": 3.7559,
+      "step": 794
+    },
+    {
+      "epoch": 0.06977681840747578,
+      "grad_norm": 0.375,
+      "learning_rate": 0.002385,
+      "loss": 3.6919,
+      "step": 795
+    },
+    {
+      "epoch": 0.06986458799037826,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0023880000000000004,
+      "loss": 3.6982,
+      "step": 796
+    },
+    {
+      "epoch": 0.06995235757328075,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0023910000000000003,
+      "loss": 3.7075,
+      "step": 797
+    },
+    {
+      "epoch": 0.07004012715618323,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0023940000000000003,
+      "loss": 3.6851,
+      "step": 798
+    },
+    {
+      "epoch": 0.07012789673908572,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0023970000000000003,
+      "loss": 3.6602,
+      "step": 799
+    },
+    {
+      "epoch": 0.0702156663219882,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0024000000000000002,
+      "loss": 3.7002,
+      "step": 800
+    },
+    {
+      "epoch": 0.07030343590489069,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002403,
+      "loss": 3.626,
+      "step": 801
+    },
+    {
+      "epoch": 0.07039120548779317,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002406,
+      "loss": 3.6548,
+      "step": 802
+    },
+    {
+      "epoch": 0.07047897507069566,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.002409,
+      "loss": 3.73,
+      "step": 803
+    },
+    {
+      "epoch": 0.07056674465359813,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.002412,
+      "loss": 3.6968,
+      "step": 804
+    },
+    {
+      "epoch": 0.07065451423650063,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.002415,
+      "loss": 3.7441,
+      "step": 805
+    },
+    {
+      "epoch": 0.07074228381940312,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.002418,
+      "loss": 3.7603,
+      "step": 806
+    },
+    {
+      "epoch": 0.0708300534023056,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.0024210000000000004,
+      "loss": 3.791,
+      "step": 807
+    },
+    {
+      "epoch": 0.07091782298520809,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0024240000000000004,
+      "loss": 3.7222,
+      "step": 808
+    },
+    {
+      "epoch": 0.07100559256811056,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0024270000000000003,
+      "loss": 3.6899,
+      "step": 809
+    },
+    {
+      "epoch": 0.07109336215101306,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0024300000000000003,
+      "loss": 3.7476,
+      "step": 810
+    },
+    {
+      "epoch": 0.07118113173391553,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0024330000000000003,
+      "loss": 3.7065,
+      "step": 811
+    },
+    {
+      "epoch": 0.07126890131681803,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0024360000000000002,
+      "loss": 3.7705,
+      "step": 812
+    },
+    {
+      "epoch": 0.0713566708997205,
+      "grad_norm": 3.5,
+      "learning_rate": 0.0024389999999999998,
+      "loss": 3.7329,
+      "step": 813
+    },
+    {
+      "epoch": 0.071444440482623,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0024419999999999997,
+      "loss": 3.7852,
+      "step": 814
+    },
+    {
+      "epoch": 0.07153221006552547,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0024449999999999997,
+      "loss": 3.7646,
+      "step": 815
+    },
+    {
+      "epoch": 0.07161997964842796,
+      "grad_norm": 3.765625,
+      "learning_rate": 0.002448,
+      "loss": 3.7314,
+      "step": 816
+    },
+    {
+      "epoch": 0.07170774923133046,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.002451,
+      "loss": 3.6938,
+      "step": 817
+    },
+    {
+      "epoch": 0.07179551881423293,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.002454,
+      "loss": 3.7383,
+      "step": 818
+    },
+    {
+      "epoch": 0.07188328839713543,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.002457,
+      "loss": 3.7593,
+      "step": 819
+    },
+    {
+      "epoch": 0.0719710579800379,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00246,
+      "loss": 3.6699,
+      "step": 820
+    },
+    {
+      "epoch": 0.0720588275629404,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.002463,
+      "loss": 3.752,
+      "step": 821
+    },
+    {
+      "epoch": 0.07214659714584287,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.002466,
+      "loss": 3.7798,
+      "step": 822
+    },
+    {
+      "epoch": 0.07223436672874536,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.002469,
+      "loss": 3.7441,
+      "step": 823
+    },
+    {
+      "epoch": 0.07232213631164784,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.002472,
+      "loss": 3.7021,
+      "step": 824
+    },
+    {
+      "epoch": 0.07240990589455033,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0024749999999999998,
+      "loss": 3.7612,
+      "step": 825
+    },
+    {
+      "epoch": 0.07249767547745281,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0024779999999999997,
+      "loss": 3.6611,
+      "step": 826
+    },
+    {
+      "epoch": 0.0725854450603553,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002481,
+      "loss": 3.6909,
+      "step": 827
+    },
+    {
+      "epoch": 0.0726732146432578,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002484,
+      "loss": 3.7026,
+      "step": 828
+    },
+    {
+      "epoch": 0.07276098422616027,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002487,
+      "loss": 3.6343,
+      "step": 829
+    },
+    {
+      "epoch": 0.07284875380906276,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00249,
+      "loss": 3.7246,
+      "step": 830
+    },
+    {
+      "epoch": 0.07293652339196524,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.002493,
+      "loss": 3.6538,
+      "step": 831
+    },
+    {
+      "epoch": 0.07302429297486773,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.002496,
+      "loss": 3.6812,
+      "step": 832
+    },
+    {
+      "epoch": 0.07311206255777021,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.002499,
+      "loss": 3.73,
+      "step": 833
+    },
+    {
+      "epoch": 0.0731998321406727,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.002502,
+      "loss": 3.7378,
+      "step": 834
+    },
+    {
+      "epoch": 0.07328760172357518,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.002505,
+      "loss": 3.7598,
+      "step": 835
+    },
+    {
+      "epoch": 0.07337537130647767,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.002508,
+      "loss": 3.7412,
+      "step": 836
+    },
+    {
+      "epoch": 0.07346314088938015,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0025109999999999998,
+      "loss": 3.7393,
+      "step": 837
+    },
+    {
+      "epoch": 0.07355091047228264,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0025139999999999997,
+      "loss": 3.7256,
+      "step": 838
+    },
+    {
+      "epoch": 0.07363868005518512,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002517,
+      "loss": 3.6777,
+      "step": 839
+    },
+    {
+      "epoch": 0.07372644963808761,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00252,
+      "loss": 3.6694,
+      "step": 840
+    },
+    {
+      "epoch": 0.0738142192209901,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.002523,
+      "loss": 3.6685,
+      "step": 841
+    },
+    {
+      "epoch": 0.07390198880389258,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002526,
+      "loss": 3.7007,
+      "step": 842
+    },
+    {
+      "epoch": 0.07398975838679507,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002529,
+      "loss": 3.7432,
+      "step": 843
+    },
+    {
+      "epoch": 0.07407752796969755,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.002532,
+      "loss": 3.6738,
+      "step": 844
+    },
+    {
+      "epoch": 0.07416529755260004,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.002535,
+      "loss": 3.646,
+      "step": 845
+    },
+    {
+      "epoch": 0.07425306713550252,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002538,
+      "loss": 3.6494,
+      "step": 846
+    },
+    {
+      "epoch": 0.07434083671840501,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.002541,
+      "loss": 3.6348,
+      "step": 847
+    },
+    {
+      "epoch": 0.07442860630130749,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.002544,
+      "loss": 3.7183,
+      "step": 848
+    },
+    {
+      "epoch": 0.07451637588420998,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002547,
+      "loss": 3.728,
+      "step": 849
+    },
+    {
+      "epoch": 0.07460414546711246,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00255,
+      "loss": 3.7329,
+      "step": 850
+    },
+    {
+      "epoch": 0.07469191505001495,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002553,
+      "loss": 3.6992,
+      "step": 851
+    },
+    {
+      "epoch": 0.07477968463291744,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002556,
+      "loss": 3.626,
+      "step": 852
+    },
+    {
+      "epoch": 0.07486745421581992,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002559,
+      "loss": 3.6987,
+      "step": 853
+    },
+    {
+      "epoch": 0.07495522379872241,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.002562,
+      "loss": 3.6509,
+      "step": 854
+    },
+    {
+      "epoch": 0.07504299338162489,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.002565,
+      "loss": 3.7505,
+      "step": 855
+    },
+    {
+      "epoch": 0.07513076296452738,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.002568,
+      "loss": 3.7656,
+      "step": 856
+    },
+    {
+      "epoch": 0.07521853254742986,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.002571,
+      "loss": 3.687,
+      "step": 857
+    },
+    {
+      "epoch": 0.07530630213033235,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.002574,
+      "loss": 3.7065,
+      "step": 858
+    },
+    {
+      "epoch": 0.07539407171323483,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.002577,
+      "loss": 3.71,
+      "step": 859
+    },
+    {
+      "epoch": 0.07548184129613732,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00258,
+      "loss": 3.7144,
+      "step": 860
+    },
+    {
+      "epoch": 0.0755696108790398,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0025830000000000002,
+      "loss": 3.6265,
+      "step": 861
+    },
+    {
+      "epoch": 0.07565738046194229,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.002586,
+      "loss": 3.6665,
+      "step": 862
+    },
+    {
+      "epoch": 0.07574515004484478,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.002589,
+      "loss": 3.6675,
+      "step": 863
+    },
+    {
+      "epoch": 0.07583291962774726,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002592,
+      "loss": 3.6807,
+      "step": 864
+    },
+    {
+      "epoch": 0.07592068921064975,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.002595,
+      "loss": 3.728,
+      "step": 865
+    },
+    {
+      "epoch": 0.07600845879355222,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002598,
+      "loss": 3.6689,
+      "step": 866
+    },
+    {
+      "epoch": 0.07609622837645472,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.002601,
+      "loss": 3.6865,
+      "step": 867
+    },
+    {
+      "epoch": 0.0761839979593572,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002604,
+      "loss": 3.6787,
+      "step": 868
+    },
+    {
+      "epoch": 0.07627176754225969,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.002607,
+      "loss": 3.6592,
+      "step": 869
+    },
+    {
+      "epoch": 0.07635953712516216,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00261,
+      "loss": 3.7295,
+      "step": 870
+    },
+    {
+      "epoch": 0.07644730670806466,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.002613,
+      "loss": 3.7124,
+      "step": 871
+    },
+    {
+      "epoch": 0.07653507629096713,
+      "grad_norm": 0.5,
+      "learning_rate": 0.002616,
+      "loss": 3.6597,
+      "step": 872
+    },
+    {
+      "epoch": 0.07662284587386962,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0026190000000000002,
+      "loss": 3.7861,
+      "step": 873
+    },
+    {
+      "epoch": 0.0767106154567721,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.002622,
+      "loss": 3.6782,
+      "step": 874
+    },
+    {
+      "epoch": 0.0767983850396746,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.002625,
+      "loss": 3.6807,
+      "step": 875
+    },
+    {
+      "epoch": 0.07688615462257709,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002628,
+      "loss": 3.6914,
+      "step": 876
+    },
+    {
+      "epoch": 0.07697392420547956,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002631,
+      "loss": 3.6938,
+      "step": 877
+    },
+    {
+      "epoch": 0.07706169378838205,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.002634,
+      "loss": 3.6724,
+      "step": 878
+    },
+    {
+      "epoch": 0.07714946337128453,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.002637,
+      "loss": 3.6084,
+      "step": 879
+    },
+    {
+      "epoch": 0.07723723295418702,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00264,
+      "loss": 3.6646,
+      "step": 880
+    },
+    {
+      "epoch": 0.0773250025370895,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.002643,
+      "loss": 3.6909,
+      "step": 881
+    },
+    {
+      "epoch": 0.077412772119992,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.002646,
+      "loss": 3.7378,
+      "step": 882
+    },
+    {
+      "epoch": 0.07750054170289447,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.002649,
+      "loss": 3.7207,
+      "step": 883
+    },
+    {
+      "epoch": 0.07758831128579696,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0026520000000000003,
+      "loss": 3.7061,
+      "step": 884
+    },
+    {
+      "epoch": 0.07767608086869944,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0026550000000000002,
+      "loss": 3.731,
+      "step": 885
+    },
+    {
+      "epoch": 0.07776385045160193,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.002658,
+      "loss": 3.6167,
+      "step": 886
+    },
+    {
+      "epoch": 0.07785162003450442,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.002661,
+      "loss": 3.6455,
+      "step": 887
+    },
+    {
+      "epoch": 0.0779393896174069,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.002664,
+      "loss": 3.6997,
+      "step": 888
+    },
+    {
+      "epoch": 0.07802715920030939,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.002667,
+      "loss": 3.6294,
+      "step": 889
+    },
+    {
+      "epoch": 0.07811492878321187,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00267,
+      "loss": 3.7241,
+      "step": 890
+    },
+    {
+      "epoch": 0.07820269836611436,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.002673,
+      "loss": 3.7808,
+      "step": 891
+    },
+    {
+      "epoch": 0.07829046794901684,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.002676,
+      "loss": 3.5918,
+      "step": 892
+    },
+    {
+      "epoch": 0.07837823753191933,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.002679,
+      "loss": 3.6763,
+      "step": 893
+    },
+    {
+      "epoch": 0.07846600711482181,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002682,
+      "loss": 3.7134,
+      "step": 894
+    },
+    {
+      "epoch": 0.0785537766977243,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.0026850000000000003,
+      "loss": 3.7212,
+      "step": 895
+    },
+    {
+      "epoch": 0.07864154628062678,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0026880000000000003,
+      "loss": 3.5791,
+      "step": 896
+    },
+    {
+      "epoch": 0.07872931586352927,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0026910000000000002,
+      "loss": 3.7041,
+      "step": 897
+    },
+    {
+      "epoch": 0.07881708544643176,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002694,
+      "loss": 3.7158,
+      "step": 898
+    },
+    {
+      "epoch": 0.07890485502933424,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002697,
+      "loss": 3.7212,
+      "step": 899
+    },
+    {
+      "epoch": 0.07899262461223673,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0027,
+      "loss": 3.604,
+      "step": 900
+    },
+    {
+      "epoch": 0.07908039419513921,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002703,
+      "loss": 3.7148,
+      "step": 901
+    },
+    {
+      "epoch": 0.0791681637780417,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002706,
+      "loss": 3.646,
+      "step": 902
+    },
+    {
+      "epoch": 0.07925593336094418,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002709,
+      "loss": 3.6235,
+      "step": 903
+    },
+    {
+      "epoch": 0.07934370294384667,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.002712,
+      "loss": 3.668,
+      "step": 904
+    },
+    {
+      "epoch": 0.07943147252674915,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002715,
+      "loss": 3.7256,
+      "step": 905
+    },
+    {
+      "epoch": 0.07951924210965164,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002718,
+      "loss": 3.6484,
+      "step": 906
+    },
+    {
+      "epoch": 0.07960701169255412,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0027210000000000003,
+      "loss": 3.6211,
+      "step": 907
+    },
+    {
+      "epoch": 0.07969478127545661,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0027240000000000003,
+      "loss": 3.7407,
+      "step": 908
+    },
+    {
+      "epoch": 0.07978255085835909,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0027270000000000003,
+      "loss": 3.6343,
+      "step": 909
+    },
+    {
+      "epoch": 0.07987032044126158,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0027300000000000002,
+      "loss": 3.6108,
+      "step": 910
+    },
+    {
+      "epoch": 0.07995809002416407,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.002733,
+      "loss": 3.6875,
+      "step": 911
+    },
+    {
+      "epoch": 0.08004585960706655,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.002736,
+      "loss": 3.585,
+      "step": 912
+    },
+    {
+      "epoch": 0.08013362918996904,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.002739,
+      "loss": 3.6724,
+      "step": 913
+    },
+    {
+      "epoch": 0.08022139877287152,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.002742,
+      "loss": 3.6455,
+      "step": 914
+    },
+    {
+      "epoch": 0.08030916835577401,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.002745,
+      "loss": 3.6523,
+      "step": 915
+    },
+    {
+      "epoch": 0.08039693793867649,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002748,
+      "loss": 3.6792,
+      "step": 916
+    },
+    {
+      "epoch": 0.08048470752157898,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002751,
+      "loss": 3.7769,
+      "step": 917
+    },
+    {
+      "epoch": 0.08057247710448145,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0027540000000000004,
+      "loss": 3.665,
+      "step": 918
+    },
+    {
+      "epoch": 0.08066024668738395,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.0027570000000000003,
+      "loss": 3.6743,
+      "step": 919
+    },
+    {
+      "epoch": 0.08074801627028642,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0027600000000000003,
+      "loss": 3.6426,
+      "step": 920
+    },
+    {
+      "epoch": 0.08083578585318892,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0027630000000000003,
+      "loss": 3.6484,
+      "step": 921
+    },
+    {
+      "epoch": 0.08092355543609141,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0027660000000000002,
+      "loss": 3.6919,
+      "step": 922
+    },
+    {
+      "epoch": 0.08101132501899388,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.002769,
+      "loss": 3.6587,
+      "step": 923
+    },
+    {
+      "epoch": 0.08109909460189638,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.002772,
+      "loss": 3.6606,
+      "step": 924
+    },
+    {
+      "epoch": 0.08118686418479885,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.002775,
+      "loss": 3.7056,
+      "step": 925
+    },
+    {
+      "epoch": 0.08127463376770135,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.002778,
+      "loss": 3.6943,
+      "step": 926
+    },
+    {
+      "epoch": 0.08136240335060382,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.002781,
+      "loss": 3.6484,
+      "step": 927
+    },
+    {
+      "epoch": 0.08145017293350632,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.002784,
+      "loss": 3.6758,
+      "step": 928
+    },
+    {
+      "epoch": 0.08153794251640879,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0027870000000000004,
+      "loss": 3.7661,
+      "step": 929
+    },
+    {
+      "epoch": 0.08162571209931128,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0027900000000000004,
+      "loss": 3.7485,
+      "step": 930
+    },
+    {
+      "epoch": 0.08171348168221376,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0027930000000000003,
+      "loss": 3.7007,
+      "step": 931
+    },
+    {
+      "epoch": 0.08180125126511625,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0027960000000000003,
+      "loss": 3.7114,
+      "step": 932
+    },
+    {
+      "epoch": 0.08188902084801875,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0027990000000000003,
+      "loss": 3.7549,
+      "step": 933
+    },
+    {
+      "epoch": 0.08197679043092122,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0028020000000000002,
+      "loss": 3.6338,
+      "step": 934
+    },
+    {
+      "epoch": 0.08206456001382371,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.002805,
+      "loss": 3.6816,
+      "step": 935
+    },
+    {
+      "epoch": 0.08215232959672619,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002808,
+      "loss": 3.6445,
+      "step": 936
+    },
+    {
+      "epoch": 0.08224009917962868,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.002811,
+      "loss": 3.6426,
+      "step": 937
+    },
+    {
+      "epoch": 0.08232786876253116,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.002814,
+      "loss": 3.6396,
+      "step": 938
+    },
+    {
+      "epoch": 0.08241563834543365,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002817,
+      "loss": 3.6426,
+      "step": 939
+    },
+    {
+      "epoch": 0.08250340792833613,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00282,
+      "loss": 3.6167,
+      "step": 940
+    },
+    {
+      "epoch": 0.08259117751123862,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002823,
+      "loss": 3.6108,
+      "step": 941
+    },
+    {
+      "epoch": 0.0826789470941411,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.002826,
+      "loss": 3.689,
+      "step": 942
+    },
+    {
+      "epoch": 0.08276671667704359,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.002829,
+      "loss": 3.6821,
+      "step": 943
+    },
+    {
+      "epoch": 0.08285448625994607,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.002832,
+      "loss": 3.6182,
+      "step": 944
+    },
+    {
+      "epoch": 0.08294225584284856,
+      "grad_norm": 4.0625,
+      "learning_rate": 0.002835,
+      "loss": 3.5884,
+      "step": 945
+    },
+    {
+      "epoch": 0.08303002542575105,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.002838,
+      "loss": 3.7207,
+      "step": 946
+    },
+    {
+      "epoch": 0.08311779500865353,
+      "grad_norm": 17.5,
+      "learning_rate": 0.0028409999999999998,
+      "loss": 4.4404,
+      "step": 947
+    },
+    {
+      "epoch": 0.08320556459155602,
+      "grad_norm": 6.59375,
+      "learning_rate": 0.0028439999999999997,
+      "loss": 3.8545,
+      "step": 948
+    },
+    {
+      "epoch": 0.0832933341744585,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.002847,
+      "loss": 3.7534,
+      "step": 949
+    },
+    {
+      "epoch": 0.08338110375736099,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00285,
+      "loss": 3.8262,
+      "step": 950
+    },
+    {
+      "epoch": 0.08346887334026347,
+      "grad_norm": 1.9921875,
+      "learning_rate": 0.002853,
+      "loss": 3.8135,
+      "step": 951
+    },
+    {
+      "epoch": 0.08355664292316596,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.002856,
+      "loss": 3.7271,
+      "step": 952
+    },
+    {
+      "epoch": 0.08364441250606844,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.002859,
+      "loss": 3.7886,
+      "step": 953
+    },
+    {
+      "epoch": 0.08373218208897093,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.002862,
+      "loss": 3.7295,
+      "step": 954
+    },
+    {
+      "epoch": 0.08381995167187341,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.002865,
+      "loss": 3.6191,
+      "step": 955
+    },
+    {
+      "epoch": 0.0839077212547759,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.002868,
+      "loss": 3.7207,
+      "step": 956
+    },
+    {
+      "epoch": 0.08399549083767839,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.002871,
+      "loss": 3.6973,
+      "step": 957
+    },
+    {
+      "epoch": 0.08408326042058087,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002874,
+      "loss": 3.7539,
+      "step": 958
+    },
+    {
+      "epoch": 0.08417103000348336,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.002877,
+      "loss": 3.7188,
+      "step": 959
+    },
+    {
+      "epoch": 0.08425879958638584,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0028799999999999997,
+      "loss": 3.73,
+      "step": 960
+    },
+    {
+      "epoch": 0.08434656916928833,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.002883,
+      "loss": 3.6167,
+      "step": 961
+    },
+    {
+      "epoch": 0.08443433875219081,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.002886,
+      "loss": 3.6514,
+      "step": 962
+    },
+    {
+      "epoch": 0.0845221083350933,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.002889,
+      "loss": 3.7183,
+      "step": 963
+    },
+    {
+      "epoch": 0.08460987791799578,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002892,
+      "loss": 3.7275,
+      "step": 964
+    },
+    {
+      "epoch": 0.08469764750089827,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002895,
+      "loss": 3.6123,
+      "step": 965
+    },
+    {
+      "epoch": 0.08478541708380075,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.002898,
+      "loss": 3.687,
+      "step": 966
+    },
+    {
+      "epoch": 0.08487318666670324,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.002901,
+      "loss": 3.6724,
+      "step": 967
+    },
+    {
+      "epoch": 0.08496095624960573,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.002904,
+      "loss": 3.7109,
+      "step": 968
+    },
+    {
+      "epoch": 0.0850487258325082,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.002907,
+      "loss": 3.5679,
+      "step": 969
+    },
+    {
+      "epoch": 0.0851364954154107,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00291,
+      "loss": 3.6958,
+      "step": 970
+    },
+    {
+      "epoch": 0.08522426499831318,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.002913,
+      "loss": 3.6597,
+      "step": 971
+    },
+    {
+      "epoch": 0.08531203458121567,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002916,
+      "loss": 3.7212,
+      "step": 972
+    },
+    {
+      "epoch": 0.08539980416411815,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.002919,
+      "loss": 3.6523,
+      "step": 973
+    },
+    {
+      "epoch": 0.08548757374702064,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002922,
+      "loss": 3.647,
+      "step": 974
+    },
+    {
+      "epoch": 0.08557534332992311,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.002925,
+      "loss": 3.6421,
+      "step": 975
+    },
+    {
+      "epoch": 0.0856631129128256,
+      "grad_norm": 17.0,
+      "learning_rate": 0.002928,
+      "loss": 3.7246,
+      "step": 976
+    },
+    {
+      "epoch": 0.08575088249572808,
+      "grad_norm": 1.125,
+      "learning_rate": 0.002931,
+      "loss": 3.6187,
+      "step": 977
+    },
+    {
+      "epoch": 0.08583865207863058,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.002934,
+      "loss": 3.7295,
+      "step": 978
+    },
+    {
+      "epoch": 0.08592642166153307,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.002937,
+      "loss": 3.7075,
+      "step": 979
+    },
+    {
+      "epoch": 0.08601419124443554,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00294,
+      "loss": 3.7109,
+      "step": 980
+    },
+    {
+      "epoch": 0.08610196082733804,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.002943,
+      "loss": 3.5815,
+      "step": 981
+    },
+    {
+      "epoch": 0.08618973041024051,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002946,
+      "loss": 3.6768,
+      "step": 982
+    },
+    {
+      "epoch": 0.086277499993143,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.0029490000000000002,
+      "loss": 3.6567,
+      "step": 983
+    },
+    {
+      "epoch": 0.08636526957604548,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002952,
+      "loss": 3.6489,
+      "step": 984
+    },
+    {
+      "epoch": 0.08645303915894798,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002955,
+      "loss": 3.709,
+      "step": 985
+    },
+    {
+      "epoch": 0.08654080874185045,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002958,
+      "loss": 3.6841,
+      "step": 986
+    },
+    {
+      "epoch": 0.08662857832475294,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.002961,
+      "loss": 3.6665,
+      "step": 987
+    },
+    {
+      "epoch": 0.08671634790765542,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002964,
+      "loss": 3.6299,
+      "step": 988
+    },
+    {
+      "epoch": 0.08680411749055791,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002967,
+      "loss": 3.5996,
+      "step": 989
+    },
+    {
+      "epoch": 0.08689188707346039,
+      "grad_norm": 24.75,
+      "learning_rate": 0.00297,
+      "loss": 3.7173,
+      "step": 990
+    },
+    {
+      "epoch": 0.08697965665636288,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.002973,
+      "loss": 3.6875,
+      "step": 991
+    },
+    {
+      "epoch": 0.08706742623926537,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.002976,
+      "loss": 3.7153,
+      "step": 992
+    },
+    {
+      "epoch": 0.08715519582216785,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.002979,
+      "loss": 3.6968,
+      "step": 993
+    },
+    {
+      "epoch": 0.08724296540507034,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.002982,
+      "loss": 3.7632,
+      "step": 994
+    },
+    {
+      "epoch": 0.08733073498797282,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0029850000000000002,
+      "loss": 3.6479,
+      "step": 995
+    },
+    {
+      "epoch": 0.08741850457087531,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.002988,
+      "loss": 3.6816,
+      "step": 996
+    },
+    {
+      "epoch": 0.08750627415377779,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.002991,
+      "loss": 3.6348,
+      "step": 997
+    },
+    {
+      "epoch": 0.08759404373668028,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.002994,
+      "loss": 3.6948,
+      "step": 998
+    },
+    {
+      "epoch": 0.08768181331958276,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002997,
+      "loss": 3.6421,
+      "step": 999
+    },
+    {
+      "epoch": 0.08776958290248525,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.003,
+      "loss": 3.584,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08776958290248525,
+      "eval_loss": 0.11389437317848206,
+      "eval_runtime": 106.9972,
+      "eval_samples_per_second": 137.845,
+      "eval_steps_per_second": 17.234,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08785735248538773,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002999999938323225,
+      "loss": 3.666,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08794512206829022,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0029999997532929035,
+      "loss": 3.6392,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08803289165119271,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0029999994449090546,
+      "loss": 3.6973,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08812066123409519,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029999990131717048,
+      "loss": 3.6641,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08820843081699768,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0029999984580808946,
+      "loss": 3.6177,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08829620039990016,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.002999997779636674,
+      "loss": 3.6816,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08838396998280265,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002999996977839106,
+      "loss": 3.6553,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08847173956570513,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.002999996052688263,
+      "loss": 3.7358,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08855950914860762,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002999995004184229,
+      "loss": 3.6221,
+      "step": 1009
+    },
+    {
+      "epoch": 0.0886472787315101,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.002999993832327101,
+      "loss": 3.6338,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08873504831441259,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002999992537116986,
+      "loss": 3.6362,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08882281789731507,
+      "grad_norm": 0.375,
+      "learning_rate": 0.002999991118554001,
+      "loss": 3.646,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08891058748021756,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002999989576638277,
+      "loss": 3.5947,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08899835706312005,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0029999879113699543,
+      "loss": 3.6348,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08908612664602253,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.002999986122749186,
+      "loss": 3.689,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08917389622892502,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0029999842107761338,
+      "loss": 3.6655,
+      "step": 1016
+    },
+    {
+      "epoch": 0.0892616658118275,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0029999821754509736,
+      "loss": 3.6191,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08934943539472999,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002999980016773891,
+      "loss": 3.6343,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08943720497763247,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002999977734745084,
+      "loss": 3.6401,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08952497456053496,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0029999753293647596,
+      "loss": 3.6064,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08961274414343744,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.002999972800633139,
+      "loss": 3.6758,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08970051372633993,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.002999970148550453,
+      "loss": 3.7158,
+      "step": 1022
+    },
+    {
+      "epoch": 0.0897882833092424,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002999967373116943,
+      "loss": 3.6021,
+      "step": 1023
+    },
+    {
+      "epoch": 0.0898760528921449,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029999644743328633,
+      "loss": 3.689,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08996382247504738,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002999961452198479,
+      "loss": 3.6118,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09005159205794987,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002999958306714066,
+      "loss": 3.6558,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09013936164085236,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0029999550378799113,
+      "loss": 3.6035,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09022713122375484,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0029999516456963144,
+      "loss": 3.6172,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09031490080665733,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0029999481301635844,
+      "loss": 3.6182,
+      "step": 1029
+    },
+    {
+      "epoch": 0.0904026703895598,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0029999444912820433,
+      "loss": 3.6841,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0904904399724623,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0029999407290520227,
+      "loss": 3.6899,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09057820955536477,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0029999368434738675,
+      "loss": 3.6436,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09066597913826727,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.002999932834547932,
+      "loss": 3.6831,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09075374872116974,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.002999928702274582,
+      "loss": 3.71,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09084151830407224,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002999924446654197,
+      "loss": 3.6523,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09092928788697471,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0029999200676871633,
+      "loss": 3.6362,
+      "step": 1036
+    },
+    {
+      "epoch": 0.0910170574698772,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029999155653738833,
+      "loss": 3.7197,
+      "step": 1037
+    },
+    {
+      "epoch": 0.0911048270527797,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029999109397147667,
+      "loss": 3.6558,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09119259663568217,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.002999906190710237,
+      "loss": 3.6309,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09128036621858467,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029999013183607277,
+      "loss": 3.647,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09136813580148714,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029998963226666848,
+      "loss": 3.6377,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09145590538438964,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029998912036285643,
+      "loss": 3.6372,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09154367496729211,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0029998859612468333,
+      "loss": 3.627,
+      "step": 1043
+    },
+    {
+      "epoch": 0.0916314445501946,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0029998805955219715,
+      "loss": 3.6733,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09171921413309708,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0029998751064544693,
+      "loss": 3.6108,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09180698371599957,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002999869494044828,
+      "loss": 3.6162,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09189475329890205,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.002999863758293561,
+      "loss": 3.6675,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09198252288180454,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002999857899201191,
+      "loss": 3.6772,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09207029246470703,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002999851916768254,
+      "loss": 3.6113,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09215806204760951,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029998458109952975,
+      "loss": 3.6729,
+      "step": 1050
+    },
+    {
+      "epoch": 0.092245831630512,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002999839581882879,
+      "loss": 3.6025,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09233360121341448,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002999833229431567,
+      "loss": 3.667,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09242137079631697,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002999826753641942,
+      "loss": 3.5806,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09250914037921945,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029998201545145963,
+      "loss": 3.6729,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09259690996212194,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029998134320501327,
+      "loss": 3.7144,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09268467954502442,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029998065862491662,
+      "loss": 3.6816,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09277244912792691,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0029997996171123205,
+      "loss": 3.7051,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09286021871082939,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0029997925246402334,
+      "loss": 3.5981,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09294798829373188,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002999785308833554,
+      "loss": 3.6689,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09303575787663436,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0029997779696929396,
+      "loss": 3.5659,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09312352745953685,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002999770507219062,
+      "loss": 3.6313,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09321129704243934,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.002999762921412603,
+      "loss": 3.6094,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09329906662534182,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002999755212274256,
+      "loss": 3.6592,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09338683620824431,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0029997473798047242,
+      "loss": 3.6226,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09347460579114679,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029997394240047245,
+      "loss": 3.6294,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09356237537404928,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002999731344874984,
+      "loss": 3.6064,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09365014495695176,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00299972314241624,
+      "loss": 3.5952,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09373791453985425,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0029997148166292413,
+      "loss": 3.6924,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09382568412275673,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002999706367514751,
+      "loss": 3.6343,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09391345370565922,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0029996977950735395,
+      "loss": 3.6719,
+      "step": 1070
+    },
+    {
+      "epoch": 0.0940012232885617,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.00299968909930639,
+      "loss": 3.6245,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09408899287146419,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002999680280214098,
+      "loss": 3.6543,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09417676245436668,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0029996713377974686,
+      "loss": 3.6558,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09426453203726916,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0029996622720573187,
+      "loss": 3.627,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09435230162017165,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0029996530829944773,
+      "loss": 3.6499,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09444007120307413,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0029996437706097836,
+      "loss": 3.6685,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09452784078597662,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0029996343349040894,
+      "loss": 3.541,
+      "step": 1077
+    },
+    {
+      "epoch": 0.0946156103688791,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0029996247758782555,
+      "loss": 3.6553,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09470337995178159,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.0029996150935331562,
+      "loss": 3.6484,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09479114953468407,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.002999605287869676,
+      "loss": 3.5708,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09487891911758656,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0029995953588887103,
+      "loss": 3.6343,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09496668870048904,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0029995853065911677,
+      "loss": 3.6763,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09505445828339153,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0029995751309779654,
+      "loss": 3.585,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09514222786629402,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0029995648320500345,
+      "loss": 3.6392,
+      "step": 1084
+    },
+    {
+      "epoch": 0.0952299974491965,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0029995544098083146,
+      "loss": 3.6509,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09531776703209899,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0029995438642537588,
+      "loss": 3.6807,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09540553661500147,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0029995331953873304,
+      "loss": 3.665,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09549330619790396,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0029995224032100037,
+      "loss": 3.7349,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09558107578080643,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.002999511487722767,
+      "loss": 3.5903,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09566884536370893,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002999500448926615,
+      "loss": 3.6128,
+      "step": 1090
+    },
+    {
+      "epoch": 0.0957566149466114,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0029994892868225576,
+      "loss": 3.6738,
+      "step": 1091
+    },
+    {
+      "epoch": 0.0958443845295139,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002999478001411615,
+      "loss": 3.6426,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09593215411241637,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002999466592694818,
+      "loss": 3.6011,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09601992369531887,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.002999455060673209,
+      "loss": 3.6729,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09610769327822134,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0029994434053478417,
+      "loss": 3.6567,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09619546286112383,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002999431626719781,
+      "loss": 3.5557,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09628323244402633,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.002999419724790104,
+      "loss": 3.5527,
+      "step": 1097
+    },
+    {
+      "epoch": 0.0963710020269288,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0029994076995598965,
+      "loss": 3.625,
+      "step": 1098
+    },
+    {
+      "epoch": 0.0964587716098313,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002999395551030259,
+      "loss": 3.6538,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09654654119273377,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0029993832792023,
+      "loss": 3.6621,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09663431077563626,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0029993708840771426,
+      "loss": 3.6631,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09672208035853874,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.002999358365655918,
+      "loss": 3.6914,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09680984994144123,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.002999345723939771,
+      "loss": 3.6865,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09689761952434371,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.0029993329589298556,
+      "loss": 3.6196,
+      "step": 1104
+    },
+    {
+      "epoch": 0.0969853891072462,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0029993200706273392,
+      "loss": 3.6572,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09707315869014868,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0029993070590333985,
+      "loss": 3.6235,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09716092827305117,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002999293924149224,
+      "loss": 3.6504,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09724869785595366,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.002999280665976014,
+      "loss": 3.5928,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09733646743885614,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0029992672845149805,
+      "loss": 3.6138,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09742423702175863,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.002999253779767347,
+      "loss": 3.6636,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09751200660466111,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0029992401517343467,
+      "loss": 3.6401,
+      "step": 1111
+    },
+    {
+      "epoch": 0.0975997761875636,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002999226400417225,
+      "loss": 3.6616,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09768754577046608,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002999212525817238,
+      "loss": 3.6187,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09777531535336857,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002999198527935654,
+      "loss": 3.5625,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09786308493627105,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0029991844067737525,
+      "loss": 3.6035,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09795085451917354,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029991701623328236,
+      "loss": 3.5967,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09803862410207602,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029991557946141677,
+      "loss": 3.6006,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09812639368497851,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002999141303619099,
+      "loss": 3.5942,
+      "step": 1118
+    },
+    {
+      "epoch": 0.098214163267881,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00299912668934894,
+      "loss": 3.5737,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09830193285078348,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002999111951805028,
+      "loss": 3.6978,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09838970243368597,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0029990970909887084,
+      "loss": 3.6426,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09847747201658845,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002999082106901339,
+      "loss": 3.6284,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09856524159949094,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029990669995442892,
+      "loss": 3.5952,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09865301118239342,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.00299905176891894,
+      "loss": 3.6245,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09874078076529591,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002999036415026683,
+      "loss": 3.6714,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09882855034819839,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029990209378689195,
+      "loss": 3.582,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09891631993110088,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002999005337447066,
+      "loss": 3.5928,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09900408951400336,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002998989613762546,
+      "loss": 3.6802,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09909185909690585,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002998973766816797,
+      "loss": 3.6211,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09917962867980833,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002998957796611268,
+      "loss": 3.6274,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09926739826271082,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0029989417031474164,
+      "loss": 3.5869,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09935516784561331,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0029989254864267135,
+      "loss": 3.5903,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09944293742851579,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0029989091464506417,
+      "loss": 3.5366,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09953070701141828,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.002998892683220693,
+      "loss": 3.5337,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09961847659432076,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.002998876096738373,
+      "loss": 3.6152,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09970624617722325,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0029988593870051957,
+      "loss": 3.6064,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09979401576012573,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0029988425540226887,
+      "loss": 3.6045,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09988178534302822,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0029988255977923908,
+      "loss": 3.603,
+      "step": 1138
+    },
+    {
+      "epoch": 0.0999695549259307,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00299880851831585,
+      "loss": 3.5308,
+      "step": 1139
+    },
+    {
+      "epoch": 0.10005732450883319,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0029987913155946275,
+      "loss": 3.543,
+      "step": 1140
+    },
+    {
+      "epoch": 0.10014509409173566,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.002998773989630295,
+      "loss": 3.6001,
+      "step": 1141
+    },
+    {
+      "epoch": 0.10023286367463816,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.002998756540424436,
+      "loss": 3.5747,
+      "step": 1142
+    },
+    {
+      "epoch": 0.10032063325754065,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002998738967978644,
+      "loss": 3.563,
+      "step": 1143
+    },
+    {
+      "epoch": 0.10040840284044313,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0029987212722945265,
+      "loss": 3.6367,
+      "step": 1144
+    },
+    {
+      "epoch": 0.10049617242334562,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002998703453373699,
+      "loss": 3.6211,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1005839420062481,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002998685511217789,
+      "loss": 3.5776,
+      "step": 1146
+    },
+    {
+      "epoch": 0.10067171158915059,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002998667445828438,
+      "loss": 3.624,
+      "step": 1147
+    },
+    {
+      "epoch": 0.10075948117205306,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.002998649257207294,
+      "loss": 3.6562,
+      "step": 1148
+    },
+    {
+      "epoch": 0.10084725075495556,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0029986309453560217,
+      "loss": 3.5806,
+      "step": 1149
+    },
+    {
+      "epoch": 0.10093502033785803,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002998612510276293,
+      "loss": 3.5264,
+      "step": 1150
+    },
+    {
+      "epoch": 0.10102278992076053,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0029985939519697915,
+      "loss": 3.6152,
+      "step": 1151
+    },
+    {
+      "epoch": 0.101110559503663,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0029985752704382146,
+      "loss": 3.6265,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1011983290865655,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0029985564656832683,
+      "loss": 3.6846,
+      "step": 1153
+    },
+    {
+      "epoch": 0.10128609866946799,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002998537537706671,
+      "loss": 3.5967,
+      "step": 1154
+    },
+    {
+      "epoch": 0.10137386825237046,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.002998518486510153,
+      "loss": 3.707,
+      "step": 1155
+    },
+    {
+      "epoch": 0.10146163783527296,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0029984993120954536,
+      "loss": 3.5596,
+      "step": 1156
+    },
+    {
+      "epoch": 0.10154940741817543,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.002998480014464326,
+      "loss": 3.5923,
+      "step": 1157
+    },
+    {
+      "epoch": 0.10163717700107792,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.002998460593618533,
+      "loss": 3.5542,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1017249465839804,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.002998441049559849,
+      "loss": 3.6089,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1018127161668829,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0029984213822900605,
+      "loss": 3.605,
+      "step": 1160
+    },
+    {
+      "epoch": 0.10190048574978537,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0029984015918109634,
+      "loss": 3.5796,
+      "step": 1161
+    },
+    {
+      "epoch": 0.10198825533268786,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002998381678124367,
+      "loss": 3.6855,
+      "step": 1162
+    },
+    {
+      "epoch": 0.10207602491559034,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0029983616412320906,
+      "loss": 3.5581,
+      "step": 1163
+    },
+    {
+      "epoch": 0.10216379449849283,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029983414811359647,
+      "loss": 3.603,
+      "step": 1164
+    },
+    {
+      "epoch": 0.10225156408139531,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029983211978378325,
+      "loss": 3.6245,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1023393336642978,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0029983007913395457,
+      "loss": 3.6157,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1024271032472003,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.00299828026164297,
+      "loss": 3.5205,
+      "step": 1167
+    },
+    {
+      "epoch": 0.10251487283010277,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.002998259608749981,
+      "loss": 3.6201,
+      "step": 1168
+    },
+    {
+      "epoch": 0.10260264241300526,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0029982388326624657,
+      "loss": 3.5986,
+      "step": 1169
+    },
+    {
+      "epoch": 0.10269041199590774,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002998217933382323,
+      "loss": 3.6548,
+      "step": 1170
+    },
+    {
+      "epoch": 0.10277818157881023,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.002998196910911461,
+      "loss": 3.5962,
+      "step": 1171
+    },
+    {
+      "epoch": 0.10286595116171271,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029981757652518025,
+      "loss": 3.5781,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1029537207446152,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.002998154496405279,
+      "loss": 3.5776,
+      "step": 1173
+    },
+    {
+      "epoch": 0.10304149032751768,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0029981331043738334,
+      "loss": 3.5469,
+      "step": 1174
+    },
+    {
+      "epoch": 0.10312925991042017,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0029981115891594213,
+      "loss": 3.5669,
+      "step": 1175
+    },
+    {
+      "epoch": 0.10321702949332265,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0029980899507640072,
+      "loss": 3.6274,
+      "step": 1176
+    },
+    {
+      "epoch": 0.10330479907622514,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002998068189189569,
+      "loss": 3.5752,
+      "step": 1177
+    },
+    {
+      "epoch": 0.10339256865912763,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0029980463044380957,
+      "loss": 3.6768,
+      "step": 1178
+    },
+    {
+      "epoch": 0.10348033824203011,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0029980242965115867,
+      "loss": 3.6045,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1035681078249326,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002998002165412052,
+      "loss": 3.6328,
+      "step": 1180
+    },
+    {
+      "epoch": 0.10365587740783508,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002997979911141515,
+      "loss": 3.6372,
+      "step": 1181
+    },
+    {
+      "epoch": 0.10374364699073757,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0029979575337020082,
+      "loss": 3.5879,
+      "step": 1182
+    },
+    {
+      "epoch": 0.10383141657364005,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002997935033095577,
+      "loss": 3.6333,
+      "step": 1183
+    },
+    {
+      "epoch": 0.10391918615654254,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002997912409324277,
+      "loss": 3.5981,
+      "step": 1184
+    },
+    {
+      "epoch": 0.10400695573944502,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029978896623901754,
+      "loss": 3.5879,
+      "step": 1185
+    },
+    {
+      "epoch": 0.10409472532234751,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0029978667922953505,
+      "loss": 3.647,
+      "step": 1186
+    },
+    {
+      "epoch": 0.10418249490524999,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0029978437990418926,
+      "loss": 3.5698,
+      "step": 1187
+    },
+    {
+      "epoch": 0.10427026448815248,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002997820682631902,
+      "loss": 3.5815,
+      "step": 1188
+    },
+    {
+      "epoch": 0.10435803407105497,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002997797443067491,
+      "loss": 3.5591,
+      "step": 1189
+    },
+    {
+      "epoch": 0.10444580365395745,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0029977740803507837,
+      "loss": 3.5732,
+      "step": 1190
+    },
+    {
+      "epoch": 0.10453357323685994,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0029977505944839143,
+      "loss": 3.5874,
+      "step": 1191
+    },
+    {
+      "epoch": 0.10462134281976242,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002997726985469029,
+      "loss": 3.5825,
+      "step": 1192
+    },
+    {
+      "epoch": 0.10470911240266491,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002997703253308285,
+      "loss": 3.6489,
+      "step": 1193
+    },
+    {
+      "epoch": 0.10479688198556739,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00299767939800385,
+      "loss": 3.5723,
+      "step": 1194
+    },
+    {
+      "epoch": 0.10488465156846988,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029976554195579046,
+      "loss": 3.6572,
+      "step": 1195
+    },
+    {
+      "epoch": 0.10497242115137236,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029976313179726397,
+      "loss": 3.6128,
+      "step": 1196
+    },
+    {
+      "epoch": 0.10506019073427485,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002997607093250258,
+      "loss": 3.5068,
+      "step": 1197
+    },
+    {
+      "epoch": 0.10514796031717732,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0029975827453929717,
+      "loss": 3.6265,
+      "step": 1198
+    },
+    {
+      "epoch": 0.10523572990007982,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.002997558274403006,
+      "loss": 3.5518,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1053234994829823,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0029975336802825978,
+      "loss": 3.6333,
+      "step": 1200
+    },
+    {
+      "epoch": 0.10541126906588479,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0029975089630339936,
+      "loss": 3.6504,
+      "step": 1201
+    },
+    {
+      "epoch": 0.10549903864878728,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0029974841226594515,
+      "loss": 3.6113,
+      "step": 1202
+    },
+    {
+      "epoch": 0.10558680823168975,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.002997459159161242,
+      "loss": 3.5811,
+      "step": 1203
+    },
+    {
+      "epoch": 0.10567457781459225,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029974340725416456,
+      "loss": 3.6025,
+      "step": 1204
+    },
+    {
+      "epoch": 0.10576234739749472,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002997408862802955,
+      "loss": 3.6377,
+      "step": 1205
+    },
+    {
+      "epoch": 0.10585011698039722,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0029973835299474737,
+      "loss": 3.6074,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1059378865632997,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029973580739775154,
+      "loss": 3.5952,
+      "step": 1207
+    },
+    {
+      "epoch": 0.10602565614620219,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029973324948954073,
+      "loss": 3.6118,
+      "step": 1208
+    },
+    {
+      "epoch": 0.10611342572910466,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.002997306792703486,
+      "loss": 3.646,
+      "step": 1209
+    },
+    {
+      "epoch": 0.10620119531200715,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0029972809674041,
+      "loss": 3.5938,
+      "step": 1210
+    },
+    {
+      "epoch": 0.10628896489490963,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0029972550189996095,
+      "loss": 3.6284,
+      "step": 1211
+    },
+    {
+      "epoch": 0.10637673447781212,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.002997228947492385,
+      "loss": 3.5811,
+      "step": 1212
+    },
+    {
+      "epoch": 0.10646450406071462,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.002997202752884809,
+      "loss": 3.6328,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1065522736436171,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002997176435179275,
+      "loss": 3.6177,
+      "step": 1214
+    },
+    {
+      "epoch": 0.10664004322651958,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.0029971499943781876,
+      "loss": 3.5679,
+      "step": 1215
+    },
+    {
+      "epoch": 0.10672781280942206,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0029971234304839627,
+      "loss": 3.5225,
+      "step": 1216
+    },
+    {
+      "epoch": 0.10681558239232455,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029970967434990274,
+      "loss": 3.5776,
+      "step": 1217
+    },
+    {
+      "epoch": 0.10690335197522703,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0029970699334258212,
+      "loss": 3.627,
+      "step": 1218
+    },
+    {
+      "epoch": 0.10699112155812952,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.002997043000266792,
+      "loss": 3.541,
+      "step": 1219
+    },
+    {
+      "epoch": 0.107078891141032,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0029970159440244025,
+      "loss": 3.6104,
+      "step": 1220
+    },
+    {
+      "epoch": 0.10716666072393449,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002996988764701124,
+      "loss": 3.5171,
+      "step": 1221
+    },
+    {
+      "epoch": 0.10725443030683697,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00299696146229944,
+      "loss": 3.6025,
+      "step": 1222
+    },
+    {
+      "epoch": 0.10734219988973946,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.002996934036821845,
+      "loss": 3.5688,
+      "step": 1223
+    },
+    {
+      "epoch": 0.10742996947264195,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029969064882708458,
+      "loss": 3.6362,
+      "step": 1224
+    },
+    {
+      "epoch": 0.10751773905554443,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002996878816648959,
+      "loss": 3.6104,
+      "step": 1225
+    },
+    {
+      "epoch": 0.10760550863844692,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0029968510219587135,
+      "loss": 3.6621,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1076932782213494,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.002996823104202648,
+      "loss": 3.5444,
+      "step": 1227
+    },
+    {
+      "epoch": 0.10778104780425189,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.002996795063383314,
+      "loss": 3.5771,
+      "step": 1228
+    },
+    {
+      "epoch": 0.10786881738715437,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.002996766899503274,
+      "loss": 3.603,
+      "step": 1229
+    },
+    {
+      "epoch": 0.10795658697005686,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0029967386125651007,
+      "loss": 3.6382,
+      "step": 1230
+    },
+    {
+      "epoch": 0.10804435655295934,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029967102025713795,
+      "loss": 3.5034,
+      "step": 1231
+    },
+    {
+      "epoch": 0.10813212613586183,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029966816695247062,
+      "loss": 3.6187,
+      "step": 1232
+    },
+    {
+      "epoch": 0.10821989571876431,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0029966530134276875,
+      "loss": 3.5845,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1083076653016668,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029966242342829423,
+      "loss": 3.5747,
+      "step": 1234
+    },
+    {
+      "epoch": 0.10839543488456928,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0029965953320931,
+      "loss": 3.5547,
+      "step": 1235
+    },
+    {
+      "epoch": 0.10848320446747177,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0029965663068608006,
+      "loss": 3.5171,
+      "step": 1236
+    },
+    {
+      "epoch": 0.10857097405037426,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002996537158588698,
+      "loss": 3.5898,
+      "step": 1237
+    },
+    {
+      "epoch": 0.10865874363327674,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.002996507887279454,
+      "loss": 3.6333,
+      "step": 1238
+    },
+    {
+      "epoch": 0.10874651321617923,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002996478492935744,
+      "loss": 3.6445,
+      "step": 1239
+    },
+    {
+      "epoch": 0.10883428279908171,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0029964489755602546,
+      "loss": 3.6631,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1089220523819842,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002996419335155681,
+      "loss": 3.5977,
+      "step": 1241
+    },
+    {
+      "epoch": 0.10900982196488668,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002996389571724733,
+      "loss": 3.583,
+      "step": 1242
+    },
+    {
+      "epoch": 0.10909759154778917,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029963596852701294,
+      "loss": 3.6157,
+      "step": 1243
+    },
+    {
+      "epoch": 0.10918536113069165,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029963296757946015,
+      "loss": 3.5708,
+      "step": 1244
+    },
+    {
+      "epoch": 0.10927313071359414,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0029962995433008913,
+      "loss": 3.626,
+      "step": 1245
+    },
+    {
+      "epoch": 0.10936090029649662,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0029962692877917516,
+      "loss": 3.5923,
+      "step": 1246
+    },
+    {
+      "epoch": 0.10944866987939911,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0029962389092699476,
+      "loss": 3.5591,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1095364394623016,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0029962084077382544,
+      "loss": 3.543,
+      "step": 1248
+    },
+    {
+      "epoch": 0.10962420904520408,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029961777831994598,
+      "loss": 3.6606,
+      "step": 1249
+    },
+    {
+      "epoch": 0.10971197862810657,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029961470356563618,
+      "loss": 3.5112,
+      "step": 1250
+    },
+    {
+      "epoch": 0.10979974821100905,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.002996116165111769,
+      "loss": 3.5693,
+      "step": 1251
+    },
+    {
+      "epoch": 0.10988751779391154,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0029960851715685032,
+      "loss": 3.5693,
+      "step": 1252
+    },
+    {
+      "epoch": 0.10997528737681402,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002996054055029396,
+      "loss": 3.5571,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11006305695971651,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.002996022815497291,
+      "loss": 3.5488,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11015082654261898,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002995991452975042,
+      "loss": 3.5732,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11023859612552148,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002995959967465515,
+      "loss": 3.5234,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11032636570842395,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029959283589715874,
+      "loss": 3.6211,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11041413529132645,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0029958966274961463,
+      "loss": 3.6626,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11050190487422894,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.002995864773042092,
+      "loss": 3.5376,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11058967445713141,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.0029958327956123346,
+      "loss": 3.5283,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1106774440400339,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.002995800695209797,
+      "loss": 3.6211,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11076521362293638,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0029957684718374105,
+      "loss": 3.5449,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11085298320583888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.002995736125498121,
+      "loss": 3.6074,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11094075278874135,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0029957036561948833,
+      "loss": 3.6323,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11102852237164385,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0029956710639306647,
+      "loss": 3.6245,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11111629195454632,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0029956383487084424,
+      "loss": 3.6138,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11120406153744881,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0029956055105312073,
+      "loss": 3.624,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11129183112035129,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.002995572549401958,
+      "loss": 3.5542,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11137960070325378,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0029955394653237077,
+      "loss": 3.6138,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11146737028615626,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002995506258299479,
+      "loss": 3.624,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11155513986905875,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.002995472928332306,
+      "loss": 3.6841,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11164290945196124,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0029954394754252334,
+      "loss": 3.6113,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11173067903486372,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002995405899581319,
+      "loss": 3.5962,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11181844861776621,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029953722008036303,
+      "loss": 3.5801,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11190621820066869,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002995338379095247,
+      "loss": 3.6934,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11199398778357118,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029953044344592588,
+      "loss": 3.6592,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11208175736647366,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0029952703668987674,
+      "loss": 3.5923,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11216952694937615,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0029952361764168855,
+      "loss": 3.6406,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11225729653227863,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029952018630167382,
+      "loss": 3.6167,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11234506611518112,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029951674267014597,
+      "loss": 3.5703,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1124328356980836,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002995132867474197,
+      "loss": 3.5942,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11252060528098609,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002995098185338108,
+      "loss": 3.5049,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11260837486388858,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0029950633802963613,
+      "loss": 3.5859,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11269614444679106,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.002995028452352138,
+      "loss": 3.542,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11278391402969355,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029949934015086285,
+      "loss": 3.5649,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11287168361259603,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0029949582277690357,
+      "loss": 3.5415,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11295945319549852,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002994922931136574,
+      "loss": 3.6206,
+      "step": 1287
+    },
+    {
+      "epoch": 0.113047222778401,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.002994887511614468,
+      "loss": 3.5449,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11313499236130349,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002994851969205955,
+      "loss": 3.5884,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11322276194420597,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029948163039142817,
+      "loss": 3.5635,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11331053152710846,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.002994780515742708,
+      "loss": 3.5693,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11339830111001094,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0029947446046945025,
+      "loss": 3.5449,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11348607069291343,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002994708570772947,
+      "loss": 3.5522,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11357384027581592,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002994672413981335,
+      "loss": 3.5913,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1136616098587184,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0029946361343229696,
+      "loss": 3.5298,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11374937944162089,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029945997318011654,
+      "loss": 3.5273,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11383714902452337,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002994563206419249,
+      "loss": 3.5654,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11392491860742586,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0029945265581805577,
+      "loss": 3.6089,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11401268819032834,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029944897870884403,
+      "loss": 3.5444,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11410045777323083,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0029944528931462565,
+      "loss": 3.6699,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1141882273561333,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.002994415876357378,
+      "loss": 3.5933,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1142759969390358,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0029943787367251863,
+      "loss": 3.5381,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11436376652193828,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029943414742530754,
+      "loss": 3.5913,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11445153610484077,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00299430408894445,
+      "loss": 3.5298,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11453930568774325,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002994266580802726,
+      "loss": 3.522,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11462707527064574,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.002994228949831331,
+      "loss": 3.6094,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11471484485354823,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002994191196033703,
+      "loss": 3.5781,
+      "step": 1307
+    },
+    {
+      "epoch": 0.1148026144364507,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002994153319413292,
+      "loss": 3.6025,
+      "step": 1308
+    },
+    {
+      "epoch": 0.1148903840193532,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002994115319973559,
+      "loss": 3.5522,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11497815360225568,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002994077197717975,
+      "loss": 3.5171,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11506592318515817,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029940389526500254,
+      "loss": 3.6226,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11515369276806064,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002994000584773203,
+      "loss": 3.6172,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11524146235096314,
+      "grad_norm": 0.625,
+      "learning_rate": 0.002993962094091014,
+      "loss": 3.5581,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11532923193386561,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.002993923480606976,
+      "loss": 3.6191,
+      "step": 1314
+    },
+    {
+      "epoch": 0.1154170015167681,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0029938847443246167,
+      "loss": 3.5308,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11550477109967058,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0029938458852474758,
+      "loss": 3.5537,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11559254068257308,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0029938069033791036,
+      "loss": 3.5625,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11568031026547557,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029937677987230626,
+      "loss": 3.5498,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11576807984837804,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0029937285712829252,
+      "loss": 3.5146,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11585584943128054,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029936892210622765,
+      "loss": 3.6274,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11594361901418301,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0029936497480647114,
+      "loss": 3.5659,
+      "step": 1321
+    },
+    {
+      "epoch": 0.1160313885970855,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002993610152293837,
+      "loss": 3.5234,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11611915817998798,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0029935704337532712,
+      "loss": 3.5601,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11620692776289047,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002993530592446643,
+      "loss": 3.6328,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11629469734579295,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002993490628377593,
+      "loss": 3.5879,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11638246692869544,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0029934505415497733,
+      "loss": 3.6245,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11647023651159792,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0029934103319668457,
+      "loss": 3.5278,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11655800609450041,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0029933699996324854,
+      "loss": 3.582,
+      "step": 1328
+    },
+    {
+      "epoch": 0.1166457756774029,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002993329544550377,
+      "loss": 3.5747,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11673354526030538,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002993288966724217,
+      "loss": 3.4795,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11682131484320787,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029932482661577135,
+      "loss": 3.5498,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11690908442611035,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002993207442854585,
+      "loss": 3.564,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11699685400901284,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002993166496818562,
+      "loss": 3.5195,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11708462359191532,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029931254280533854,
+      "loss": 3.5513,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11717239317481781,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002993084236562809,
+      "loss": 3.5596,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11726016275772029,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002993042922350595,
+      "loss": 3.4824,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11734793234062278,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0029930014854205192,
+      "loss": 3.644,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11743570192352526,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029929599257763674,
+      "loss": 3.5879,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11752347150642775,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029929182434219378,
+      "loss": 3.561,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11761124108933023,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029928764383610382,
+      "loss": 3.6216,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11769901067223272,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002992834510597489,
+      "loss": 3.5942,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11778678025513521,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.002992792460135121,
+      "loss": 3.5894,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11787454983803769,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002992750286977777,
+      "loss": 3.5176,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11796231942094018,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029927079911293095,
+      "loss": 3.5615,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11805008900384266,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.002992665572593584,
+      "loss": 3.5815,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11813785858674515,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029926230313744766,
+      "loss": 3.5918,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11822562816964763,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0029925803674758734,
+      "loss": 3.6523,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11831339775255012,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0029925375809016735,
+      "loss": 3.6191,
+      "step": 1348
+    },
+    {
+      "epoch": 0.1184011673354526,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002992494671655786,
+      "loss": 3.5571,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11848893691835509,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029924516397421322,
+      "loss": 3.4976,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11857670650125757,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.002992408485164644,
+      "loss": 3.5337,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11866447608416006,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0029923652079272643,
+      "loss": 3.4746,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11875224566706255,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.002992321808033947,
+      "loss": 3.5386,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11884001524996503,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.002992278285488658,
+      "loss": 3.5322,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11892778483286752,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002992234640295375,
+      "loss": 3.6255,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11901555441577,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.002992190872458085,
+      "loss": 3.5942,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11910332399867249,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0029921469819807873,
+      "loss": 3.5869,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11919109358157497,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0029921029688674923,
+      "loss": 3.6313,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11927886316447746,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0029920588331222224,
+      "loss": 3.5264,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11936663274737994,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002992014574749009,
+      "loss": 3.5566,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11945440233028243,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.002991970193751898,
+      "loss": 3.5415,
+      "step": 1361
+    },
+    {
+      "epoch": 0.1195421719131849,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0029919256901349426,
+      "loss": 3.5864,
+      "step": 1362
+    },
+    {
+      "epoch": 0.1196299414960874,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00299188106390221,
+      "loss": 3.5903,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11971771107898989,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0029918363150577784,
+      "loss": 3.5146,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11980548066189237,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.002991791443605736,
+      "loss": 3.5942,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11989325024479486,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029917464495501835,
+      "loss": 3.5464,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11998101982769734,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002991701332895231,
+      "loss": 3.5337,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12006878941059983,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029916560936450024,
+      "loss": 3.6211,
+      "step": 1368
+    },
+    {
+      "epoch": 0.1201565589935023,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029916107318036303,
+      "loss": 3.6074,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1202443285764048,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00299156524737526,
+      "loss": 3.5493,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12033209815930727,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0029915196403640467,
+      "loss": 3.5312,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12041986774220977,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002991473910774159,
+      "loss": 3.5215,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12050763732511224,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0029914280586097746,
+      "loss": 3.6108,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12059540690801474,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002991382083875083,
+      "loss": 3.5933,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12068317649091721,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029913359865742857,
+      "loss": 3.5269,
+      "step": 1375
+    },
+    {
+      "epoch": 0.1207709460738197,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029912897667115942,
+      "loss": 3.5366,
+      "step": 1376
+    },
+    {
+      "epoch": 0.1208587156567222,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029912434242912317,
+      "loss": 3.5903,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12094648523962467,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0029911969593174326,
+      "loss": 3.5698,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12103425482252717,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029911503717944435,
+      "loss": 3.4722,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12112202440542964,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00299110366172652,
+      "loss": 3.5449,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12120979398833213,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029910568291179303,
+      "loss": 3.5376,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12129756357123461,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029910098739729545,
+      "loss": 3.5933,
+      "step": 1382
+    },
+    {
+      "epoch": 0.1213853331541371,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0029909627962958828,
+      "loss": 3.5015,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12147310273703958,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002990915596091016,
+      "loss": 3.5767,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12156087231994207,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002990868273362668,
+      "loss": 3.5503,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12164864190284455,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029908208281151615,
+      "loss": 3.5,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12173641148574704,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0029907732603528334,
+      "loss": 3.54,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12182418106864953,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029907255700800284,
+      "loss": 3.6021,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12191195065155201,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0029906777573011053,
+      "loss": 3.5884,
+      "step": 1389
+    },
+    {
+      "epoch": 0.1219997202344545,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002990629822020433,
+      "loss": 3.5459,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12208748981735698,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029905817642423907,
+      "loss": 3.6094,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12217525940025947,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002990533583971369,
+      "loss": 3.5342,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12226302898316195,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029904852812117724,
+      "loss": 3.6021,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12235079856606444,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0029904368559680126,
+      "loss": 3.5225,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12243856814896692,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029903883082445155,
+      "loss": 3.5171,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12252633773186941,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029903396380457164,
+      "loss": 3.5649,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12261410731477189,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.002990290845376062,
+      "loss": 3.54,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12270187689767438,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0029902419302400116,
+      "loss": 3.5098,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12278964648057687,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.002990192892642034,
+      "loss": 3.5063,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12287741606347935,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029901437325866114,
+      "loss": 3.5234,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12296518564638184,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0029900944500782334,
+      "loss": 3.5352,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12305295522928432,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0029900450451214046,
+      "loss": 3.646,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12314072481218681,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.002989995517720639,
+      "loss": 3.5181,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12322849439508929,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0029899458678804623,
+      "loss": 3.541,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12331626397799178,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00298989609560541,
+      "loss": 3.5356,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12340403356089426,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0029898462009000314,
+      "loss": 3.5913,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12349180314379675,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002989796183768885,
+      "loss": 3.5913,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12357957272669923,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029897460442165407,
+      "loss": 3.5898,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12366734230960172,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029896957822475806,
+      "loss": 3.5513,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1237551118925042,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0029896453978665964,
+      "loss": 3.5908,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12384288147540669,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002989594891078192,
+      "loss": 3.6099,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12393065105830918,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029895442618869832,
+      "loss": 3.4771,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12401842064121166,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029894935102975956,
+      "loss": 3.5889,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12410619022411415,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029894426363146665,
+      "loss": 3.5103,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12419395980701663,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.002989391639942844,
+      "loss": 3.5928,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12428172938991912,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0029893405211867887,
+      "loss": 3.5879,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1243694989728216,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0029892892800511706,
+      "loss": 3.6099,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12445726855572409,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.002989237916540673,
+      "loss": 3.5688,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12454503813862657,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.002989186430659987,
+      "loss": 3.5664,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12463280772152906,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0029891348224138185,
+      "loss": 3.5825,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12472057730443153,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.002989083091806884,
+      "loss": 3.5996,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12480834688733403,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029890312388439083,
+      "loss": 3.5752,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12489611647023652,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.002988979263529631,
+      "loss": 3.5356,
+      "step": 1423
+    },
+    {
+      "epoch": 0.124983886053139,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0029889271658687998,
+      "loss": 3.5176,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12507165563604147,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0029888749458661756,
+      "loss": 3.584,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12515942521894396,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029888226035265308,
+      "loss": 3.5537,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12524719480184646,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029887701388546465,
+      "loss": 3.5981,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12533496438474895,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029887175518553176,
+      "loss": 3.5894,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1254227339676514,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029886648425333493,
+      "loss": 3.603,
+      "step": 1429
+    },
+    {
+      "epoch": 0.1255105035505539,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029886120108935568,
+      "loss": 3.584,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1255982731334564,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0029885590569407685,
+      "loss": 3.5791,
+      "step": 1431
+    },
+    {
+      "epoch": 0.1256860427163589,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002988505980679822,
+      "loss": 3.5176,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12577381229926135,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002988452782115568,
+      "loss": 3.5513,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12586158188216384,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029883994612528667,
+      "loss": 3.5923,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12594935146506633,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002988346018096591,
+      "loss": 3.5991,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12603712104796883,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029882924526516232,
+      "loss": 3.561,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12612489063087132,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002988238764922858,
+      "loss": 3.4897,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12621266021377378,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0029881849549152012,
+      "loss": 3.606,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12630042979667627,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0029881310226335697,
+      "loss": 3.5547,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12638819937957876,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0029880769680828917,
+      "loss": 3.543,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12647596896248126,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029880227912681056,
+      "loss": 3.502,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12656373854538372,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002987968492194162,
+      "loss": 3.5474,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1266515081282862,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0029879140708660226,
+      "loss": 3.4326,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1267392777111887,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0029878595272886596,
+      "loss": 3.5464,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1268270472940912,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0029878048614670567,
+      "loss": 3.5728,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12691481687699369,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0029877500734062103,
+      "loss": 3.5171,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12700258645989615,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029876951631111243,
+      "loss": 3.6523,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12709035604279864,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0029876401305868176,
+      "loss": 3.5244,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12717812562570113,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029875849758383187,
+      "loss": 3.6323,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12726589520860362,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0029875296988706667,
+      "loss": 3.5283,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1273536647915061,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002987474299688912,
+      "loss": 3.5527,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12744143437440858,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002987418778298118,
+      "loss": 3.5352,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12752920395731107,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0029873631347033564,
+      "loss": 3.5693,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12761697354021356,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029873073689097125,
+      "loss": 3.5483,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12770474312311603,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029872514809222814,
+      "loss": 3.5552,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12779251270601852,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0029871954707461697,
+      "loss": 3.5259,
+      "step": 1456
+    },
+    {
+      "epoch": 0.127880282288921,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029871393383864952,
+      "loss": 3.6387,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1279680518718235,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002987083083848387,
+      "loss": 3.561,
+      "step": 1458
+    },
+    {
+      "epoch": 0.128055821454726,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.002987026707136985,
+      "loss": 3.5581,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12814359103762846,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002986970208257441,
+      "loss": 3.502,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12823136062053095,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.002986913587214917,
+      "loss": 3.5371,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12831913020343344,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029868568440145874,
+      "loss": 3.4937,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12840689978633593,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.002986799978661636,
+      "loss": 3.5107,
+      "step": 1463
+    },
+    {
+      "epoch": 0.1284946693692384,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002986742991161259,
+      "loss": 3.5459,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1285824389521409,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.002986685881518664,
+      "loss": 3.5386,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12867020853504338,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.002986628649739069,
+      "loss": 3.5122,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12875797811794587,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0029865712958277034,
+      "loss": 3.5308,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12884574770084833,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0029865138197898073,
+      "loss": 3.5591,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12893351728375083,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029864562216306337,
+      "loss": 3.521,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12902128686665332,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002986398501355445,
+      "loss": 3.5234,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1291090564495558,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.002986340658969514,
+      "loss": 3.5059,
+      "step": 1471
+    },
+    {
+      "epoch": 0.1291968260324583,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.002986282694478128,
+      "loss": 3.6118,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12928459561536076,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002986224607886582,
+      "loss": 3.562,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12937236519826326,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002986166399200184,
+      "loss": 3.604,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12946013478116575,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002986108068424253,
+      "loss": 3.4727,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12954790436406824,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029860496155641176,
+      "loss": 3.5386,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1296356739469707,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002985991040625121,
+      "loss": 3.5562,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1297234435298732,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002985932343612613,
+      "loss": 3.6108,
+      "step": 1478
+    },
+    {
+      "epoch": 0.1298112131127757,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0029858735245319582,
+      "loss": 3.5938,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12989898269567818,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.002985814583388531,
+      "loss": 3.6196,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12998675227858067,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002985755520187717,
+      "loss": 3.4839,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13007452186148313,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002985696334934913,
+      "loss": 3.5566,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13016229144438562,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002985637027635527,
+      "loss": 3.5596,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13025006102728812,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0029855775982949775,
+      "loss": 3.4639,
+      "step": 1484
+    },
+    {
+      "epoch": 0.1303378306101906,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002985518046918695,
+      "loss": 3.6108,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13042560019309307,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0029854583735121213,
+      "loss": 3.4692,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13051336977599556,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029853985780807085,
+      "loss": 3.5142,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13060113935889806,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0029853386606299207,
+      "loss": 3.5645,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13068890894180055,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029852786211652326,
+      "loss": 3.5181,
+      "step": 1489
+    },
+    {
+      "epoch": 0.130776678524703,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029852184596921303,
+      "loss": 3.502,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1308644481076055,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0029851581762161098,
+      "loss": 3.5347,
+      "step": 1491
+    },
+    {
+      "epoch": 0.130952217690508,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002985097770742681,
+      "loss": 3.4697,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13103998727341049,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029850372432773625,
+      "loss": 3.5562,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13112775685631298,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002984976593825685,
+      "loss": 3.417,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13121552643921544,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029849158223931903,
+      "loss": 3.5205,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13130329602211793,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.002984854928985431,
+      "loss": 3.4775,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13139106560502042,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0029847939136079717,
+      "loss": 3.5142,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13147883518792292,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002984732776266387,
+      "loss": 3.5596,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13156660477082538,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002984671516966263,
+      "loss": 3.5889,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13165437435372787,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.002984610135713198,
+      "loss": 3.6089,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13174214393663036,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0029845486325128,
+      "loss": 3.457,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13182991351953285,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002984487007370689,
+      "loss": 3.583,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13191768310243532,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029844252602924956,
+      "loss": 3.5859,
+      "step": 1503
+    },
+    {
+      "epoch": 0.1320054526853378,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0029843633912838615,
+      "loss": 3.5327,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1320932222682403,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029843014003504407,
+      "loss": 3.4868,
+      "step": 1505
+    },
+    {
+      "epoch": 0.1321809918511428,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002984239287497897,
+      "loss": 3.5386,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13226876143404528,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0029841770527319063,
+      "loss": 3.5767,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13235653101694775,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0029841146960581546,
+      "loss": 3.6064,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13244430059985024,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029840522174823395,
+      "loss": 3.5161,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13253207018275273,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029839896170101706,
+      "loss": 3.5977,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13261983976565522,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002983926894647367,
+      "loss": 3.5557,
+      "step": 1511
+    },
+    {
+      "epoch": 0.1327076093485577,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029838640503996607,
+      "loss": 3.522,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13279537893146018,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.002983801084272793,
+      "loss": 3.4814,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13288314851436267,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029837379962725187,
+      "loss": 3.4771,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13297091809726516,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002983674786404601,
+      "loss": 3.5229,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13305868768016765,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002983611454674816,
+      "loss": 3.4595,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13314645726307012,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002983548001088951,
+      "loss": 3.603,
+      "step": 1517
+    },
+    {
+      "epoch": 0.1332342268459726,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.002983484425652803,
+      "loss": 3.5293,
+      "step": 1518
+    },
+    {
+      "epoch": 0.1333219964288751,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002983420728372182,
+      "loss": 3.5654,
+      "step": 1519
+    },
+    {
+      "epoch": 0.1334097660117776,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029833569092529077,
+      "loss": 3.4932,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13349753559468006,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002983292968300812,
+      "loss": 3.5498,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13358530517758255,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0029832289055217362,
+      "loss": 3.5723,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13367307476048504,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002983164720921535,
+      "loss": 3.4775,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13376084434338753,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.002983100414506073,
+      "loss": 3.498,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13384861392629,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0029830359862812257,
+      "loss": 3.4751,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13393638350919249,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0029829714362528802,
+      "loss": 3.4653,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13402415309209498,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.002982906764426935,
+      "loss": 3.5444,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13411192267499747,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002982841970809299,
+      "loss": 3.4956,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13419969225789996,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002982777055405892,
+      "loss": 3.5708,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13428746184080242,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0029827120182226467,
+      "loss": 3.436,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13437523142370492,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0029826468592655054,
+      "loss": 3.5288,
+      "step": 1531
+    },
+    {
+      "epoch": 0.1344630010066074,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.002982581578540421,
+      "loss": 3.6108,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1345507705895099,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.00298251617605336,
+      "loss": 3.4951,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13463854017241236,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002982450651810297,
+      "loss": 3.4678,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13472630975531485,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0029823850058172193,
+      "loss": 3.5249,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13481407933821735,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0029823192380801257,
+      "loss": 3.5078,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13490184892111984,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0029822533486050257,
+      "loss": 3.623,
+      "step": 1537
+    },
+    {
+      "epoch": 0.1349896185040223,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0029821873373979393,
+      "loss": 3.4209,
+      "step": 1538
+    },
+    {
+      "epoch": 0.1350773880869248,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0029821212044648985,
+      "loss": 3.521,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13516515766982728,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0029820549498119457,
+      "loss": 3.5181,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13525292725272978,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.002981988573445135,
+      "loss": 3.4883,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13534069683563227,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0029819220753705316,
+      "loss": 3.5078,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13542846641853473,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0029818554555942115,
+      "loss": 3.5869,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13551623600143722,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029817887141222617,
+      "loss": 3.4458,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13560400558433972,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.002981721850960781,
+      "loss": 3.5776,
+      "step": 1545
+    },
+    {
+      "epoch": 0.1356917751672422,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0029816548661158785,
+      "loss": 3.5454,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13577954475014467,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0029815877595936752,
+      "loss": 3.4912,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13586731433304716,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0029815205314003023,
+      "loss": 3.5303,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13595508391594965,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002981453181541903,
+      "loss": 3.478,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13604285349885215,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0029813857100246316,
+      "loss": 3.5107,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13613062308175464,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029813181168546524,
+      "loss": 3.6113,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1362183926646571,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.002981250402038142,
+      "loss": 3.6372,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1363061622475596,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0029811825655812873,
+      "loss": 3.5415,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13639393183046208,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029811146074902877,
+      "loss": 3.5747,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13648170141336458,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0029810465277713516,
+      "loss": 3.481,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13656947099626704,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029809783264307006,
+      "loss": 3.4883,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13665724057916953,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002980910003474565,
+      "loss": 3.6069,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13674501016207202,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.00298084155890919,
+      "loss": 3.5854,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13683277974497451,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002980772992740828,
+      "loss": 3.501,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13692054932787698,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029807043049757437,
+      "loss": 3.5244,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13700831891077947,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029806354956202147,
+      "loss": 3.4834,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13709608849368196,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029805665646805276,
+      "loss": 3.5811,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13718385807658445,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0029804975121629803,
+      "loss": 3.5303,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13727162765948694,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002980428338073883,
+      "loss": 3.6274,
+      "step": 1564
+    },
+    {
+      "epoch": 0.1373593972423894,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029803590424195567,
+      "loss": 3.498,
+      "step": 1565
+    },
+    {
+      "epoch": 0.1374471668252919,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002980289625206332,
+      "loss": 3.5825,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1375349364081944,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029802200864405526,
+      "loss": 3.4995,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13762270599109688,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029801504261285725,
+      "loss": 3.4429,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13771047557399935,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002980080644276757,
+      "loss": 3.5498,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13779824515690184,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.002980010740891481,
+      "loss": 3.4844,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13788601473980433,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0029799407159791327,
+      "loss": 3.4932,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13797378432270682,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0029798705695461107,
+      "loss": 3.4902,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13806155390560929,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0029798003015988244,
+      "loss": 3.4824,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13814932348851178,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002979729912143694,
+      "loss": 3.5464,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13823709307141427,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.002979659401187151,
+      "loss": 3.5103,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13832486265431676,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002979588768735639,
+      "loss": 3.542,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13841263223721925,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0029795180147956114,
+      "loss": 3.5078,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13850040182012172,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002979447139373533,
+      "loss": 3.5581,
+      "step": 1578
+    },
+    {
+      "epoch": 0.1385881714030242,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029793761424758806,
+      "loss": 3.5298,
+      "step": 1579
+    },
+    {
+      "epoch": 0.1386759409859267,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002979305024109141,
+      "loss": 3.5391,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1387637105688292,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.002979233784279812,
+      "loss": 3.5181,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13885148015173165,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002979162422994404,
+      "loss": 3.5151,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13893924973463415,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0029790909402594365,
+      "loss": 3.5859,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13902701931753664,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.002979019336081442,
+      "loss": 3.479,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13911478890043913,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002978947610466962,
+      "loss": 3.5127,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13920255848334162,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.002978875763422552,
+      "loss": 3.5562,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13929032806624408,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0029788037949547755,
+      "loss": 3.5293,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13937809764914658,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002978731705070209,
+      "loss": 3.6045,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13946586723204907,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0029786594937754396,
+      "loss": 3.5396,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13955363681495156,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002978587161077065,
+      "loss": 3.4902,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13964140639785402,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002978514706981695,
+      "loss": 3.478,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13972917598075651,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029784421314959495,
+      "loss": 3.5903,
+      "step": 1592
+    },
+    {
+      "epoch": 0.139816945563659,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00297836943462646,
+      "loss": 3.5396,
+      "step": 1593
+    },
+    {
+      "epoch": 0.1399047151465615,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0029782966163798695,
+      "loss": 3.5869,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13999248472946396,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002978223676762831,
+      "loss": 3.4922,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14008025431236645,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00297815061578201,
+      "loss": 3.541,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14016802389526895,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002978077433444082,
+      "loss": 3.4971,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14025579347817144,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002978004129755733,
+      "loss": 3.543,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14034356306107393,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002977930704723662,
+      "loss": 3.5283,
+      "step": 1599
+    },
+    {
+      "epoch": 0.1404313326439764,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002977857158354578,
+      "loss": 3.5112,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14051910222687888,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029777834906552004,
+      "loss": 3.5806,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14060687180978138,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029777097016322617,
+      "loss": 3.5142,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14069464139268387,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002977635791292503,
+      "loss": 3.4399,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14078241097558633,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002977561759642678,
+      "loss": 3.4771,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14087018055848882,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0029774876066895515,
+      "loss": 3.585,
+      "step": 1605
+    },
+    {
+      "epoch": 0.1409579501413913,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029774133324398993,
+      "loss": 3.5273,
+      "step": 1606
+    },
+    {
+      "epoch": 0.1410457197242938,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0029773389369005076,
+      "loss": 3.481,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14113348930719627,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002977264420078174,
+      "loss": 3.4863,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14122125889009876,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002977189781979708,
+      "loss": 3.48,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14130902847300125,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0029771150226119286,
+      "loss": 3.5444,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14139679805590374,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029770401419816676,
+      "loss": 3.4653,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14148456763880624,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002976965140095767,
+      "loss": 3.4712,
+      "step": 1612
+    },
+    {
+      "epoch": 0.1415723372217087,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029768900169610795,
+      "loss": 3.626,
+      "step": 1613
+    },
+    {
+      "epoch": 0.1416601068046112,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0029768147725844694,
+      "loss": 3.5513,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14174787638751368,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002976739406972812,
+      "loss": 3.5625,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14183564597041617,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0029766639201329937,
+      "loss": 3.5024,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14192341555331864,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029765883120719126,
+      "loss": 3.4463,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14201118513622113,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0029765125827964766,
+      "loss": 3.5073,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14209895471912362,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0029764367323136054,
+      "loss": 3.5752,
+      "step": 1619
+    },
+    {
+      "epoch": 0.1421867243020261,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0029763607606302298,
+      "loss": 3.5146,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1422744938849286,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002976284667753291,
+      "loss": 3.5249,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14236226346783107,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029762084536897426,
+      "loss": 3.459,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14245003305073356,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0029761321184465484,
+      "loss": 3.5654,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14253780263363605,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0029760556620306825,
+      "loss": 3.4248,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14262557221653854,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0029759790844491327,
+      "loss": 3.5161,
+      "step": 1625
+    },
+    {
+      "epoch": 0.142713341799441,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0029759023857088944,
+      "loss": 3.502,
+      "step": 1626
+    },
+    {
+      "epoch": 0.1428011113823435,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029758255658169767,
+      "loss": 3.5688,
+      "step": 1627
+    },
+    {
+      "epoch": 0.142888880965246,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0029757486247803983,
+      "loss": 3.4727,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14297665054814848,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00297567156260619,
+      "loss": 3.5332,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14306442013105095,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002975594379301393,
+      "loss": 3.4424,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14315218971395344,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00297551707487306,
+      "loss": 3.6143,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14323995929685593,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0029754396493282542,
+      "loss": 3.5469,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14332772887975842,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0029753621026740503,
+      "loss": 3.5708,
+      "step": 1633
+    },
+    {
+      "epoch": 0.1434154984626609,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0029752844349175346,
+      "loss": 3.48,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14350326804556338,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0029752066460658026,
+      "loss": 3.5854,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14359103762846587,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.002975128736125963,
+      "loss": 3.5693,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14367880721136836,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.002975050705105135,
+      "loss": 3.5308,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14376657679427085,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0029749725530104475,
+      "loss": 3.46,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14385434637717331,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.0029748942798490416,
+      "loss": 3.4834,
+      "step": 1639
+    },
+    {
+      "epoch": 0.1439421159600758,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00297481588562807,
+      "loss": 3.5088,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1440298855429783,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0029747373703546956,
+      "loss": 3.5093,
+      "step": 1641
+    },
+    {
+      "epoch": 0.1441176551258808,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0029746587340360926,
+      "loss": 3.5249,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14420542470878325,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0029745799766794463,
+      "loss": 3.5269,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14429319429168574,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0029745010982919523,
+      "loss": 3.5269,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14438096387458824,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.002974422098880819,
+      "loss": 3.5474,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14446873345749073,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0029743429784532638,
+      "loss": 3.5073,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14455650304039322,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0029742637370165173,
+      "loss": 3.5142,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14464427262329568,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.002974184374577819,
+      "loss": 3.5156,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14473204220619817,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029741048911444213,
+      "loss": 3.5308,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14481981178910067,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0029740252867235858,
+      "loss": 3.5342,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14490758137200316,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002973945561322587,
+      "loss": 3.5508,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14499535095490562,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002973865714948709,
+      "loss": 3.4805,
+      "step": 1652
+    },
+    {
+      "epoch": 0.1450831205378081,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029737857476092493,
+      "loss": 3.4849,
+      "step": 1653
+    },
+    {
+      "epoch": 0.1451708901207106,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002973705659311513,
+      "loss": 3.4966,
+      "step": 1654
+    },
+    {
+      "epoch": 0.1452586597036131,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029736254500628183,
+      "loss": 3.4561,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1453464292865156,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029735451198704944,
+      "loss": 3.4658,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14543419886941805,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029734646687418816,
+      "loss": 3.5122,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14552196845232054,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0029733840966843306,
+      "loss": 3.4985,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14560973803522304,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002973303403705204,
+      "loss": 3.4468,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14569750761812553,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029732225898118736,
+      "loss": 3.4741,
+      "step": 1660
+    },
+    {
+      "epoch": 0.145785277201028,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002973141655011725,
+      "loss": 3.417,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14587304678393048,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002973060599312153,
+      "loss": 3.4927,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14596081636683297,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0029729794227205644,
+      "loss": 3.5205,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14604858594973547,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0029728981252443755,
+      "loss": 3.4355,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14613635553263793,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002972816706891015,
+      "loss": 3.563,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14622412511554042,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0029727351676679227,
+      "loss": 3.5327,
+      "step": 1666
+    },
+    {
+      "epoch": 0.1463118946984429,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0029726535075825494,
+      "loss": 3.5742,
+      "step": 1667
+    },
+    {
+      "epoch": 0.1463996642813454,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0029725717266423563,
+      "loss": 3.4531,
+      "step": 1668
+    },
+    {
+      "epoch": 0.1464874338642479,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0029724898248548156,
+      "loss": 3.4912,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14657520344715036,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002972407802227411,
+      "loss": 3.5259,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14666297303005285,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0029723256587676377,
+      "loss": 3.5225,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14675074261295534,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002972243394483001,
+      "loss": 3.4541,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14683851219585783,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029721610093810177,
+      "loss": 3.5068,
+      "step": 1673
+    },
+    {
+      "epoch": 0.1469262817787603,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002972078503469215,
+      "loss": 3.4761,
+      "step": 1674
+    },
+    {
+      "epoch": 0.1470140513616628,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029719958767551327,
+      "loss": 3.5171,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14710182094456528,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029719131292463203,
+      "loss": 3.4854,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14718959052746777,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0029718302609503386,
+      "loss": 3.5435,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14727736011037024,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029717472718747587,
+      "loss": 3.4834,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14736512969327273,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0029716641620271653,
+      "loss": 3.522,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14745289927617522,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002971580931415151,
+      "loss": 3.5117,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1475406688590777,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0029714975800463218,
+      "loss": 3.5137,
+      "step": 1681
+    },
+    {
+      "epoch": 0.1476284384419802,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.002971414107928293,
+      "loss": 3.4277,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14771620802488267,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029713305150686916,
+      "loss": 3.4712,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14780397760778516,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002971246801475157,
+      "loss": 3.5337,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14789174719068765,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0029711629671553366,
+      "loss": 3.5737,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14797951677359014,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029710790121168914,
+      "loss": 3.5039,
+      "step": 1686
+    },
+    {
+      "epoch": 0.1480672863564926,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0029709949363674933,
+      "loss": 3.5244,
+      "step": 1687
+    },
+    {
+      "epoch": 0.1481550559393951,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0029709107399148236,
+      "loss": 3.4751,
+      "step": 1688
+    },
+    {
+      "epoch": 0.1482428255222976,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002970826422766576,
+      "loss": 3.5112,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14833059510520008,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029707419849304547,
+      "loss": 3.4766,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14841836468810257,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002970657426414175,
+      "loss": 3.4668,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14850613427100504,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0029705727472254637,
+      "loss": 3.5439,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14859390385390753,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0029704879473720575,
+      "loss": 3.4873,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14868167343681002,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002970403026861705,
+      "loss": 3.4922,
+      "step": 1694
+    },
+    {
+      "epoch": 0.1487694430197125,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002970317985702166,
+      "loss": 3.5073,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14885721260261497,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002970232823901211,
+      "loss": 3.4863,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14894498218551747,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029701475414666205,
+      "loss": 3.4482,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14903275176841996,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0029700621384061883,
+      "loss": 3.5034,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14912052135132245,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.002969976614727717,
+      "loss": 3.5586,
+      "step": 1699
+    },
+    {
+      "epoch": 0.1492082909342249,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0029698909704390223,
+      "loss": 3.4639,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1492960605171274,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0029698052055479283,
+      "loss": 3.4673,
+      "step": 1701
+    },
+    {
+      "epoch": 0.1493838301000299,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0029697193200622723,
+      "loss": 3.5151,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1494715996829324,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.002969633313989902,
+      "loss": 3.5093,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14955936926583488,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002969547187338676,
+      "loss": 3.5015,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14964713884873734,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.002969460940116464,
+      "loss": 3.5513,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14973490843163983,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0029693745723311468,
+      "loss": 3.5039,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14982267801454233,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0029692880839906157,
+      "loss": 3.4888,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14991044759744482,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029692014751027734,
+      "loss": 3.4834,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14999821718034728,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0029691147456755336,
+      "loss": 3.418,
+      "step": 1709
+    },
+    {
+      "epoch": 0.15008598676324977,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002969027895716822,
+      "loss": 3.4629,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15017375634615227,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0029689409252345728,
+      "loss": 3.5259,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15026152592905476,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029688538342367335,
+      "loss": 3.4683,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15034929551195722,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.002968766622731262,
+      "loss": 3.4204,
+      "step": 1713
+    },
+    {
+      "epoch": 0.1504370650948597,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002968679290726127,
+      "loss": 3.5112,
+      "step": 1714
+    },
+    {
+      "epoch": 0.1505248346777622,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002968591838229308,
+      "loss": 3.6064,
+      "step": 1715
+    },
+    {
+      "epoch": 0.1506126042606647,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0029685042652487963,
+      "loss": 3.5518,
+      "step": 1716
+    },
+    {
+      "epoch": 0.1507003738435672,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029684165717925935,
+      "loss": 3.4746,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15078814342646965,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002968328757868712,
+      "loss": 3.4614,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15087591300937214,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029682408234851763,
+      "loss": 3.481,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15096368259227463,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0029681527686500203,
+      "loss": 3.5049,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15105145217517713,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002968064593371291,
+      "loss": 3.5151,
+      "step": 1721
+    },
+    {
+      "epoch": 0.1511392217580796,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0029679762976570446,
+      "loss": 3.5063,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15122699134098208,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0029678878815153488,
+      "loss": 3.5737,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15131476092388457,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002967799344954283,
+      "loss": 3.4536,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15140253050678706,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0029677106879819363,
+      "loss": 3.4536,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15149030008968956,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0029676219106064103,
+      "loss": 3.5439,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15157806967259202,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029675330128358167,
+      "loss": 3.5039,
+      "step": 1727
+    },
+    {
+      "epoch": 0.1516658392554945,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0029674439946782774,
+      "loss": 3.4463,
+      "step": 1728
+    },
+    {
+      "epoch": 0.151753608838397,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002967354856141928,
+      "loss": 3.5049,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1518413784212995,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0029672655972349114,
+      "loss": 3.4668,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15192914800420196,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029671762179653846,
+      "loss": 3.5347,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15201691758710445,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0029670867183415146,
+      "loss": 3.4629,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15210468717000694,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002966997098371479,
+      "loss": 3.5054,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15219245675290943,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002966907358063467,
+      "loss": 3.5845,
+      "step": 1734
+    },
+    {
+      "epoch": 0.1522802263358119,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002966817497425677,
+      "loss": 3.5986,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1523679959187144,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029667275164663217,
+      "loss": 3.4277,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15245576550161688,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002966637415193622,
+      "loss": 3.52,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15254353508451937,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029665471936158104,
+      "loss": 3.459,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15263130466742186,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0029664568517411313,
+      "loss": 3.4038,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15271907425032433,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00296636638957784,
+      "loss": 3.5142,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15280684383322682,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0029662758071342006,
+      "loss": 3.5156,
+      "step": 1741
+    },
+    {
+      "epoch": 0.1528946134161293,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002966185104418491,
+      "loss": 3.5303,
+      "step": 1742
+    },
+    {
+      "epoch": 0.1529823829990318,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029660942814389996,
+      "loss": 3.5186,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15307015258193427,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002966003338204024,
+      "loss": 3.4951,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15315792216483676,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029659122747218744,
+      "loss": 3.5454,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15324569174773925,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029658210910008718,
+      "loss": 3.4443,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15333346133064174,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002965729787049348,
+      "loss": 3.4839,
+      "step": 1747
+    },
+    {
+      "epoch": 0.1534212309135442,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002965638362875645,
+      "loss": 3.438,
+      "step": 1748
+    },
+    {
+      "epoch": 0.1535090004964467,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002965546818488117,
+      "loss": 3.5322,
+      "step": 1749
+    },
+    {
+      "epoch": 0.1535967700793492,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029654551538951285,
+      "loss": 3.5176,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15368453966225168,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029653633691050555,
+      "loss": 3.5439,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15377230924515417,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002965271464126284,
+      "loss": 3.4873,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15386007882805663,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002965179438967213,
+      "loss": 3.5103,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15394784841095913,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0029650872936362494,
+      "loss": 3.5728,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15403561799386162,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029649950281418137,
+      "loss": 3.5181,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1541233875767641,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0029649026424923364,
+      "loss": 3.5137,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15421115715966657,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002964810136696259,
+      "loss": 3.4165,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15429892674256906,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002964717510762034,
+      "loss": 3.5918,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15438669632547156,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.002964624764698125,
+      "loss": 3.4434,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15447446590837405,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0029645318985130065,
+      "loss": 3.4478,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15456223549127654,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002964438912215164,
+      "loss": 3.604,
+      "step": 1761
+    },
+    {
+      "epoch": 0.154650005074179,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.002964345805813093,
+      "loss": 3.4492,
+      "step": 1762
+    },
+    {
+      "epoch": 0.1547377746570815,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002964252579315303,
+      "loss": 3.5171,
+      "step": 1763
+    },
+    {
+      "epoch": 0.154825544239984,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0029641592327303103,
+      "loss": 3.5542,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15491331382288648,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.002964065766066645,
+      "loss": 3.4697,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15500108340578894,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029639721793328485,
+      "loss": 3.5684,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15508885298869143,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0029638784725374706,
+      "loss": 3.4844,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15517662257159393,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0029637846456890735,
+      "loss": 3.4238,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15526439215449642,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002963690698796231,
+      "loss": 3.4766,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15535216173739888,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002963596631867528,
+      "loss": 3.5229,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15543993132030137,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029635024449115592,
+      "loss": 3.4761,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15552770090320386,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00296340813793693,
+      "loss": 3.5063,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15561547048610636,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0029633137109522584,
+      "loss": 3.4976,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15570324006900885,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0029632191639661722,
+      "loss": 3.4883,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1557910096519113,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00296312449698731,
+      "loss": 3.5049,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1558787792348138,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029630297100243217,
+      "loss": 3.5679,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1559665488177163,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00296293480308587,
+      "loss": 3.478,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15605431840061879,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029628397761806246,
+      "loss": 3.5249,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15614208798352125,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029627446293172704,
+      "loss": 3.4751,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15622985756642374,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0029626493625044994,
+      "loss": 3.5015,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15631762714932623,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0029625539757510175,
+      "loss": 3.4941,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15640539673222872,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00296245846906554,
+      "loss": 3.5063,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1564931663151312,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0029623628424567945,
+      "loss": 3.4595,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15658093589803368,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0029622670959335174,
+      "loss": 3.5127,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15666870548093617,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0029621712295044584,
+      "loss": 3.5029,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15675647506383866,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0029620752431783765,
+      "loss": 3.4902,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15684424464674115,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029619791369640425,
+      "loss": 3.4888,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15693201422964362,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029618829108702385,
+      "loss": 3.5269,
+      "step": 1788
+    },
+    {
+      "epoch": 0.1570197838125461,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0029617865649057555,
+      "loss": 3.54,
+      "step": 1789
+    },
+    {
+      "epoch": 0.1571075533954486,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002961690099079398,
+      "loss": 3.4795,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1571953229783511,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029615935133999803,
+      "loss": 3.502,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15728309256125356,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.002961496807876327,
+      "loss": 3.5298,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15737086214415605,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.002961399982517276,
+      "loss": 3.4429,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15745863172705854,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0029613030373316722,
+      "loss": 3.5127,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15754640130996103,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002961205972328376,
+      "loss": 3.4712,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15763417089286352,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002961108787516255,
+      "loss": 3.4502,
+      "step": 1796
+    },
+    {
+      "epoch": 0.157721940475766,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0029610114829041897,
+      "loss": 3.4878,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15780971005866848,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002960914058501072,
+      "loss": 3.5581,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15789747964157097,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0029608165143158024,
+      "loss": 3.5181,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15798524922447346,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.002960718850357295,
+      "loss": 3.4897,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15807301880737593,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0029606210666344728,
+      "loss": 3.4185,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15816078839027842,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.002960523163156271,
+      "loss": 3.5229,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1582485579731809,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0029604251399316353,
+      "loss": 3.4819,
+      "step": 1803
+    },
+    {
+      "epoch": 0.1583363275560834,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.0029603269969695222,
+      "loss": 3.5176,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15842409713898586,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029602287342789,
+      "loss": 3.5366,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15851186672188836,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.002960130351868746,
+      "loss": 3.4644,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15859963630479085,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0029600318497480507,
+      "loss": 3.5889,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15868740588769334,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029599332279258146,
+      "loss": 3.4609,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15877517547059583,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.002959834486411048,
+      "loss": 3.4771,
+      "step": 1809
+    },
+    {
+      "epoch": 0.1588629450534983,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002959735625212775,
+      "loss": 3.5332,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1589507146364008,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.002959636644340027,
+      "loss": 3.5308,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15903848421930328,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.002959537543801849,
+      "loss": 3.5518,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15912625380220577,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.002959438323607296,
+      "loss": 3.4429,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15921402338510823,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0029593389837654343,
+      "loss": 3.46,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15930179296801072,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029592395242853406,
+      "loss": 3.5249,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15938956255091322,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002959139945176103,
+      "loss": 3.5132,
+      "step": 1816
+    },
+    {
+      "epoch": 0.1594773321338157,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0029590402464468204,
+      "loss": 3.501,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15956510171671817,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029589404281066024,
+      "loss": 3.4878,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15965287129962066,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0029588404901645694,
+      "loss": 3.5195,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15974064088252315,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002958740432629854,
+      "loss": 3.4653,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15982841046542565,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0029586402555115974,
+      "loss": 3.5459,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15991618004832814,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002958539958818954,
+      "loss": 3.5366,
+      "step": 1822
+    },
+    {
+      "epoch": 0.1600039496312306,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029584395425610883,
+      "loss": 3.459,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1600917192141331,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002958339006747175,
+      "loss": 3.4536,
+      "step": 1824
+    },
+    {
+      "epoch": 0.16017948879703559,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029582383513864006,
+      "loss": 3.5039,
+      "step": 1825
+    },
+    {
+      "epoch": 0.16026725837993808,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0029581375764879624,
+      "loss": 3.5137,
+      "step": 1826
+    },
+    {
+      "epoch": 0.16035502796284054,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0029580366820610685,
+      "loss": 3.4946,
+      "step": 1827
+    },
+    {
+      "epoch": 0.16044279754574303,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029579356681149385,
+      "loss": 3.5415,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16053056712864552,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002957834534658801,
+      "loss": 3.4819,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16061833671154802,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002957733281701898,
+      "loss": 3.4844,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1607061062944505,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002957631909253481,
+      "loss": 3.501,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16079387587735297,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002957530417322812,
+      "loss": 3.5239,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16088164546025546,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029574288059191663,
+      "loss": 3.4995,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16096941504315795,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0029573270750518268,
+      "loss": 3.374,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16105718462606045,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0029572252247300895,
+      "loss": 3.4609,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1611449542089629,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0029571232549632604,
+      "loss": 3.5259,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1612327237918654,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.002957021165760658,
+      "loss": 3.4634,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1613204933747679,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029569189571316094,
+      "loss": 3.522,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16140826295767038,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.002956816629085454,
+      "loss": 3.4717,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16149603254057285,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0029567141816315414,
+      "loss": 3.4771,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16158380212347534,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029566116147792336,
+      "loss": 3.4917,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16167157170637783,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.002956508928537901,
+      "loss": 3.4609,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16175934128928032,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002956406122916928,
+      "loss": 3.5586,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16184711087218281,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0029563031979257066,
+      "loss": 3.4258,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16193488045508528,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0029562001535736425,
+      "loss": 3.5005,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16202265003798777,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002956096989870151,
+      "loss": 3.5259,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16211041962089026,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029559937068246583,
+      "loss": 3.4639,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16219818920379275,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029558903044466017,
+      "loss": 3.5469,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16228595878669522,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002955786782745429,
+      "loss": 3.5269,
+      "step": 1849
+    },
+    {
+      "epoch": 0.1623737283695977,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029556831417306,
+      "loss": 3.3975,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1624614979525002,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029555793814115845,
+      "loss": 3.4209,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1625492675354027,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0029554755017978628,
+      "loss": 3.5386,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16263703711830516,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0029553715028989274,
+      "loss": 3.501,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16272480670120765,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0029552673847242808,
+      "loss": 3.4619,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16281257628411014,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.002955163147283436,
+      "loss": 3.4634,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16290034586701263,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029550587905859183,
+      "loss": 3.5635,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16298811544991512,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029549543146412632,
+      "loss": 3.5068,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16307588503281759,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002954849719459016,
+      "loss": 3.4712,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16316365461572008,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0029547450050487352,
+      "loss": 3.4717,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16325142419862257,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029546401714199877,
+      "loss": 3.4878,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16333919378152506,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.002954535218582353,
+      "loss": 3.5176,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16342696336442752,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0029544301465454204,
+      "loss": 3.5894,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16351473294733002,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002954324955318791,
+      "loss": 3.5625,
+      "step": 1863
+    },
+    {
+      "epoch": 0.1636025025302325,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002954219644912077,
+      "loss": 3.4141,
+      "step": 1864
+    },
+    {
+      "epoch": 0.163690272113135,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0029541142153349004,
+      "loss": 3.5337,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1637780416960375,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002954008666596895,
+      "loss": 3.4697,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16386581127893995,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0029539029987077038,
+      "loss": 3.4546,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16395358086184245,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029537972116769834,
+      "loss": 3.4385,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16404135044474494,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0029536913055144,
+      "loss": 3.5347,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16412912002764743,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0029535852802296293,
+      "loss": 3.4731,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1642168896105499,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.00295347913583236,
+      "loss": 3.4858,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16430465919345238,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002953372872332291,
+      "loss": 3.52,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16439242877635488,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002953266489739131,
+      "loss": 3.5596,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16448019835925737,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0029531599880626017,
+      "loss": 3.4839,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16456796794215983,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0029530533673124337,
+      "loss": 3.4282,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16465573752506232,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.00295294662749837,
+      "loss": 3.4399,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16474350710796482,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.002952839768630162,
+      "loss": 3.4951,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1648312766908673,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029527327907175756,
+      "loss": 3.5449,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1649190462737698,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002952625693770385,
+      "loss": 3.4839,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16500681585667226,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002952518477798376,
+      "loss": 3.519,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16509458543957475,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029524111428113446,
+      "loss": 3.4302,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16518235502247725,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029523036888190993,
+      "loss": 3.4409,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16527012460537974,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0029521961158314584,
+      "loss": 3.5225,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1653578941882822,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029520884238582504,
+      "loss": 3.4683,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1654456637711847,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002951980612909316,
+      "loss": 3.5122,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16553343335408718,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029518726829945064,
+      "loss": 3.5469,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16562120293698968,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002951764634123683,
+      "loss": 3.5493,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16570897251989214,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002951656466306719,
+      "loss": 3.5186,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16579674210279463,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002951548179553497,
+      "loss": 3.5312,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16588451168569712,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002951439773873913,
+      "loss": 3.501,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16597228126859961,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0029513312492778715,
+      "loss": 3.4565,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1660600508515021,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0029512226057752887,
+      "loss": 3.5942,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16614782043440457,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.002951113843376092,
+      "loss": 3.5083,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16623559001730706,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.0029510049620902193,
+      "loss": 3.4468,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16632335960020955,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029508959619276185,
+      "loss": 3.478,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16641112918311204,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0029507868428982507,
+      "loss": 3.5342,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1664988987660145,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.002950677605012085,
+      "loss": 3.4961,
+      "step": 1897
+    },
+    {
+      "epoch": 0.166586668348917,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0029505682482791045,
+      "loss": 3.5479,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1666744379318195,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0029504587727093,
+      "loss": 3.4507,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16676220751472198,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0029503491783126755,
+      "loss": 3.4888,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16684997709762447,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0029502394650992446,
+      "loss": 3.478,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16693774668052694,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002950129633079032,
+      "loss": 3.4634,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16702551626342943,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002950019682262073,
+      "loss": 3.519,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16711328584633192,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002949909612658415,
+      "loss": 3.4673,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1672010554292344,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002949799424278115,
+      "loss": 3.4009,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16728882501213688,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.002949689117131242,
+      "loss": 3.5552,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16737659459503937,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029495786912278736,
+      "loss": 3.5059,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16746436417794186,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029494681465781006,
+      "loss": 3.4648,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16755213376084435,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002949357483192024,
+      "loss": 3.5054,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16763990334374682,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0029492467010797546,
+      "loss": 3.4316,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1677276729266493,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002949135800251416,
+      "loss": 3.5103,
+      "step": 1911
+    },
+    {
+      "epoch": 0.1678154425095518,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0029490247807171404,
+      "loss": 3.5137,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1679032120924543,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029489136424870727,
+      "loss": 3.4312,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16799098167535678,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002948802385571368,
+      "loss": 3.481,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16807875125825925,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0029486910099801923,
+      "loss": 3.5249,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16816652084116174,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.002948579515723722,
+      "loss": 3.5405,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16825429042406423,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0029484679028121442,
+      "loss": 3.5132,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16834206000696672,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029483561712556577,
+      "loss": 3.5034,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16842982958986918,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002948244321064473,
+      "loss": 3.5234,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16851759917277168,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002948132352248808,
+      "loss": 3.4648,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16860536875567417,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029480202648188942,
+      "loss": 3.502,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16869313833857666,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0029479080587849747,
+      "loss": 3.4517,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16878090792147912,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002947795734157301,
+      "loss": 3.4775,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16886867750438161,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0029476832909461365,
+      "loss": 3.4976,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1689564470872841,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002947570729161756,
+      "loss": 3.5327,
+      "step": 1925
+    },
+    {
+      "epoch": 0.1690442166701866,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029474580488144443,
+      "loss": 3.4575,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1691319862530891,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002947345249914497,
+      "loss": 3.4839,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16921975583599155,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002947232332472222,
+      "loss": 3.5815,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16930752541889404,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029471192964979362,
+      "loss": 3.416,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16939529500179654,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002947006142001967,
+      "loss": 3.4214,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16948306458469903,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002946892868994655,
+      "loss": 3.5396,
+      "step": 1931
+    },
+    {
+      "epoch": 0.1695708341676015,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0029467794774863504,
+      "loss": 3.5117,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16965860375050398,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029466659674874138,
+      "loss": 3.4819,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16974637333340648,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002946552339008216,
+      "loss": 3.4756,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16983414291630897,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002946438592059141,
+      "loss": 3.4429,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16992191249921146,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0029463247266505815,
+      "loss": 3.501,
+      "step": 1936
+    },
+    {
+      "epoch": 0.17000968208211392,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002946210742792942,
+      "loss": 3.4927,
+      "step": 1937
+    },
+    {
+      "epoch": 0.1700974516650164,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0029460966404966367,
+      "loss": 3.4844,
+      "step": 1938
+    },
+    {
+      "epoch": 0.1701852212479189,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0029459824197720927,
+      "loss": 3.5063,
+      "step": 1939
+    },
+    {
+      "epoch": 0.1702729908308214,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002945868080629746,
+      "loss": 3.4404,
+      "step": 1940
+    },
+    {
+      "epoch": 0.17036076041372386,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.002945753623080044,
+      "loss": 3.5254,
+      "step": 1941
+    },
+    {
+      "epoch": 0.17044852999662635,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.0029456390471334455,
+      "loss": 3.5503,
+      "step": 1942
+    },
+    {
+      "epoch": 0.17053629957952884,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0029455243528004192,
+      "loss": 3.4814,
+      "step": 1943
+    },
+    {
+      "epoch": 0.17062406916243134,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0029454095400914455,
+      "loss": 3.4937,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1707118387453338,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.002945294609017015,
+      "loss": 3.4961,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1707996083282363,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029451795595876293,
+      "loss": 3.5161,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17088737791113878,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.002945064391813801,
+      "loss": 3.459,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17097514749404127,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0029449491057060525,
+      "loss": 3.4575,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17106291707694377,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0029448337012749188,
+      "loss": 3.4629,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17115068665984623,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.002944718178530944,
+      "loss": 3.4814,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17123845624274872,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0029446025374846845,
+      "loss": 3.4863,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1713262258256512,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0029444867781467064,
+      "loss": 3.4985,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1714139954085537,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0029443709005275874,
+      "loss": 3.397,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17150176499145617,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.002944254904637915,
+      "loss": 3.5024,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17158953457435866,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.002944138790488288,
+      "loss": 3.5093,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17167730415726115,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0029440225580893158,
+      "loss": 3.4829,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17176507374016364,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00294390620745162,
+      "loss": 3.4321,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17185284332306613,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0029437897385858314,
+      "loss": 3.5142,
+      "step": 1958
+    },
+    {
+      "epoch": 0.1719406129059686,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.002943673151502592,
+      "loss": 3.499,
+      "step": 1959
+    },
+    {
+      "epoch": 0.1720283824888711,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002943556446212555,
+      "loss": 3.4258,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17211615207177358,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0029434396227263843,
+      "loss": 3.4517,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17220392165467607,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0029433226810547533,
+      "loss": 3.48,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17229169123757854,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029432056212083486,
+      "loss": 3.4062,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17237946082048103,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0029430884431978655,
+      "loss": 3.4707,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17246723040338352,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002942971147034011,
+      "loss": 3.4316,
+      "step": 1965
+    },
+    {
+      "epoch": 0.172554999986286,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002942853732727503,
+      "loss": 3.4502,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17264276956918848,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0029427362002890692,
+      "loss": 3.5254,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17273053915209097,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002942618549729451,
+      "loss": 3.4414,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17281830873499346,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0029425007810593967,
+      "loss": 3.4751,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17290607831789595,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029423828942896674,
+      "loss": 3.4897,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17299384790079844,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.002942264889431035,
+      "loss": 3.4497,
+      "step": 1971
+    },
+    {
+      "epoch": 0.1730816174837009,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002942146766494282,
+      "loss": 3.373,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1731693870666034,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0029420285254902014,
+      "loss": 3.4814,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1732571566495059,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002941910166429598,
+      "loss": 3.5225,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17334492623240838,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0029417916893232855,
+      "loss": 3.4331,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17343269581531084,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0029416730941820898,
+      "loss": 3.4727,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17352046539821334,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002941554381016848,
+      "loss": 3.4551,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17360823498111583,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002941435549838407,
+      "loss": 3.4883,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17369600456401832,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002941316600657624,
+      "loss": 3.5317,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17378377414692078,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0029411975334853687,
+      "loss": 3.4731,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17387154372982327,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00294107834833252,
+      "loss": 3.4751,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17395931331272577,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002940959045209969,
+      "loss": 3.4922,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17404708289562826,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0029408396241286153,
+      "loss": 3.4263,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17413485247853075,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029407200850993724,
+      "loss": 3.562,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1742226220614332,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0029406004281331624,
+      "loss": 3.5059,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1743103916443357,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002940480653240918,
+      "loss": 3.4736,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1743981612272382,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002940360760433584,
+      "loss": 3.499,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1744859308101407,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0029402407497221158,
+      "loss": 3.5,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17457370039304315,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0029401206211174783,
+      "loss": 3.4736,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17466146997594564,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.002940000374630648,
+      "loss": 3.459,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17474923955884814,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0029398800102726132,
+      "loss": 3.5039,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17483700914175063,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029397595280543708,
+      "loss": 3.4185,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17492477872465312,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002939638927986931,
+      "loss": 3.5347,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17501254830755558,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0029395182100813116,
+      "loss": 3.4907,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17510031789045807,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002939397374348544,
+      "loss": 3.4419,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17518808747336057,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0029392764207996696,
+      "loss": 3.4551,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17527585705626306,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0029391553494457394,
+      "loss": 3.4458,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17536362663916552,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0029390341602978168,
+      "loss": 3.4741,
+      "step": 1998
+    },
+    {
+      "epoch": 0.175451396222068,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002938912853366975,
+      "loss": 3.4922,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1755391658049705,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002938791428664298,
+      "loss": 3.4937,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1755391658049705,
+      "eval_loss": 0.10803784430027008,
+      "eval_runtime": 106.5654,
+      "eval_samples_per_second": 138.403,
+      "eval_steps_per_second": 17.304,
+      "step": 2000
+    },
+    {
+      "epoch": 0.175626935387873,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002938669886200881,
+      "loss": 3.4902,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17571470497077546,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.002938548225987829,
+      "loss": 3.4663,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17580247455367795,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029384264480362596,
+      "loss": 3.4551,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17589024413658044,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0029383045523572993,
+      "loss": 3.4751,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17597801371948293,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0029381825389620865,
+      "loss": 3.5132,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17606578330238543,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002938060407861769,
+      "loss": 3.4116,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1761535528852879,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.002937938159067507,
+      "loss": 3.4521,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17624132246819038,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0029378157925904714,
+      "loss": 3.5376,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17632909205109287,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002937693308441842,
+      "loss": 3.4844,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17641686163399536,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002937570706632811,
+      "loss": 3.4912,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17650463121689783,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002937447987174581,
+      "loss": 3.5557,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17659240079980032,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0029373251500783653,
+      "loss": 3.4722,
+      "step": 2012
+    },
+    {
+      "epoch": 0.1766801703827028,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.002937202195355387,
+      "loss": 3.4658,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1767679399656053,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029370791230168823,
+      "loss": 3.4751,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17685570954850777,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0029369559330740956,
+      "loss": 3.5049,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17694347913141026,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002936832625538284,
+      "loss": 3.4448,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17703124871431275,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.0029367092004207135,
+      "loss": 3.4458,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17711901829721524,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0029365856577326624,
+      "loss": 3.4922,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17720678788011773,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002936461997485419,
+      "loss": 3.4561,
+      "step": 2019
+    },
+    {
+      "epoch": 0.1772945574630202,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0029363382196902826,
+      "loss": 3.4619,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1773823270459227,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029362143243585634,
+      "loss": 3.4346,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17747009662882518,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002936090311501581,
+      "loss": 3.4692,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17755786621172767,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002935966181130669,
+      "loss": 3.4766,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17764563579463014,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.002935841933257167,
+      "loss": 3.4453,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17773340537753263,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00293571756789243,
+      "loss": 3.5054,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17782117496043512,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00293559308504782,
+      "loss": 3.4492,
+      "step": 2026
+    },
+    {
+      "epoch": 0.1779089445433376,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0029354684847347125,
+      "loss": 3.4805,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1779967141262401,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029353437669644923,
+      "loss": 3.4619,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17808448370914257,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002935218931748555,
+      "loss": 3.4648,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17817225329204506,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002935093979098308,
+      "loss": 3.5249,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17826002287494755,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002934968909025167,
+      "loss": 3.4517,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17834779245785004,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.002934843721540561,
+      "loss": 3.3975,
+      "step": 2032
+    },
+    {
+      "epoch": 0.1784355620407525,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029347184166559296,
+      "loss": 3.3975,
+      "step": 2033
+    },
+    {
+      "epoch": 0.178523331623655,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0029345929943827205,
+      "loss": 3.4956,
+      "step": 2034
+    },
+    {
+      "epoch": 0.1786111012065575,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002934467454732395,
+      "loss": 3.5049,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17869887078945998,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0029343417977164238,
+      "loss": 3.4922,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17878664037236244,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002934216023346289,
+      "loss": 3.4673,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17887440995526493,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002934090131633482,
+      "loss": 3.4692,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17896217953816743,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002933964122589507,
+      "loss": 3.4067,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17904994912106992,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002933837996225877,
+      "loss": 3.4595,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1791377187039724,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0029337117525541166,
+      "loss": 3.4551,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17922548828687487,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002933585391585762,
+      "loss": 3.5,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17931325786977736,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002933458913332358,
+      "loss": 3.4995,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17940102745267986,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002933332317805462,
+      "loss": 3.4839,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17948879703558235,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002933205605016641,
+      "loss": 3.4751,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1795765666184848,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0029330787749774737,
+      "loss": 3.48,
+      "step": 2046
+    },
+    {
+      "epoch": 0.1796643362013873,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0029329518276995483,
+      "loss": 3.4966,
+      "step": 2047
+    },
+    {
+      "epoch": 0.1797521057842898,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0029328247631944643,
+      "loss": 3.3945,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1798398753671923,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002932697581473833,
+      "loss": 3.4634,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17992764495009475,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002932570282549274,
+      "loss": 3.5034,
+      "step": 2050
+    },
+    {
+      "epoch": 0.18001541453299724,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029324428664324197,
+      "loss": 3.4131,
+      "step": 2051
+    },
+    {
+      "epoch": 0.18010318411589973,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002932315333134913,
+      "loss": 3.4644,
+      "step": 2052
+    },
+    {
+      "epoch": 0.18019095369880223,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0029321876826684063,
+      "loss": 3.4077,
+      "step": 2053
+    },
+    {
+      "epoch": 0.18027872328170472,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029320599150445634,
+      "loss": 3.4409,
+      "step": 2054
+    },
+    {
+      "epoch": 0.18036649286460718,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0029319320302750593,
+      "loss": 3.4224,
+      "step": 2055
+    },
+    {
+      "epoch": 0.18045426244750967,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0029318040283715786,
+      "loss": 3.5952,
+      "step": 2056
+    },
+    {
+      "epoch": 0.18054203203041216,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002931675909345817,
+      "loss": 3.5415,
+      "step": 2057
+    },
+    {
+      "epoch": 0.18062980161331466,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.002931547673209482,
+      "loss": 3.4814,
+      "step": 2058
+    },
+    {
+      "epoch": 0.18071757119621712,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0029314193199742908,
+      "loss": 3.4731,
+      "step": 2059
+    },
+    {
+      "epoch": 0.1808053407791196,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002931290849651971,
+      "loss": 3.5718,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1808931103620221,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0029311622622542615,
+      "loss": 3.4531,
+      "step": 2061
+    },
+    {
+      "epoch": 0.1809808799449246,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029310335577929117,
+      "loss": 3.4385,
+      "step": 2062
+    },
+    {
+      "epoch": 0.18106864952782709,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029309047362796816,
+      "loss": 3.5044,
+      "step": 2063
+    },
+    {
+      "epoch": 0.18115641911072955,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002930775797726342,
+      "loss": 3.5122,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18124418869363204,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029306467421446746,
+      "loss": 3.4189,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18133195827653453,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0029305175695464713,
+      "loss": 3.4487,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18141972785943702,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002930388279943536,
+      "loss": 3.5112,
+      "step": 2067
+    },
+    {
+      "epoch": 0.1815074974423395,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.002930258873347681,
+      "loss": 3.3931,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18159526702524198,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029301293497707307,
+      "loss": 3.5283,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18168303660814447,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029299997092245207,
+      "loss": 3.4883,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18177080619104696,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0029298699517208964,
+      "loss": 3.48,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18185857577394943,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002929740077271714,
+      "loss": 3.5723,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18194634535685192,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0029296100858888408,
+      "loss": 3.4614,
+      "step": 2073
+    },
+    {
+      "epoch": 0.1820341149397544,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002929479977584154,
+      "loss": 3.4033,
+      "step": 2074
+    },
+    {
+      "epoch": 0.1821218845226569,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0029293497523695425,
+      "loss": 3.5215,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1822096541055594,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.002929219410256905,
+      "loss": 3.5264,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18229742368846186,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029290889512581515,
+      "loss": 3.3921,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18238519327136435,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029289583753852023,
+      "loss": 3.4336,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18247296285426684,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002928827682649989,
+      "loss": 3.4448,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18256073243716933,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0029286968730644525,
+      "loss": 3.4385,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1826485020200718,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0029285659466405456,
+      "loss": 3.479,
+      "step": 2081
+    },
+    {
+      "epoch": 0.1827362716029743,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0029284349033902317,
+      "loss": 3.4302,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18282404118587678,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029283037433254847,
+      "loss": 3.4639,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18291181076877927,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002928172466458288,
+      "loss": 3.4341,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18299958035168173,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002928041072800638,
+      "loss": 3.4185,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18308734993458423,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002927909562364541,
+      "loss": 3.4106,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18317511951748672,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0029277779351620115,
+      "loss": 3.4727,
+      "step": 2087
+    },
+    {
+      "epoch": 0.1832628891003892,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029276461912050786,
+      "loss": 3.4565,
+      "step": 2088
+    },
+    {
+      "epoch": 0.1833506586832917,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0029275143305057785,
+      "loss": 3.4419,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18343842826619416,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029273823530761617,
+      "loss": 3.4326,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18352619784909666,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0029272502589282852,
+      "loss": 3.4248,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18361396743199915,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0029271180480742206,
+      "loss": 3.4634,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18370173701490164,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029269857205260473,
+      "loss": 3.3569,
+      "step": 2093
+    },
+    {
+      "epoch": 0.1837895065978041,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002926853276295857,
+      "loss": 3.4683,
+      "step": 2094
+    },
+    {
+      "epoch": 0.1838772761807066,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.002926720715395751,
+      "loss": 3.4204,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1839650457636091,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0029265880378378426,
+      "loss": 3.4976,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18405281534651158,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0029264552436342544,
+      "loss": 3.4707,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18414058492941407,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0029263223327971205,
+      "loss": 3.4678,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18422835451231653,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0029261893053385848,
+      "loss": 3.5073,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18431612409521902,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002926056161270803,
+      "loss": 3.3975,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18440389367812152,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00292592290060594,
+      "loss": 3.5142,
+      "step": 2101
+    },
+    {
+      "epoch": 0.184491663261024,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0029257895233561738,
+      "loss": 3.4653,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18457943284392647,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00292565602953369,
+      "loss": 3.4092,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18466720242682896,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029255224191506875,
+      "loss": 3.4507,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18475497200973146,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002925388692219374,
+      "loss": 3.501,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18484274159263395,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002925254848751968,
+      "loss": 3.4243,
+      "step": 2106
+    },
+    {
+      "epoch": 0.1849305111755364,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0029251208887607,
+      "loss": 3.4619,
+      "step": 2107
+    },
+    {
+      "epoch": 0.1850182807584389,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0029249868122578106,
+      "loss": 3.5132,
+      "step": 2108
+    },
+    {
+      "epoch": 0.1851060503413414,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0029248526192555495,
+      "loss": 3.3867,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18519381992424389,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029247183097661795,
+      "loss": 3.3833,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18528158950714638,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0029245838838019727,
+      "loss": 3.4702,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18536935909004884,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0029244493413752115,
+      "loss": 3.5259,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18545712867295133,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.00292431468249819,
+      "loss": 3.6514,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18554489825585382,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002924179907183211,
+      "loss": 3.4302,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18563266783875632,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0029240450154425918,
+      "loss": 3.4575,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18572043742165878,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0029239100072886563,
+      "loss": 3.4155,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18580820700456127,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029237748827337403,
+      "loss": 3.5249,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18589597658746376,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0029236396417901913,
+      "loss": 3.458,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18598374617036625,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0029235042844703665,
+      "loss": 3.4849,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18607151575326872,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0029233688107866336,
+      "loss": 3.4194,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1861592853361712,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0029232332207513715,
+      "loss": 3.4756,
+      "step": 2121
+    },
+    {
+      "epoch": 0.1862470549190737,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029230975143769693,
+      "loss": 3.4126,
+      "step": 2122
+    },
+    {
+      "epoch": 0.1863348245019762,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002922961691675827,
+      "loss": 3.4731,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18642259408487868,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0029228257526603555,
+      "loss": 3.502,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18651036366778115,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002922689697342975,
+      "loss": 3.3804,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18659813325068364,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002922553525736118,
+      "loss": 3.4321,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18668590283358613,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0029224172378522267,
+      "loss": 3.499,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18677367241648862,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0029222808337037547,
+      "loss": 3.4312,
+      "step": 2128
+    },
+    {
+      "epoch": 0.1868614419993911,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0029221443133031653,
+      "loss": 3.5122,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18694921158229358,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0029220076766629318,
+      "loss": 3.4336,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18703698116519607,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0029218709237955405,
+      "loss": 3.4434,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18712475074809856,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0029217340547134864,
+      "loss": 3.4229,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18721252033100105,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0029215970694292754,
+      "loss": 3.3945,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18730028991390352,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002921459967955425,
+      "loss": 3.4824,
+      "step": 2134
+    },
+    {
+      "epoch": 0.187388059496806,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0029213227503044614,
+      "loss": 3.4556,
+      "step": 2135
+    },
+    {
+      "epoch": 0.1874758290797085,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0029211854164889234,
+      "loss": 3.375,
+      "step": 2136
+    },
+    {
+      "epoch": 0.187563598662611,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0029210479665213594,
+      "loss": 3.4873,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18765136824551346,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0029209104004143285,
+      "loss": 3.4614,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18773913782841595,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002920772718180401,
+      "loss": 3.4243,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18782690741131844,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.002920634919832157,
+      "loss": 3.5059,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18791467699422093,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0029204970053821875,
+      "loss": 3.4209,
+      "step": 2141
+    },
+    {
+      "epoch": 0.1880024465771234,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002920358974843094,
+      "loss": 3.5254,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18809021616002589,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.002920220828227489,
+      "loss": 3.4893,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18817798574292838,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0029200825655479956,
+      "loss": 3.4448,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18826575532583087,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.002919944186817247,
+      "loss": 3.5205,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18835352490873336,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.0029198056920478875,
+      "loss": 3.4556,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18844129449163582,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002919667081252571,
+      "loss": 3.4077,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18852906407453832,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0029195283544439636,
+      "loss": 3.4634,
+      "step": 2148
+    },
+    {
+      "epoch": 0.1886168336574408,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002919389511634741,
+      "loss": 3.4556,
+      "step": 2149
+    },
+    {
+      "epoch": 0.1887046032403433,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00291925055283759,
+      "loss": 3.46,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18879237282324576,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002919111478065207,
+      "loss": 3.4414,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18888014240614825,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0029189722873303,
+      "loss": 3.4263,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18896791198905075,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029188329806455874,
+      "loss": 3.4546,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18905568157195324,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0029186935580237977,
+      "loss": 3.4272,
+      "step": 2154
+    },
+    {
+      "epoch": 0.1891434511548557,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0029185540194776706,
+      "loss": 3.4277,
+      "step": 2155
+    },
+    {
+      "epoch": 0.1892312207377582,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0029184143650199564,
+      "loss": 3.522,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18931899032066069,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029182745946634155,
+      "loss": 3.4473,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18940675990356318,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002918134708420819,
+      "loss": 3.3955,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18949452948646567,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0029179947063049487,
+      "loss": 3.4072,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18958229906936813,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0029178545883285973,
+      "loss": 3.4575,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18967006865227062,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002917714354504568,
+      "loss": 3.4346,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18975783823517312,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0029175740048456742,
+      "loss": 3.4395,
+      "step": 2162
+    },
+    {
+      "epoch": 0.1898456078180756,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002917433539364739,
+      "loss": 3.4399,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18993337740097807,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0029172929580745994,
+      "loss": 3.4146,
+      "step": 2164
+    },
+    {
+      "epoch": 0.19002114698388056,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002917152260988099,
+      "loss": 3.5176,
+      "step": 2165
+    },
+    {
+      "epoch": 0.19010891656678305,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0029170114481180936,
+      "loss": 3.4482,
+      "step": 2166
+    },
+    {
+      "epoch": 0.19019668614968555,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002916870519477451,
+      "loss": 3.4028,
+      "step": 2167
+    },
+    {
+      "epoch": 0.19028445573258804,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029167294750790467,
+      "loss": 3.4268,
+      "step": 2168
+    },
+    {
+      "epoch": 0.1903722253154905,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0029165883149357696,
+      "loss": 3.418,
+      "step": 2169
+    },
+    {
+      "epoch": 0.190459994898393,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002916447039060518,
+      "loss": 3.4585,
+      "step": 2170
+    },
+    {
+      "epoch": 0.19054776448129548,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0029163056474661993,
+      "loss": 3.4556,
+      "step": 2171
+    },
+    {
+      "epoch": 0.19063553406419798,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002916164140165734,
+      "loss": 3.4546,
+      "step": 2172
+    },
+    {
+      "epoch": 0.19072330364710044,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002916022517172052,
+      "loss": 3.4966,
+      "step": 2173
+    },
+    {
+      "epoch": 0.19081107323000293,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0029158807784980937,
+      "loss": 3.4243,
+      "step": 2174
+    },
+    {
+      "epoch": 0.19089884281290542,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00291573892415681,
+      "loss": 3.4365,
+      "step": 2175
+    },
+    {
+      "epoch": 0.19098661239580791,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0029155969541611627,
+      "loss": 3.4868,
+      "step": 2176
+    },
+    {
+      "epoch": 0.19107438197871038,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0029154548685241238,
+      "loss": 3.4028,
+      "step": 2177
+    },
+    {
+      "epoch": 0.19116215156161287,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002915312667258677,
+      "loss": 3.4253,
+      "step": 2178
+    },
+    {
+      "epoch": 0.19124992114451536,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002915170350377814,
+      "loss": 3.4438,
+      "step": 2179
+    },
+    {
+      "epoch": 0.19133769072741785,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00291502791789454,
+      "loss": 3.3887,
+      "step": 2180
+    },
+    {
+      "epoch": 0.19142546031032034,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029148853698218696,
+      "loss": 3.4868,
+      "step": 2181
+    },
+    {
+      "epoch": 0.1915132298932228,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0029147427061728265,
+      "loss": 3.5698,
+      "step": 2182
+    },
+    {
+      "epoch": 0.1916009994761253,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0029145999269604476,
+      "loss": 3.4629,
+      "step": 2183
+    },
+    {
+      "epoch": 0.1916887690590278,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0029144570321977792,
+      "loss": 3.4595,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19177653864193028,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.002914314021897877,
+      "loss": 3.5801,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19186430822483275,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0029141708960738087,
+      "loss": 3.4429,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19195207780773524,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002914027654738652,
+      "loss": 3.4946,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19203984739063773,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.002913884297905496,
+      "loss": 3.4551,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19212761697354022,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0029137408255874393,
+      "loss": 3.4712,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19221538655644269,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.002913597237797591,
+      "loss": 3.374,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19230315613934518,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002913453534549071,
+      "loss": 3.4355,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19239092572224767,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002913309715855011,
+      "loss": 3.4326,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19247869530515016,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002913165781728551,
+      "loss": 3.4365,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19256646488805265,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002913021732182843,
+      "loss": 3.4517,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19265423447095512,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0029128775672310494,
+      "loss": 3.439,
+      "step": 2195
+    },
+    {
+      "epoch": 0.1927420040538576,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002912733286886343,
+      "loss": 3.4673,
+      "step": 2196
+    },
+    {
+      "epoch": 0.1928297736367601,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002912588891161907,
+      "loss": 3.3501,
+      "step": 2197
+    },
+    {
+      "epoch": 0.1929175432196626,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002912444380070936,
+      "loss": 3.4453,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19300531280256505,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002912299753626633,
+      "loss": 3.4609,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19309308238546755,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002912155011842214,
+      "loss": 3.415,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19318085196837004,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002912010154730904,
+      "loss": 3.5195,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19326862155127253,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.002911865182305939,
+      "loss": 3.5522,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19335639113417502,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0029117200945805665,
+      "loss": 3.4551,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19344416071707748,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0029115748915680426,
+      "loss": 3.5024,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19353193029997998,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.002911429573281635,
+      "loss": 3.4102,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19361969988288247,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002911284139734622,
+      "loss": 3.4062,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19370746946578496,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.002911138590940293,
+      "loss": 3.4312,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19379523904868742,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0029109929269119463,
+      "loss": 3.4761,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19388300863158991,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0029108471476628917,
+      "loss": 3.4702,
+      "step": 2209
+    },
+    {
+      "epoch": 0.1939707782144924,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0029107012532064505,
+      "loss": 3.4165,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1940585477973949,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0029105552435559523,
+      "loss": 3.4697,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19414631738029736,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0029104091187247385,
+      "loss": 3.4839,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19423408696319985,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.002910262878726162,
+      "loss": 3.5264,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19432185654610235,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0029101165235735846,
+      "loss": 3.5059,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19440962612900484,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0029099700532803784,
+      "loss": 3.3975,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19449739571190733,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0029098234678599285,
+      "loss": 3.4805,
+      "step": 2216
+    },
+    {
+      "epoch": 0.1945851652948098,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002909676767325628,
+      "loss": 3.4053,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19467293487771228,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0029095299516908807,
+      "loss": 3.3901,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19476070446061478,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.002909383020969103,
+      "loss": 3.3916,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19484847404351727,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0029092359751737187,
+      "loss": 3.4482,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19493624362641973,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0029090888143181656,
+      "loss": 3.4468,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19502401320932222,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0029089415384158887,
+      "loss": 3.4526,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19511178279222471,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.002908794147480346,
+      "loss": 3.4121,
+      "step": 2223
+    },
+    {
+      "epoch": 0.1951995523751272,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0029086466415250045,
+      "loss": 3.4795,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19528732195802967,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002908499020563343,
+      "loss": 3.396,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19537509154093216,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029083512846088495,
+      "loss": 3.4219,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19546286112383465,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002908203433675023,
+      "loss": 3.4932,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19555063070673714,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002908055467775373,
+      "loss": 3.4478,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19563840028963964,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0029079073869234203,
+      "loss": 3.4097,
+      "step": 2229
+    },
+    {
+      "epoch": 0.1957261698725421,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002907759191132695,
+      "loss": 3.3662,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1958139394554446,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0029076108804167384,
+      "loss": 3.4858,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19590170903834708,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002907462454789101,
+      "loss": 3.48,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19598947862124957,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029073139142633464,
+      "loss": 3.5522,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19607724820415204,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002907165258853047,
+      "loss": 3.4077,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19616501778705453,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0029070164885717847,
+      "loss": 3.4443,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19625278736995702,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0029068676034331546,
+      "loss": 3.4937,
+      "step": 2236
+    },
+    {
+      "epoch": 0.1963405569528595,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0029067186034507594,
+      "loss": 3.4155,
+      "step": 2237
+    },
+    {
+      "epoch": 0.196428326535762,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002906569488638215,
+      "loss": 3.4482,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19651609611866447,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0029064202590091456,
+      "loss": 3.4824,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19660386570156696,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002906270914577187,
+      "loss": 3.4355,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19669163528446945,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.002906121455355985,
+      "loss": 3.4634,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19677940486737194,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0029059718813591964,
+      "loss": 3.4624,
+      "step": 2242
+    },
+    {
+      "epoch": 0.1968671744502744,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0029058221926004883,
+      "loss": 3.4155,
+      "step": 2243
+    },
+    {
+      "epoch": 0.1969549440331769,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002905672389093538,
+      "loss": 3.4956,
+      "step": 2244
+    },
+    {
+      "epoch": 0.1970427136160794,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.002905522470852034,
+      "loss": 3.4185,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19713048319898188,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0029053724378896736,
+      "loss": 3.46,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19721825278188435,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002905222290220167,
+      "loss": 3.4629,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19730602236478684,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0029050720278572333,
+      "loss": 3.5049,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19739379194768933,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002904921650814602,
+      "loss": 3.3745,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19748156153059182,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.0029047711591060144,
+      "loss": 3.4321,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1975693311134943,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00290462055274522,
+      "loss": 3.5278,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19765710069639678,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0029044698317459814,
+      "loss": 3.4204,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19774487027929927,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0029043189961220697,
+      "loss": 3.4165,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19783263986220176,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0029041680458872675,
+      "loss": 3.4624,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19792040944510425,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0029040169810553677,
+      "loss": 3.4453,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19800817902800671,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0029038658016401733,
+      "loss": 3.498,
+      "step": 2256
+    },
+    {
+      "epoch": 0.1980959486109092,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0029037145076554983,
+      "loss": 3.4434,
+      "step": 2257
+    },
+    {
+      "epoch": 0.1981837181938117,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0029035630991151662,
+      "loss": 3.4155,
+      "step": 2258
+    },
+    {
+      "epoch": 0.1982714877767142,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0029034115760330124,
+      "loss": 3.4575,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19835925735961665,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.002903259938422882,
+      "loss": 3.4502,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19844702694251914,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.00290310818629863,
+      "loss": 3.3965,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19853479652542164,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0029029563196741235,
+      "loss": 3.5376,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19862256610832413,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0029028043385632375,
+      "loss": 3.479,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19871033569122662,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.00290265224297986,
+      "loss": 3.4287,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19879810527412908,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0029025000329378882,
+      "loss": 3.5015,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19888587485703157,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00290234770845123,
+      "loss": 3.4316,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19897364443993407,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.002902195269533804,
+      "loss": 3.4346,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19906141402283656,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0029020427161995386,
+      "loss": 3.4316,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19914918360573902,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002901890048462373,
+      "loss": 3.4224,
+      "step": 2269
+    },
+    {
+      "epoch": 0.1992369531886415,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002901737266336257,
+      "loss": 3.4771,
+      "step": 2270
+    },
+    {
+      "epoch": 0.199324722771544,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0029015843698351517,
+      "loss": 3.46,
+      "step": 2271
+    },
+    {
+      "epoch": 0.1994124923544465,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002901431358973026,
+      "loss": 3.4277,
+      "step": 2272
+    },
+    {
+      "epoch": 0.199500261937349,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0029012782337638618,
+      "loss": 3.4268,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19958803152025145,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002901124994221651,
+      "loss": 3.3931,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19967580110315394,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0029009716403603954,
+      "loss": 3.4702,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19976357068605644,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002900818172194107,
+      "loss": 3.4385,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19985134026895893,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0029006645897368087,
+      "loss": 3.3843,
+      "step": 2277
+    },
+    {
+      "epoch": 0.1999391098518614,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002900510893002534,
+      "loss": 3.439,
+      "step": 2278
+    },
+    {
+      "epoch": 0.20002687943476388,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0029003570820053267,
+      "loss": 3.4648,
+      "step": 2279
+    },
+    {
+      "epoch": 0.20011464901766637,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0029002031567592406,
+      "loss": 3.4482,
+      "step": 2280
+    },
+    {
+      "epoch": 0.20020241860056887,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002900049117278341,
+      "loss": 3.5005,
+      "step": 2281
+    },
+    {
+      "epoch": 0.20029018818347133,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002899894963576702,
+      "loss": 3.4634,
+      "step": 2282
+    },
+    {
+      "epoch": 0.20037795776637382,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.00289974069566841,
+      "loss": 3.436,
+      "step": 2283
+    },
+    {
+      "epoch": 0.2004657273492763,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002899586313567561,
+      "loss": 3.4941,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2005534969321788,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0028994318172882597,
+      "loss": 3.499,
+      "step": 2285
+    },
+    {
+      "epoch": 0.2006412665150813,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0028992772068446248,
+      "loss": 3.5347,
+      "step": 2286
+    },
+    {
+      "epoch": 0.20072903609798376,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0028991224822507828,
+      "loss": 3.4521,
+      "step": 2287
+    },
+    {
+      "epoch": 0.20081680568088625,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002898967643520871,
+      "loss": 3.4697,
+      "step": 2288
+    },
+    {
+      "epoch": 0.20090457526378874,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0028988126906690373,
+      "loss": 3.4673,
+      "step": 2289
+    },
+    {
+      "epoch": 0.20099234484669123,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028986576237094415,
+      "loss": 3.4282,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2010801144295937,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0028985024426562518,
+      "loss": 3.4717,
+      "step": 2291
+    },
+    {
+      "epoch": 0.2011678840124962,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0028983471475236473,
+      "loss": 3.4692,
+      "step": 2292
+    },
+    {
+      "epoch": 0.20125565359539868,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002898191738325818,
+      "loss": 3.4702,
+      "step": 2293
+    },
+    {
+      "epoch": 0.20134342317830117,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.002898036215076964,
+      "loss": 3.4619,
+      "step": 2294
+    },
+    {
+      "epoch": 0.20143119276120364,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0028978805777912964,
+      "loss": 3.4219,
+      "step": 2295
+    },
+    {
+      "epoch": 0.20151896234410613,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0028977248264830356,
+      "loss": 3.437,
+      "step": 2296
+    },
+    {
+      "epoch": 0.20160673192700862,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0028975689611664135,
+      "loss": 3.5034,
+      "step": 2297
+    },
+    {
+      "epoch": 0.2016945015099111,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0028974129818556717,
+      "loss": 3.4717,
+      "step": 2298
+    },
+    {
+      "epoch": 0.2017822710928136,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.002897256888565062,
+      "loss": 3.3833,
+      "step": 2299
+    },
+    {
+      "epoch": 0.20187004067571607,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.002897100681308849,
+      "loss": 3.5122,
+      "step": 2300
+    },
+    {
+      "epoch": 0.20195781025861856,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028969443601013036,
+      "loss": 3.3984,
+      "step": 2301
+    },
+    {
+      "epoch": 0.20204557984152105,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0028967879249567107,
+      "loss": 3.5039,
+      "step": 2302
+    },
+    {
+      "epoch": 0.20213334942442354,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028966313758893632,
+      "loss": 3.3872,
+      "step": 2303
+    },
+    {
+      "epoch": 0.202221119007326,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002896474712913567,
+      "loss": 3.4946,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2023088885902285,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0028963179360436356,
+      "loss": 3.5171,
+      "step": 2305
+    },
+    {
+      "epoch": 0.202396658173131,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0028961610452938945,
+      "loss": 3.4087,
+      "step": 2306
+    },
+    {
+      "epoch": 0.20248442775603348,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028960040406786794,
+      "loss": 3.4688,
+      "step": 2307
+    },
+    {
+      "epoch": 0.20257219733893597,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0028958469222123357,
+      "loss": 3.3843,
+      "step": 2308
+    },
+    {
+      "epoch": 0.20265996692183844,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002895689689909221,
+      "loss": 3.4883,
+      "step": 2309
+    },
+    {
+      "epoch": 0.20274773650474093,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.002895532343783701,
+      "loss": 3.4023,
+      "step": 2310
+    },
+    {
+      "epoch": 0.20283550608764342,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0028953748838501533,
+      "loss": 3.5459,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2029232756705459,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0028952173101229653,
+      "loss": 3.4268,
+      "step": 2312
+    },
+    {
+      "epoch": 0.20301104525344837,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0028950596226165358,
+      "loss": 3.4956,
+      "step": 2313
+    },
+    {
+      "epoch": 0.20309881483635087,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0028949018213452714,
+      "loss": 3.3984,
+      "step": 2314
+    },
+    {
+      "epoch": 0.20318658441925336,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002894743906323593,
+      "loss": 3.4775,
+      "step": 2315
+    },
+    {
+      "epoch": 0.20327435400215585,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002894585877565928,
+      "loss": 3.4385,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2033621235850583,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0028944277350867174,
+      "loss": 3.4282,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2034498931679608,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.00289426947890041,
+      "loss": 3.4233,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2035376627508633,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0028941111090214668,
+      "loss": 3.5132,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2036254323337658,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0028939526254643582,
+      "loss": 3.4189,
+      "step": 2320
+    },
+    {
+      "epoch": 0.20371320191666828,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002893794028243566,
+      "loss": 3.4399,
+      "step": 2321
+    },
+    {
+      "epoch": 0.20380097149957074,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00289363531737358,
+      "loss": 3.4937,
+      "step": 2322
+    },
+    {
+      "epoch": 0.20388874108247323,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002893476492868904,
+      "loss": 3.4194,
+      "step": 2323
+    },
+    {
+      "epoch": 0.20397651066537573,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0028933175547440496,
+      "loss": 3.4517,
+      "step": 2324
+    },
+    {
+      "epoch": 0.20406428024827822,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0028931585030135395,
+      "loss": 3.4766,
+      "step": 2325
+    },
+    {
+      "epoch": 0.20415204983118068,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002892999337691906,
+      "loss": 3.4443,
+      "step": 2326
+    },
+    {
+      "epoch": 0.20423981941408317,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0028928400587936937,
+      "loss": 3.4175,
+      "step": 2327
+    },
+    {
+      "epoch": 0.20432758899698567,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0028926806663334558,
+      "loss": 3.4194,
+      "step": 2328
+    },
+    {
+      "epoch": 0.20441535857988816,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002892521160325756,
+      "loss": 3.4009,
+      "step": 2329
+    },
+    {
+      "epoch": 0.20450312816279062,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00289236154078517,
+      "loss": 3.4878,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2045908977456931,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028922018077262818,
+      "loss": 3.4453,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2046786673285956,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0028920419611636865,
+      "loss": 3.438,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2047664369114981,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0028918820011119904,
+      "loss": 3.5034,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2048542064944006,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0028917219275858093,
+      "loss": 3.5103,
+      "step": 2334
+    },
+    {
+      "epoch": 0.20494197607730305,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0028915617405997703,
+      "loss": 3.479,
+      "step": 2335
+    },
+    {
+      "epoch": 0.20502974566020554,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0028914014401685085,
+      "loss": 3.4551,
+      "step": 2336
+    },
+    {
+      "epoch": 0.20511751524310803,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0028912410263066727,
+      "loss": 3.4141,
+      "step": 2337
+    },
+    {
+      "epoch": 0.20520528482601053,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002891080499028919,
+      "loss": 3.4663,
+      "step": 2338
+    },
+    {
+      "epoch": 0.205293054408913,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0028909198583499164,
+      "loss": 3.4326,
+      "step": 2339
+    },
+    {
+      "epoch": 0.20538082399181548,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002890759104284343,
+      "loss": 3.4248,
+      "step": 2340
+    },
+    {
+      "epoch": 0.20546859357471797,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.002890598236846887,
+      "loss": 3.4585,
+      "step": 2341
+    },
+    {
+      "epoch": 0.20555636315762046,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.002890437256052247,
+      "loss": 3.4766,
+      "step": 2342
+    },
+    {
+      "epoch": 0.20564413274052296,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0028902761619151326,
+      "loss": 3.3848,
+      "step": 2343
+    },
+    {
+      "epoch": 0.20573190232342542,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.002890114954450264,
+      "loss": 3.4536,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2058196719063279,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0028899536336723713,
+      "loss": 3.4189,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2059074414892304,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0028897921995961933,
+      "loss": 3.4263,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2059952110721329,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002889630652236482,
+      "loss": 3.4282,
+      "step": 2347
+    },
+    {
+      "epoch": 0.20608298065503536,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028894689916079985,
+      "loss": 3.5073,
+      "step": 2348
+    },
+    {
+      "epoch": 0.20617075023793785,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.002889307217725514,
+      "loss": 3.3706,
+      "step": 2349
+    },
+    {
+      "epoch": 0.20625851982084034,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.00288914533060381,
+      "loss": 3.4287,
+      "step": 2350
+    },
+    {
+      "epoch": 0.20634628940374283,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002888983330257679,
+      "loss": 3.4678,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2064340589866453,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002888821216701923,
+      "loss": 3.4551,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2065218285695478,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0028886589899513554,
+      "loss": 3.4834,
+      "step": 2353
+    },
+    {
+      "epoch": 0.20660959815245028,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002888496650020799,
+      "loss": 3.4683,
+      "step": 2354
+    },
+    {
+      "epoch": 0.20669736773535277,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0028883341969250867,
+      "loss": 3.4409,
+      "step": 2355
+    },
+    {
+      "epoch": 0.20678513731825526,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002888171630679064,
+      "loss": 3.5059,
+      "step": 2356
+    },
+    {
+      "epoch": 0.20687290690115773,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002888008951297583,
+      "loss": 3.4058,
+      "step": 2357
+    },
+    {
+      "epoch": 0.20696067648406022,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002887846158795509,
+      "loss": 3.4458,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2070484460669627,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002887683253187718,
+      "loss": 3.499,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2071362156498652,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0028875202344890937,
+      "loss": 3.438,
+      "step": 2360
+    },
+    {
+      "epoch": 0.20722398523276767,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0028873571027145327,
+      "loss": 3.4663,
+      "step": 2361
+    },
+    {
+      "epoch": 0.20731175481567016,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.00288719385787894,
+      "loss": 3.415,
+      "step": 2362
+    },
+    {
+      "epoch": 0.20739952439857265,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002887030499997231,
+      "loss": 3.4365,
+      "step": 2363
+    },
+    {
+      "epoch": 0.20748729398147514,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0028868670290843343,
+      "loss": 3.4541,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2075750635643776,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0028867034451551857,
+      "loss": 3.4653,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2076628331472801,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002886539748224732,
+      "loss": 3.4951,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2077506027301826,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0028863759383079314,
+      "loss": 3.4609,
+      "step": 2367
+    },
+    {
+      "epoch": 0.20783837231308508,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002886212015419751,
+      "loss": 3.4009,
+      "step": 2368
+    },
+    {
+      "epoch": 0.20792614189598757,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0028860479795751693,
+      "loss": 3.4448,
+      "step": 2369
+    },
+    {
+      "epoch": 0.20801391147889003,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0028858838307891745,
+      "loss": 3.4902,
+      "step": 2370
+    },
+    {
+      "epoch": 0.20810168106179253,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0028857195690767654,
+      "loss": 3.437,
+      "step": 2371
+    },
+    {
+      "epoch": 0.20818945064469502,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0028855551944529517,
+      "loss": 3.4888,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2082772202275975,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0028853907069327518,
+      "loss": 3.3804,
+      "step": 2373
+    },
+    {
+      "epoch": 0.20836498981049997,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0028852261065311965,
+      "loss": 3.4805,
+      "step": 2374
+    },
+    {
+      "epoch": 0.20845275939340246,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002885061393263325,
+      "loss": 3.4531,
+      "step": 2375
+    },
+    {
+      "epoch": 0.20854052897630496,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.002884896567144188,
+      "loss": 3.4077,
+      "step": 2376
+    },
+    {
+      "epoch": 0.20862829855920745,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002884731628188846,
+      "loss": 3.4448,
+      "step": 2377
+    },
+    {
+      "epoch": 0.20871606814210994,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0028845665764123704,
+      "loss": 3.5073,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2088038377250124,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0028844014118298415,
+      "loss": 3.4341,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2088916073079149,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028842361344563515,
+      "loss": 3.4751,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2089793768908174,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.002884070744307003,
+      "loss": 3.4058,
+      "step": 2381
+    },
+    {
+      "epoch": 0.20906714647371988,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.002883905241396907,
+      "loss": 3.52,
+      "step": 2382
+    },
+    {
+      "epoch": 0.20915491605662234,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0028837396257411873,
+      "loss": 3.48,
+      "step": 2383
+    },
+    {
+      "epoch": 0.20924268563952483,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0028835738973549746,
+      "loss": 3.4458,
+      "step": 2384
+    },
+    {
+      "epoch": 0.20933045522242733,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0028834080562534143,
+      "loss": 3.4141,
+      "step": 2385
+    },
+    {
+      "epoch": 0.20941822480532982,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002883242102451658,
+      "loss": 3.4731,
+      "step": 2386
+    },
+    {
+      "epoch": 0.20950599438823228,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002883076035964871,
+      "loss": 3.4185,
+      "step": 2387
+    },
+    {
+      "epoch": 0.20959376397113477,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002882909856808226,
+      "loss": 3.4399,
+      "step": 2388
+    },
+    {
+      "epoch": 0.20968153355403726,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0028827435649969074,
+      "loss": 3.4312,
+      "step": 2389
+    },
+    {
+      "epoch": 0.20976930313693976,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.002882577160546111,
+      "loss": 3.5225,
+      "step": 2390
+    },
+    {
+      "epoch": 0.20985707271984225,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0028824106434710405,
+      "loss": 3.3955,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2099448423027447,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0028822440137869112,
+      "loss": 3.4634,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2100326118856472,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0028820772715089493,
+      "loss": 3.4316,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2101203814685497,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002881910416652389,
+      "loss": 3.4263,
+      "step": 2394
+    },
+    {
+      "epoch": 0.21020815105145219,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0028817434492324776,
+      "loss": 3.4219,
+      "step": 2395
+    },
+    {
+      "epoch": 0.21029592063435465,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0028815763692644717,
+      "loss": 3.4932,
+      "step": 2396
+    },
+    {
+      "epoch": 0.21038369021725714,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0028814091767636373,
+      "loss": 3.4399,
+      "step": 2397
+    },
+    {
+      "epoch": 0.21047145980015963,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0028812418717452507,
+      "loss": 3.3999,
+      "step": 2398
+    },
+    {
+      "epoch": 0.21055922938306212,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0028810744542246,
+      "loss": 3.4419,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2106469989659646,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0028809069242169824,
+      "loss": 3.4209,
+      "step": 2400
+    },
+    {
+      "epoch": 0.21073476854886708,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0028807392817377054,
+      "loss": 3.4194,
+      "step": 2401
+    },
+    {
+      "epoch": 0.21082253813176957,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0028805715268020874,
+      "loss": 3.4775,
+      "step": 2402
+    },
+    {
+      "epoch": 0.21091030771467206,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0028804036594254558,
+      "loss": 3.4927,
+      "step": 2403
+    },
+    {
+      "epoch": 0.21099807729757455,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0028802356796231504,
+      "loss": 3.4565,
+      "step": 2404
+    },
+    {
+      "epoch": 0.21108584688047702,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0028800675874105187,
+      "loss": 3.416,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2111736164633795,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0028798993828029213,
+      "loss": 3.4688,
+      "step": 2406
+    },
+    {
+      "epoch": 0.211261386046282,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0028797310658157257,
+      "loss": 3.3716,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2113491556291845,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0028795626364643136,
+      "loss": 3.4014,
+      "step": 2408
+    },
+    {
+      "epoch": 0.21143692521208696,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0028793940947640733,
+      "loss": 3.4014,
+      "step": 2409
+    },
+    {
+      "epoch": 0.21152469479498945,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0028792254407304053,
+      "loss": 3.4087,
+      "step": 2410
+    },
+    {
+      "epoch": 0.21161246437789194,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0028790566743787205,
+      "loss": 3.333,
+      "step": 2411
+    },
+    {
+      "epoch": 0.21170023396079443,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002878887795724439,
+      "loss": 3.4766,
+      "step": 2412
+    },
+    {
+      "epoch": 0.21178800354369692,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002878718804782993,
+      "loss": 3.457,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2118757731265994,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002878549701569822,
+      "loss": 3.4653,
+      "step": 2414
+    },
+    {
+      "epoch": 0.21196354270950188,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.002878380486100378,
+      "loss": 3.457,
+      "step": 2415
+    },
+    {
+      "epoch": 0.21205131229240437,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0028782111583901234,
+      "loss": 3.4307,
+      "step": 2416
+    },
+    {
+      "epoch": 0.21213908187530686,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.00287804171845453,
+      "loss": 3.4927,
+      "step": 2417
+    },
+    {
+      "epoch": 0.21222685145820933,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.002877872166309079,
+      "loss": 3.4785,
+      "step": 2418
+    },
+    {
+      "epoch": 0.21231462104111182,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0028777025019692637,
+      "loss": 3.4463,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2124023906240143,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0028775327254505875,
+      "loss": 3.4453,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2124901602069168,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0028773628367685625,
+      "loss": 3.4727,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21257792978981926,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002877192835938712,
+      "loss": 3.4712,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21266569937272176,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0028770227229765696,
+      "loss": 3.4614,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21275346895562425,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.002876852497897679,
+      "loss": 3.4429,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21284123853852674,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0028766821607175944,
+      "loss": 3.4248,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21292900812142923,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.002876511711451879,
+      "loss": 3.4917,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2130167777043317,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0028763411501161094,
+      "loss": 3.478,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2131045472872342,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002876170476725869,
+      "loss": 3.3965,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21319231687013668,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002875999691296752,
+      "loss": 3.4756,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21328008645303917,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0028758287938443644,
+      "loss": 3.4561,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21336785603594163,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0028756577843843216,
+      "loss": 3.5249,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21345562561884412,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.002875486662932249,
+      "loss": 3.3701,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21354339520174662,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0028753154295037835,
+      "loss": 3.4336,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2136311647846491,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0028751440841145695,
+      "loss": 3.4404,
+      "step": 2434
+    },
+    {
+      "epoch": 0.21371893436755157,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002874972626780265,
+      "loss": 3.4189,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21380670395045406,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.002874801057516535,
+      "loss": 3.4692,
+      "step": 2436
+    },
+    {
+      "epoch": 0.21389447353335655,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0028746293763390576,
+      "loss": 3.4106,
+      "step": 2437
+    },
+    {
+      "epoch": 0.21398224311625905,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.002874457583263519,
+      "loss": 3.5063,
+      "step": 2438
+    },
+    {
+      "epoch": 0.21407001269916154,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002874285678305617,
+      "loss": 3.3936,
+      "step": 2439
+    },
+    {
+      "epoch": 0.214157782282064,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0028741136614810585,
+      "loss": 3.3892,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2142455518649665,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002873941532805562,
+      "loss": 3.5195,
+      "step": 2441
+    },
+    {
+      "epoch": 0.21433332144786899,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002873769292294855,
+      "loss": 3.3906,
+      "step": 2442
+    },
+    {
+      "epoch": 0.21442109103077148,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0028735969399646755,
+      "loss": 3.4077,
+      "step": 2443
+    },
+    {
+      "epoch": 0.21450886061367394,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0028734244758307713,
+      "loss": 3.48,
+      "step": 2444
+    },
+    {
+      "epoch": 0.21459663019657643,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0028732518999089024,
+      "loss": 3.5264,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21468439977947892,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002873079212214836,
+      "loss": 3.4292,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21477216936238142,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002872906412764353,
+      "loss": 3.3848,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2148599389452839,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.00287273350157324,
+      "loss": 3.3955,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21494770852818637,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002872560478657299,
+      "loss": 3.4937,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21503547811108886,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0028723873440323383,
+      "loss": 3.4204,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21512324769399135,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0028722140977141776,
+      "loss": 3.3975,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21521101727689385,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0028720407397186476,
+      "loss": 3.4092,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2152987868597963,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002871867270061588,
+      "loss": 3.4858,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2153865564426988,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0028716936887588497,
+      "loss": 3.4526,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2154743260256013,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002871519995826293,
+      "loss": 3.4341,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21556209560850378,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028713461912797883,
+      "loss": 3.3979,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21564986519140625,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002871172275135218,
+      "loss": 3.4644,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21573763477430874,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028709982474084722,
+      "loss": 3.4556,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21582540435721123,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002870824108115452,
+      "loss": 3.4531,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21591317394011372,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002870649857272071,
+      "loss": 3.4585,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21600094352301621,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0028704754948942488,
+      "loss": 3.4072,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21608871310591868,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0028703010209979187,
+      "loss": 3.4458,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21617648268882117,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0028701264355990223,
+      "loss": 3.4634,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21626425227172366,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0028699517387135126,
+      "loss": 3.4399,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21635202185462615,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0028697769303573516,
+      "loss": 3.418,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21643979143752862,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002869602010546512,
+      "loss": 3.4546,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2165275610204311,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0028694269792969777,
+      "loss": 3.4209,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2166153306033336,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0028692518366247406,
+      "loss": 3.3745,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2167031001862361,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0028690765825458045,
+      "loss": 3.46,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21679086976913856,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.002868901217076183,
+      "loss": 3.4111,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21687863935204105,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0028687257402319,
+      "loss": 3.4761,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21696640893494354,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0028685501520289884,
+      "loss": 3.4316,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21705417851784603,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0028683744524834937,
+      "loss": 3.4961,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21714194810074852,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002868198641611469,
+      "loss": 3.4399,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21722971768365099,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0028680227194289786,
+      "loss": 3.4761,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21731748726655348,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002867846685952098,
+      "loss": 3.3774,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21740525684945597,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0028676705411969108,
+      "loss": 3.3774,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21749302643235846,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0028674942851795123,
+      "loss": 3.4194,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21758079601526092,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002867317917916008,
+      "loss": 3.4238,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21766856559816342,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0028671414394225124,
+      "loss": 3.4023,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2177563351810659,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0028669648497151516,
+      "loss": 3.4663,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2178441047639684,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0028667881488100605,
+      "loss": 3.4346,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2179318743468709,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0028666113367233847,
+      "loss": 3.4834,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21801964392977335,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0028664344134712813,
+      "loss": 3.4619,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21810741351267585,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0028662573790699144,
+      "loss": 3.4067,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21819518309557834,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002866080233535461,
+      "loss": 3.4316,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21828295267848083,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0028659029768841084,
+      "loss": 3.3687,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2183707222613833,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0028657256091320516,
+      "loss": 3.4829,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21845849184428578,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.002865548130295498,
+      "loss": 3.4253,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21854626142718828,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002865370540390664,
+      "loss": 3.4331,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21863403101009077,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0028651928394337767,
+      "loss": 3.4067,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21872180059299323,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002865015027441074,
+      "loss": 3.373,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21880957017589572,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002864837104428802,
+      "loss": 3.4033,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21889733975879822,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0028646590704132177,
+      "loss": 3.4424,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2189851093417007,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00286448092541059,
+      "loss": 3.4707,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2190728789246032,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0028643026694371958,
+      "loss": 3.4062,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21916064850750566,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0028641243025093227,
+      "loss": 3.4058,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21924841809040815,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0028639458246432687,
+      "loss": 3.4316,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21933618767331065,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0028637672358553423,
+      "loss": 3.4287,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21942395725621314,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0028635885361618612,
+      "loss": 3.3682,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2195117268391156,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002863409725579154,
+      "loss": 3.4468,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2195994964220181,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002863230804123559,
+      "loss": 3.4883,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21968726600492058,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002863051771811425,
+      "loss": 3.3164,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21977503558782308,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0028628726286591107,
+      "loss": 3.4922,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21986280517072554,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002862693374682984,
+      "loss": 3.3794,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21995057475362803,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.002862514009899426,
+      "loss": 3.4805,
+      "step": 2506
+    },
+    {
+      "epoch": 0.22003834433653052,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0028623345343248233,
+      "loss": 3.436,
+      "step": 2507
+    },
+    {
+      "epoch": 0.22012611391943301,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0028621549479755774,
+      "loss": 3.4829,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2202138835023355,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002861975250868096,
+      "loss": 3.3848,
+      "step": 2509
+    },
+    {
+      "epoch": 0.22030165308523797,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002861795443018799,
+      "loss": 3.4673,
+      "step": 2510
+    },
+    {
+      "epoch": 0.22038942266814046,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002861615524444116,
+      "loss": 3.4312,
+      "step": 2511
+    },
+    {
+      "epoch": 0.22047719225104295,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0028614354951604867,
+      "loss": 3.416,
+      "step": 2512
+    },
+    {
+      "epoch": 0.22056496183394544,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002861255355184362,
+      "loss": 3.4409,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2206527314168479,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0028610751045322,
+      "loss": 3.4106,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2207405009997504,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0028608947432204723,
+      "loss": 3.3721,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2208282705826529,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.002860714271265658,
+      "loss": 3.418,
+      "step": 2516
+    },
+    {
+      "epoch": 0.22091604016555538,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.0028605336886842474,
+      "loss": 3.4668,
+      "step": 2517
+    },
+    {
+      "epoch": 0.22100380974845787,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0028603529954927412,
+      "loss": 3.4214,
+      "step": 2518
+    },
+    {
+      "epoch": 0.22109157933136034,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0028601721917076504,
+      "loss": 3.4077,
+      "step": 2519
+    },
+    {
+      "epoch": 0.22117934891426283,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0028599912773454948,
+      "loss": 3.4478,
+      "step": 2520
+    },
+    {
+      "epoch": 0.22126711849716532,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0028598102524228053,
+      "loss": 3.4204,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2213548880800678,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002859629116956123,
+      "loss": 3.4243,
+      "step": 2522
+    },
+    {
+      "epoch": 0.22144265766297028,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002859447870961998,
+      "loss": 3.4629,
+      "step": 2523
+    },
+    {
+      "epoch": 0.22153042724587277,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0028592665144569927,
+      "loss": 3.3994,
+      "step": 2524
+    },
+    {
+      "epoch": 0.22161819682877526,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002859085047457676,
+      "loss": 3.3896,
+      "step": 2525
+    },
+    {
+      "epoch": 0.22170596641167775,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0028589034699806315,
+      "loss": 3.4531,
+      "step": 2526
+    },
+    {
+      "epoch": 0.22179373599458022,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.002858721782042449,
+      "loss": 3.4619,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2218815055774827,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0028585399836597303,
+      "loss": 3.4253,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2219692751603852,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0028583580748490866,
+      "loss": 3.4458,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2220570447432877,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00285817605562714,
+      "loss": 3.4341,
+      "step": 2530
+    },
+    {
+      "epoch": 0.22214481432619018,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002857993926010522,
+      "loss": 3.4663,
+      "step": 2531
+    },
+    {
+      "epoch": 0.22223258390909265,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0028578116860158735,
+      "loss": 3.4121,
+      "step": 2532
+    },
+    {
+      "epoch": 0.22232035349199514,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002857629335659847,
+      "loss": 3.5317,
+      "step": 2533
+    },
+    {
+      "epoch": 0.22240812307489763,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0028574468749591048,
+      "loss": 3.4668,
+      "step": 2534
+    },
+    {
+      "epoch": 0.22249589265780012,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002857264303930318,
+      "loss": 3.4609,
+      "step": 2535
+    },
+    {
+      "epoch": 0.22258366224070258,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002857081622590169,
+      "loss": 3.4302,
+      "step": 2536
+    },
+    {
+      "epoch": 0.22267143182360508,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0028568988309553506,
+      "loss": 3.4307,
+      "step": 2537
+    },
+    {
+      "epoch": 0.22275920140650757,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002856715929042564,
+      "loss": 3.4878,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22284697098941006,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0028565329168685222,
+      "loss": 3.4292,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22293474057231252,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0028563497944499474,
+      "loss": 3.3599,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22302251015521501,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.002856166561803572,
+      "loss": 3.3892,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2231102797381175,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0028559832189461376,
+      "loss": 3.3945,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22319804932102,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0028557997658943987,
+      "loss": 3.4219,
+      "step": 2543
+    },
+    {
+      "epoch": 0.2232858189039225,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0028556162026651165,
+      "loss": 3.4282,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22337358848682495,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.002855432529275064,
+      "loss": 3.4712,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22346135806972744,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002855248745741025,
+      "loss": 3.4575,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22354912765262994,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0028550648520797904,
+      "loss": 3.395,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22363689723553243,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0028548808483081648,
+      "loss": 3.5122,
+      "step": 2548
+    },
+    {
+      "epoch": 0.2237246668184349,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0028546967344429603,
+      "loss": 3.4648,
+      "step": 2549
+    },
+    {
+      "epoch": 0.22381243640133738,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0028545125105010004,
+      "loss": 3.4531,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22390020598423988,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002854328176499118,
+      "loss": 3.3857,
+      "step": 2551
+    },
+    {
+      "epoch": 0.22398797556714237,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0028541437324541563,
+      "loss": 3.4209,
+      "step": 2552
+    },
+    {
+      "epoch": 0.22407574515004486,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002853959178382968,
+      "loss": 3.502,
+      "step": 2553
+    },
+    {
+      "epoch": 0.22416351473294732,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.0028537745143024174,
+      "loss": 3.4678,
+      "step": 2554
+    },
+    {
+      "epoch": 0.2242512843158498,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0028535897402293775,
+      "loss": 3.4595,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2243390538987523,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.002853404856180731,
+      "loss": 3.4336,
+      "step": 2556
+    },
+    {
+      "epoch": 0.2244268234816548,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0028532198621733715,
+      "loss": 3.4336,
+      "step": 2557
+    },
+    {
+      "epoch": 0.22451459306455726,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0028530347582242026,
+      "loss": 3.4165,
+      "step": 2558
+    },
+    {
+      "epoch": 0.22460236264745975,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002852849544350139,
+      "loss": 3.3853,
+      "step": 2559
+    },
+    {
+      "epoch": 0.22469013223036224,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0028526642205681015,
+      "loss": 3.3735,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22477790181326474,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002852478786895026,
+      "loss": 3.4316,
+      "step": 2561
+    },
+    {
+      "epoch": 0.2248656713961672,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.002852293243347856,
+      "loss": 3.3975,
+      "step": 2562
+    },
+    {
+      "epoch": 0.2249534409790697,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002852107589943544,
+      "loss": 3.4482,
+      "step": 2563
+    },
+    {
+      "epoch": 0.22504121056197218,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002851921826699054,
+      "loss": 3.4077,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22512898014487467,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0028517359536313607,
+      "loss": 3.3501,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22521674972777717,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002851549970757447,
+      "loss": 3.4336,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22530451931067963,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0028513638780943073,
+      "loss": 3.3784,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22539228889358212,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002851177675658945,
+      "loss": 3.3843,
+      "step": 2568
+    },
+    {
+      "epoch": 0.2254800584764846,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0028509913634683737,
+      "loss": 3.4443,
+      "step": 2569
+    },
+    {
+      "epoch": 0.2255678280593871,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002850804941539618,
+      "loss": 3.3677,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22565559764228957,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0028506184098897112,
+      "loss": 3.5356,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22574336722519206,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002850431768535698,
+      "loss": 3.5063,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22583113680809455,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0028502450174946316,
+      "loss": 3.355,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22591890639099704,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0028500581567835765,
+      "loss": 3.4697,
+      "step": 2574
+    },
+    {
+      "epoch": 0.2260066759738995,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.002849871186419607,
+      "loss": 3.4785,
+      "step": 2575
+    },
+    {
+      "epoch": 0.226094445556802,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002849684106419806,
+      "loss": 3.4062,
+      "step": 2576
+    },
+    {
+      "epoch": 0.2261822151397045,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0028494969168012686,
+      "loss": 3.4727,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22626998472260698,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0028493096175810986,
+      "loss": 3.4385,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22635775430550947,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0028491222087764097,
+      "loss": 3.4585,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22644552388841194,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002848934690404327,
+      "loss": 3.3901,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22653329347131443,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0028487470624819836,
+      "loss": 3.3833,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22662106305421692,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002848559325026524,
+      "loss": 3.3511,
+      "step": 2582
+    },
+    {
+      "epoch": 0.2267088326371194,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002848371478055102,
+      "loss": 3.4424,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22679660222002188,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0028481835215848827,
+      "loss": 3.3223,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22688437180292437,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0028479954556330394,
+      "loss": 3.4419,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22697214138582686,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0028478072802167564,
+      "loss": 3.4595,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22705991096872935,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002847618995353228,
+      "loss": 3.4082,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22714768055163184,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002847430601059658,
+      "loss": 3.5122,
+      "step": 2588
+    },
+    {
+      "epoch": 0.2272354501345343,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002847242097353261,
+      "loss": 3.4219,
+      "step": 2589
+    },
+    {
+      "epoch": 0.2273232197174368,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002847053484251261,
+      "loss": 3.4434,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2274109893003393,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0028468647617708926,
+      "loss": 3.4648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22749875888324178,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.002846675929929399,
+      "loss": 3.4888,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22758652846614424,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0028464869887440346,
+      "loss": 3.4106,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22767429804904674,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0028462979382320643,
+      "loss": 3.4912,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22776206763194923,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002846108778410761,
+      "loss": 3.4805,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22784983721485172,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.00284591950929741,
+      "loss": 3.3896,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22793760679775418,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0028457301309093045,
+      "loss": 3.436,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22802537638065667,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002845540643263749,
+      "loss": 3.4053,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22811314596355917,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0028453510463780575,
+      "loss": 3.4136,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22820091554646166,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028451613402695536,
+      "loss": 3.4272,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22828868512936415,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0028449715249555716,
+      "loss": 3.3843,
+      "step": 2601
+    },
+    {
+      "epoch": 0.2283764547122666,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0028447816004534563,
+      "loss": 3.4248,
+      "step": 2602
+    },
+    {
+      "epoch": 0.2284642242951691,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002844591566780561,
+      "loss": 3.4565,
+      "step": 2603
+    },
+    {
+      "epoch": 0.2285519938780716,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002844401423954249,
+      "loss": 3.4258,
+      "step": 2604
+    },
+    {
+      "epoch": 0.2286397634609741,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002844211171991895,
+      "loss": 3.4375,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22872753304387655,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0028440208109108823,
+      "loss": 3.3477,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22881530262677904,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.002843830340728606,
+      "loss": 3.4448,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22890307220968154,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002843639761462469,
+      "loss": 3.4102,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22899084179258403,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002843449073129885,
+      "loss": 3.4556,
+      "step": 2609
+    },
+    {
+      "epoch": 0.2290786113754865,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0028432582757482775,
+      "loss": 3.3809,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22916638095838898,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0028430673693350815,
+      "loss": 3.4941,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22925415054129147,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0028428763539077397,
+      "loss": 3.4248,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22934192012419397,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0028426852294837057,
+      "loss": 3.3477,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22942968970709646,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0028424939960804436,
+      "loss": 3.3931,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22951745928999892,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002842302653715427,
+      "loss": 3.4297,
+      "step": 2615
+    },
+    {
+      "epoch": 0.2296052288729014,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0028421112024061383,
+      "loss": 3.3154,
+      "step": 2616
+    },
+    {
+      "epoch": 0.2296929984558039,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002841919642170073,
+      "loss": 3.4248,
+      "step": 2617
+    },
+    {
+      "epoch": 0.2297807680387064,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028417279730247325,
+      "loss": 3.4292,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22986853762160886,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0028415361949876315,
+      "loss": 3.4575,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22995630720451135,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0028413443080762923,
+      "loss": 3.4268,
+      "step": 2620
+    },
+    {
+      "epoch": 0.23004407678741384,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0028411523123082496,
+      "loss": 3.417,
+      "step": 2621
+    },
+    {
+      "epoch": 0.23013184637031633,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002840960207701045,
+      "loss": 3.4536,
+      "step": 2622
+    },
+    {
+      "epoch": 0.23021961595321883,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0028407679942722337,
+      "loss": 3.4087,
+      "step": 2623
+    },
+    {
+      "epoch": 0.2303073855361213,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002840575672039376,
+      "loss": 3.4194,
+      "step": 2624
+    },
+    {
+      "epoch": 0.23039515511902378,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.0028403832410200482,
+      "loss": 3.4146,
+      "step": 2625
+    },
+    {
+      "epoch": 0.23048292470192627,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0028401907012318307,
+      "loss": 3.4263,
+      "step": 2626
+    },
+    {
+      "epoch": 0.23057069428482876,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0028399980526923174,
+      "loss": 3.4453,
+      "step": 2627
+    },
+    {
+      "epoch": 0.23065846386773123,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0028398052954191114,
+      "loss": 3.4702,
+      "step": 2628
+    },
+    {
+      "epoch": 0.23074623345063372,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0028396124294298255,
+      "loss": 3.4082,
+      "step": 2629
+    },
+    {
+      "epoch": 0.2308340030335362,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002839419454742082,
+      "loss": 3.3911,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2309217726164387,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0028392263713735134,
+      "loss": 3.3789,
+      "step": 2631
+    },
+    {
+      "epoch": 0.23100954219934117,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002839033179341763,
+      "loss": 3.4771,
+      "step": 2632
+    },
+    {
+      "epoch": 0.23109731178224366,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0028388398786644833,
+      "loss": 3.4023,
+      "step": 2633
+    },
+    {
+      "epoch": 0.23118508136514615,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0028386464693593363,
+      "loss": 3.481,
+      "step": 2634
+    },
+    {
+      "epoch": 0.23127285094804864,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0028384529514439948,
+      "loss": 3.4839,
+      "step": 2635
+    },
+    {
+      "epoch": 0.23136062053095113,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0028382593249361406,
+      "loss": 3.3848,
+      "step": 2636
+    },
+    {
+      "epoch": 0.2314483901138536,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0028380655898534663,
+      "loss": 3.3794,
+      "step": 2637
+    },
+    {
+      "epoch": 0.2315361596967561,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0028378717462136742,
+      "loss": 3.5459,
+      "step": 2638
+    },
+    {
+      "epoch": 0.23162392927965858,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0028376777940344764,
+      "loss": 3.374,
+      "step": 2639
+    },
+    {
+      "epoch": 0.23171169886256107,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0028374837333335944,
+      "loss": 3.3901,
+      "step": 2640
+    },
+    {
+      "epoch": 0.23179946844546354,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00283728956412876,
+      "loss": 3.4341,
+      "step": 2641
+    },
+    {
+      "epoch": 0.23188723802836603,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028370952864377164,
+      "loss": 3.3584,
+      "step": 2642
+    },
+    {
+      "epoch": 0.23197500761126852,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0028369009002782134,
+      "loss": 3.4663,
+      "step": 2643
+    },
+    {
+      "epoch": 0.232062777194171,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0028367064056680135,
+      "loss": 3.4097,
+      "step": 2644
+    },
+    {
+      "epoch": 0.23215054677707347,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002836511802624889,
+      "loss": 3.397,
+      "step": 2645
+    },
+    {
+      "epoch": 0.23223831635997597,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0028363170911666204,
+      "loss": 3.4053,
+      "step": 2646
+    },
+    {
+      "epoch": 0.23232608594287846,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0028361222713109997,
+      "loss": 3.4087,
+      "step": 2647
+    },
+    {
+      "epoch": 0.23241385552578095,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0028359273430758275,
+      "loss": 3.478,
+      "step": 2648
+    },
+    {
+      "epoch": 0.23250162510868344,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0028357323064789157,
+      "loss": 3.4097,
+      "step": 2649
+    },
+    {
+      "epoch": 0.2325893946915859,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0028355371615380845,
+      "loss": 3.4688,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2326771642744884,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0028353419082711656,
+      "loss": 3.4561,
+      "step": 2651
+    },
+    {
+      "epoch": 0.2327649338573909,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002835146546696,
+      "loss": 3.394,
+      "step": 2652
+    },
+    {
+      "epoch": 0.23285270344029338,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002834951076830438,
+      "loss": 3.4443,
+      "step": 2653
+    },
+    {
+      "epoch": 0.23294047302319584,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0028347554986923396,
+      "loss": 3.395,
+      "step": 2654
+    },
+    {
+      "epoch": 0.23302824260609833,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002834559812299577,
+      "loss": 3.4209,
+      "step": 2655
+    },
+    {
+      "epoch": 0.23311601218900083,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0028343640176700297,
+      "loss": 3.3545,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23320378177190332,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002834168114821588,
+      "loss": 3.4077,
+      "step": 2657
+    },
+    {
+      "epoch": 0.2332915513548058,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0028339721037721525,
+      "loss": 3.415,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23337932093770827,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0028337759845396327,
+      "loss": 3.3911,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23346709052061076,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0028335797571419495,
+      "loss": 3.4341,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23355486010351326,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002833383421597032,
+      "loss": 3.4575,
+      "step": 2661
+    },
+    {
+      "epoch": 0.23364262968641575,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0028331869779228205,
+      "loss": 3.3823,
+      "step": 2662
+    },
+    {
+      "epoch": 0.2337303992693182,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002832990426137264,
+      "loss": 3.3613,
+      "step": 2663
+    },
+    {
+      "epoch": 0.2338181688522207,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.002832793766258323,
+      "loss": 3.4121,
+      "step": 2664
+    },
+    {
+      "epoch": 0.2339059384351232,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002832596998303966,
+      "loss": 3.4082,
+      "step": 2665
+    },
+    {
+      "epoch": 0.2339937080180257,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002832400122292173,
+      "loss": 3.4038,
+      "step": 2666
+    },
+    {
+      "epoch": 0.23408147760092815,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0028322031382409327,
+      "loss": 3.3906,
+      "step": 2667
+    },
+    {
+      "epoch": 0.23416924718383064,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002832006046168244,
+      "loss": 3.4028,
+      "step": 2668
+    },
+    {
+      "epoch": 0.23425701676673313,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002831808846092116,
+      "loss": 3.4619,
+      "step": 2669
+    },
+    {
+      "epoch": 0.23434478634963563,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002831611538030567,
+      "loss": 3.4248,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23443255593253812,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0028314141220016267,
+      "loss": 3.4155,
+      "step": 2671
+    },
+    {
+      "epoch": 0.23452032551544058,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002831216598023333,
+      "loss": 3.4399,
+      "step": 2672
+    },
+    {
+      "epoch": 0.23460809509834307,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.002831018966113734,
+      "loss": 3.4316,
+      "step": 2673
+    },
+    {
+      "epoch": 0.23469586468124556,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0028308212262908882,
+      "loss": 3.4658,
+      "step": 2674
+    },
+    {
+      "epoch": 0.23478363426414806,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0028306233785728636,
+      "loss": 3.4277,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23487140384705052,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0028304254229777385,
+      "loss": 3.4844,
+      "step": 2676
+    },
+    {
+      "epoch": 0.234959173429953,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0028302273595236,
+      "loss": 3.4082,
+      "step": 2677
+    },
+    {
+      "epoch": 0.2350469430128555,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002830029188228546,
+      "loss": 3.3384,
+      "step": 2678
+    },
+    {
+      "epoch": 0.235134712595758,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.002829830909110685,
+      "loss": 3.5176,
+      "step": 2679
+    },
+    {
+      "epoch": 0.23522248217866046,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0028296325221881325,
+      "loss": 3.4253,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23531025176156295,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002829434027479017,
+      "loss": 3.3823,
+      "step": 2681
+    },
+    {
+      "epoch": 0.23539802134446544,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0028292354250014752,
+      "loss": 3.4453,
+      "step": 2682
+    },
+    {
+      "epoch": 0.23548579092736793,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0028290367147736544,
+      "loss": 3.3408,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23557356051027042,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002828837896813711,
+      "loss": 3.4077,
+      "step": 2684
+    },
+    {
+      "epoch": 0.2356613300931729,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0028286389711398114,
+      "loss": 3.3794,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23574909967607538,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0028284399377701322,
+      "loss": 3.3994,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23583686925897787,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00282824079672286,
+      "loss": 3.4194,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23592463884188036,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0028280415480161905,
+      "loss": 3.5024,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23601240842478283,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0028278421916683293,
+      "loss": 3.4204,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23610017800768532,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0028276427276974934,
+      "loss": 3.4336,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2361879475905878,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0028274431561219073,
+      "loss": 3.5093,
+      "step": 2691
+    },
+    {
+      "epoch": 0.2362757171734903,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0028272434769598073,
+      "loss": 3.4258,
+      "step": 2692
+    },
+    {
+      "epoch": 0.2363634867563928,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002827043690229438,
+      "loss": 3.3862,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23645125633929526,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0028268437959490544,
+      "loss": 3.3657,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23653902592219775,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0028266437941369222,
+      "loss": 3.4204,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23662679550510024,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0028264436848113154,
+      "loss": 3.4897,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23671456508800273,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002826243467990519,
+      "loss": 3.4224,
+      "step": 2697
+    },
+    {
+      "epoch": 0.2368023346709052,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0028260431436928275,
+      "loss": 3.4292,
+      "step": 2698
+    },
+    {
+      "epoch": 0.2368901042538077,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.002825842711936545,
+      "loss": 3.4082,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23697787383671018,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0028256421727399857,
+      "loss": 3.4849,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23706564341961267,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.002825441526121473,
+      "loss": 3.3843,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23715341300251513,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002825240772099341,
+      "loss": 3.3936,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23724118258541763,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.002825039910691933,
+      "loss": 3.3657,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23732895216832012,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0028248389419176026,
+      "loss": 3.4346,
+      "step": 2704
+    },
+    {
+      "epoch": 0.2374167217512226,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002824637865794713,
+      "loss": 3.4038,
+      "step": 2705
+    },
+    {
+      "epoch": 0.2375044913341251,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002824436682341636,
+      "loss": 3.5483,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23759226091702756,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002824235391576756,
+      "loss": 3.4243,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23768003049993006,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.002824033993518464,
+      "loss": 3.4233,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23776780008283255,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0028238324881851634,
+      "loss": 3.4995,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23785556966573504,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.002823630875595266,
+      "loss": 3.4761,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2379433392486375,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0028234291557671944,
+      "loss": 3.4053,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23803110883154,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0028232273287193787,
+      "loss": 3.4883,
+      "step": 2712
+    },
+    {
+      "epoch": 0.2381188784144425,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028230253944702623,
+      "loss": 3.4053,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23820664799734498,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028228233530382947,
+      "loss": 3.3472,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23829441758024744,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002822621204441939,
+      "loss": 3.4624,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23838218716314993,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0028224189486996645,
+      "loss": 3.374,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23846995674605242,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002822216585829953,
+      "loss": 3.4312,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23855772632895492,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0028220141158512946,
+      "loss": 3.4487,
+      "step": 2718
+    },
+    {
+      "epoch": 0.2386454959118574,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002821811538782189,
+      "loss": 3.4761,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23873326549475987,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0028216088546411475,
+      "loss": 3.3848,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23882103507766236,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0028214060634466887,
+      "loss": 3.4028,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23890880466056486,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002821203165217343,
+      "loss": 3.3545,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23899657424346735,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.00282100015997165,
+      "loss": 3.458,
+      "step": 2723
+    },
+    {
+      "epoch": 0.2390843438263698,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0028207970477281577,
+      "loss": 3.3511,
+      "step": 2724
+    },
+    {
+      "epoch": 0.2391721134092723,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0028205938285054267,
+      "loss": 3.439,
+      "step": 2725
+    },
+    {
+      "epoch": 0.2392598829921748,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0028203905023220246,
+      "loss": 3.4082,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23934765257507729,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0028201870691965305,
+      "loss": 3.439,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23943542215797978,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0028199835291475325,
+      "loss": 3.4053,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23952319174088224,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0028197798821936286,
+      "loss": 3.3174,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23961096132378473,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028195761283534263,
+      "loss": 3.4385,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23969873090668722,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0028193722676455443,
+      "loss": 3.376,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23978650048958972,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002819168300088609,
+      "loss": 3.394,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23987427007249218,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0028189642257012574,
+      "loss": 3.4731,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23996203965539467,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0028187600445021372,
+      "loss": 3.3579,
+      "step": 2734
+    },
+    {
+      "epoch": 0.24004980923829716,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002818555756509904,
+      "loss": 3.3428,
+      "step": 2735
+    },
+    {
+      "epoch": 0.24013757882119965,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.0028183513617432254,
+      "loss": 3.3687,
+      "step": 2736
+    },
+    {
+      "epoch": 0.24022534840410212,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0028181468602207767,
+      "loss": 3.4526,
+      "step": 2737
+    },
+    {
+      "epoch": 0.2403131179870046,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0028179422519612443,
+      "loss": 3.3408,
+      "step": 2738
+    },
+    {
+      "epoch": 0.2404008875699071,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002817737536983324,
+      "loss": 3.3447,
+      "step": 2739
+    },
+    {
+      "epoch": 0.2404886571528096,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0028175327153057207,
+      "loss": 3.4058,
+      "step": 2740
+    },
+    {
+      "epoch": 0.24057642673571208,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002817327786947149,
+      "loss": 3.4224,
+      "step": 2741
+    },
+    {
+      "epoch": 0.24066419631861455,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0028171227519263355,
+      "loss": 3.3672,
+      "step": 2742
+    },
+    {
+      "epoch": 0.24075196590151704,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0028169176102620136,
+      "loss": 3.3921,
+      "step": 2743
+    },
+    {
+      "epoch": 0.24083973548441953,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002816712361972928,
+      "loss": 3.4854,
+      "step": 2744
+    },
+    {
+      "epoch": 0.24092750506732202,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002816507007077833,
+      "loss": 3.4023,
+      "step": 2745
+    },
+    {
+      "epoch": 0.2410152746502245,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002816301545595492,
+      "loss": 3.4399,
+      "step": 2746
+    },
+    {
+      "epoch": 0.24110304423312698,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.00281609597754468,
+      "loss": 3.3975,
+      "step": 2747
+    },
+    {
+      "epoch": 0.24119081381602947,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0028158903029441786,
+      "loss": 3.4546,
+      "step": 2748
+    },
+    {
+      "epoch": 0.24127858339893196,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002815684521812782,
+      "loss": 3.4116,
+      "step": 2749
+    },
+    {
+      "epoch": 0.24136635298183443,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0028154786341692928,
+      "loss": 3.4082,
+      "step": 2750
+    },
+    {
+      "epoch": 0.24145412256473692,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0028152726400325232,
+      "loss": 3.394,
+      "step": 2751
+    },
+    {
+      "epoch": 0.2415418921476394,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002815066539421296,
+      "loss": 3.4668,
+      "step": 2752
+    },
+    {
+      "epoch": 0.2416296617305419,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0028148603323544436,
+      "loss": 3.4517,
+      "step": 2753
+    },
+    {
+      "epoch": 0.2417174313134444,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0028146540188508065,
+      "loss": 3.375,
+      "step": 2754
+    },
+    {
+      "epoch": 0.24180520089634686,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0028144475989292372,
+      "loss": 3.4038,
+      "step": 2755
+    },
+    {
+      "epoch": 0.24189297047924935,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0028142410726085965,
+      "loss": 3.3721,
+      "step": 2756
+    },
+    {
+      "epoch": 0.24198074006215184,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0028140344399077556,
+      "loss": 3.3525,
+      "step": 2757
+    },
+    {
+      "epoch": 0.24206850964505433,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0028138277008455947,
+      "loss": 3.3887,
+      "step": 2758
+    },
+    {
+      "epoch": 0.2421562792279568,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0028136208554410042,
+      "loss": 3.4717,
+      "step": 2759
+    },
+    {
+      "epoch": 0.24224404881085929,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.002813413903712885,
+      "loss": 3.3975,
+      "step": 2760
+    },
+    {
+      "epoch": 0.24233181839376178,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0028132068456801465,
+      "loss": 3.3643,
+      "step": 2761
+    },
+    {
+      "epoch": 0.24241958797666427,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.002812999681361708,
+      "loss": 3.4932,
+      "step": 2762
+    },
+    {
+      "epoch": 0.24250735755956676,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0028127924107764985,
+      "loss": 3.4248,
+      "step": 2763
+    },
+    {
+      "epoch": 0.24259512714246922,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0028125850339434572,
+      "loss": 3.4302,
+      "step": 2764
+    },
+    {
+      "epoch": 0.24268289672537172,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.002812377550881533,
+      "loss": 3.5288,
+      "step": 2765
+    },
+    {
+      "epoch": 0.2427706663082742,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0028121699616096836,
+      "loss": 3.375,
+      "step": 2766
+    },
+    {
+      "epoch": 0.2428584358911767,
+      "grad_norm": 0.375,
+      "learning_rate": 0.002811962266146878,
+      "loss": 3.4077,
+      "step": 2767
+    },
+    {
+      "epoch": 0.24294620547407916,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0028117544645120934,
+      "loss": 3.4248,
+      "step": 2768
+    },
+    {
+      "epoch": 0.24303397505698165,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0028115465567243162,
+      "loss": 3.4307,
+      "step": 2769
+    },
+    {
+      "epoch": 0.24312174463988415,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0028113385428025456,
+      "loss": 3.4048,
+      "step": 2770
+    },
+    {
+      "epoch": 0.24320951422278664,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0028111304227657874,
+      "loss": 3.4482,
+      "step": 2771
+    },
+    {
+      "epoch": 0.2432972838056891,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0028109221966330583,
+      "loss": 3.4854,
+      "step": 2772
+    },
+    {
+      "epoch": 0.2433850533885916,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0028107138644233843,
+      "loss": 3.3218,
+      "step": 2773
+    },
+    {
+      "epoch": 0.24347282297149409,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0028105054261558017,
+      "loss": 3.417,
+      "step": 2774
+    },
+    {
+      "epoch": 0.24356059255439658,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0028102968818493556,
+      "loss": 3.3887,
+      "step": 2775
+    },
+    {
+      "epoch": 0.24364836213729907,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002810088231523102,
+      "loss": 3.4106,
+      "step": 2776
+    },
+    {
+      "epoch": 0.24373613172020153,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0028098794751961055,
+      "loss": 3.417,
+      "step": 2777
+    },
+    {
+      "epoch": 0.24382390130310402,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002809670612887441,
+      "loss": 3.3843,
+      "step": 2778
+    },
+    {
+      "epoch": 0.24391167088600652,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0028094616446161925,
+      "loss": 3.3843,
+      "step": 2779
+    },
+    {
+      "epoch": 0.243999440468909,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0028092525704014547,
+      "loss": 3.458,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24408721005181147,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002809043390262331,
+      "loss": 3.4507,
+      "step": 2781
+    },
+    {
+      "epoch": 0.24417497963471396,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0028088341042179347,
+      "loss": 3.4102,
+      "step": 2782
+    },
+    {
+      "epoch": 0.24426274921761645,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0028086247122873884,
+      "loss": 3.4268,
+      "step": 2783
+    },
+    {
+      "epoch": 0.24435051880051895,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0028084152144898255,
+      "loss": 3.4561,
+      "step": 2784
+    },
+    {
+      "epoch": 0.2444382883834214,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0028082056108443888,
+      "loss": 3.4272,
+      "step": 2785
+    },
+    {
+      "epoch": 0.2445260579663239,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.00280799590137023,
+      "loss": 3.4595,
+      "step": 2786
+    },
+    {
+      "epoch": 0.2446138275492264,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002807786086086511,
+      "loss": 3.4346,
+      "step": 2787
+    },
+    {
+      "epoch": 0.24470159713212888,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0028075761650124026,
+      "loss": 3.4971,
+      "step": 2788
+    },
+    {
+      "epoch": 0.24478936671503138,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002807366138167087,
+      "loss": 3.4443,
+      "step": 2789
+    },
+    {
+      "epoch": 0.24487713629793384,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0028071560055697537,
+      "loss": 3.4146,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24496490588083633,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0028069457672396047,
+      "loss": 3.3877,
+      "step": 2791
+    },
+    {
+      "epoch": 0.24505267546373882,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002806735423195849,
+      "loss": 3.4604,
+      "step": 2792
+    },
+    {
+      "epoch": 0.24514044504664131,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0028065249734577067,
+      "loss": 3.395,
+      "step": 2793
+    },
+    {
+      "epoch": 0.24522821462954378,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0028063144180444067,
+      "loss": 3.3926,
+      "step": 2794
+    },
+    {
+      "epoch": 0.24531598421244627,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002806103756975189,
+      "loss": 3.4375,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24540375379534876,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0028058929902693014,
+      "loss": 3.3555,
+      "step": 2796
+    },
+    {
+      "epoch": 0.24549152337825125,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0028056821179460035,
+      "loss": 3.3882,
+      "step": 2797
+    },
+    {
+      "epoch": 0.24557929296115374,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002805471140024562,
+      "loss": 3.3755,
+      "step": 2798
+    },
+    {
+      "epoch": 0.2456670625440562,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0028052600565242557,
+      "loss": 3.4766,
+      "step": 2799
+    },
+    {
+      "epoch": 0.2457548321269587,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002805048867464371,
+      "loss": 3.3906,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2458426017098612,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0028048375728642054,
+      "loss": 3.4185,
+      "step": 2801
+    },
+    {
+      "epoch": 0.24593037129276368,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0028046261727430657,
+      "loss": 3.3945,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24601814087566615,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0028044146671202677,
+      "loss": 3.5122,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24610591045856864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0028042030560151377,
+      "loss": 3.4185,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24619368004147113,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0028039913394470105,
+      "loss": 3.3823,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24628144962437362,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002803779517435232,
+      "loss": 3.4731,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24636921920727609,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002803567589999157,
+      "loss": 3.3569,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24645698879017858,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00280335555715815,
+      "loss": 3.3765,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24654475837308107,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0028031434189315844,
+      "loss": 3.4106,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24663252795598356,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0028029311753388444,
+      "loss": 3.4248,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24672029753888605,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002802718826399323,
+      "loss": 3.395,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24680806712178852,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002802506372132424,
+      "loss": 3.416,
+      "step": 2812
+    },
+    {
+      "epoch": 0.246895836704691,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.00280229381255756,
+      "loss": 3.4082,
+      "step": 2813
+    },
+    {
+      "epoch": 0.2469836062875935,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0028020811476941513,
+      "loss": 3.4077,
+      "step": 2814
+    },
+    {
+      "epoch": 0.247071375870496,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002801868377561632,
+      "loss": 3.4272,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24715914545339845,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0028016555021794422,
+      "loss": 3.4414,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24724691503630095,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002801442521567034,
+      "loss": 3.4429,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24733468461920344,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0028012294357438665,
+      "loss": 3.4409,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24742245420210593,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0028010162447294116,
+      "loss": 3.3862,
+      "step": 2819
+    },
+    {
+      "epoch": 0.2475102237850084,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0028008029485431483,
+      "loss": 3.417,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24759799336791088,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002800589547204566,
+      "loss": 3.481,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24768576295081338,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002800376040733165,
+      "loss": 3.5127,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24777353253371587,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0028001624291484526,
+      "loss": 3.3867,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24786130211661836,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.002799948712469948,
+      "loss": 3.5073,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24794907169952082,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0027997348907171787,
+      "loss": 3.4028,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24803684128242331,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0027995209639096823,
+      "loss": 3.4058,
+      "step": 2826
+    },
+    {
+      "epoch": 0.2481246108653258,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002799306932067006,
+      "loss": 3.4062,
+      "step": 2827
+    },
+    {
+      "epoch": 0.2482123804482283,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002799092795208707,
+      "loss": 3.4204,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24830015003113076,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0027988785533543514,
+      "loss": 3.3911,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24838791961403325,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002798664206523514,
+      "loss": 3.3921,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24847568919693575,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.0027984497547357823,
+      "loss": 3.4424,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24856345877983824,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0027982351980107493,
+      "loss": 3.3848,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24865122836274073,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0027980205363680216,
+      "loss": 3.3994,
+      "step": 2833
+    },
+    {
+      "epoch": 0.2487389979456432,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.002797805769827212,
+      "loss": 3.4014,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24882676752854568,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0027975908984079452,
+      "loss": 3.4302,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24891453711144818,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.002797375922129854,
+      "loss": 3.4199,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24900230669435067,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0027971608410125824,
+      "loss": 3.3662,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24909007627725313,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0027969456550757826,
+      "loss": 3.4775,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24917784586015562,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0027967303643391165,
+      "loss": 3.3921,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24926561544305811,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0027965149688222554,
+      "loss": 3.3691,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2493533850259606,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002796299468544882,
+      "loss": 3.4316,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24944115460886307,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.002796083863526686,
+      "loss": 3.4468,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24952892419176556,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0027958681537873687,
+      "loss": 3.4219,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24961669377466805,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002795652339346639,
+      "loss": 3.3989,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24970446335757054,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.002795436420224218,
+      "loss": 3.3848,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24979223294047304,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0027952203964398343,
+      "loss": 3.3457,
+      "step": 2846
+    },
+    {
+      "epoch": 0.2498800025233755,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.002795004268013226,
+      "loss": 3.4424,
+      "step": 2847
+    },
+    {
+      "epoch": 0.249967772106278,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0027947880349641417,
+      "loss": 3.4155,
+      "step": 2848
+    },
+    {
+      "epoch": 0.2500555416891805,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.0027945716973123394,
+      "loss": 3.4712,
+      "step": 2849
+    },
+    {
+      "epoch": 0.25014331127208295,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002794355255077587,
+      "loss": 3.3887,
+      "step": 2850
+    },
+    {
+      "epoch": 0.25023108085498547,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0027941387082796613,
+      "loss": 3.4282,
+      "step": 2851
+    },
+    {
+      "epoch": 0.25031885043788793,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002793922056938348,
+      "loss": 3.3896,
+      "step": 2852
+    },
+    {
+      "epoch": 0.2504066200207904,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0027937053010734444,
+      "loss": 3.4619,
+      "step": 2853
+    },
+    {
+      "epoch": 0.2504943896036929,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002793488440704755,
+      "loss": 3.3687,
+      "step": 2854
+    },
+    {
+      "epoch": 0.2505821591865954,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002793271475852096,
+      "loss": 3.4116,
+      "step": 2855
+    },
+    {
+      "epoch": 0.2506699287694979,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0027930544065352912,
+      "loss": 3.4565,
+      "step": 2856
+    },
+    {
+      "epoch": 0.25075769835240036,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.002792837232774175,
+      "loss": 3.4644,
+      "step": 2857
+    },
+    {
+      "epoch": 0.2508454679353028,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0027926199545885923,
+      "loss": 3.4404,
+      "step": 2858
+    },
+    {
+      "epoch": 0.25093323751820534,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002792402571998395,
+      "loss": 3.3979,
+      "step": 2859
+    },
+    {
+      "epoch": 0.2510210071011078,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002792185085023447,
+      "loss": 3.3755,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2511087766840103,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0027919674936836203,
+      "loss": 3.418,
+      "step": 2861
+    },
+    {
+      "epoch": 0.2511965462669128,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0027917497979987973,
+      "loss": 3.4004,
+      "step": 2862
+    },
+    {
+      "epoch": 0.25128431584981525,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0027915319979888685,
+      "loss": 3.4204,
+      "step": 2863
+    },
+    {
+      "epoch": 0.2513720854327178,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0027913140936737363,
+      "loss": 3.4731,
+      "step": 2864
+    },
+    {
+      "epoch": 0.25145985501562024,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00279109608507331,
+      "loss": 3.5166,
+      "step": 2865
+    },
+    {
+      "epoch": 0.2515476245985227,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002790877972207511,
+      "loss": 3.4014,
+      "step": 2866
+    },
+    {
+      "epoch": 0.2516353941814252,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002790659755096268,
+      "loss": 3.4062,
+      "step": 2867
+    },
+    {
+      "epoch": 0.2517231637643277,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.00279044143375952,
+      "loss": 3.4404,
+      "step": 2868
+    },
+    {
+      "epoch": 0.2518109333472302,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.002790223008217216,
+      "loss": 3.5015,
+      "step": 2869
+    },
+    {
+      "epoch": 0.25189870293013267,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0027900044784893146,
+      "loss": 3.4277,
+      "step": 2870
+    },
+    {
+      "epoch": 0.25198647251303513,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002789785844595783,
+      "loss": 3.4219,
+      "step": 2871
+    },
+    {
+      "epoch": 0.25207424209593765,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.002789567106556599,
+      "loss": 3.4873,
+      "step": 2872
+    },
+    {
+      "epoch": 0.2521620116788401,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0027893482643917486,
+      "loss": 3.396,
+      "step": 2873
+    },
+    {
+      "epoch": 0.25224978126174263,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0027891293181212286,
+      "loss": 3.3853,
+      "step": 2874
+    },
+    {
+      "epoch": 0.2523375508446451,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.002788910267765044,
+      "loss": 3.5117,
+      "step": 2875
+    },
+    {
+      "epoch": 0.25242532042754756,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0027886911133432114,
+      "loss": 3.4331,
+      "step": 2876
+    },
+    {
+      "epoch": 0.2525130900104501,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0027884718548757547,
+      "loss": 3.3916,
+      "step": 2877
+    },
+    {
+      "epoch": 0.25260085959335254,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.002788252492382708,
+      "loss": 3.4893,
+      "step": 2878
+    },
+    {
+      "epoch": 0.25268862917625506,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0027880330258841157,
+      "loss": 3.4233,
+      "step": 2879
+    },
+    {
+      "epoch": 0.25277639875915753,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002787813455400031,
+      "loss": 3.4839,
+      "step": 2880
+    },
+    {
+      "epoch": 0.25286416834206,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0027875937809505175,
+      "loss": 3.3887,
+      "step": 2881
+    },
+    {
+      "epoch": 0.2529519379249625,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0027873740025556454,
+      "loss": 3.3853,
+      "step": 2882
+    },
+    {
+      "epoch": 0.253039707507865,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.002787154120235498,
+      "loss": 3.4106,
+      "step": 2883
+    },
+    {
+      "epoch": 0.25312747709076744,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002786934134010167,
+      "loss": 3.4194,
+      "step": 2884
+    },
+    {
+      "epoch": 0.25321524667366996,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0027867140438997515,
+      "loss": 3.5464,
+      "step": 2885
+    },
+    {
+      "epoch": 0.2533030162565724,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0027864938499243637,
+      "loss": 3.4199,
+      "step": 2886
+    },
+    {
+      "epoch": 0.25339078583947494,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002786273552104122,
+      "loss": 3.4082,
+      "step": 2887
+    },
+    {
+      "epoch": 0.2534785554223774,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.002786053150459157,
+      "loss": 3.4766,
+      "step": 2888
+    },
+    {
+      "epoch": 0.25356632500527987,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002785832645009606,
+      "loss": 3.4014,
+      "step": 2889
+    },
+    {
+      "epoch": 0.2536540945881824,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.002785612035775618,
+      "loss": 3.5146,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25374186417108485,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0027853913227773503,
+      "loss": 3.5034,
+      "step": 2891
+    },
+    {
+      "epoch": 0.25382963375398737,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0027851705060349713,
+      "loss": 3.4746,
+      "step": 2892
+    },
+    {
+      "epoch": 0.25391740333688984,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0027849495855686566,
+      "loss": 3.4673,
+      "step": 2893
+    },
+    {
+      "epoch": 0.2540051729197923,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002784728561398592,
+      "loss": 3.3662,
+      "step": 2894
+    },
+    {
+      "epoch": 0.2540929425026948,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0027845074335449745,
+      "loss": 3.3701,
+      "step": 2895
+    },
+    {
+      "epoch": 0.2541807120855973,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0027842862020280076,
+      "loss": 3.4761,
+      "step": 2896
+    },
+    {
+      "epoch": 0.25426848166849975,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0027840648668679075,
+      "loss": 3.3989,
+      "step": 2897
+    },
+    {
+      "epoch": 0.25435625125140227,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002783843428084897,
+      "loss": 3.3813,
+      "step": 2898
+    },
+    {
+      "epoch": 0.25444402083430473,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0027836218856992106,
+      "loss": 3.3413,
+      "step": 2899
+    },
+    {
+      "epoch": 0.25453179041720725,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0027834002397310906,
+      "loss": 3.4541,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2546195600001097,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0027831784902007895,
+      "loss": 3.4644,
+      "step": 2901
+    },
+    {
+      "epoch": 0.2547073295830122,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0027829566371285696,
+      "loss": 3.4346,
+      "step": 2902
+    },
+    {
+      "epoch": 0.2547950991659147,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.002782734680534702,
+      "loss": 3.3906,
+      "step": 2903
+    },
+    {
+      "epoch": 0.25488286874881716,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002782512620439468,
+      "loss": 3.3418,
+      "step": 2904
+    },
+    {
+      "epoch": 0.2549706383317197,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0027822904568631573,
+      "loss": 3.4341,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25505840791462214,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0027820681898260693,
+      "loss": 3.376,
+      "step": 2906
+    },
+    {
+      "epoch": 0.2551461774975246,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.0027818458193485146,
+      "loss": 3.4092,
+      "step": 2907
+    },
+    {
+      "epoch": 0.2552339470804271,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0027816233454508107,
+      "loss": 3.3433,
+      "step": 2908
+    },
+    {
+      "epoch": 0.2553217166633296,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0027814007681532856,
+      "loss": 3.4121,
+      "step": 2909
+    },
+    {
+      "epoch": 0.25540948624623205,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0027811780874762784,
+      "loss": 3.4697,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2554972558291346,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.002780955303440134,
+      "loss": 3.4351,
+      "step": 2911
+    },
+    {
+      "epoch": 0.25558502541203704,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00278073241606521,
+      "loss": 3.4565,
+      "step": 2912
+    },
+    {
+      "epoch": 0.25567279499493956,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0027805094253718727,
+      "loss": 3.4448,
+      "step": 2913
+    },
+    {
+      "epoch": 0.255760564577842,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0027802863313804963,
+      "loss": 3.4175,
+      "step": 2914
+    },
+    {
+      "epoch": 0.2558483341607445,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.002780063134111467,
+      "loss": 3.3916,
+      "step": 2915
+    },
+    {
+      "epoch": 0.255936103743647,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002779839833585177,
+      "loss": 3.3892,
+      "step": 2916
+    },
+    {
+      "epoch": 0.25602387332654947,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.002779616429822031,
+      "loss": 3.313,
+      "step": 2917
+    },
+    {
+      "epoch": 0.256111642909452,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002779392922842443,
+      "loss": 3.4609,
+      "step": 2918
+    },
+    {
+      "epoch": 0.25619941249235445,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0027791693126668344,
+      "loss": 3.4624,
+      "step": 2919
+    },
+    {
+      "epoch": 0.2562871820752569,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0027789455993156376,
+      "loss": 3.4023,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25637495165815943,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002778721782809293,
+      "loss": 3.3379,
+      "step": 2921
+    },
+    {
+      "epoch": 0.2564627212410619,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002778497863168253,
+      "loss": 3.3545,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25655049082396436,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002778273840412976,
+      "loss": 3.4805,
+      "step": 2923
+    },
+    {
+      "epoch": 0.2566382604068669,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002778049714563934,
+      "loss": 3.3921,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25672602998976934,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002777825485641603,
+      "loss": 3.437,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25681379957267186,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002777601153666474,
+      "loss": 3.4111,
+      "step": 2926
+    },
+    {
+      "epoch": 0.2569015691555743,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002777376718659044,
+      "loss": 3.4639,
+      "step": 2927
+    },
+    {
+      "epoch": 0.2569893387384768,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0027771521806398197,
+      "loss": 3.3926,
+      "step": 2928
+    },
+    {
+      "epoch": 0.2570771083213793,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002776927539629319,
+      "loss": 3.3838,
+      "step": 2929
+    },
+    {
+      "epoch": 0.2571648779042818,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.002776702795648067,
+      "loss": 3.4468,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2572526474871843,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0027764779487166,
+      "loss": 3.4385,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25734041707008676,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002776252998855462,
+      "loss": 3.4282,
+      "step": 2932
+    },
+    {
+      "epoch": 0.2574281866529892,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.002776027946085209,
+      "loss": 3.4116,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25751595623589174,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0027758027904264022,
+      "loss": 3.3516,
+      "step": 2934
+    },
+    {
+      "epoch": 0.2576037258187942,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0027755775318996177,
+      "loss": 3.3799,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25769149540169667,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002775352170525436,
+      "loss": 3.4136,
+      "step": 2936
+    },
+    {
+      "epoch": 0.2577792649845992,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002775126706324449,
+      "loss": 3.4146,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25786703456750165,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0027749011393172593,
+      "loss": 3.4683,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25795480415040417,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.002774675469524477,
+      "loss": 3.4497,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25804257373330663,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002774449696966722,
+      "loss": 3.3721,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2581303433162091,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002774223821664624,
+      "loss": 3.4175,
+      "step": 2941
+    },
+    {
+      "epoch": 0.2582181128991116,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002773997843638822,
+      "loss": 3.3838,
+      "step": 2942
+    },
+    {
+      "epoch": 0.2583058824820141,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0027737717629099643,
+      "loss": 3.4146,
+      "step": 2943
+    },
+    {
+      "epoch": 0.2583936520649166,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0027735455794987087,
+      "loss": 3.4658,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25848142164781907,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002773319293425722,
+      "loss": 3.4419,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25856919123072153,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00277309290471168,
+      "loss": 3.3931,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25865696081362405,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00277286641337727,
+      "loss": 3.4746,
+      "step": 2947
+    },
+    {
+      "epoch": 0.2587447303965265,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0027726398194431864,
+      "loss": 3.4683,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25883249997942903,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.002772413122930133,
+      "loss": 3.4097,
+      "step": 2949
+    },
+    {
+      "epoch": 0.2589202695623315,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0027721863238588254,
+      "loss": 3.4697,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25900803914523396,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0027719594222499855,
+      "loss": 3.3628,
+      "step": 2951
+    },
+    {
+      "epoch": 0.2590958087281365,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002771732418124347,
+      "loss": 3.4531,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25918357831103894,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.002771505311502651,
+      "loss": 3.3701,
+      "step": 2953
+    },
+    {
+      "epoch": 0.2592713478939414,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0027712781024056493,
+      "loss": 3.3286,
+      "step": 2954
+    },
+    {
+      "epoch": 0.2593591174768439,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0027710507908541033,
+      "loss": 3.3384,
+      "step": 2955
+    },
+    {
+      "epoch": 0.2594468870597464,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0027708233768687822,
+      "loss": 3.3984,
+      "step": 2956
+    },
+    {
+      "epoch": 0.2595346566426489,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002770595860470466,
+      "loss": 3.3853,
+      "step": 2957
+    },
+    {
+      "epoch": 0.2596224262255514,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0027703682416799435,
+      "loss": 3.356,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25971019580845384,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002770140520518012,
+      "loss": 3.4492,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25979796539135636,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002769912697005481,
+      "loss": 3.4844,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2598857349742588,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002769684771163166,
+      "loss": 3.4829,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25997350455716134,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0027694567430118938,
+      "loss": 3.4644,
+      "step": 2962
+    },
+    {
+      "epoch": 0.2600612741400638,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002769228612572499,
+      "loss": 3.4395,
+      "step": 2963
+    },
+    {
+      "epoch": 0.26014904372296627,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002769000379865828,
+      "loss": 3.4033,
+      "step": 2964
+    },
+    {
+      "epoch": 0.2602368133058688,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0027687720449127346,
+      "loss": 3.3486,
+      "step": 2965
+    },
+    {
+      "epoch": 0.26032458288877125,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002768543607734082,
+      "loss": 3.4111,
+      "step": 2966
+    },
+    {
+      "epoch": 0.2604123524716737,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.002768315068350744,
+      "loss": 3.4888,
+      "step": 2967
+    },
+    {
+      "epoch": 0.26050012205457623,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002768086426783602,
+      "loss": 3.4434,
+      "step": 2968
+    },
+    {
+      "epoch": 0.2605878916374787,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0027678576830535485,
+      "loss": 3.3345,
+      "step": 2969
+    },
+    {
+      "epoch": 0.2606756612203812,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0027676288371814833,
+      "loss": 3.4321,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2607634308032837,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0027673998891883187,
+      "loss": 3.4126,
+      "step": 2971
+    },
+    {
+      "epoch": 0.26085120038618614,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0027671708390949725,
+      "loss": 3.3857,
+      "step": 2972
+    },
+    {
+      "epoch": 0.26093896996908866,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0027669416869223746,
+      "loss": 3.4497,
+      "step": 2973
+    },
+    {
+      "epoch": 0.2610267395519911,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002766712432691463,
+      "loss": 3.2974,
+      "step": 2974
+    },
+    {
+      "epoch": 0.26111450913489365,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002766483076423186,
+      "loss": 3.3745,
+      "step": 2975
+    },
+    {
+      "epoch": 0.2612022787177961,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0027662536181385,
+      "loss": 3.4956,
+      "step": 2976
+    },
+    {
+      "epoch": 0.2612900483006986,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00276602405785837,
+      "loss": 3.3999,
+      "step": 2977
+    },
+    {
+      "epoch": 0.2613778178836011,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002765794395603774,
+      "loss": 3.375,
+      "step": 2978
+    },
+    {
+      "epoch": 0.26146558746650356,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0027655646313956964,
+      "loss": 3.4199,
+      "step": 2979
+    },
+    {
+      "epoch": 0.261553357049406,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0027653347652551305,
+      "loss": 3.4453,
+      "step": 2980
+    },
+    {
+      "epoch": 0.26164112663230854,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.00276510479720308,
+      "loss": 3.2944,
+      "step": 2981
+    },
+    {
+      "epoch": 0.261728896215211,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0027648747272605583,
+      "loss": 3.4727,
+      "step": 2982
+    },
+    {
+      "epoch": 0.2618166657981135,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0027646445554485874,
+      "loss": 3.3647,
+      "step": 2983
+    },
+    {
+      "epoch": 0.261904435381016,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002764414281788199,
+      "loss": 3.3491,
+      "step": 2984
+    },
+    {
+      "epoch": 0.26199220496391845,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002764183906300433,
+      "loss": 3.4219,
+      "step": 2985
+    },
+    {
+      "epoch": 0.26207997454682097,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0027639534290063404,
+      "loss": 3.4775,
+      "step": 2986
+    },
+    {
+      "epoch": 0.26216774412972343,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0027637228499269805,
+      "loss": 3.4282,
+      "step": 2987
+    },
+    {
+      "epoch": 0.26225551371262595,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0027634921690834215,
+      "loss": 3.4248,
+      "step": 2988
+    },
+    {
+      "epoch": 0.2623432832955284,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002763261386496742,
+      "loss": 3.3818,
+      "step": 2989
+    },
+    {
+      "epoch": 0.2624310528784309,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0027630305021880287,
+      "loss": 3.3535,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2625188224613334,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002762799516178379,
+      "loss": 3.4146,
+      "step": 2991
+    },
+    {
+      "epoch": 0.26260659204423586,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002762568428488898,
+      "loss": 3.5044,
+      "step": 2992
+    },
+    {
+      "epoch": 0.26269436162713833,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0027623372391407013,
+      "loss": 3.3193,
+      "step": 2993
+    },
+    {
+      "epoch": 0.26278213121004085,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002762105948154913,
+      "loss": 3.3491,
+      "step": 2994
+    },
+    {
+      "epoch": 0.2628699007929433,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0027618745555526673,
+      "loss": 3.4014,
+      "step": 2995
+    },
+    {
+      "epoch": 0.26295767037584583,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0027616430613551075,
+      "loss": 3.4116,
+      "step": 2996
+    },
+    {
+      "epoch": 0.2630454399587483,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0027614114655833853,
+      "loss": 3.4404,
+      "step": 2997
+    },
+    {
+      "epoch": 0.26313320954165076,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002761179768258662,
+      "loss": 3.3862,
+      "step": 2998
+    },
+    {
+      "epoch": 0.2632209791245533,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002760947969402109,
+      "loss": 3.4053,
+      "step": 2999
+    },
+    {
+      "epoch": 0.26330874870745574,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0027607160690349066,
+      "loss": 3.3486,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26330874870745574,
+      "eval_loss": 0.10586483031511307,
+      "eval_runtime": 106.5257,
+      "eval_samples_per_second": 138.455,
+      "eval_steps_per_second": 17.31,
+      "step": 3000
+    },
+    {
+      "epoch": 0.26339651829035826,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002760484067178244,
+      "loss": 3.3779,
+      "step": 3001
+    },
+    {
+      "epoch": 0.2634842878732607,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0027602519638533197,
+      "loss": 3.4062,
+      "step": 3002
+    },
+    {
+      "epoch": 0.2635720574561632,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002760019759081342,
+      "loss": 3.3989,
+      "step": 3003
+    },
+    {
+      "epoch": 0.2636598270390657,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002759787452883528,
+      "loss": 3.4092,
+      "step": 3004
+    },
+    {
+      "epoch": 0.26374759662196817,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0027595550452811047,
+      "loss": 3.3955,
+      "step": 3005
+    },
+    {
+      "epoch": 0.26383536620487064,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002759322536295307,
+      "loss": 3.4277,
+      "step": 3006
+    },
+    {
+      "epoch": 0.26392313578777316,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002759089925947381,
+      "loss": 3.4468,
+      "step": 3007
+    },
+    {
+      "epoch": 0.2640109053706756,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0027588572142585796,
+      "loss": 3.3926,
+      "step": 3008
+    },
+    {
+      "epoch": 0.26409867495357814,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002758624401250168,
+      "loss": 3.4297,
+      "step": 3009
+    },
+    {
+      "epoch": 0.2641864445364806,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0027583914869434172,
+      "loss": 3.4482,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26427421411938307,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0027581584713596108,
+      "loss": 3.3989,
+      "step": 3011
+    },
+    {
+      "epoch": 0.2643619837022856,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0027579253545200397,
+      "loss": 3.3931,
+      "step": 3012
+    },
+    {
+      "epoch": 0.26444975328518805,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002757692136446004,
+      "loss": 3.4502,
+      "step": 3013
+    },
+    {
+      "epoch": 0.26453752286809057,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0027574588171588145,
+      "loss": 3.4263,
+      "step": 3014
+    },
+    {
+      "epoch": 0.26462529245099303,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0027572253966797896,
+      "loss": 3.4326,
+      "step": 3015
+    },
+    {
+      "epoch": 0.2647130620338955,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0027569918750302573,
+      "loss": 3.4277,
+      "step": 3016
+    },
+    {
+      "epoch": 0.264800831616798,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002756758252231556,
+      "loss": 3.397,
+      "step": 3017
+    },
+    {
+      "epoch": 0.2648886011997005,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002756524528305032,
+      "loss": 3.3291,
+      "step": 3018
+    },
+    {
+      "epoch": 0.264976370782603,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0027562907032720415,
+      "loss": 3.4224,
+      "step": 3019
+    },
+    {
+      "epoch": 0.26506414036550546,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0027560567771539495,
+      "loss": 3.4229,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2651519099484079,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0027558227499721306,
+      "loss": 3.4209,
+      "step": 3021
+    },
+    {
+      "epoch": 0.26523967953131045,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0027555886217479697,
+      "loss": 3.3442,
+      "step": 3022
+    },
+    {
+      "epoch": 0.2653274491142129,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0027553543925028586,
+      "loss": 3.3613,
+      "step": 3023
+    },
+    {
+      "epoch": 0.2654152186971154,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0027551200622581993,
+      "loss": 3.394,
+      "step": 3024
+    },
+    {
+      "epoch": 0.2655029882800179,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0027548856310354044,
+      "loss": 3.4419,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26559075786292036,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0027546510988558934,
+      "loss": 3.418,
+      "step": 3026
+    },
+    {
+      "epoch": 0.2656785274458229,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.002754416465741097,
+      "loss": 3.4482,
+      "step": 3027
+    },
+    {
+      "epoch": 0.26576629702872534,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0027541817317124537,
+      "loss": 3.4302,
+      "step": 3028
+    },
+    {
+      "epoch": 0.2658540666116278,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0027539468967914126,
+      "loss": 3.4419,
+      "step": 3029
+    },
+    {
+      "epoch": 0.2659418361945303,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0027537119609994313,
+      "loss": 3.2969,
+      "step": 3030
+    },
+    {
+      "epoch": 0.2660296057774328,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0027534769243579755,
+      "loss": 3.3452,
+      "step": 3031
+    },
+    {
+      "epoch": 0.2661173753603353,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0027532417868885222,
+      "loss": 3.4072,
+      "step": 3032
+    },
+    {
+      "epoch": 0.26620514494323777,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002753006548612556,
+      "loss": 3.3926,
+      "step": 3033
+    },
+    {
+      "epoch": 0.26629291452614023,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0027527712095515725,
+      "loss": 3.3613,
+      "step": 3034
+    },
+    {
+      "epoch": 0.26638068410904275,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002752535769727074,
+      "loss": 3.3984,
+      "step": 3035
+    },
+    {
+      "epoch": 0.2664684536919452,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0027523002291605734,
+      "loss": 3.3208,
+      "step": 3036
+    },
+    {
+      "epoch": 0.2665562232748477,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0027520645878735934,
+      "loss": 3.4224,
+      "step": 3037
+    },
+    {
+      "epoch": 0.2666439928577502,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0027518288458876655,
+      "loss": 3.4917,
+      "step": 3038
+    },
+    {
+      "epoch": 0.26673176244065266,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0027515930032243293,
+      "loss": 3.4258,
+      "step": 3039
+    },
+    {
+      "epoch": 0.2668195320235552,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0027513570599051345,
+      "loss": 3.4116,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26690730160645765,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0027511210159516407,
+      "loss": 3.4292,
+      "step": 3041
+    },
+    {
+      "epoch": 0.2669950711893601,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0027508848713854155,
+      "loss": 3.4766,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26708284077226263,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0027506486262280356,
+      "loss": 3.4136,
+      "step": 3043
+    },
+    {
+      "epoch": 0.2671706103551651,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0027504122805010888,
+      "loss": 3.3696,
+      "step": 3044
+    },
+    {
+      "epoch": 0.2672583799380676,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0027501758342261696,
+      "loss": 3.4707,
+      "step": 3045
+    },
+    {
+      "epoch": 0.2673461495209701,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0027499392874248832,
+      "loss": 3.3784,
+      "step": 3046
+    },
+    {
+      "epoch": 0.26743391910387254,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.002749702640118843,
+      "loss": 3.4268,
+      "step": 3047
+    },
+    {
+      "epoch": 0.26752168868677506,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0027494658923296732,
+      "loss": 3.4004,
+      "step": 3048
+    },
+    {
+      "epoch": 0.2676094582696775,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0027492290440790054,
+      "loss": 3.4277,
+      "step": 3049
+    },
+    {
+      "epoch": 0.26769722785258,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002748992095388482,
+      "loss": 3.415,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2677849974354825,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0027487550462797525,
+      "loss": 3.4087,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26787276701838497,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002748517896774477,
+      "loss": 3.4028,
+      "step": 3052
+    },
+    {
+      "epoch": 0.2679605366012875,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0027482806468943255,
+      "loss": 3.394,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26804830618418996,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0027480432966609752,
+      "loss": 3.3677,
+      "step": 3054
+    },
+    {
+      "epoch": 0.2681360757670924,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0027478058460961145,
+      "loss": 3.3623,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26822384534999494,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002747568295221439,
+      "loss": 3.4463,
+      "step": 3056
+    },
+    {
+      "epoch": 0.2683116149328974,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002747330644058655,
+      "loss": 3.417,
+      "step": 3057
+    },
+    {
+      "epoch": 0.2683993845157999,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0027470928926294773,
+      "loss": 3.4116,
+      "step": 3058
+    },
+    {
+      "epoch": 0.2684871540987024,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0027468550409556307,
+      "loss": 3.4756,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26857492368160485,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0027466170890588465,
+      "loss": 3.3628,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26866269326450737,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0027463790369608686,
+      "loss": 3.4438,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26875046284740983,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0027461408846834487,
+      "loss": 3.3853,
+      "step": 3062
+    },
+    {
+      "epoch": 0.2688382324303123,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0027459026322483463,
+      "loss": 3.416,
+      "step": 3063
+    },
+    {
+      "epoch": 0.2689260020132148,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0027456642796773317,
+      "loss": 3.3975,
+      "step": 3064
+    },
+    {
+      "epoch": 0.2690137715961173,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002745425826992185,
+      "loss": 3.4458,
+      "step": 3065
+    },
+    {
+      "epoch": 0.2691015411790198,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0027451872742146927,
+      "loss": 3.3423,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26918931076192226,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0027449486213666527,
+      "loss": 3.395,
+      "step": 3067
+    },
+    {
+      "epoch": 0.2692770803448247,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002744709868469872,
+      "loss": 3.4331,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26936484992772725,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002744471015546166,
+      "loss": 3.3335,
+      "step": 3069
+    },
+    {
+      "epoch": 0.2694526195106297,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0027442320626173584,
+      "loss": 3.3516,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26954038909353223,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002743993009705284,
+      "loss": 3.4409,
+      "step": 3071
+    },
+    {
+      "epoch": 0.2696281586764347,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0027437538568317855,
+      "loss": 3.4355,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26971592825933716,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.002743514604018715,
+      "loss": 3.3955,
+      "step": 3073
+    },
+    {
+      "epoch": 0.2698036978422397,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0027432752512879344,
+      "loss": 3.4326,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26989146742514214,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002743035798661313,
+      "loss": 3.4458,
+      "step": 3075
+    },
+    {
+      "epoch": 0.2699792370080446,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0027427962461607307,
+      "loss": 3.2974,
+      "step": 3076
+    },
+    {
+      "epoch": 0.2700670065909471,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002742556593808076,
+      "loss": 3.3862,
+      "step": 3077
+    },
+    {
+      "epoch": 0.2701547761738496,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002742316841625247,
+      "loss": 3.3628,
+      "step": 3078
+    },
+    {
+      "epoch": 0.2702425457567521,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0027420769896341506,
+      "loss": 3.4346,
+      "step": 3079
+    },
+    {
+      "epoch": 0.27033031533965457,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0027418370378567026,
+      "loss": 3.4443,
+      "step": 3080
+    },
+    {
+      "epoch": 0.27041808492255703,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0027415969863148286,
+      "loss": 3.4033,
+      "step": 3081
+    },
+    {
+      "epoch": 0.27050585450545955,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.002741356835030462,
+      "loss": 3.4111,
+      "step": 3082
+    },
+    {
+      "epoch": 0.270593624088362,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0027411165840255467,
+      "loss": 3.3442,
+      "step": 3083
+    },
+    {
+      "epoch": 0.27068139367126454,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0027408762333220346,
+      "loss": 3.3516,
+      "step": 3084
+    },
+    {
+      "epoch": 0.270769163254167,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.002740635782941888,
+      "loss": 3.4795,
+      "step": 3085
+    },
+    {
+      "epoch": 0.27085693283706946,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.002740395232907077,
+      "loss": 3.4512,
+      "step": 3086
+    },
+    {
+      "epoch": 0.270944702419972,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0027401545832395815,
+      "loss": 3.3384,
+      "step": 3087
+    },
+    {
+      "epoch": 0.27103247200287445,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002739913833961391,
+      "loss": 3.4346,
+      "step": 3088
+    },
+    {
+      "epoch": 0.27112024158577697,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.0027396729850945024,
+      "loss": 3.418,
+      "step": 3089
+    },
+    {
+      "epoch": 0.27120801116867943,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002739432036660924,
+      "loss": 3.3613,
+      "step": 3090
+    },
+    {
+      "epoch": 0.2712957807515819,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0027391909886826703,
+      "loss": 3.4014,
+      "step": 3091
+    },
+    {
+      "epoch": 0.2713835503344844,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0027389498411817684,
+      "loss": 3.4087,
+      "step": 3092
+    },
+    {
+      "epoch": 0.2714713199173869,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0027387085941802517,
+      "loss": 3.4126,
+      "step": 3093
+    },
+    {
+      "epoch": 0.27155908950028934,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002738467247700164,
+      "loss": 3.4658,
+      "step": 3094
+    },
+    {
+      "epoch": 0.27164685908319186,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.002738225801763558,
+      "loss": 3.3931,
+      "step": 3095
+    },
+    {
+      "epoch": 0.2717346286660943,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002737984256392494,
+      "loss": 3.4429,
+      "step": 3096
+    },
+    {
+      "epoch": 0.27182239824899684,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.002737742611609045,
+      "loss": 3.3691,
+      "step": 3097
+    },
+    {
+      "epoch": 0.2719101678318993,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0027375008674352884,
+      "loss": 3.3579,
+      "step": 3098
+    },
+    {
+      "epoch": 0.27199793741480177,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.002737259023893315,
+      "loss": 3.4248,
+      "step": 3099
+    },
+    {
+      "epoch": 0.2720857069977043,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0027370170810052216,
+      "loss": 3.3921,
+      "step": 3100
+    },
+    {
+      "epoch": 0.27217347658060675,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0027367750387931157,
+      "loss": 3.4419,
+      "step": 3101
+    },
+    {
+      "epoch": 0.2722612461635093,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0027365328972791137,
+      "loss": 3.3994,
+      "step": 3102
+    },
+    {
+      "epoch": 0.27234901574641174,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0027362906564853406,
+      "loss": 3.4604,
+      "step": 3103
+    },
+    {
+      "epoch": 0.2724367853293142,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0027360483164339303,
+      "loss": 3.4468,
+      "step": 3104
+    },
+    {
+      "epoch": 0.2725245549122167,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0027358058771470265,
+      "loss": 3.4629,
+      "step": 3105
+    },
+    {
+      "epoch": 0.2726123244951192,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0027355633386467815,
+      "loss": 3.4624,
+      "step": 3106
+    },
+    {
+      "epoch": 0.27270009407802165,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002735320700955357,
+      "loss": 3.3599,
+      "step": 3107
+    },
+    {
+      "epoch": 0.27278786366092417,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002735077964094923,
+      "loss": 3.4438,
+      "step": 3108
+    },
+    {
+      "epoch": 0.27287563324382663,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0027348351280876595,
+      "loss": 3.4224,
+      "step": 3109
+    },
+    {
+      "epoch": 0.27296340282672915,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0027345921929557548,
+      "loss": 3.3765,
+      "step": 3110
+    },
+    {
+      "epoch": 0.2730511724096316,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002734349158721407,
+      "loss": 3.3506,
+      "step": 3111
+    },
+    {
+      "epoch": 0.2731389419925341,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0027341060254068234,
+      "loss": 3.376,
+      "step": 3112
+    },
+    {
+      "epoch": 0.2732267115754366,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0027338627930342183,
+      "loss": 3.5161,
+      "step": 3113
+    },
+    {
+      "epoch": 0.27331448115833906,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0027336194616258175,
+      "loss": 3.415,
+      "step": 3114
+    },
+    {
+      "epoch": 0.2734022507412416,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0027333760312038555,
+      "loss": 3.4038,
+      "step": 3115
+    },
+    {
+      "epoch": 0.27349002032414405,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0027331325017905736,
+      "loss": 3.4033,
+      "step": 3116
+    },
+    {
+      "epoch": 0.2735777899070465,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002732888873408226,
+      "loss": 3.4707,
+      "step": 3117
+    },
+    {
+      "epoch": 0.27366555948994903,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0027326451460790717,
+      "loss": 3.4302,
+      "step": 3118
+    },
+    {
+      "epoch": 0.2737533290728515,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002732401319825382,
+      "loss": 3.3521,
+      "step": 3119
+    },
+    {
+      "epoch": 0.27384109865575396,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0027321573946694363,
+      "loss": 3.3643,
+      "step": 3120
+    },
+    {
+      "epoch": 0.2739288682386565,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002731913370633521,
+      "loss": 3.5078,
+      "step": 3121
+    },
+    {
+      "epoch": 0.27401663782155894,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0027316692477399357,
+      "loss": 3.4229,
+      "step": 3122
+    },
+    {
+      "epoch": 0.27410440740446146,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002731425026010985,
+      "loss": 3.397,
+      "step": 3123
+    },
+    {
+      "epoch": 0.2741921769873639,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0027311807054689846,
+      "loss": 3.4023,
+      "step": 3124
+    },
+    {
+      "epoch": 0.2742799465702664,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0027309362861362586,
+      "loss": 3.3584,
+      "step": 3125
+    },
+    {
+      "epoch": 0.2743677161531689,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002730691768035141,
+      "loss": 3.3823,
+      "step": 3126
+    },
+    {
+      "epoch": 0.27445548573607137,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002730447151187973,
+      "loss": 3.3711,
+      "step": 3127
+    },
+    {
+      "epoch": 0.2745432553189739,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0027302024356171076,
+      "loss": 3.416,
+      "step": 3128
+    },
+    {
+      "epoch": 0.27463102490187635,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0027299576213449033,
+      "loss": 3.4536,
+      "step": 3129
+    },
+    {
+      "epoch": 0.2747187944847788,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0027297127083937313,
+      "loss": 3.4053,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27480656406768134,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.002729467696785969,
+      "loss": 3.3784,
+      "step": 3131
+    },
+    {
+      "epoch": 0.2748943336505838,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002729222586544004,
+      "loss": 3.2852,
+      "step": 3132
+    },
+    {
+      "epoch": 0.27498210323348626,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002728977377690233,
+      "loss": 3.3193,
+      "step": 3133
+    },
+    {
+      "epoch": 0.2750698728163888,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002728732070247061,
+      "loss": 3.3545,
+      "step": 3134
+    },
+    {
+      "epoch": 0.27515764239929125,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002728486664236903,
+      "loss": 3.3691,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27524541198219377,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0027282411596821817,
+      "loss": 3.3789,
+      "step": 3136
+    },
+    {
+      "epoch": 0.27533318156509623,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002727995556605331,
+      "loss": 3.3511,
+      "step": 3137
+    },
+    {
+      "epoch": 0.2754209511479987,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0027277498550287906,
+      "loss": 3.4951,
+      "step": 3138
+    },
+    {
+      "epoch": 0.2755087207309012,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0027275040549750123,
+      "loss": 3.3496,
+      "step": 3139
+    },
+    {
+      "epoch": 0.2755964903138037,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.002727258156466456,
+      "loss": 3.4277,
+      "step": 3140
+    },
+    {
+      "epoch": 0.2756842598967062,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0027270121595255885,
+      "loss": 3.4795,
+      "step": 3141
+    },
+    {
+      "epoch": 0.27577202947960866,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0027267660641748883,
+      "loss": 3.4277,
+      "step": 3142
+    },
+    {
+      "epoch": 0.2758597990625111,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0027265198704368417,
+      "loss": 3.4028,
+      "step": 3143
+    },
+    {
+      "epoch": 0.27594756864541364,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.002726273578333945,
+      "loss": 3.4526,
+      "step": 3144
+    },
+    {
+      "epoch": 0.2760353382283161,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0027260271878887005,
+      "loss": 3.3833,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27612310781121857,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.002725780699123624,
+      "loss": 3.4385,
+      "step": 3146
+    },
+    {
+      "epoch": 0.2762108773941211,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0027255341120612365,
+      "loss": 3.3193,
+      "step": 3147
+    },
+    {
+      "epoch": 0.27629864697702355,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0027252874267240704,
+      "loss": 3.4062,
+      "step": 3148
+    },
+    {
+      "epoch": 0.2763864165599261,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0027250406431346646,
+      "loss": 3.4565,
+      "step": 3149
+    },
+    {
+      "epoch": 0.27647418614282854,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0027247937613155697,
+      "loss": 3.3975,
+      "step": 3150
+    },
+    {
+      "epoch": 0.276561955725731,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0027245467812893435,
+      "loss": 3.3584,
+      "step": 3151
+    },
+    {
+      "epoch": 0.2766497253086335,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0027242997030785545,
+      "loss": 3.5273,
+      "step": 3152
+    },
+    {
+      "epoch": 0.276737494891536,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.002724052526705777,
+      "loss": 3.3838,
+      "step": 3153
+    },
+    {
+      "epoch": 0.2768252644744385,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002723805252193597,
+      "loss": 3.4067,
+      "step": 3154
+    },
+    {
+      "epoch": 0.27691303405734097,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0027235578795646097,
+      "loss": 3.3872,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27700080364024343,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0027233104088414165,
+      "loss": 3.356,
+      "step": 3156
+    },
+    {
+      "epoch": 0.27708857322314595,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0027230628400466317,
+      "loss": 3.3589,
+      "step": 3157
+    },
+    {
+      "epoch": 0.2771763428060484,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0027228151732028743,
+      "loss": 3.4878,
+      "step": 3158
+    },
+    {
+      "epoch": 0.27726411238895093,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0027225674083327755,
+      "loss": 3.4194,
+      "step": 3159
+    },
+    {
+      "epoch": 0.2773518819718534,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002722319545458975,
+      "loss": 3.4146,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27743965155475586,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0027220715846041185,
+      "loss": 3.3892,
+      "step": 3161
+    },
+    {
+      "epoch": 0.2775274211376584,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0027218235257908657,
+      "loss": 3.3931,
+      "step": 3162
+    },
+    {
+      "epoch": 0.27761519072056084,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0027215753690418803,
+      "loss": 3.3169,
+      "step": 3163
+    },
+    {
+      "epoch": 0.2777029603034633,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002721327114379838,
+      "loss": 3.4648,
+      "step": 3164
+    },
+    {
+      "epoch": 0.27779072988636583,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0027210787618274224,
+      "loss": 3.3633,
+      "step": 3165
+    },
+    {
+      "epoch": 0.2778784994692683,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002720830311407327,
+      "loss": 3.3838,
+      "step": 3166
+    },
+    {
+      "epoch": 0.2779662690521708,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.002720581763142252,
+      "loss": 3.3965,
+      "step": 3167
+    },
+    {
+      "epoch": 0.2780540386350733,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002720333117054909,
+      "loss": 3.3667,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27814180821797574,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.002720084373168018,
+      "loss": 3.4492,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27822957780087826,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002719835531504306,
+      "loss": 3.3643,
+      "step": 3170
+    },
+    {
+      "epoch": 0.2783173473837807,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0027195865920865113,
+      "loss": 3.3691,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27840511696668324,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0027193375549373804,
+      "loss": 3.3379,
+      "step": 3172
+    },
+    {
+      "epoch": 0.2784928865495857,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0027190884200796682,
+      "loss": 3.3286,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27858065613248817,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002718839187536139,
+      "loss": 3.4346,
+      "step": 3174
+    },
+    {
+      "epoch": 0.2786684257153907,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0027185898573295663,
+      "loss": 3.3799,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27875619529829315,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002718340429482731,
+      "loss": 3.3325,
+      "step": 3176
+    },
+    {
+      "epoch": 0.2788439648811956,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.002718090904018426,
+      "loss": 3.3608,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27893173446409814,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002717841280959449,
+      "loss": 3.418,
+      "step": 3178
+    },
+    {
+      "epoch": 0.2790195040470006,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.0027175915603286104,
+      "loss": 3.3823,
+      "step": 3179
+    },
+    {
+      "epoch": 0.2791072736299031,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0027173417421487277,
+      "loss": 3.4658,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2791950432128056,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.002717091826442627,
+      "loss": 3.397,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27928281279570805,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0027168418132331434,
+      "loss": 3.3491,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27937058237861057,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0027165917025431226,
+      "loss": 3.5068,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27945835196151303,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0027163414943954176,
+      "loss": 3.4473,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27954612154441555,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.00271609118881289,
+      "loss": 3.4502,
+      "step": 3185
+    },
+    {
+      "epoch": 0.279633891127318,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.002715840785818412,
+      "loss": 3.4844,
+      "step": 3186
+    },
+    {
+      "epoch": 0.2797216607102205,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0027155902854348623,
+      "loss": 3.3589,
+      "step": 3187
+    },
+    {
+      "epoch": 0.279809430293123,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0027153396876851313,
+      "loss": 3.4492,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27989719987602546,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.002715088992592116,
+      "loss": 3.4175,
+      "step": 3189
+    },
+    {
+      "epoch": 0.2799849694589279,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.002714838200178724,
+      "loss": 3.4438,
+      "step": 3190
+    },
+    {
+      "epoch": 0.28007273904183044,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.00271458731046787,
+      "loss": 3.3774,
+      "step": 3191
+    },
+    {
+      "epoch": 0.2801605086247329,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0027143363234824786,
+      "loss": 3.4072,
+      "step": 3192
+    },
+    {
+      "epoch": 0.2802482782076354,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0027140852392454834,
+      "loss": 3.4263,
+      "step": 3193
+    },
+    {
+      "epoch": 0.2803360477905379,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0027138340577798274,
+      "loss": 3.4307,
+      "step": 3194
+    },
+    {
+      "epoch": 0.28042381737344035,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0027135827791084613,
+      "loss": 3.373,
+      "step": 3195
+    },
+    {
+      "epoch": 0.2805115869563429,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002713331403254345,
+      "loss": 3.4272,
+      "step": 3196
+    },
+    {
+      "epoch": 0.28059935653924534,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0027130799302404474,
+      "loss": 3.4087,
+      "step": 3197
+    },
+    {
+      "epoch": 0.28068712612214786,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002712828360089747,
+      "loss": 3.3872,
+      "step": 3198
+    },
+    {
+      "epoch": 0.2807748957050503,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0027125766928252294,
+      "loss": 3.4023,
+      "step": 3199
+    },
+    {
+      "epoch": 0.2808626652879528,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0027123249284698916,
+      "loss": 3.3516,
+      "step": 3200
+    },
+    {
+      "epoch": 0.2809504348708553,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0027120730670467377,
+      "loss": 3.3701,
+      "step": 3201
+    },
+    {
+      "epoch": 0.28103820445375777,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0027118211085787793,
+      "loss": 3.4531,
+      "step": 3202
+    },
+    {
+      "epoch": 0.28112597403666023,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002711569053089041,
+      "loss": 3.3828,
+      "step": 3203
+    },
+    {
+      "epoch": 0.28121374361956275,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0027113169006005525,
+      "loss": 3.4009,
+      "step": 3204
+    },
+    {
+      "epoch": 0.2813015132024652,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0027110646511363545,
+      "loss": 3.4126,
+      "step": 3205
+    },
+    {
+      "epoch": 0.28138928278536773,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0027108123047194952,
+      "loss": 3.3657,
+      "step": 3206
+    },
+    {
+      "epoch": 0.2814770523682702,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0027105598613730323,
+      "loss": 3.3486,
+      "step": 3207
+    },
+    {
+      "epoch": 0.28156482195117266,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0027103073211200324,
+      "loss": 3.4199,
+      "step": 3208
+    },
+    {
+      "epoch": 0.2816525915340752,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0027100546839835707,
+      "loss": 3.3608,
+      "step": 3209
+    },
+    {
+      "epoch": 0.28174036111697764,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002709801949986732,
+      "loss": 3.3896,
+      "step": 3210
+    },
+    {
+      "epoch": 0.28182813069988016,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0027095491191526087,
+      "loss": 3.377,
+      "step": 3211
+    },
+    {
+      "epoch": 0.2819159002827826,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.002709296191504303,
+      "loss": 3.4277,
+      "step": 3212
+    },
+    {
+      "epoch": 0.2820036698656851,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002709043167064925,
+      "loss": 3.3828,
+      "step": 3213
+    },
+    {
+      "epoch": 0.2820914394485876,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0027087900458575963,
+      "loss": 3.4609,
+      "step": 3214
+    },
+    {
+      "epoch": 0.2821792090314901,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0027085368279054435,
+      "loss": 3.5044,
+      "step": 3215
+    },
+    {
+      "epoch": 0.28226697861439254,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0027082835132316036,
+      "loss": 3.4028,
+      "step": 3216
+    },
+    {
+      "epoch": 0.28235474819729506,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0027080301018592243,
+      "loss": 3.3711,
+      "step": 3217
+    },
+    {
+      "epoch": 0.2824425177801975,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0027077765938114595,
+      "loss": 3.4067,
+      "step": 3218
+    },
+    {
+      "epoch": 0.28253028736310004,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0027075229891114732,
+      "loss": 3.3179,
+      "step": 3219
+    },
+    {
+      "epoch": 0.2826180569460025,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002707269287782438,
+      "loss": 3.4194,
+      "step": 3220
+    },
+    {
+      "epoch": 0.28270582652890497,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0027070154898475355,
+      "loss": 3.4194,
+      "step": 3221
+    },
+    {
+      "epoch": 0.2827935961118075,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002706761595329956,
+      "loss": 3.3955,
+      "step": 3222
+    },
+    {
+      "epoch": 0.28288136569470995,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0027065076042528985,
+      "loss": 3.4316,
+      "step": 3223
+    },
+    {
+      "epoch": 0.28296913527761247,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0027062535166395706,
+      "loss": 3.4507,
+      "step": 3224
+    },
+    {
+      "epoch": 0.28305690486051494,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0027059993325131895,
+      "loss": 3.3525,
+      "step": 3225
+    },
+    {
+      "epoch": 0.2831446744434174,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.002705745051896981,
+      "loss": 3.4175,
+      "step": 3226
+    },
+    {
+      "epoch": 0.2832324440263199,
+      "grad_norm": 0.8203125,
+      "learning_rate": 0.0027054906748141786,
+      "loss": 3.499,
+      "step": 3227
+    },
+    {
+      "epoch": 0.2833202136092224,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0027052362012880264,
+      "loss": 3.4233,
+      "step": 3228
+    },
+    {
+      "epoch": 0.2834079831921249,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.002704981631341776,
+      "loss": 3.3994,
+      "step": 3229
+    },
+    {
+      "epoch": 0.28349575277502737,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0027047269649986883,
+      "loss": 3.3706,
+      "step": 3230
+    },
+    {
+      "epoch": 0.28358352235792983,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.002704472202282032,
+      "loss": 3.3457,
+      "step": 3231
+    },
+    {
+      "epoch": 0.28367129194083235,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002704217343215087,
+      "loss": 3.4048,
+      "step": 3232
+    },
+    {
+      "epoch": 0.2837590615237348,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00270396238782114,
+      "loss": 3.3354,
+      "step": 3233
+    },
+    {
+      "epoch": 0.2838468311066373,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002703707336123487,
+      "loss": 3.4121,
+      "step": 3234
+    },
+    {
+      "epoch": 0.2839346006895398,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.002703452188145433,
+      "loss": 3.3408,
+      "step": 3235
+    },
+    {
+      "epoch": 0.28402237027244226,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0027031969439102917,
+      "loss": 3.4512,
+      "step": 3236
+    },
+    {
+      "epoch": 0.2841101398553448,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0027029416034413845,
+      "loss": 3.396,
+      "step": 3237
+    },
+    {
+      "epoch": 0.28419790943824724,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002702686166762044,
+      "loss": 3.4097,
+      "step": 3238
+    },
+    {
+      "epoch": 0.2842856790211497,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002702430633895609,
+      "loss": 3.4077,
+      "step": 3239
+    },
+    {
+      "epoch": 0.2843734486040522,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0027021750048654295,
+      "loss": 3.4741,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2844612181869547,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0027019192796948624,
+      "loss": 3.4629,
+      "step": 3241
+    },
+    {
+      "epoch": 0.2845489877698572,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0027016634584072744,
+      "loss": 3.3818,
+      "step": 3242
+    },
+    {
+      "epoch": 0.2846367573527597,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0027014075410260402,
+      "loss": 3.4023,
+      "step": 3243
+    },
+    {
+      "epoch": 0.28472452693566214,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0027011515275745437,
+      "loss": 3.3994,
+      "step": 3244
+    },
+    {
+      "epoch": 0.28481229651856466,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.0027008954180761783,
+      "loss": 3.3555,
+      "step": 3245
+    },
+    {
+      "epoch": 0.2849000661014671,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002700639212554345,
+      "loss": 3.2881,
+      "step": 3246
+    },
+    {
+      "epoch": 0.2849878356843696,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.002700382911032454,
+      "loss": 3.3364,
+      "step": 3247
+    },
+    {
+      "epoch": 0.2850756052672721,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002700126513533925,
+      "loss": 3.4028,
+      "step": 3248
+    },
+    {
+      "epoch": 0.28516337485017457,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002699870020082185,
+      "loss": 3.3809,
+      "step": 3249
+    },
+    {
+      "epoch": 0.2852511444330771,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002699613430700671,
+      "loss": 3.3926,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28533891401597955,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0026993567454128287,
+      "loss": 3.355,
+      "step": 3251
+    },
+    {
+      "epoch": 0.285426683598882,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002699099964242111,
+      "loss": 3.3721,
+      "step": 3252
+    },
+    {
+      "epoch": 0.28551445318178453,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0026988430872119814,
+      "loss": 3.4482,
+      "step": 3253
+    },
+    {
+      "epoch": 0.285602222764687,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0026985861143459123,
+      "loss": 3.3696,
+      "step": 3254
+    },
+    {
+      "epoch": 0.2856899923475895,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002698329045667383,
+      "loss": 3.4824,
+      "step": 3255
+    },
+    {
+      "epoch": 0.285777761930492,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0026980718811998833,
+      "loss": 3.4111,
+      "step": 3256
+    },
+    {
+      "epoch": 0.28586553151339444,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002697814620966911,
+      "loss": 3.4385,
+      "step": 3257
+    },
+    {
+      "epoch": 0.28595330109629696,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0026975572649919725,
+      "loss": 3.4087,
+      "step": 3258
+    },
+    {
+      "epoch": 0.2860410706791994,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.002697299813298583,
+      "loss": 3.4766,
+      "step": 3259
+    },
+    {
+      "epoch": 0.2861288402621019,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0026970422659102667,
+      "loss": 3.4351,
+      "step": 3260
+    },
+    {
+      "epoch": 0.2862166098450044,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.002696784622850557,
+      "loss": 3.376,
+      "step": 3261
+    },
+    {
+      "epoch": 0.2863043794279069,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0026965268841429952,
+      "loss": 3.4868,
+      "step": 3262
+    },
+    {
+      "epoch": 0.2863921490108094,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0026962690498111316,
+      "loss": 3.4624,
+      "step": 3263
+    },
+    {
+      "epoch": 0.28647991859371186,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0026960111198785254,
+      "loss": 3.4653,
+      "step": 3264
+    },
+    {
+      "epoch": 0.2865676881766143,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0026957530943687445,
+      "loss": 3.311,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28665545775951684,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0026954949733053652,
+      "loss": 3.3716,
+      "step": 3266
+    },
+    {
+      "epoch": 0.2867432273424193,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002695236756711973,
+      "loss": 3.3994,
+      "step": 3267
+    },
+    {
+      "epoch": 0.2868309969253218,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.002694978444612162,
+      "loss": 3.3936,
+      "step": 3268
+    },
+    {
+      "epoch": 0.2869187665082243,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002694720037029535,
+      "loss": 3.4146,
+      "step": 3269
+    },
+    {
+      "epoch": 0.28700653609112675,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002694461533987703,
+      "loss": 3.3403,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28709430567402927,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002694202935510286,
+      "loss": 3.3594,
+      "step": 3271
+    },
+    {
+      "epoch": 0.28718207525693173,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0026939442416209144,
+      "loss": 3.3354,
+      "step": 3272
+    },
+    {
+      "epoch": 0.2872698448398342,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.0026936854523432244,
+      "loss": 3.4175,
+      "step": 3273
+    },
+    {
+      "epoch": 0.2873576144227367,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002693426567700863,
+      "loss": 3.396,
+      "step": 3274
+    },
+    {
+      "epoch": 0.2874453840056392,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002693167587717485,
+      "loss": 3.354,
+      "step": 3275
+    },
+    {
+      "epoch": 0.2875331535885417,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0026929085124167544,
+      "loss": 3.4233,
+      "step": 3276
+    },
+    {
+      "epoch": 0.28762092317144416,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002692649341822343,
+      "loss": 3.4399,
+      "step": 3277
+    },
+    {
+      "epoch": 0.28770869275434663,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.002692390075957933,
+      "loss": 3.4082,
+      "step": 3278
+    },
+    {
+      "epoch": 0.28779646233724915,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0026921307148472136,
+      "loss": 3.311,
+      "step": 3279
+    },
+    {
+      "epoch": 0.2878842319201516,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.002691871258513884,
+      "loss": 3.3687,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28797200150305413,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0026916117069816507,
+      "loss": 3.4014,
+      "step": 3281
+    },
+    {
+      "epoch": 0.2880597710859566,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0026913520602742306,
+      "loss": 3.3506,
+      "step": 3282
+    },
+    {
+      "epoch": 0.28814754066885906,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002691092318415348,
+      "loss": 3.3906,
+      "step": 3283
+    },
+    {
+      "epoch": 0.2882353102517616,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0026908324814287356,
+      "loss": 3.3384,
+      "step": 3284
+    },
+    {
+      "epoch": 0.28832307983466404,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0026905725493381366,
+      "loss": 3.3066,
+      "step": 3285
+    },
+    {
+      "epoch": 0.2884108494175665,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002690312522167301,
+      "loss": 3.4053,
+      "step": 3286
+    },
+    {
+      "epoch": 0.288498619000469,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0026900523999399887,
+      "loss": 3.3311,
+      "step": 3287
+    },
+    {
+      "epoch": 0.2885863885833715,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002689792182679968,
+      "loss": 3.3242,
+      "step": 3288
+    },
+    {
+      "epoch": 0.288674158166274,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.002689531870411015,
+      "loss": 3.4199,
+      "step": 3289
+    },
+    {
+      "epoch": 0.2887619277491765,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0026892714631569165,
+      "loss": 3.4463,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28884969733207894,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0026890109609414648,
+      "loss": 3.4497,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28893746691498146,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0026887503637884644,
+      "loss": 3.4473,
+      "step": 3292
+    },
+    {
+      "epoch": 0.2890252364978839,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0026884896717217256,
+      "loss": 3.374,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28911300608078644,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00268822888476507,
+      "loss": 3.3726,
+      "step": 3294
+    },
+    {
+      "epoch": 0.2892007756636889,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002687968002942326,
+      "loss": 3.3921,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28928854524659137,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0026877070262773303,
+      "loss": 3.3398,
+      "step": 3296
+    },
+    {
+      "epoch": 0.2893763148294939,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0026874459547939302,
+      "loss": 3.3374,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28946408441239635,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00268718478851598,
+      "loss": 3.3833,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28955185399529887,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0026869235274673427,
+      "loss": 3.2783,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28963962357820133,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.002686662171671892,
+      "loss": 3.3843,
+      "step": 3300
+    },
+    {
+      "epoch": 0.2897273931611038,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0026864007211535074,
+      "loss": 3.4292,
+      "step": 3301
+    },
+    {
+      "epoch": 0.2898151627440063,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0026861391759360797,
+      "loss": 3.3452,
+      "step": 3302
+    },
+    {
+      "epoch": 0.2899029323269088,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0026858775360435057,
+      "loss": 3.4517,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28999070190981124,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002685615801499693,
+      "loss": 3.3564,
+      "step": 3304
+    },
+    {
+      "epoch": 0.29007847149271376,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0026853539723285568,
+      "loss": 3.4219,
+      "step": 3305
+    },
+    {
+      "epoch": 0.2901662410756162,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002685092048554022,
+      "loss": 3.373,
+      "step": 3306
+    },
+    {
+      "epoch": 0.29025401065851875,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0026848300302000196,
+      "loss": 3.3325,
+      "step": 3307
+    },
+    {
+      "epoch": 0.2903417802414212,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0026845679172904932,
+      "loss": 3.4141,
+      "step": 3308
+    },
+    {
+      "epoch": 0.2904295498243237,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002684305709849391,
+      "loss": 3.3613,
+      "step": 3309
+    },
+    {
+      "epoch": 0.2905173194072262,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0026840434079006725,
+      "loss": 3.3076,
+      "step": 3310
+    },
+    {
+      "epoch": 0.29060508899012866,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002683781011468305,
+      "loss": 3.3623,
+      "step": 3311
+    },
+    {
+      "epoch": 0.2906928585730312,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002683518520576265,
+      "loss": 3.3569,
+      "step": 3312
+    },
+    {
+      "epoch": 0.29078062815593364,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002683255935248536,
+      "loss": 3.3315,
+      "step": 3313
+    },
+    {
+      "epoch": 0.2908683977388361,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0026829932555091115,
+      "loss": 3.3872,
+      "step": 3314
+    },
+    {
+      "epoch": 0.2909561673217386,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0026827304813819937,
+      "loss": 3.3765,
+      "step": 3315
+    },
+    {
+      "epoch": 0.2910439369046411,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002682467612891193,
+      "loss": 3.418,
+      "step": 3316
+    },
+    {
+      "epoch": 0.29113170648754355,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0026822046500607284,
+      "loss": 3.3521,
+      "step": 3317
+    },
+    {
+      "epoch": 0.29121947607044607,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0026819415929146273,
+      "loss": 3.4087,
+      "step": 3318
+    },
+    {
+      "epoch": 0.29130724565334853,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002681678441476927,
+      "loss": 3.3594,
+      "step": 3319
+    },
+    {
+      "epoch": 0.29139501523625105,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002681415195771671,
+      "loss": 3.332,
+      "step": 3320
+    },
+    {
+      "epoch": 0.2914827848191535,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0026811518558229135,
+      "loss": 3.3496,
+      "step": 3321
+    },
+    {
+      "epoch": 0.291570554402056,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002680888421654717,
+      "loss": 3.3853,
+      "step": 3322
+    },
+    {
+      "epoch": 0.2916583239849585,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002680624893291152,
+      "loss": 3.4204,
+      "step": 3323
+    },
+    {
+      "epoch": 0.29174609356786096,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002680361270756298,
+      "loss": 3.4248,
+      "step": 3324
+    },
+    {
+      "epoch": 0.2918338631507635,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002680097554074243,
+      "loss": 3.4165,
+      "step": 3325
+    },
+    {
+      "epoch": 0.29192163273366595,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.002679833743269083,
+      "loss": 3.3716,
+      "step": 3326
+    },
+    {
+      "epoch": 0.2920094023165684,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0026795698383649236,
+      "loss": 3.3862,
+      "step": 3327
+    },
+    {
+      "epoch": 0.29209717189947093,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0026793058393858784,
+      "loss": 3.3906,
+      "step": 3328
+    },
+    {
+      "epoch": 0.2921849414823734,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0026790417463560708,
+      "loss": 3.3579,
+      "step": 3329
+    },
+    {
+      "epoch": 0.29227271106527586,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0026787775592996307,
+      "loss": 3.3647,
+      "step": 3330
+    },
+    {
+      "epoch": 0.2923604806481784,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0026785132782406983,
+      "loss": 3.3721,
+      "step": 3331
+    },
+    {
+      "epoch": 0.29244825023108084,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0026782489032034207,
+      "loss": 3.394,
+      "step": 3332
+    },
+    {
+      "epoch": 0.29253601981398336,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0026779844342119555,
+      "loss": 3.4155,
+      "step": 3333
+    },
+    {
+      "epoch": 0.2926237893968858,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002677719871290468,
+      "loss": 3.3999,
+      "step": 3334
+    },
+    {
+      "epoch": 0.2927115589797883,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0026774552144631317,
+      "loss": 3.312,
+      "step": 3335
+    },
+    {
+      "epoch": 0.2927993285626908,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00267719046375413,
+      "loss": 3.3931,
+      "step": 3336
+    },
+    {
+      "epoch": 0.29288709814559327,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.0026769256191876522,
+      "loss": 3.3901,
+      "step": 3337
+    },
+    {
+      "epoch": 0.2929748677284958,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0026766606807879,
+      "loss": 3.3525,
+      "step": 3338
+    },
+    {
+      "epoch": 0.29306263731139826,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0026763956485790804,
+      "loss": 3.4189,
+      "step": 3339
+    },
+    {
+      "epoch": 0.2931504068943007,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.002676130522585411,
+      "loss": 3.4141,
+      "step": 3340
+    },
+    {
+      "epoch": 0.29323817647720324,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0026758653028311156,
+      "loss": 3.4458,
+      "step": 3341
+    },
+    {
+      "epoch": 0.2933259460601057,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0026755999893404298,
+      "loss": 3.3691,
+      "step": 3342
+    },
+    {
+      "epoch": 0.29341371564300817,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002675334582137595,
+      "loss": 3.4814,
+      "step": 3343
+    },
+    {
+      "epoch": 0.2935014852259107,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002675069081246863,
+      "loss": 3.3701,
+      "step": 3344
+    },
+    {
+      "epoch": 0.29358925480881315,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0026748034866924935,
+      "loss": 3.3979,
+      "step": 3345
+    },
+    {
+      "epoch": 0.29367702439171567,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0026745377984987535,
+      "loss": 3.377,
+      "step": 3346
+    },
+    {
+      "epoch": 0.29376479397461813,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002674272016689921,
+      "loss": 3.3628,
+      "step": 3347
+    },
+    {
+      "epoch": 0.2938525635575206,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0026740061412902807,
+      "loss": 3.3486,
+      "step": 3348
+    },
+    {
+      "epoch": 0.2939403331404231,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0026737401723241263,
+      "loss": 3.3506,
+      "step": 3349
+    },
+    {
+      "epoch": 0.2940281027233256,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002673474109815761,
+      "loss": 3.3799,
+      "step": 3350
+    },
+    {
+      "epoch": 0.2941158723062281,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0026732079537894944,
+      "loss": 3.3687,
+      "step": 3351
+    },
+    {
+      "epoch": 0.29420364188913056,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0026729417042696472,
+      "loss": 3.3457,
+      "step": 3352
+    },
+    {
+      "epoch": 0.294291411472033,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0026726753612805465,
+      "loss": 3.3315,
+      "step": 3353
+    },
+    {
+      "epoch": 0.29437918105493555,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0026724089248465295,
+      "loss": 3.3594,
+      "step": 3354
+    },
+    {
+      "epoch": 0.294466950637838,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0026721423949919412,
+      "loss": 3.3862,
+      "step": 3355
+    },
+    {
+      "epoch": 0.2945547202207405,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002671875771741135,
+      "loss": 3.3892,
+      "step": 3356
+    },
+    {
+      "epoch": 0.294642489803643,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0026716090551184733,
+      "loss": 3.4019,
+      "step": 3357
+    },
+    {
+      "epoch": 0.29473025938654546,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0026713422451483263,
+      "loss": 3.375,
+      "step": 3358
+    },
+    {
+      "epoch": 0.294818028969448,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0026710753418550735,
+      "loss": 3.3457,
+      "step": 3359
+    },
+    {
+      "epoch": 0.29490579855235044,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0026708083452631035,
+      "loss": 3.377,
+      "step": 3360
+    },
+    {
+      "epoch": 0.2949935681352529,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002670541255396811,
+      "loss": 3.4175,
+      "step": 3361
+    },
+    {
+      "epoch": 0.2950813377181554,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0026702740722806028,
+      "loss": 3.4727,
+      "step": 3362
+    },
+    {
+      "epoch": 0.2951691073010579,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0026700067959388903,
+      "loss": 3.395,
+      "step": 3363
+    },
+    {
+      "epoch": 0.2952568768839604,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002669739426396096,
+      "loss": 3.3584,
+      "step": 3364
+    },
+    {
+      "epoch": 0.29534464646686287,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0026694719636766507,
+      "loss": 3.3301,
+      "step": 3365
+    },
+    {
+      "epoch": 0.29543241604976533,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0026692044078049926,
+      "loss": 3.4229,
+      "step": 3366
+    },
+    {
+      "epoch": 0.29552018563266785,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0026689367588055694,
+      "loss": 3.3477,
+      "step": 3367
+    },
+    {
+      "epoch": 0.2956079552155703,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002668669016702837,
+      "loss": 3.3232,
+      "step": 3368
+    },
+    {
+      "epoch": 0.29569572479847284,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00266840118152126,
+      "loss": 3.3354,
+      "step": 3369
+    },
+    {
+      "epoch": 0.2957834943813753,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002668133253285311,
+      "loss": 3.3867,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29587126396427776,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.002667865232019472,
+      "loss": 3.2666,
+      "step": 3371
+    },
+    {
+      "epoch": 0.2959590335471803,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0026675971177482316,
+      "loss": 3.4092,
+      "step": 3372
+    },
+    {
+      "epoch": 0.29604680313008275,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.0026673289104960896,
+      "loss": 3.4165,
+      "step": 3373
+    },
+    {
+      "epoch": 0.2961345727129852,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0026670606102875524,
+      "loss": 3.3989,
+      "step": 3374
+    },
+    {
+      "epoch": 0.29622234229588773,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0026667922171471348,
+      "loss": 3.3687,
+      "step": 3375
+    },
+    {
+      "epoch": 0.2963101118787902,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0026665237310993612,
+      "loss": 3.3267,
+      "step": 3376
+    },
+    {
+      "epoch": 0.2963978814616927,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0026662551521687648,
+      "loss": 3.3799,
+      "step": 3377
+    },
+    {
+      "epoch": 0.2964856510445952,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002665986480379885,
+      "loss": 3.3726,
+      "step": 3378
+    },
+    {
+      "epoch": 0.29657342062749764,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.002665717715757272,
+      "loss": 3.3608,
+      "step": 3379
+    },
+    {
+      "epoch": 0.29666119021040016,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0026654488583254836,
+      "loss": 3.3867,
+      "step": 3380
+    },
+    {
+      "epoch": 0.2967489597933026,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0026651799081090855,
+      "loss": 3.396,
+      "step": 3381
+    },
+    {
+      "epoch": 0.29683672937620514,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0026649108651326534,
+      "loss": 3.3506,
+      "step": 3382
+    },
+    {
+      "epoch": 0.2969244989591076,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.00266464172942077,
+      "loss": 3.3242,
+      "step": 3383
+    },
+    {
+      "epoch": 0.29701226854201007,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002664372500998027,
+      "loss": 3.376,
+      "step": 3384
+    },
+    {
+      "epoch": 0.2971000381249126,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0026641031798890244,
+      "loss": 3.4692,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29718780770781505,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0026638337661183714,
+      "loss": 3.3784,
+      "step": 3386
+    },
+    {
+      "epoch": 0.2972755772907175,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0026635642597106855,
+      "loss": 3.4194,
+      "step": 3387
+    },
+    {
+      "epoch": 0.29736334687362004,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.0026632946606905913,
+      "loss": 3.3779,
+      "step": 3388
+    },
+    {
+      "epoch": 0.2974511164565225,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0026630249690827233,
+      "loss": 3.3203,
+      "step": 3389
+    },
+    {
+      "epoch": 0.297538886039425,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0026627551849117237,
+      "loss": 3.377,
+      "step": 3390
+    },
+    {
+      "epoch": 0.2976266556223275,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002662485308202245,
+      "loss": 3.3823,
+      "step": 3391
+    },
+    {
+      "epoch": 0.29771442520522995,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0026622153389789446,
+      "loss": 3.3896,
+      "step": 3392
+    },
+    {
+      "epoch": 0.29780219478813247,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002661945277266491,
+      "loss": 3.4893,
+      "step": 3393
+    },
+    {
+      "epoch": 0.29788996437103493,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0026616751230895616,
+      "loss": 3.3174,
+      "step": 3394
+    },
+    {
+      "epoch": 0.29797773395393745,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0026614048764728405,
+      "loss": 3.3975,
+      "step": 3395
+    },
+    {
+      "epoch": 0.2980655035368399,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0026611345374410204,
+      "loss": 3.415,
+      "step": 3396
+    },
+    {
+      "epoch": 0.2981532731197424,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002660864106018804,
+      "loss": 3.3872,
+      "step": 3397
+    },
+    {
+      "epoch": 0.2982410427026449,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002660593582230901,
+      "loss": 3.3555,
+      "step": 3398
+    },
+    {
+      "epoch": 0.29832881228554736,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0026603229661020292,
+      "loss": 3.4072,
+      "step": 3399
+    },
+    {
+      "epoch": 0.2984165818684498,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0026600522576569166,
+      "loss": 3.3418,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29850435145135235,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0026597814569202984,
+      "loss": 3.4429,
+      "step": 3401
+    },
+    {
+      "epoch": 0.2985921210342548,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0026595105639169183,
+      "loss": 3.3154,
+      "step": 3402
+    },
+    {
+      "epoch": 0.29867989061715733,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002659239578671529,
+      "loss": 3.4023,
+      "step": 3403
+    },
+    {
+      "epoch": 0.2987676602000598,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0026589685012088906,
+      "loss": 3.3564,
+      "step": 3404
+    },
+    {
+      "epoch": 0.29885542978296226,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002658697331553773,
+      "loss": 3.3511,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2989431993658648,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0026584260697309534,
+      "loss": 3.4443,
+      "step": 3406
+    },
+    {
+      "epoch": 0.29903096894876724,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0026581547157652173,
+      "loss": 3.3735,
+      "step": 3407
+    },
+    {
+      "epoch": 0.29911873853166976,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.00265788326968136,
+      "loss": 3.4009,
+      "step": 3408
+    },
+    {
+      "epoch": 0.2992065081145722,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002657611731504184,
+      "loss": 3.3496,
+      "step": 3409
+    },
+    {
+      "epoch": 0.2992942776974747,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0026573401012585005,
+      "loss": 3.4575,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2993820472803772,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002657068378969129,
+      "loss": 3.4521,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29946981686327967,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0026567965646608982,
+      "loss": 3.3628,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29955758644618213,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0026565246583586436,
+      "loss": 3.3403,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29964535602908465,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002656252660087211,
+      "loss": 3.3657,
+      "step": 3414
+    },
+    {
+      "epoch": 0.2997331256119871,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0026559805698714535,
+      "loss": 3.4194,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29982089519488964,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002655708387736232,
+      "loss": 3.3271,
+      "step": 3416
+    },
+    {
+      "epoch": 0.2999086647777921,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002655436113706418,
+      "loss": 3.3564,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29999643436069456,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.002655163747806889,
+      "loss": 3.3779,
+      "step": 3418
+    },
+    {
+      "epoch": 0.3000842039435971,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0026548912900625325,
+      "loss": 3.4238,
+      "step": 3419
+    },
+    {
+      "epoch": 0.30017197352649955,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.002654618740498243,
+      "loss": 3.3398,
+      "step": 3420
+    },
+    {
+      "epoch": 0.30025974310940207,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0026543460991389246,
+      "loss": 3.3975,
+      "step": 3421
+    },
+    {
+      "epoch": 0.30034751269230453,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0026540733660094895,
+      "loss": 3.353,
+      "step": 3422
+    },
+    {
+      "epoch": 0.300435282275207,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0026538005411348583,
+      "loss": 3.4297,
+      "step": 3423
+    },
+    {
+      "epoch": 0.3005230518581095,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.002653527624539959,
+      "loss": 3.4355,
+      "step": 3424
+    },
+    {
+      "epoch": 0.300610821441012,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0026532546162497295,
+      "loss": 3.3462,
+      "step": 3425
+    },
+    {
+      "epoch": 0.30069859102391444,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0026529815162891156,
+      "loss": 3.3745,
+      "step": 3426
+    },
+    {
+      "epoch": 0.30078636060681696,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.00265270832468307,
+      "loss": 3.4824,
+      "step": 3427
+    },
+    {
+      "epoch": 0.3008741301897194,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002652435041456557,
+      "loss": 3.3628,
+      "step": 3428
+    },
+    {
+      "epoch": 0.30096189977262194,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0026521616666345454,
+      "loss": 3.3247,
+      "step": 3429
+    },
+    {
+      "epoch": 0.3010496693555244,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0026518882002420155,
+      "loss": 3.3486,
+      "step": 3430
+    },
+    {
+      "epoch": 0.30113743893842687,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0026516146423039544,
+      "loss": 3.3398,
+      "step": 3431
+    },
+    {
+      "epoch": 0.3012252085213294,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002651340992845358,
+      "loss": 3.3672,
+      "step": 3432
+    },
+    {
+      "epoch": 0.30131297810423185,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00265106725189123,
+      "loss": 3.4731,
+      "step": 3433
+    },
+    {
+      "epoch": 0.3014007476871344,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0026507934194665832,
+      "loss": 3.3418,
+      "step": 3434
+    },
+    {
+      "epoch": 0.30148851727003684,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.002650519495596439,
+      "loss": 3.3853,
+      "step": 3435
+    },
+    {
+      "epoch": 0.3015762868529393,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002650245480305826,
+      "loss": 3.3701,
+      "step": 3436
+    },
+    {
+      "epoch": 0.3016640564358418,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0026499713736197823,
+      "loss": 3.4243,
+      "step": 3437
+    },
+    {
+      "epoch": 0.3017518260187443,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0026496971755633533,
+      "loss": 3.3721,
+      "step": 3438
+    },
+    {
+      "epoch": 0.3018395956016468,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002649422886161593,
+      "loss": 3.3584,
+      "step": 3439
+    },
+    {
+      "epoch": 0.30192736518454927,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0026491485054395652,
+      "loss": 3.4409,
+      "step": 3440
+    },
+    {
+      "epoch": 0.30201513476745173,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0026488740334223403,
+      "loss": 3.3711,
+      "step": 3441
+    },
+    {
+      "epoch": 0.30210290435035425,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0026485994701349975,
+      "loss": 3.4053,
+      "step": 3442
+    },
+    {
+      "epoch": 0.3021906739332567,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.002648324815602625,
+      "loss": 3.3379,
+      "step": 3443
+    },
+    {
+      "epoch": 0.3022784435161592,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002648050069850318,
+      "loss": 3.3857,
+      "step": 3444
+    },
+    {
+      "epoch": 0.3023662130990617,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0026477752329031813,
+      "loss": 3.4204,
+      "step": 3445
+    },
+    {
+      "epoch": 0.30245398268196416,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0026475003047863274,
+      "loss": 3.4233,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3025417522648667,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0026472252855248776,
+      "loss": 3.4199,
+      "step": 3447
+    },
+    {
+      "epoch": 0.30262952184776915,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0026469501751439613,
+      "loss": 3.355,
+      "step": 3448
+    },
+    {
+      "epoch": 0.3027172914306716,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002646674973668716,
+      "loss": 3.4004,
+      "step": 3449
+    },
+    {
+      "epoch": 0.30280506101357413,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0026463996811242865,
+      "loss": 3.4995,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3028928305964766,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0026461242975358294,
+      "loss": 3.3662,
+      "step": 3451
+    },
+    {
+      "epoch": 0.3029806001793791,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002645848822928506,
+      "loss": 3.376,
+      "step": 3452
+    },
+    {
+      "epoch": 0.3030683697622816,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002645573257327487,
+      "loss": 3.4224,
+      "step": 3453
+    },
+    {
+      "epoch": 0.30315613934518404,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002645297600757952,
+      "loss": 3.4316,
+      "step": 3454
+    },
+    {
+      "epoch": 0.30324390892808656,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002645021853245089,
+      "loss": 3.3955,
+      "step": 3455
+    },
+    {
+      "epoch": 0.303331678510989,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002644746014814093,
+      "loss": 3.3623,
+      "step": 3456
+    },
+    {
+      "epoch": 0.3034194480938915,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0026444700854901695,
+      "loss": 3.3267,
+      "step": 3457
+    },
+    {
+      "epoch": 0.303507217676794,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0026441940652985295,
+      "loss": 3.3862,
+      "step": 3458
+    },
+    {
+      "epoch": 0.30359498725969647,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002643917954264394,
+      "loss": 3.3423,
+      "step": 3459
+    },
+    {
+      "epoch": 0.303682756842599,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002643641752412994,
+      "loss": 3.3628,
+      "step": 3460
+    },
+    {
+      "epoch": 0.30377052642550145,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0026433654597695644,
+      "loss": 3.3608,
+      "step": 3461
+    },
+    {
+      "epoch": 0.3038582960084039,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002643089076359352,
+      "loss": 3.3486,
+      "step": 3462
+    },
+    {
+      "epoch": 0.30394606559130644,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002642812602207611,
+      "loss": 3.3843,
+      "step": 3463
+    },
+    {
+      "epoch": 0.3040338351742089,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002642536037339603,
+      "loss": 3.3291,
+      "step": 3464
+    },
+    {
+      "epoch": 0.3041216047571114,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0026422593817805994,
+      "loss": 3.3979,
+      "step": 3465
+    },
+    {
+      "epoch": 0.3042093743400139,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002641982635555878,
+      "loss": 3.3003,
+      "step": 3466
+    },
+    {
+      "epoch": 0.30429714392291635,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002641705798690727,
+      "loss": 3.3276,
+      "step": 3467
+    },
+    {
+      "epoch": 0.30438491350581887,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0026414288712104414,
+      "loss": 3.3667,
+      "step": 3468
+    },
+    {
+      "epoch": 0.30447268308872133,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002641151853140325,
+      "loss": 3.3716,
+      "step": 3469
+    },
+    {
+      "epoch": 0.3045604526716238,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002640874744505689,
+      "loss": 3.4419,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3046482222545263,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0026405975453318545,
+      "loss": 3.3145,
+      "step": 3471
+    },
+    {
+      "epoch": 0.3047359918374288,
+      "grad_norm": 0.125,
+      "learning_rate": 0.00264032025564415,
+      "loss": 3.3242,
+      "step": 3472
+    },
+    {
+      "epoch": 0.3048237614203313,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0026400428754679113,
+      "loss": 3.3726,
+      "step": 3473
+    },
+    {
+      "epoch": 0.30491153100323376,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002639765404828485,
+      "loss": 3.374,
+      "step": 3474
+    },
+    {
+      "epoch": 0.3049993005861362,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0026394878437512237,
+      "loss": 3.3433,
+      "step": 3475
+    },
+    {
+      "epoch": 0.30508707016903874,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0026392101922614888,
+      "loss": 3.3169,
+      "step": 3476
+    },
+    {
+      "epoch": 0.3051748397519412,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0026389324503846502,
+      "loss": 3.4189,
+      "step": 3477
+    },
+    {
+      "epoch": 0.3052626093348437,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0026386546181460862,
+      "loss": 3.3809,
+      "step": 3478
+    },
+    {
+      "epoch": 0.3053503789177462,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.002638376695571183,
+      "loss": 3.3115,
+      "step": 3479
+    },
+    {
+      "epoch": 0.30543814850064865,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0026380986826853355,
+      "loss": 3.3052,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3055259180835512,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0026378205795139463,
+      "loss": 3.3647,
+      "step": 3481
+    },
+    {
+      "epoch": 0.30561368766645364,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002637542386082426,
+      "loss": 3.4033,
+      "step": 3482
+    },
+    {
+      "epoch": 0.3057014572493561,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002637264102416195,
+      "loss": 3.3149,
+      "step": 3483
+    },
+    {
+      "epoch": 0.3057892268322586,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0026369857285406806,
+      "loss": 3.4253,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3058769964151611,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002636707264481318,
+      "loss": 3.4419,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3059647659980636,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0026364287102635525,
+      "loss": 3.415,
+      "step": 3486
+    },
+    {
+      "epoch": 0.30605253558096607,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0026361500659128354,
+      "loss": 3.4048,
+      "step": 3487
+    },
+    {
+      "epoch": 0.30614030516386853,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0026358713314546273,
+      "loss": 3.438,
+      "step": 3488
+    },
+    {
+      "epoch": 0.30622807474677105,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002635592506914398,
+      "loss": 3.3188,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3063158443296735,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0026353135923176236,
+      "loss": 3.4448,
+      "step": 3490
+    },
+    {
+      "epoch": 0.30640361391257603,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0026350345876897897,
+      "loss": 3.3433,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3064913834954785,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.00263475549305639,
+      "loss": 3.3516,
+      "step": 3492
+    },
+    {
+      "epoch": 0.30657915307838096,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002634476308442925,
+      "loss": 3.2939,
+      "step": 3493
+    },
+    {
+      "epoch": 0.3066669226612835,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002634197033874906,
+      "loss": 3.355,
+      "step": 3494
+    },
+    {
+      "epoch": 0.30675469224418594,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.002633917669377851,
+      "loss": 3.4434,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3068424618270884,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0026336382149772858,
+      "loss": 3.4458,
+      "step": 3496
+    },
+    {
+      "epoch": 0.30693023140999093,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0026333586706987454,
+      "loss": 3.3848,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3070180009928934,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0026330790365677722,
+      "loss": 3.3848,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3071057705757959,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002632799312609918,
+      "loss": 3.3979,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3071935401586984,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002632519498850741,
+      "loss": 3.293,
+      "step": 3500
+    },
+    {
+      "epoch": 0.30728130974160084,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002632239595315809,
+      "loss": 3.3838,
+      "step": 3501
+    },
+    {
+      "epoch": 0.30736907932450336,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002631959602030698,
+      "loss": 3.439,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3074568489074058,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002631679519020992,
+      "loss": 3.2729,
+      "step": 3503
+    },
+    {
+      "epoch": 0.30754461849030834,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.002631399346312282,
+      "loss": 3.3784,
+      "step": 3504
+    },
+    {
+      "epoch": 0.3076323880732108,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0026311190839301696,
+      "loss": 3.3867,
+      "step": 3505
+    },
+    {
+      "epoch": 0.30772015765611327,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0026308387319002617,
+      "loss": 3.3628,
+      "step": 3506
+    },
+    {
+      "epoch": 0.3078079272390158,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0026305582902481756,
+      "loss": 3.333,
+      "step": 3507
+    },
+    {
+      "epoch": 0.30789569682191825,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0026302777589995367,
+      "loss": 3.3633,
+      "step": 3508
+    },
+    {
+      "epoch": 0.30798346640482077,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.0026299971381799772,
+      "loss": 3.3979,
+      "step": 3509
+    },
+    {
+      "epoch": 0.30807123598772324,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0026297164278151386,
+      "loss": 3.3574,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3081590055706257,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0026294356279306695,
+      "loss": 3.3843,
+      "step": 3511
+    },
+    {
+      "epoch": 0.3082467751535282,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002629154738552229,
+      "loss": 3.29,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3083345447364307,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002628873759705481,
+      "loss": 3.3989,
+      "step": 3513
+    },
+    {
+      "epoch": 0.30842231431933315,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0026285926914161005,
+      "loss": 3.3921,
+      "step": 3514
+    },
+    {
+      "epoch": 0.30851008390223567,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.00262831153370977,
+      "loss": 3.4492,
+      "step": 3515
+    },
+    {
+      "epoch": 0.30859785348513813,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0026280302866121784,
+      "loss": 3.4272,
+      "step": 3516
+    },
+    {
+      "epoch": 0.30868562306804065,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002627748950149025,
+      "loss": 3.4106,
+      "step": 3517
+    },
+    {
+      "epoch": 0.3087733926509431,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002627467524346016,
+      "loss": 3.3633,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3088611622338456,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002627186009228866,
+      "loss": 3.373,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3089489318167481,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002626904404823298,
+      "loss": 3.4585,
+      "step": 3520
+    },
+    {
+      "epoch": 0.30903670139965056,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002626622711155043,
+      "loss": 3.3589,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3091244709825531,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0026263409282498407,
+      "loss": 3.2935,
+      "step": 3522
+    },
+    {
+      "epoch": 0.30921224056545554,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002626059056133438,
+      "loss": 3.4341,
+      "step": 3523
+    },
+    {
+      "epoch": 0.309300010148358,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.00262577709483159,
+      "loss": 3.3521,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3093877797312605,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0026254950443700617,
+      "loss": 3.3447,
+      "step": 3525
+    },
+    {
+      "epoch": 0.309475549314163,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0026252129047746237,
+      "loss": 3.4395,
+      "step": 3526
+    },
+    {
+      "epoch": 0.30956331889706545,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0026249306760710565,
+      "loss": 3.3848,
+      "step": 3527
+    },
+    {
+      "epoch": 0.309651088479968,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002624648358285147,
+      "loss": 3.3818,
+      "step": 3528
+    },
+    {
+      "epoch": 0.30973885806287044,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0026243659514426933,
+      "loss": 3.396,
+      "step": 3529
+    },
+    {
+      "epoch": 0.30982662764577296,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002624083455569498,
+      "loss": 3.4204,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3099143972286754,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.002623800870691375,
+      "loss": 3.4829,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3100021668115779,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002623518196834144,
+      "loss": 3.3604,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3100899363944804,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0026232354340236345,
+      "loss": 3.3931,
+      "step": 3533
+    },
+    {
+      "epoch": 0.31017770597738287,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0026229525822856824,
+      "loss": 3.3862,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3102654755602854,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002622669641646134,
+      "loss": 3.4048,
+      "step": 3535
+    },
+    {
+      "epoch": 0.31035324514318785,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0026223866121308412,
+      "loss": 3.4033,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3104410147260903,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0026221034937656666,
+      "loss": 3.4629,
+      "step": 3537
+    },
+    {
+      "epoch": 0.31052878430899283,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002621820286576478,
+      "loss": 3.3472,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3106165538918953,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0026215369905891535,
+      "loss": 3.4214,
+      "step": 3539
+    },
+    {
+      "epoch": 0.31070432347479776,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0026212536058295792,
+      "loss": 3.3096,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3107920930577003,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002620970132323649,
+      "loss": 3.4302,
+      "step": 3541
+    },
+    {
+      "epoch": 0.31087986264060274,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0026206865700972633,
+      "loss": 3.3818,
+      "step": 3542
+    },
+    {
+      "epoch": 0.31096763222350526,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002620402919176333,
+      "loss": 3.5059,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3110554018064077,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0026201191795867766,
+      "loss": 3.3682,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3111431713893102,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0026198353513545197,
+      "loss": 3.3101,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3112309409722127,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002619551434505496,
+      "loss": 3.3721,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3113187105551152,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002619267429065649,
+      "loss": 3.3472,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3114064801380177,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0026189833350609283,
+      "loss": 3.3384,
+      "step": 3548
+    },
+    {
+      "epoch": 0.31149424972092016,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0026186991525172926,
+      "loss": 3.437,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3115820193038226,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002618414881460708,
+      "loss": 3.417,
+      "step": 3550
+    },
+    {
+      "epoch": 0.31166978888672514,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0026181305219171503,
+      "loss": 3.3027,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3117575584696276,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002617846073912602,
+      "loss": 3.394,
+      "step": 3552
+    },
+    {
+      "epoch": 0.31184532805253007,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0026175615374730535,
+      "loss": 3.3315,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3119330976354326,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0026172769126245037,
+      "loss": 3.3262,
+      "step": 3554
+    },
+    {
+      "epoch": 0.31202086721833505,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.00261699219939296,
+      "loss": 3.4644,
+      "step": 3555
+    },
+    {
+      "epoch": 0.31210863680123757,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002616707397804437,
+      "loss": 3.3174,
+      "step": 3556
+    },
+    {
+      "epoch": 0.31219640638414003,
+      "grad_norm": 0.08544921875,
+      "learning_rate": 0.0026164225078849587,
+      "loss": 3.3267,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3122841759670425,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002616137529660556,
+      "loss": 3.3716,
+      "step": 3558
+    },
+    {
+      "epoch": 0.312371945549945,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002615852463157268,
+      "loss": 3.4521,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3124597151328475,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002615567308401142,
+      "loss": 3.3306,
+      "step": 3560
+    },
+    {
+      "epoch": 0.31254748471575,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0026152820654182334,
+      "loss": 3.3428,
+      "step": 3561
+    },
+    {
+      "epoch": 0.31263525429865247,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0026149967342346063,
+      "loss": 3.3979,
+      "step": 3562
+    },
+    {
+      "epoch": 0.31272302388155493,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002614711314876332,
+      "loss": 3.4287,
+      "step": 3563
+    },
+    {
+      "epoch": 0.31281079346445745,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.00261442580736949,
+      "loss": 3.3521,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3128985630473599,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002614140211740168,
+      "loss": 3.354,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3129863326302624,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0026138545280144617,
+      "loss": 3.416,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3130741022131649,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0026135687562184747,
+      "loss": 3.3706,
+      "step": 3567
+    },
+    {
+      "epoch": 0.31316187179606736,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002613282896378319,
+      "loss": 3.3613,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3132496413789699,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002612996948520115,
+      "loss": 3.3916,
+      "step": 3569
+    },
+    {
+      "epoch": 0.31333741096187234,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.00261271091266999,
+      "loss": 3.3823,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3134251805447748,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00261242478885408,
+      "loss": 3.4326,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3135129501276773,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002612138577098529,
+      "loss": 3.436,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3136007197105798,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002611852277429489,
+      "loss": 3.4004,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3136884892934823,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0026115658898731202,
+      "loss": 3.3799,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3137762588763848,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0026112794144555907,
+      "loss": 3.4316,
+      "step": 3575
+    },
+    {
+      "epoch": 0.31386402845928724,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0026109928512030767,
+      "loss": 3.3364,
+      "step": 3576
+    },
+    {
+      "epoch": 0.31395179804218976,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0026107062001417625,
+      "loss": 3.3789,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3140395676250922,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0026104194612978394,
+      "loss": 3.4111,
+      "step": 3578
+    },
+    {
+      "epoch": 0.31412733720799474,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0026101326346975088,
+      "loss": 3.4688,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3142151067908972,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0026098457203669774,
+      "loss": 3.3486,
+      "step": 3580
+    },
+    {
+      "epoch": 0.31430287637379967,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002609558718332463,
+      "loss": 3.3774,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3143906459567022,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.002609271628620189,
+      "loss": 3.4136,
+      "step": 3582
+    },
+    {
+      "epoch": 0.31447841553960465,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0026089844512563877,
+      "loss": 3.3765,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3145661851225071,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0026086971862672994,
+      "loss": 3.3555,
+      "step": 3584
+    },
+    {
+      "epoch": 0.31465395470540963,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002608409833679173,
+      "loss": 3.3306,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3147417242883121,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.002608122393518263,
+      "loss": 3.3413,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3148294938712146,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0026078348658108356,
+      "loss": 3.3926,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3149172634541171,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0026075472505831623,
+      "loss": 3.4531,
+      "step": 3588
+    },
+    {
+      "epoch": 0.31500503303701954,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0026072595478615233,
+      "loss": 3.354,
+      "step": 3589
+    },
+    {
+      "epoch": 0.31509280261992206,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002606971757672207,
+      "loss": 3.3345,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3151805722028245,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.00260668388004151,
+      "loss": 3.3882,
+      "step": 3591
+    },
+    {
+      "epoch": 0.31526834178572705,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.002606395914995735,
+      "loss": 3.4067,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3153561113686295,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0026061078625611964,
+      "loss": 3.3774,
+      "step": 3593
+    },
+    {
+      "epoch": 0.315443880951532,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002605819722764213,
+      "loss": 3.3804,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3155316505344345,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002605531495631114,
+      "loss": 3.3804,
+      "step": 3595
+    },
+    {
+      "epoch": 0.31561942011733696,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002605243181188234,
+      "loss": 3.3735,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3157071897002394,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0026049547794619187,
+      "loss": 3.334,
+      "step": 3597
+    },
+    {
+      "epoch": 0.31579495928314194,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0026046662904785195,
+      "loss": 3.3091,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3158827288660444,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002604377714264397,
+      "loss": 3.3481,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3159704984489469,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002604089050845918,
+      "loss": 3.3936,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3160582680318494,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.00260380030024946,
+      "loss": 3.3477,
+      "step": 3601
+    },
+    {
+      "epoch": 0.31614603761475185,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0026035114625014067,
+      "loss": 3.3521,
+      "step": 3602
+    },
+    {
+      "epoch": 0.31623380719765437,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00260322253762815,
+      "loss": 3.3599,
+      "step": 3603
+    },
+    {
+      "epoch": 0.31632157678055683,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002602933525656089,
+      "loss": 3.3921,
+      "step": 3604
+    },
+    {
+      "epoch": 0.31640934636345935,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002602644426611633,
+      "loss": 3.3511,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3164971159463618,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0026023552405211965,
+      "loss": 3.3638,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3165848855292643,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0026020659674112043,
+      "loss": 3.3867,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3166726551121668,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0026017766073080876,
+      "loss": 3.312,
+      "step": 3608
+    },
+    {
+      "epoch": 0.31676042469506926,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002601487160238286,
+      "loss": 3.3628,
+      "step": 3609
+    },
+    {
+      "epoch": 0.31684819427797173,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.002601197626228248,
+      "loss": 3.3154,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31693596386087425,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002600908005304428,
+      "loss": 3.3926,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3170237334437767,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.002600618297493291,
+      "loss": 3.3472,
+      "step": 3612
+    },
+    {
+      "epoch": 0.31711150302667923,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0026003285028213074,
+      "loss": 3.3271,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3171992726095817,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0026000386213149563,
+      "loss": 3.3555,
+      "step": 3614
+    },
+    {
+      "epoch": 0.31728704219248416,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0025997486530007262,
+      "loss": 3.3711,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3173748117753867,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.002599458597905112,
+      "loss": 3.4673,
+      "step": 3616
+    },
+    {
+      "epoch": 0.31746258135828914,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.0025991684560546157,
+      "loss": 3.3608,
+      "step": 3617
+    },
+    {
+      "epoch": 0.31755035094119166,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0025988782274757494,
+      "loss": 3.3071,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3176381205240941,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0025985879121950325,
+      "loss": 3.3159,
+      "step": 3619
+    },
+    {
+      "epoch": 0.3177258901069966,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002598297510238992,
+      "loss": 3.2656,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3178136596898991,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.002598007021634162,
+      "loss": 3.395,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3179014292728016,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0025977164464070857,
+      "loss": 3.353,
+      "step": 3622
+    },
+    {
+      "epoch": 0.31798919885570404,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0025974257845843137,
+      "loss": 3.3604,
+      "step": 3623
+    },
+    {
+      "epoch": 0.31807696843860656,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002597135036192405,
+      "loss": 3.3545,
+      "step": 3624
+    },
+    {
+      "epoch": 0.318164738021509,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002596844201257926,
+      "loss": 3.4028,
+      "step": 3625
+    },
+    {
+      "epoch": 0.31825250760441154,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0025965532798074517,
+      "loss": 3.3701,
+      "step": 3626
+    },
+    {
+      "epoch": 0.318340277187314,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025962622718675625,
+      "loss": 3.4229,
+      "step": 3627
+    },
+    {
+      "epoch": 0.31842804677021647,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0025959711774648513,
+      "loss": 3.3892,
+      "step": 3628
+    },
+    {
+      "epoch": 0.318515816353119,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0025956799966259146,
+      "loss": 3.3604,
+      "step": 3629
+    },
+    {
+      "epoch": 0.31860358593602145,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002595388729377359,
+      "loss": 3.3447,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31869135551892397,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002595097375745798,
+      "loss": 3.3794,
+      "step": 3631
+    },
+    {
+      "epoch": 0.31877912510182643,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002594805935757854,
+      "loss": 3.4375,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3188668946847289,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0025945144094401567,
+      "loss": 3.3379,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3189546642676314,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025942227968193433,
+      "loss": 3.4297,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3190424338505339,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0025939310979220605,
+      "loss": 3.3569,
+      "step": 3635
+    },
+    {
+      "epoch": 0.31913020343343634,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0025936393127749596,
+      "loss": 3.3286,
+      "step": 3636
+    },
+    {
+      "epoch": 0.31921797301633886,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0025933474414047038,
+      "loss": 3.3408,
+      "step": 3637
+    },
+    {
+      "epoch": 0.3193057425992413,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002593055483837961,
+      "loss": 3.3491,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31939351218214385,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0025927634401014094,
+      "loss": 3.3447,
+      "step": 3639
+    },
+    {
+      "epoch": 0.3194812817650463,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0025924713102217327,
+      "loss": 3.397,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3195690513479488,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002592179094225625,
+      "loss": 3.312,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3196568209308513,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002591886792139786,
+      "loss": 3.356,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31974459051375376,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0025915944039909245,
+      "loss": 3.3979,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3198323600966563,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002591301929805756,
+      "loss": 3.3823,
+      "step": 3644
+    },
+    {
+      "epoch": 0.31992012967955874,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0025910093696110063,
+      "loss": 3.3643,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3200078992624612,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.002590716723433407,
+      "loss": 3.3784,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3200956688453637,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0025904239912996983,
+      "loss": 3.335,
+      "step": 3647
+    },
+    {
+      "epoch": 0.3201834384282662,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0025901311732366267,
+      "loss": 3.3545,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3202712080111687,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002589838269270949,
+      "loss": 3.4692,
+      "step": 3649
+    },
+    {
+      "epoch": 0.32035897759407117,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0025895452794294283,
+      "loss": 3.3257,
+      "step": 3650
+    },
+    {
+      "epoch": 0.32044674717697363,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.002589252203738836,
+      "loss": 3.3413,
+      "step": 3651
+    },
+    {
+      "epoch": 0.32053451675987615,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002588959042225952,
+      "loss": 3.3643,
+      "step": 3652
+    },
+    {
+      "epoch": 0.3206222863427786,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0025886657949175623,
+      "loss": 3.4126,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3207100559256811,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0025883724618404624,
+      "loss": 3.3315,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3207978255085836,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002588079043021455,
+      "loss": 3.3608,
+      "step": 3655
+    },
+    {
+      "epoch": 0.32088559509148606,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00258778553848735,
+      "loss": 3.3525,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3209733646743886,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0025874919482649665,
+      "loss": 3.3203,
+      "step": 3657
+    },
+    {
+      "epoch": 0.32106113425729105,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002587198272381131,
+      "loss": 3.394,
+      "step": 3658
+    },
+    {
+      "epoch": 0.3211489038401935,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0025869045108626765,
+      "loss": 3.3823,
+      "step": 3659
+    },
+    {
+      "epoch": 0.32123667342309603,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0025866106637364456,
+      "loss": 3.3906,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3213244430059985,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0025863167310292883,
+      "loss": 3.3472,
+      "step": 3661
+    },
+    {
+      "epoch": 0.321412212588901,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0025860227127680615,
+      "loss": 3.3462,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3214999821718035,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0025857286089796304,
+      "loss": 3.3257,
+      "step": 3663
+    },
+    {
+      "epoch": 0.32158775175470594,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0025854344196908686,
+      "loss": 3.3213,
+      "step": 3664
+    },
+    {
+      "epoch": 0.32167552133760846,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0025851401449286573,
+      "loss": 3.3511,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3217632909205109,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0025848457847198842,
+      "loss": 3.4033,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3218510605034134,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0025845513390914464,
+      "loss": 3.3369,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3219388300863159,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0025842568080702492,
+      "loss": 3.4277,
+      "step": 3668
+    },
+    {
+      "epoch": 0.32202659966921837,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0025839621916832036,
+      "loss": 3.4404,
+      "step": 3669
+    },
+    {
+      "epoch": 0.3221143692521209,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0025836674899572306,
+      "loss": 3.2837,
+      "step": 3670
+    },
+    {
+      "epoch": 0.32220213883502336,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0025833727029192562,
+      "loss": 3.3999,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3222899084179258,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002583077830596218,
+      "loss": 3.3804,
+      "step": 3672
+    },
+    {
+      "epoch": 0.32237767800082834,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0025827828730150583,
+      "loss": 3.4863,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3224654475837308,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0025824878302027282,
+      "loss": 3.2817,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3225532171666333,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0025821927021861873,
+      "loss": 3.3794,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3226409867495358,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.002581897488992402,
+      "loss": 3.3481,
+      "step": 3676
+    },
+    {
+      "epoch": 0.32272875633243825,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.002581602190648346,
+      "loss": 3.394,
+      "step": 3677
+    },
+    {
+      "epoch": 0.32281652591534077,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0025813068071810033,
+      "loss": 3.3477,
+      "step": 3678
+    },
+    {
+      "epoch": 0.32290429549824323,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0025810113386173627,
+      "loss": 3.3916,
+      "step": 3679
+    },
+    {
+      "epoch": 0.3229920650811457,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0025807157849844224,
+      "loss": 3.3984,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3230798346640482,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002580420146309188,
+      "loss": 3.354,
+      "step": 3681
+    },
+    {
+      "epoch": 0.3231676042469507,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0025801244226186727,
+      "loss": 3.4233,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3232553738298532,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.002579828613939898,
+      "loss": 3.2812,
+      "step": 3683
+    },
+    {
+      "epoch": 0.32334314341275566,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0025795327202998933,
+      "loss": 3.4385,
+      "step": 3684
+    },
+    {
+      "epoch": 0.3234309129956581,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002579236741725694,
+      "loss": 3.3794,
+      "step": 3685
+    },
+    {
+      "epoch": 0.32351868257856065,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0025789406782443453,
+      "loss": 3.3696,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3236064521614631,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.002578644529882899,
+      "loss": 3.3965,
+      "step": 3687
+    },
+    {
+      "epoch": 0.32369422174436563,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0025783482966684162,
+      "loss": 3.4902,
+      "step": 3688
+    },
+    {
+      "epoch": 0.3237819913272681,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0025780519786279634,
+      "loss": 3.3857,
+      "step": 3689
+    },
+    {
+      "epoch": 0.32386976091017056,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0025777555757886166,
+      "loss": 3.3574,
+      "step": 3690
+    },
+    {
+      "epoch": 0.3239575304930731,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0025774590881774587,
+      "loss": 3.3906,
+      "step": 3691
+    },
+    {
+      "epoch": 0.32404530007597554,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0025771625158215812,
+      "loss": 3.3066,
+      "step": 3692
+    },
+    {
+      "epoch": 0.324133069658878,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002576865858748082,
+      "loss": 3.3306,
+      "step": 3693
+    },
+    {
+      "epoch": 0.3242208392417805,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0025765691169840686,
+      "loss": 3.439,
+      "step": 3694
+    },
+    {
+      "epoch": 0.324308608824683,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002576272290556654,
+      "loss": 3.4189,
+      "step": 3695
+    },
+    {
+      "epoch": 0.3243963784075855,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002575975379492961,
+      "loss": 3.3589,
+      "step": 3696
+    },
+    {
+      "epoch": 0.32448414799048797,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0025756783838201186,
+      "loss": 3.3516,
+      "step": 3697
+    },
+    {
+      "epoch": 0.32457191757339043,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0025753813035652657,
+      "loss": 3.3516,
+      "step": 3698
+    },
+    {
+      "epoch": 0.32465968715629295,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0025750841387555457,
+      "loss": 3.4097,
+      "step": 3699
+    },
+    {
+      "epoch": 0.3247474567391954,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002574786889418112,
+      "loss": 3.3545,
+      "step": 3700
+    },
+    {
+      "epoch": 0.32483522632209794,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025744895555801246,
+      "loss": 3.3599,
+      "step": 3701
+    },
+    {
+      "epoch": 0.3249229959050004,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002574192137268753,
+      "loss": 3.4072,
+      "step": 3702
+    },
+    {
+      "epoch": 0.32501076548790286,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0025738946345111725,
+      "loss": 3.3545,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3250985350708054,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0025735970473345668,
+      "loss": 3.3242,
+      "step": 3704
+    },
+    {
+      "epoch": 0.32518630465370785,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0025732993757661277,
+      "loss": 3.4258,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3252740742366103,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002573001619833054,
+      "loss": 3.3726,
+      "step": 3706
+    },
+    {
+      "epoch": 0.32536184381951283,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002572703779562552,
+      "loss": 3.3682,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3254496134024153,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0025724058549818373,
+      "loss": 3.3584,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3255373829853178,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0025721078461181323,
+      "loss": 3.2876,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3256251525682203,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0025718097529986656,
+      "loss": 3.3735,
+      "step": 3710
+    },
+    {
+      "epoch": 0.32571292215112274,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002571511575650676,
+      "loss": 3.4883,
+      "step": 3711
+    },
+    {
+      "epoch": 0.32580069173402526,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0025712133141014087,
+      "loss": 3.3906,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3258884613169277,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0025709149683781165,
+      "loss": 3.3657,
+      "step": 3713
+    },
+    {
+      "epoch": 0.32597623089983024,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00257061653850806,
+      "loss": 3.3589,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3260640004827327,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0025703180245185086,
+      "loss": 3.3906,
+      "step": 3715
+    },
+    {
+      "epoch": 0.32615177006563517,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.002570019426436737,
+      "loss": 3.3691,
+      "step": 3716
+    },
+    {
+      "epoch": 0.3262395396485377,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0025697207442900295,
+      "loss": 3.3887,
+      "step": 3717
+    },
+    {
+      "epoch": 0.32632730923144015,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0025694219781056777,
+      "loss": 3.3511,
+      "step": 3718
+    },
+    {
+      "epoch": 0.3264150788143427,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0025691231279109817,
+      "loss": 3.3496,
+      "step": 3719
+    },
+    {
+      "epoch": 0.32650284839724514,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0025688241937332473,
+      "loss": 3.3843,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3265906179801476,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0025685251755997885,
+      "loss": 3.3906,
+      "step": 3721
+    },
+    {
+      "epoch": 0.3266783875630501,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0025682260735379286,
+      "loss": 3.3618,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3267661571459526,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0025679268875749973,
+      "loss": 3.3989,
+      "step": 3723
+    },
+    {
+      "epoch": 0.32685392672885505,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0025676276177383313,
+      "loss": 3.3892,
+      "step": 3724
+    },
+    {
+      "epoch": 0.32694169631175757,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0025673282640552766,
+      "loss": 3.4131,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32702946589466003,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002567028826553186,
+      "loss": 3.3823,
+      "step": 3726
+    },
+    {
+      "epoch": 0.32711723547756255,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002566729305259419,
+      "loss": 3.3242,
+      "step": 3727
+    },
+    {
+      "epoch": 0.327205005060465,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002566429700201345,
+      "loss": 3.311,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3272927746433675,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0025661300114063396,
+      "loss": 3.4565,
+      "step": 3729
+    },
+    {
+      "epoch": 0.32738054422627,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002565830238901786,
+      "loss": 3.3618,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32746831380917246,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002565530382715075,
+      "loss": 3.3042,
+      "step": 3731
+    },
+    {
+      "epoch": 0.327556083392075,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002565230442873606,
+      "loss": 3.4116,
+      "step": 3732
+    },
+    {
+      "epoch": 0.32764385297497745,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0025649304194047848,
+      "loss": 3.4487,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3277316225578799,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0025646303123360256,
+      "loss": 3.373,
+      "step": 3734
+    },
+    {
+      "epoch": 0.32781939214078243,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0025643301216947507,
+      "loss": 3.354,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3279071617236849,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0025640298475083888,
+      "loss": 3.375,
+      "step": 3736
+    },
+    {
+      "epoch": 0.32799493130658736,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0025637294898043766,
+      "loss": 3.3281,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3280827008894899,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.002563429048610159,
+      "loss": 3.3418,
+      "step": 3738
+    },
+    {
+      "epoch": 0.32817047047239234,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002563128523953189,
+      "loss": 3.4751,
+      "step": 3739
+    },
+    {
+      "epoch": 0.32825824005529486,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002562827915860925,
+      "loss": 3.3882,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3283460096381973,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0025625272243608357,
+      "loss": 3.3696,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3284337792210998,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0025622264494803955,
+      "loss": 3.3374,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3285215488040023,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0025619255912470874,
+      "loss": 3.2822,
+      "step": 3743
+    },
+    {
+      "epoch": 0.32860931838690477,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0025616246496884014,
+      "loss": 3.3374,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3286970879698073,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0025613236248318355,
+      "loss": 3.4409,
+      "step": 3745
+    },
+    {
+      "epoch": 0.32878485755270975,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002561022516704896,
+      "loss": 3.2661,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3288726271356122,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002560721325335095,
+      "loss": 3.4326,
+      "step": 3747
+    },
+    {
+      "epoch": 0.32896039671851474,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0025604200507499536,
+      "loss": 3.334,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3290481663014172,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0025601186929770006,
+      "loss": 3.3564,
+      "step": 3749
+    },
+    {
+      "epoch": 0.32913593588431966,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002559817252043771,
+      "loss": 3.3853,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3292237054672222,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0025595157279778096,
+      "loss": 3.4102,
+      "step": 3751
+    },
+    {
+      "epoch": 0.32931147505012465,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0025592141208066665,
+      "loss": 3.3369,
+      "step": 3752
+    },
+    {
+      "epoch": 0.32939924463302717,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002558912430557901,
+      "loss": 3.3535,
+      "step": 3753
+    },
+    {
+      "epoch": 0.32948701421592963,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0025586106572590794,
+      "loss": 3.2495,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3295747837988321,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0025583088009377758,
+      "loss": 3.3418,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3296625533817346,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.002558006861621571,
+      "loss": 3.3433,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3297503229646371,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0025577048393380553,
+      "loss": 3.3633,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3298380925475396,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002557402734114824,
+      "loss": 3.3794,
+      "step": 3758
+    },
+    {
+      "epoch": 0.32992586213044206,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0025571005459794818,
+      "loss": 3.3755,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3300136317133445,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0025567982749596413,
+      "loss": 3.3281,
+      "step": 3760
+    },
+    {
+      "epoch": 0.33010140129624704,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0025564959210829215,
+      "loss": 3.3662,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3301891708791495,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0025561934843769487,
+      "loss": 3.3428,
+      "step": 3762
+    },
+    {
+      "epoch": 0.33027694046205197,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0025558909648693584,
+      "loss": 3.3633,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3303647100449545,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.002555588362587792,
+      "loss": 3.4219,
+      "step": 3764
+    },
+    {
+      "epoch": 0.33045247962785695,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0025552856775598997,
+      "loss": 3.269,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3305402492107595,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0025549829098133387,
+      "loss": 3.3496,
+      "step": 3766
+    },
+    {
+      "epoch": 0.33062801879366194,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.002554680059375773,
+      "loss": 3.3862,
+      "step": 3767
+    },
+    {
+      "epoch": 0.3307157883765644,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0025543771262748764,
+      "loss": 3.3804,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3308035579594669,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002554074110538328,
+      "loss": 3.3853,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3308913275423694,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002553771012193815,
+      "loss": 3.3691,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3309790971252719,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0025534678312690326,
+      "loss": 3.3696,
+      "step": 3771
+    },
+    {
+      "epoch": 0.33106686670817437,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002553164567791683,
+      "loss": 3.3276,
+      "step": 3772
+    },
+    {
+      "epoch": 0.33115463629107683,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002552861221789477,
+      "loss": 3.3916,
+      "step": 3773
+    },
+    {
+      "epoch": 0.33124240587397935,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002552557793290132,
+      "loss": 3.4048,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3313301754568818,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.002552254282321373,
+      "loss": 3.4116,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3314179450397843,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002551950688910933,
+      "loss": 3.3838,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3315057146226868,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002551647013086552,
+      "loss": 3.4014,
+      "step": 3777
+    },
+    {
+      "epoch": 0.33159348420558926,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.002551343254875978,
+      "loss": 3.3662,
+      "step": 3778
+    },
+    {
+      "epoch": 0.3316812537884918,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002551039414306966,
+      "loss": 3.4429,
+      "step": 3779
+    },
+    {
+      "epoch": 0.33176902337139424,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0025507354914072787,
+      "loss": 3.3916,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3318567929542967,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0025504314862046873,
+      "loss": 3.3364,
+      "step": 3781
+    },
+    {
+      "epoch": 0.33194456253719923,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0025501273987269682,
+      "loss": 3.314,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3320323321201017,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0025498232290019087,
+      "loss": 3.3643,
+      "step": 3783
+    },
+    {
+      "epoch": 0.3321201017030042,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0025495189770572994,
+      "loss": 3.3584,
+      "step": 3784
+    },
+    {
+      "epoch": 0.3322078712859067,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0025492146429209428,
+      "loss": 3.4326,
+      "step": 3785
+    },
+    {
+      "epoch": 0.33229564086880914,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0025489102266206456,
+      "loss": 3.3906,
+      "step": 3786
+    },
+    {
+      "epoch": 0.33238341045171166,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0025486057281842237,
+      "loss": 3.3286,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3324711800346141,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0025483011476395002,
+      "loss": 3.4004,
+      "step": 3788
+    },
+    {
+      "epoch": 0.33255894961751664,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0025479964850143055,
+      "loss": 3.3921,
+      "step": 3789
+    },
+    {
+      "epoch": 0.3326467192004191,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002547691740336477,
+      "loss": 3.3374,
+      "step": 3790
+    },
+    {
+      "epoch": 0.33273448878332157,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0025473869136338604,
+      "loss": 3.2788,
+      "step": 3791
+    },
+    {
+      "epoch": 0.3328222583662241,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0025470820049343086,
+      "loss": 3.3984,
+      "step": 3792
+    },
+    {
+      "epoch": 0.33291002794912655,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002546777014265682,
+      "loss": 3.3521,
+      "step": 3793
+    },
+    {
+      "epoch": 0.332997797532029,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002546471941655849,
+      "loss": 3.3765,
+      "step": 3794
+    },
+    {
+      "epoch": 0.33308556711493154,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002546166787132685,
+      "loss": 3.3252,
+      "step": 3795
+    },
+    {
+      "epoch": 0.333173336697834,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0025458615507240718,
+      "loss": 3.4092,
+      "step": 3796
+    },
+    {
+      "epoch": 0.3332611062807365,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0025455562324579006,
+      "loss": 3.2979,
+      "step": 3797
+    },
+    {
+      "epoch": 0.333348875863639,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0025452508323620692,
+      "loss": 3.4253,
+      "step": 3798
+    },
+    {
+      "epoch": 0.33343664544654145,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0025449453504644832,
+      "loss": 3.3164,
+      "step": 3799
+    },
+    {
+      "epoch": 0.33352441502944397,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002544639786793055,
+      "loss": 3.3867,
+      "step": 3800
+    },
+    {
+      "epoch": 0.33361218461234643,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0025443341413757046,
+      "loss": 3.3462,
+      "step": 3801
+    },
+    {
+      "epoch": 0.33369995419524895,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.00254402841424036,
+      "loss": 3.3354,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3337877237781514,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0025437226054149564,
+      "loss": 3.4102,
+      "step": 3803
+    },
+    {
+      "epoch": 0.3338754933610539,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002543416714927437,
+      "loss": 3.2915,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3339632629439564,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002543110742805751,
+      "loss": 3.3784,
+      "step": 3805
+    },
+    {
+      "epoch": 0.33405103252685886,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002542804689077857,
+      "loss": 3.3169,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3341388021097613,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002542498553771719,
+      "loss": 3.3262,
+      "step": 3807
+    },
+    {
+      "epoch": 0.33422657169266384,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00254219233691531,
+      "loss": 3.3633,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3343143412755663,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0025418860385366102,
+      "loss": 3.4077,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3344021108584688,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0025415796586636067,
+      "loss": 3.3247,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3344898804413713,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002541273197324294,
+      "loss": 3.3159,
+      "step": 3811
+    },
+    {
+      "epoch": 0.33457765002427375,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002540966654546676,
+      "loss": 3.3667,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3346654196071763,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0025406600303587604,
+      "loss": 3.3403,
+      "step": 3813
+    },
+    {
+      "epoch": 0.33475318919007874,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002540353324788565,
+      "loss": 3.3726,
+      "step": 3814
+    },
+    {
+      "epoch": 0.33484095877298126,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002540046537864115,
+      "loss": 3.3525,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3349287283558837,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002539739669613442,
+      "loss": 3.3379,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3350164979387862,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0025394327200645855,
+      "loss": 3.3828,
+      "step": 3817
+    },
+    {
+      "epoch": 0.3351042675216887,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002539125689245592,
+      "loss": 3.3765,
+      "step": 3818
+    },
+    {
+      "epoch": 0.33519203710459117,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0025388185771845167,
+      "loss": 3.3682,
+      "step": 3819
+    },
+    {
+      "epoch": 0.33527980668749363,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002538511383909421,
+      "loss": 3.3809,
+      "step": 3820
+    },
+    {
+      "epoch": 0.33536757627039615,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002538204109448374,
+      "loss": 3.4277,
+      "step": 3821
+    },
+    {
+      "epoch": 0.3354553458532986,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0025378967538294515,
+      "loss": 3.2861,
+      "step": 3822
+    },
+    {
+      "epoch": 0.33554311543620113,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0025375893170807395,
+      "loss": 3.3354,
+      "step": 3823
+    },
+    {
+      "epoch": 0.3356308850191036,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0025372817992303275,
+      "loss": 3.3213,
+      "step": 3824
+    },
+    {
+      "epoch": 0.33571865460200606,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0025369742003063145,
+      "loss": 3.4551,
+      "step": 3825
+    },
+    {
+      "epoch": 0.3358064241849086,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0025366665203368083,
+      "loss": 3.3052,
+      "step": 3826
+    },
+    {
+      "epoch": 0.33589419376781104,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0025363587593499213,
+      "loss": 3.3154,
+      "step": 3827
+    },
+    {
+      "epoch": 0.33598196335071356,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002536050917373774,
+      "loss": 3.3667,
+      "step": 3828
+    },
+    {
+      "epoch": 0.336069732933616,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0025357429944364964,
+      "loss": 3.3115,
+      "step": 3829
+    },
+    {
+      "epoch": 0.3361575025165185,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0025354349905662233,
+      "loss": 3.3135,
+      "step": 3830
+    },
+    {
+      "epoch": 0.336245272099421,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0025351269057910983,
+      "loss": 3.3193,
+      "step": 3831
+    },
+    {
+      "epoch": 0.3363330416823235,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002534818740139272,
+      "loss": 3.3862,
+      "step": 3832
+    },
+    {
+      "epoch": 0.33642081126522594,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0025345104936389017,
+      "loss": 3.3784,
+      "step": 3833
+    },
+    {
+      "epoch": 0.33650858084812846,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0025342021663181534,
+      "loss": 3.3511,
+      "step": 3834
+    },
+    {
+      "epoch": 0.3365963504310309,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0025338937582052006,
+      "loss": 3.4102,
+      "step": 3835
+    },
+    {
+      "epoch": 0.33668412001393344,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0025335852693282226,
+      "loss": 3.2949,
+      "step": 3836
+    },
+    {
+      "epoch": 0.3367718895968359,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002533276699715407,
+      "loss": 3.3887,
+      "step": 3837
+    },
+    {
+      "epoch": 0.33685965917973837,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0025329680493949487,
+      "loss": 3.4521,
+      "step": 3838
+    },
+    {
+      "epoch": 0.3369474287626409,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0025326593183950504,
+      "loss": 3.4214,
+      "step": 3839
+    },
+    {
+      "epoch": 0.33703519834554335,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0025323505067439216,
+      "loss": 3.4482,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33712296792844587,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0025320416144697794,
+      "loss": 3.3789,
+      "step": 3841
+    },
+    {
+      "epoch": 0.33721073751134834,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002531732641600848,
+      "loss": 3.3667,
+      "step": 3842
+    },
+    {
+      "epoch": 0.3372985070942508,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002531423588165359,
+      "loss": 3.3105,
+      "step": 3843
+    },
+    {
+      "epoch": 0.3373862766771533,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0025311144541915525,
+      "loss": 3.3218,
+      "step": 3844
+    },
+    {
+      "epoch": 0.3374740462600558,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.002530805239707674,
+      "loss": 3.3208,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33756181584295825,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0025304959447419776,
+      "loss": 3.3955,
+      "step": 3846
+    },
+    {
+      "epoch": 0.33764958542586077,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002530186569322725,
+      "loss": 3.3696,
+      "step": 3847
+    },
+    {
+      "epoch": 0.33773735500876323,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002529877113478184,
+      "loss": 3.3418,
+      "step": 3848
+    },
+    {
+      "epoch": 0.33782512459166575,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0025295675772366313,
+      "loss": 3.3677,
+      "step": 3849
+    },
+    {
+      "epoch": 0.3379128941745682,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0025292579606263493,
+      "loss": 3.3975,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3380006637574707,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002528948263675629,
+      "loss": 3.3208,
+      "step": 3851
+    },
+    {
+      "epoch": 0.3380884333403732,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0025286384864127683,
+      "loss": 3.3267,
+      "step": 3852
+    },
+    {
+      "epoch": 0.33817620292327566,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0025283286288660728,
+      "loss": 3.3564,
+      "step": 3853
+    },
+    {
+      "epoch": 0.3382639725061782,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002528018691063855,
+      "loss": 3.3496,
+      "step": 3854
+    },
+    {
+      "epoch": 0.33835174208908064,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0025277086730344346,
+      "loss": 3.3765,
+      "step": 3855
+    },
+    {
+      "epoch": 0.3384395116719831,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002527398574806139,
+      "loss": 3.3359,
+      "step": 3856
+    },
+    {
+      "epoch": 0.3385272812548856,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0025270883964073026,
+      "loss": 3.3662,
+      "step": 3857
+    },
+    {
+      "epoch": 0.3386150508377881,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0025267781378662674,
+      "loss": 3.3096,
+      "step": 3858
+    },
+    {
+      "epoch": 0.3387028204206906,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002526467799211383,
+      "loss": 3.3457,
+      "step": 3859
+    },
+    {
+      "epoch": 0.3387905900035931,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002526157380471006,
+      "loss": 3.2812,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33887835958649554,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0025258468816735,
+      "loss": 3.4634,
+      "step": 3861
+    },
+    {
+      "epoch": 0.33896612916939806,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0025255363028472357,
+      "loss": 3.3691,
+      "step": 3862
+    },
+    {
+      "epoch": 0.3390538987523005,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002525225644020593,
+      "loss": 3.4355,
+      "step": 3863
+    },
+    {
+      "epoch": 0.339141668335203,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0025249149052219565,
+      "loss": 3.3462,
+      "step": 3864
+    },
+    {
+      "epoch": 0.3392294379181055,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0025246040864797194,
+      "loss": 3.3931,
+      "step": 3865
+    },
+    {
+      "epoch": 0.33931720750100797,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002524293187822283,
+      "loss": 3.3945,
+      "step": 3866
+    },
+    {
+      "epoch": 0.3394049770839105,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0025239822092780542,
+      "loss": 3.3784,
+      "step": 3867
+    },
+    {
+      "epoch": 0.33949274666681295,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0025236711508754483,
+      "loss": 3.3872,
+      "step": 3868
+    },
+    {
+      "epoch": 0.3395805162497154,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002523360012642888,
+      "loss": 3.3037,
+      "step": 3869
+    },
+    {
+      "epoch": 0.33966828583261793,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0025230487946088028,
+      "loss": 3.2866,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3397560554155204,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0025227374968016286,
+      "loss": 3.3037,
+      "step": 3871
+    },
+    {
+      "epoch": 0.3398438249984229,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0025224261192498113,
+      "loss": 3.4087,
+      "step": 3872
+    },
+    {
+      "epoch": 0.3399315945813254,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002522114661981801,
+      "loss": 3.3403,
+      "step": 3873
+    },
+    {
+      "epoch": 0.34001936416422784,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0025218031250260576,
+      "loss": 3.3164,
+      "step": 3874
+    },
+    {
+      "epoch": 0.34010713374713036,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0025214915084110466,
+      "loss": 3.4102,
+      "step": 3875
+    },
+    {
+      "epoch": 0.3401949033300328,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0025211798121652414,
+      "loss": 3.3501,
+      "step": 3876
+    },
+    {
+      "epoch": 0.3402826729129353,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002520868036317122,
+      "loss": 3.3774,
+      "step": 3877
+    },
+    {
+      "epoch": 0.3403704424958378,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0025205561808951775,
+      "loss": 3.3579,
+      "step": 3878
+    },
+    {
+      "epoch": 0.3404582120787403,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0025202442459279022,
+      "loss": 3.3135,
+      "step": 3879
+    },
+    {
+      "epoch": 0.3405459816616428,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0025199322314437985,
+      "loss": 3.4321,
+      "step": 3880
+    },
+    {
+      "epoch": 0.34063375124454526,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002519620137471377,
+      "loss": 3.4282,
+      "step": 3881
+    },
+    {
+      "epoch": 0.3407215208274477,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0025193079640391538,
+      "loss": 3.4565,
+      "step": 3882
+    },
+    {
+      "epoch": 0.34080929041035024,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025189957111756535,
+      "loss": 3.3721,
+      "step": 3883
+    },
+    {
+      "epoch": 0.3408970599932527,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0025186833789094076,
+      "loss": 3.3374,
+      "step": 3884
+    },
+    {
+      "epoch": 0.3409848295761552,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0025183709672689544,
+      "loss": 3.3452,
+      "step": 3885
+    },
+    {
+      "epoch": 0.3410725991590577,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00251805847628284,
+      "loss": 3.3589,
+      "step": 3886
+    },
+    {
+      "epoch": 0.34116036874196015,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.002517745905979618,
+      "loss": 3.332,
+      "step": 3887
+    },
+    {
+      "epoch": 0.34124813832486267,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0025174332563878492,
+      "loss": 3.3872,
+      "step": 3888
+    },
+    {
+      "epoch": 0.34133590790776513,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0025171205275360997,
+      "loss": 3.4302,
+      "step": 3889
+    },
+    {
+      "epoch": 0.3414236774906676,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0025168077194529467,
+      "loss": 3.3364,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3415114470735701,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002516494832166971,
+      "loss": 3.3442,
+      "step": 3891
+    },
+    {
+      "epoch": 0.3415992166564726,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002516181865706762,
+      "loss": 3.4443,
+      "step": 3892
+    },
+    {
+      "epoch": 0.3416869862393751,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002515868820100917,
+      "loss": 3.4316,
+      "step": 3893
+    },
+    {
+      "epoch": 0.34177475582227757,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00251555569537804,
+      "loss": 3.3828,
+      "step": 3894
+    },
+    {
+      "epoch": 0.34186252540518003,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0025152424915667413,
+      "loss": 3.3511,
+      "step": 3895
+    },
+    {
+      "epoch": 0.34195029498808255,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00251492920869564,
+      "loss": 3.418,
+      "step": 3896
+    },
+    {
+      "epoch": 0.342038064570985,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.002514615846793362,
+      "loss": 3.3989,
+      "step": 3897
+    },
+    {
+      "epoch": 0.34212583415388753,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0025143024058885393,
+      "loss": 3.3745,
+      "step": 3898
+    },
+    {
+      "epoch": 0.34221360373679,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002513988886009812,
+      "loss": 3.4482,
+      "step": 3899
+    },
+    {
+      "epoch": 0.34230137331969246,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0025136752871858276,
+      "loss": 3.3589,
+      "step": 3900
+    },
+    {
+      "epoch": 0.342389142902595,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0025133616094452408,
+      "loss": 3.4282,
+      "step": 3901
+    },
+    {
+      "epoch": 0.34247691248549744,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002513047852816713,
+      "loss": 3.3472,
+      "step": 3902
+    },
+    {
+      "epoch": 0.3425646820683999,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002512734017328913,
+      "loss": 3.396,
+      "step": 3903
+    },
+    {
+      "epoch": 0.3426524516513024,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0025124201030105172,
+      "loss": 3.3677,
+      "step": 3904
+    },
+    {
+      "epoch": 0.3427402212342049,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0025121061098902085,
+      "loss": 3.3442,
+      "step": 3905
+    },
+    {
+      "epoch": 0.3428279908171074,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002511792037996678,
+      "loss": 3.292,
+      "step": 3906
+    },
+    {
+      "epoch": 0.3429157604000099,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0025114778873586226,
+      "loss": 3.3462,
+      "step": 3907
+    },
+    {
+      "epoch": 0.34300352998291234,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0025111636580047473,
+      "loss": 3.3618,
+      "step": 3908
+    },
+    {
+      "epoch": 0.34309129956581486,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002510849349963765,
+      "loss": 3.2822,
+      "step": 3909
+    },
+    {
+      "epoch": 0.3431790691487173,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002510534963264394,
+      "loss": 3.3784,
+      "step": 3910
+    },
+    {
+      "epoch": 0.34326683873161984,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.002510220497935362,
+      "loss": 3.3477,
+      "step": 3911
+    },
+    {
+      "epoch": 0.3433546083145223,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002509905954005401,
+      "loss": 3.3491,
+      "step": 3912
+    },
+    {
+      "epoch": 0.34344237789742477,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002509591331503253,
+      "loss": 3.3784,
+      "step": 3913
+    },
+    {
+      "epoch": 0.3435301474803273,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002509276630457665,
+      "loss": 3.3848,
+      "step": 3914
+    },
+    {
+      "epoch": 0.34361791706322975,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0025089618508973934,
+      "loss": 3.354,
+      "step": 3915
+    },
+    {
+      "epoch": 0.34370568664613227,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.002508646992851201,
+      "loss": 3.4814,
+      "step": 3916
+    },
+    {
+      "epoch": 0.34379345622903473,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0025083320563478546,
+      "loss": 3.3052,
+      "step": 3917
+    },
+    {
+      "epoch": 0.3438812258119372,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0025080170414161335,
+      "loss": 3.4229,
+      "step": 3918
+    },
+    {
+      "epoch": 0.3439689953948397,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0025077019480848205,
+      "loss": 3.3564,
+      "step": 3919
+    },
+    {
+      "epoch": 0.3440567649777422,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0025073867763827067,
+      "loss": 3.3926,
+      "step": 3920
+    },
+    {
+      "epoch": 0.34414453456064464,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.00250707152633859,
+      "loss": 3.3477,
+      "step": 3921
+    },
+    {
+      "epoch": 0.34423230414354716,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0025067561979812765,
+      "loss": 3.3247,
+      "step": 3922
+    },
+    {
+      "epoch": 0.3443200737264496,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0025064407913395785,
+      "loss": 3.3643,
+      "step": 3923
+    },
+    {
+      "epoch": 0.34440784330935215,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002506125306442315,
+      "loss": 3.3457,
+      "step": 3924
+    },
+    {
+      "epoch": 0.3444956128922546,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0025058097433183135,
+      "loss": 3.4019,
+      "step": 3925
+    },
+    {
+      "epoch": 0.3445833824751571,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002505494101996408,
+      "loss": 3.2998,
+      "step": 3926
+    },
+    {
+      "epoch": 0.3446711520580596,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002505178382505439,
+      "loss": 3.4121,
+      "step": 3927
+    },
+    {
+      "epoch": 0.34475892164096206,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002504862584874255,
+      "loss": 3.3452,
+      "step": 3928
+    },
+    {
+      "epoch": 0.3448466912238646,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0025045467091317115,
+      "loss": 3.3438,
+      "step": 3929
+    },
+    {
+      "epoch": 0.34493446080676704,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0025042307553066706,
+      "loss": 3.3218,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3450222303896695,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0025039147234280027,
+      "loss": 3.3115,
+      "step": 3931
+    },
+    {
+      "epoch": 0.345109999972572,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002503598613524584,
+      "loss": 3.3389,
+      "step": 3932
+    },
+    {
+      "epoch": 0.3451977695554745,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0025032824256252992,
+      "loss": 3.3232,
+      "step": 3933
+    },
+    {
+      "epoch": 0.34528553913837695,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0025029661597590386,
+      "loss": 3.2437,
+      "step": 3934
+    },
+    {
+      "epoch": 0.34537330872127947,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0025026498159546997,
+      "loss": 3.3184,
+      "step": 3935
+    },
+    {
+      "epoch": 0.34546107830418193,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0025023333942411886,
+      "loss": 3.3721,
+      "step": 3936
+    },
+    {
+      "epoch": 0.34554884788708445,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.002502016894647418,
+      "loss": 3.416,
+      "step": 3937
+    },
+    {
+      "epoch": 0.3456366174699869,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.002501700317202307,
+      "loss": 3.3833,
+      "step": 3938
+    },
+    {
+      "epoch": 0.3457243870528894,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0025013836619347824,
+      "loss": 3.4048,
+      "step": 3939
+    },
+    {
+      "epoch": 0.3458121566357919,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002501066928873778,
+      "loss": 3.3525,
+      "step": 3940
+    },
+    {
+      "epoch": 0.34589992621869436,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002500750118048234,
+      "loss": 3.374,
+      "step": 3941
+    },
+    {
+      "epoch": 0.3459876958015969,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0025004332294870993,
+      "loss": 3.25,
+      "step": 3942
+    },
+    {
+      "epoch": 0.34607546538449935,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002500116263219328,
+      "loss": 3.3677,
+      "step": 3943
+    },
+    {
+      "epoch": 0.3461632349674018,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0024997992192738834,
+      "loss": 3.3755,
+      "step": 3944
+    },
+    {
+      "epoch": 0.34625100455030433,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.002499482097679734,
+      "loss": 3.3198,
+      "step": 3945
+    },
+    {
+      "epoch": 0.3463387741332068,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002499164898465856,
+      "loss": 3.4302,
+      "step": 3946
+    },
+    {
+      "epoch": 0.34642654371610926,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0024988476216612336,
+      "loss": 3.3403,
+      "step": 3947
+    },
+    {
+      "epoch": 0.3465143132990118,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002498530267294856,
+      "loss": 3.3765,
+      "step": 3948
+    },
+    {
+      "epoch": 0.34660208288191424,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0024982128353957227,
+      "loss": 3.3804,
+      "step": 3949
+    },
+    {
+      "epoch": 0.34668985246481676,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0024978953259928364,
+      "loss": 3.3833,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3467776220477192,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0024975777391152104,
+      "loss": 3.3276,
+      "step": 3951
+    },
+    {
+      "epoch": 0.3468653916306217,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002497260074791863,
+      "loss": 3.3311,
+      "step": 3952
+    },
+    {
+      "epoch": 0.3469531612135242,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0024969423330518203,
+      "loss": 3.3716,
+      "step": 3953
+    },
+    {
+      "epoch": 0.34704093079642667,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002496624513924115,
+      "loss": 3.333,
+      "step": 3954
+    },
+    {
+      "epoch": 0.3471287003793292,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0024963066174377873,
+      "loss": 3.4004,
+      "step": 3955
+    },
+    {
+      "epoch": 0.34721646996223166,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0024959886436218843,
+      "loss": 3.4136,
+      "step": 3956
+    },
+    {
+      "epoch": 0.3473042395451341,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0024956705925054607,
+      "loss": 3.4331,
+      "step": 3957
+    },
+    {
+      "epoch": 0.34739200912803664,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0024953524641175774,
+      "loss": 3.3091,
+      "step": 3958
+    },
+    {
+      "epoch": 0.3474797787109391,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002495034258487303,
+      "loss": 3.3779,
+      "step": 3959
+    },
+    {
+      "epoch": 0.34756754829384157,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0024947159756437123,
+      "loss": 3.4639,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3476553178767441,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0024943976156158883,
+      "loss": 3.3052,
+      "step": 3961
+    },
+    {
+      "epoch": 0.34774308745964655,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0024940791784329205,
+      "loss": 3.3345,
+      "step": 3962
+    },
+    {
+      "epoch": 0.34783085704254907,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.002493760664123905,
+      "loss": 3.416,
+      "step": 3963
+    },
+    {
+      "epoch": 0.34791862662545153,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0024934420727179456,
+      "loss": 3.3501,
+      "step": 3964
+    },
+    {
+      "epoch": 0.348006396208354,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002493123404244154,
+      "loss": 3.3145,
+      "step": 3965
+    },
+    {
+      "epoch": 0.3480941657912565,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002492804658731646,
+      "loss": 3.3784,
+      "step": 3966
+    },
+    {
+      "epoch": 0.348181935374159,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.002492485836209548,
+      "loss": 3.3789,
+      "step": 3967
+    },
+    {
+      "epoch": 0.3482697049570615,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0024921669367069907,
+      "loss": 3.4053,
+      "step": 3968
+    },
+    {
+      "epoch": 0.34835747453996396,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0024918479602531133,
+      "loss": 3.2651,
+      "step": 3969
+    },
+    {
+      "epoch": 0.3484452441228664,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002491528906877062,
+      "loss": 3.3428,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34853301370576895,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.002491209776607989,
+      "loss": 3.4028,
+      "step": 3971
+    },
+    {
+      "epoch": 0.3486207832886714,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0024908905694750543,
+      "loss": 3.3823,
+      "step": 3972
+    },
+    {
+      "epoch": 0.3487085528715739,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002490571285507425,
+      "loss": 3.373,
+      "step": 3973
+    },
+    {
+      "epoch": 0.3487963224544764,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002490251924734275,
+      "loss": 3.4189,
+      "step": 3974
+    },
+    {
+      "epoch": 0.34888409203737886,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0024899324871847854,
+      "loss": 3.3159,
+      "step": 3975
+    },
+    {
+      "epoch": 0.3489718616202814,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002489612972888144,
+      "loss": 3.3066,
+      "step": 3976
+    },
+    {
+      "epoch": 0.34905963120318384,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0024892933818735463,
+      "loss": 3.3613,
+      "step": 3977
+    },
+    {
+      "epoch": 0.3491474007860863,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0024889737141701933,
+      "loss": 3.3818,
+      "step": 3978
+    },
+    {
+      "epoch": 0.3492351703689888,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0024886539698072948,
+      "loss": 3.4204,
+      "step": 3979
+    },
+    {
+      "epoch": 0.3493229399518913,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002488334148814066,
+      "loss": 3.3853,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3494107095347938,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002488014251219731,
+      "loss": 3.3701,
+      "step": 3981
+    },
+    {
+      "epoch": 0.34949847911769627,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002487694277053519,
+      "loss": 3.3462,
+      "step": 3982
+    },
+    {
+      "epoch": 0.34958624870059873,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.002487374226344667,
+      "loss": 3.375,
+      "step": 3983
+    },
+    {
+      "epoch": 0.34967401828350125,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.00248705409912242,
+      "loss": 3.3735,
+      "step": 3984
+    },
+    {
+      "epoch": 0.3497617878664037,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0024867338954160274,
+      "loss": 3.3857,
+      "step": 3985
+    },
+    {
+      "epoch": 0.34984955744930624,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002486413615254748,
+      "loss": 3.3345,
+      "step": 3986
+    },
+    {
+      "epoch": 0.3499373270322087,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.002486093258667847,
+      "loss": 3.3267,
+      "step": 3987
+    },
+    {
+      "epoch": 0.35002509661511116,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0024857728256845965,
+      "loss": 3.375,
+      "step": 3988
+    },
+    {
+      "epoch": 0.3501128661980137,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0024854523163342743,
+      "loss": 3.3833,
+      "step": 3989
+    },
+    {
+      "epoch": 0.35020063578091615,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002485131730646167,
+      "loss": 3.2983,
+      "step": 3990
+    },
+    {
+      "epoch": 0.3502884053638186,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002484811068649568,
+      "loss": 3.3145,
+      "step": 3991
+    },
+    {
+      "epoch": 0.35037617494672113,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0024844903303737763,
+      "loss": 3.3608,
+      "step": 3992
+    },
+    {
+      "epoch": 0.3504639445296236,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002484169515848099,
+      "loss": 3.3379,
+      "step": 3993
+    },
+    {
+      "epoch": 0.3505517141125261,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0024838486251018504,
+      "loss": 3.3555,
+      "step": 3994
+    },
+    {
+      "epoch": 0.3506394836954286,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.00248352765816435,
+      "loss": 3.3071,
+      "step": 3995
+    },
+    {
+      "epoch": 0.35072725327833104,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002483206615064927,
+      "loss": 3.3208,
+      "step": 3996
+    },
+    {
+      "epoch": 0.35081502286123356,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002482885495832915,
+      "loss": 3.3369,
+      "step": 3997
+    },
+    {
+      "epoch": 0.350902792444136,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0024825643004976563,
+      "loss": 3.4297,
+      "step": 3998
+    },
+    {
+      "epoch": 0.35099056202703854,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0024822430290884984,
+      "loss": 3.3364,
+      "step": 3999
+    },
+    {
+      "epoch": 0.351078331609941,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0024819216816347983,
+      "loss": 3.4019,
+      "step": 4000
+    },
+    {
+      "epoch": 0.351078331609941,
+      "eval_loss": 0.10440438985824585,
+      "eval_runtime": 106.6382,
+      "eval_samples_per_second": 138.309,
+      "eval_steps_per_second": 17.292,
+      "step": 4000
+    },
+    {
+      "epoch": 0.35116610119284347,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002481600258165918,
+      "loss": 3.3491,
+      "step": 4001
+    },
+    {
+      "epoch": 0.351253870775746,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002481278758711226,
+      "loss": 3.3682,
+      "step": 4002
+    },
+    {
+      "epoch": 0.35134164035864845,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.0024809571833001,
+      "loss": 3.3867,
+      "step": 4003
+    },
+    {
+      "epoch": 0.3514294099415509,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002480635531961922,
+      "loss": 3.2983,
+      "step": 4004
+    },
+    {
+      "epoch": 0.35151717952445344,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0024803138047260836,
+      "loss": 3.354,
+      "step": 4005
+    },
+    {
+      "epoch": 0.3516049491073559,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0024799920016219806,
+      "loss": 3.3789,
+      "step": 4006
+    },
+    {
+      "epoch": 0.3516927186902584,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0024796701226790186,
+      "loss": 3.3413,
+      "step": 4007
+    },
+    {
+      "epoch": 0.3517804882731609,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0024793481679266077,
+      "loss": 3.4062,
+      "step": 4008
+    },
+    {
+      "epoch": 0.35186825785606335,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002479026137394166,
+      "loss": 3.3696,
+      "step": 4009
+    },
+    {
+      "epoch": 0.35195602743896587,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002478704031111118,
+      "loss": 3.3599,
+      "step": 4010
+    },
+    {
+      "epoch": 0.35204379702186833,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.0024783818491068963,
+      "loss": 3.2886,
+      "step": 4011
+    },
+    {
+      "epoch": 0.35213156660477085,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0024780595914109397,
+      "loss": 3.356,
+      "step": 4012
+    },
+    {
+      "epoch": 0.3522193361876733,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002477737258052693,
+      "loss": 3.3608,
+      "step": 4013
+    },
+    {
+      "epoch": 0.3523071057705758,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0024774148490616095,
+      "loss": 3.2969,
+      "step": 4014
+    },
+    {
+      "epoch": 0.3523948753534783,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0024770923644671485,
+      "loss": 3.4277,
+      "step": 4015
+    },
+    {
+      "epoch": 0.35248264493638076,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0024767698042987758,
+      "loss": 3.3286,
+      "step": 4016
+    },
+    {
+      "epoch": 0.3525704145192832,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0024764471685859656,
+      "loss": 3.333,
+      "step": 4017
+    },
+    {
+      "epoch": 0.35265818410218575,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0024761244573581974,
+      "loss": 3.3604,
+      "step": 4018
+    },
+    {
+      "epoch": 0.3527459536850882,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002475801670644959,
+      "loss": 3.3403,
+      "step": 4019
+    },
+    {
+      "epoch": 0.35283372326799073,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0024754788084757435,
+      "loss": 3.3467,
+      "step": 4020
+    },
+    {
+      "epoch": 0.3529214928508932,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0024751558708800523,
+      "loss": 3.2798,
+      "step": 4021
+    },
+    {
+      "epoch": 0.35300926243379566,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0024748328578873932,
+      "loss": 3.291,
+      "step": 4022
+    },
+    {
+      "epoch": 0.3530970320166982,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0024745097695272806,
+      "loss": 3.2847,
+      "step": 4023
+    },
+    {
+      "epoch": 0.35318480159960064,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0024741866058292364,
+      "loss": 3.3535,
+      "step": 4024
+    },
+    {
+      "epoch": 0.35327257118250316,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002473863366822789,
+      "loss": 3.3423,
+      "step": 4025
+    },
+    {
+      "epoch": 0.3533603407654056,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0024735400525374733,
+      "loss": 3.3423,
+      "step": 4026
+    },
+    {
+      "epoch": 0.3534481103483081,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002473216663002832,
+      "loss": 3.3525,
+      "step": 4027
+    },
+    {
+      "epoch": 0.3535358799312106,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0024728931982484137,
+      "loss": 3.4404,
+      "step": 4028
+    },
+    {
+      "epoch": 0.35362364951411307,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002472569658303775,
+      "loss": 3.3291,
+      "step": 4029
+    },
+    {
+      "epoch": 0.35371141909701553,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0024722460431984786,
+      "loss": 3.3994,
+      "step": 4030
+    },
+    {
+      "epoch": 0.35379918867991805,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0024719223529620932,
+      "loss": 3.4009,
+      "step": 4031
+    },
+    {
+      "epoch": 0.3538869582628205,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0024715985876241965,
+      "loss": 3.3926,
+      "step": 4032
+    },
+    {
+      "epoch": 0.35397472784572304,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0024712747472143714,
+      "loss": 3.2725,
+      "step": 4033
+    },
+    {
+      "epoch": 0.3540624974286255,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0024709508317622077,
+      "loss": 3.4341,
+      "step": 4034
+    },
+    {
+      "epoch": 0.35415026701152796,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0024706268412973036,
+      "loss": 3.3867,
+      "step": 4035
+    },
+    {
+      "epoch": 0.3542380365944305,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0024703027758492626,
+      "loss": 3.3774,
+      "step": 4036
+    },
+    {
+      "epoch": 0.35432580617733295,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0024699786354476956,
+      "loss": 3.4346,
+      "step": 4037
+    },
+    {
+      "epoch": 0.35441357576023547,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0024696544201222195,
+      "loss": 3.3184,
+      "step": 4038
+    },
+    {
+      "epoch": 0.35450134534313793,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.00246933012990246,
+      "loss": 3.3242,
+      "step": 4039
+    },
+    {
+      "epoch": 0.3545891149260404,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0024690057648180477,
+      "loss": 3.3677,
+      "step": 4040
+    },
+    {
+      "epoch": 0.3546768845089429,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.002468681324898621,
+      "loss": 3.3687,
+      "step": 4041
+    },
+    {
+      "epoch": 0.3547646540918454,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0024683568101738253,
+      "loss": 3.354,
+      "step": 4042
+    },
+    {
+      "epoch": 0.35485242367474784,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0024680322206733123,
+      "loss": 3.3154,
+      "step": 4043
+    },
+    {
+      "epoch": 0.35494019325765036,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.00246770755642674,
+      "loss": 3.3667,
+      "step": 4044
+    },
+    {
+      "epoch": 0.3550279628405528,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002467382817463775,
+      "loss": 3.3301,
+      "step": 4045
+    },
+    {
+      "epoch": 0.35511573242345534,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0024670580038140895,
+      "loss": 3.2964,
+      "step": 4046
+    },
+    {
+      "epoch": 0.3552035020063578,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.002466733115507362,
+      "loss": 3.3618,
+      "step": 4047
+    },
+    {
+      "epoch": 0.35529127158926027,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002466408152573279,
+      "loss": 3.3301,
+      "step": 4048
+    },
+    {
+      "epoch": 0.3553790411721628,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0024660831150415336,
+      "loss": 3.3779,
+      "step": 4049
+    },
+    {
+      "epoch": 0.35546681075506525,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002465758002941825,
+      "loss": 3.397,
+      "step": 4050
+    },
+    {
+      "epoch": 0.3555545803379678,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002465432816303859,
+      "loss": 3.3374,
+      "step": 4051
+    },
+    {
+      "epoch": 0.35564234992087024,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0024651075551573504,
+      "loss": 3.3735,
+      "step": 4052
+    },
+    {
+      "epoch": 0.3557301195037727,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0024647822195320184,
+      "loss": 3.4175,
+      "step": 4053
+    },
+    {
+      "epoch": 0.3558178890866752,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0024644568094575907,
+      "loss": 3.2998,
+      "step": 4054
+    },
+    {
+      "epoch": 0.3559056586695777,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0024641313249638,
+      "loss": 3.3247,
+      "step": 4055
+    },
+    {
+      "epoch": 0.3559934282524802,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0024638057660803865,
+      "loss": 3.3057,
+      "step": 4056
+    },
+    {
+      "epoch": 0.35608119783538267,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.002463480132837099,
+      "loss": 3.3154,
+      "step": 4057
+    },
+    {
+      "epoch": 0.35616896741828513,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0024631544252636905,
+      "loss": 3.3535,
+      "step": 4058
+    },
+    {
+      "epoch": 0.35625673700118765,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0024628286433899223,
+      "loss": 3.3423,
+      "step": 4059
+    },
+    {
+      "epoch": 0.3563445065840901,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0024625027872455614,
+      "loss": 3.356,
+      "step": 4060
+    },
+    {
+      "epoch": 0.3564322761669926,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002462176856860383,
+      "loss": 3.3662,
+      "step": 4061
+    },
+    {
+      "epoch": 0.3565200457498951,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.002461850852264168,
+      "loss": 3.4956,
+      "step": 4062
+    },
+    {
+      "epoch": 0.35660781533279756,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0024615247734867045,
+      "loss": 3.4023,
+      "step": 4063
+    },
+    {
+      "epoch": 0.3566955849157001,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0024611986205577874,
+      "loss": 3.3955,
+      "step": 4064
+    },
+    {
+      "epoch": 0.35678335449860255,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0024608723935072184,
+      "loss": 3.3071,
+      "step": 4065
+    },
+    {
+      "epoch": 0.356871124081505,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002460546092364805,
+      "loss": 3.2974,
+      "step": 4066
+    },
+    {
+      "epoch": 0.35695889366440753,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0024602197171603632,
+      "loss": 3.4062,
+      "step": 4067
+    },
+    {
+      "epoch": 0.35704666324731,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002459893267923715,
+      "loss": 3.3481,
+      "step": 4068
+    },
+    {
+      "epoch": 0.3571344328302125,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0024595667446846883,
+      "loss": 3.3481,
+      "step": 4069
+    },
+    {
+      "epoch": 0.357222202413115,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0024592401474731185,
+      "loss": 3.3594,
+      "step": 4070
+    },
+    {
+      "epoch": 0.35730997199601744,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0024589134763188486,
+      "loss": 3.3696,
+      "step": 4071
+    },
+    {
+      "epoch": 0.35739774157891996,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0024585867312517267,
+      "loss": 3.4019,
+      "step": 4072
+    },
+    {
+      "epoch": 0.3574855111618224,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0024582599123016092,
+      "loss": 3.3076,
+      "step": 4073
+    },
+    {
+      "epoch": 0.3575732807447249,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002457933019498358,
+      "loss": 3.3159,
+      "step": 4074
+    },
+    {
+      "epoch": 0.3576610503276274,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0024576060528718425,
+      "loss": 3.3833,
+      "step": 4075
+    },
+    {
+      "epoch": 0.35774881991052987,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0024572790124519385,
+      "loss": 3.3569,
+      "step": 4076
+    },
+    {
+      "epoch": 0.3578365894934324,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0024569518982685288,
+      "loss": 3.3462,
+      "step": 4077
+    },
+    {
+      "epoch": 0.35792435907633485,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0024566247103515026,
+      "loss": 3.373,
+      "step": 4078
+    },
+    {
+      "epoch": 0.3580121286592373,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0024562974487307558,
+      "loss": 3.2705,
+      "step": 4079
+    },
+    {
+      "epoch": 0.35809989824213984,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.0024559701134361923,
+      "loss": 3.4648,
+      "step": 4080
+    },
+    {
+      "epoch": 0.3581876678250423,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0024556427044977206,
+      "loss": 3.2603,
+      "step": 4081
+    },
+    {
+      "epoch": 0.3582754374079448,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.002455315221945258,
+      "loss": 3.3228,
+      "step": 4082
+    },
+    {
+      "epoch": 0.3583632069908473,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0024549876658087268,
+      "loss": 3.3447,
+      "step": 4083
+    },
+    {
+      "epoch": 0.35845097657374975,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002454660036118057,
+      "loss": 3.2705,
+      "step": 4084
+    },
+    {
+      "epoch": 0.35853874615665227,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0024543323329031847,
+      "loss": 3.3564,
+      "step": 4085
+    },
+    {
+      "epoch": 0.35862651573955473,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0024540045561940544,
+      "loss": 3.3403,
+      "step": 4086
+    },
+    {
+      "epoch": 0.3587142853224572,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002453676706020614,
+      "loss": 3.3315,
+      "step": 4087
+    },
+    {
+      "epoch": 0.3588020549053597,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0024533487824128226,
+      "loss": 3.3755,
+      "step": 4088
+    },
+    {
+      "epoch": 0.3588898244882622,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002453020785400642,
+      "loss": 3.3628,
+      "step": 4089
+    },
+    {
+      "epoch": 0.3589775940711647,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002452692715014042,
+      "loss": 3.3311,
+      "step": 4090
+    },
+    {
+      "epoch": 0.35906536365406716,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002452364571283001,
+      "loss": 3.3413,
+      "step": 4091
+    },
+    {
+      "epoch": 0.3591531332369696,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0024520363542375008,
+      "loss": 3.4136,
+      "step": 4092
+    },
+    {
+      "epoch": 0.35924090281987214,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0024517080639075324,
+      "loss": 3.4302,
+      "step": 4093
+    },
+    {
+      "epoch": 0.3593286724027746,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002451379700323092,
+      "loss": 3.3306,
+      "step": 4094
+    },
+    {
+      "epoch": 0.3594164419856771,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0024510512635141837,
+      "loss": 3.2734,
+      "step": 4095
+    },
+    {
+      "epoch": 0.3595042115685796,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002450722753510818,
+      "loss": 3.313,
+      "step": 4096
+    },
+    {
+      "epoch": 0.35959198115148205,
+      "grad_norm": 0.0830078125,
+      "learning_rate": 0.0024503941703430125,
+      "loss": 3.3521,
+      "step": 4097
+    },
+    {
+      "epoch": 0.3596797507343846,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002450065514040789,
+      "loss": 3.3594,
+      "step": 4098
+    },
+    {
+      "epoch": 0.35976752031728704,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002449736784634179,
+      "loss": 3.3359,
+      "step": 4099
+    },
+    {
+      "epoch": 0.3598552899001895,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002449407982153219,
+      "loss": 3.3359,
+      "step": 4100
+    },
+    {
+      "epoch": 0.359943059483092,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0024490791066279524,
+      "loss": 3.3809,
+      "step": 4101
+    },
+    {
+      "epoch": 0.3600308290659945,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0024487501580884303,
+      "loss": 3.3203,
+      "step": 4102
+    },
+    {
+      "epoch": 0.360118598648897,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.00244842113656471,
+      "loss": 3.3789,
+      "step": 4103
+    },
+    {
+      "epoch": 0.36020636823179947,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0024480920420868546,
+      "loss": 3.3423,
+      "step": 4104
+    },
+    {
+      "epoch": 0.36029413781470193,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002447762874684934,
+      "loss": 3.3906,
+      "step": 4105
+    },
+    {
+      "epoch": 0.36038190739760445,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0024474336343890254,
+      "loss": 3.3384,
+      "step": 4106
+    },
+    {
+      "epoch": 0.3604696769805069,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0024471043212292132,
+      "loss": 3.2891,
+      "step": 4107
+    },
+    {
+      "epoch": 0.36055744656340943,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002446774935235587,
+      "loss": 3.3662,
+      "step": 4108
+    },
+    {
+      "epoch": 0.3606452161463119,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0024464454764382445,
+      "loss": 3.3691,
+      "step": 4109
+    },
+    {
+      "epoch": 0.36073298572921436,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0024461159448672886,
+      "loss": 3.4082,
+      "step": 4110
+    },
+    {
+      "epoch": 0.3608207553121169,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0024457863405528296,
+      "loss": 3.2632,
+      "step": 4111
+    },
+    {
+      "epoch": 0.36090852489501934,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0024454566635249858,
+      "loss": 3.2935,
+      "step": 4112
+    },
+    {
+      "epoch": 0.3609962944779218,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002445126913813879,
+      "loss": 3.4355,
+      "step": 4113
+    },
+    {
+      "epoch": 0.36108406406082433,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00244479709144964,
+      "loss": 3.4155,
+      "step": 4114
+    },
+    {
+      "epoch": 0.3611718336437268,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002444467196462406,
+      "loss": 3.3506,
+      "step": 4115
+    },
+    {
+      "epoch": 0.3612596032266293,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00244413722888232,
+      "loss": 3.3569,
+      "step": 4116
+    },
+    {
+      "epoch": 0.3613473728095318,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002443807188739533,
+      "loss": 3.3682,
+      "step": 4117
+    },
+    {
+      "epoch": 0.36143514239243424,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0024434770760642013,
+      "loss": 3.355,
+      "step": 4118
+    },
+    {
+      "epoch": 0.36152291197533676,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0024431468908864868,
+      "loss": 3.376,
+      "step": 4119
+    },
+    {
+      "epoch": 0.3616106815582392,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002442816633236562,
+      "loss": 3.3667,
+      "step": 4120
+    },
+    {
+      "epoch": 0.36169845114114174,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002442486303144602,
+      "loss": 3.3374,
+      "step": 4121
+    },
+    {
+      "epoch": 0.3617862207240442,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002442155900640791,
+      "loss": 3.3711,
+      "step": 4122
+    },
+    {
+      "epoch": 0.36187399030694667,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002441825425755318,
+      "loss": 3.3237,
+      "step": 4123
+    },
+    {
+      "epoch": 0.3619617598898492,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.00244149487851838,
+      "loss": 3.4106,
+      "step": 4124
+    },
+    {
+      "epoch": 0.36204952947275165,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002441164258960179,
+      "loss": 3.3462,
+      "step": 4125
+    },
+    {
+      "epoch": 0.36213729905565417,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0024408335671109263,
+      "loss": 3.3677,
+      "step": 4126
+    },
+    {
+      "epoch": 0.36222506863855664,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0024405028030008375,
+      "loss": 3.2915,
+      "step": 4127
+    },
+    {
+      "epoch": 0.3623128382214591,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0024401719666601357,
+      "loss": 3.2837,
+      "step": 4128
+    },
+    {
+      "epoch": 0.3624006078043616,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.00243984105811905,
+      "loss": 3.3784,
+      "step": 4129
+    },
+    {
+      "epoch": 0.3624883773872641,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002439510077407817,
+      "loss": 3.3428,
+      "step": 4130
+    },
+    {
+      "epoch": 0.36257614697016655,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002439179024556679,
+      "loss": 3.4277,
+      "step": 4131
+    },
+    {
+      "epoch": 0.36266391655306907,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.002438847899595885,
+      "loss": 3.3306,
+      "step": 4132
+    },
+    {
+      "epoch": 0.36275168613597153,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0024385167025556915,
+      "loss": 3.2749,
+      "step": 4133
+    },
+    {
+      "epoch": 0.36283945571887405,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0024381854334663607,
+      "loss": 3.377,
+      "step": 4134
+    },
+    {
+      "epoch": 0.3629272253017765,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002437854092358162,
+      "loss": 3.4185,
+      "step": 4135
+    },
+    {
+      "epoch": 0.363014994884679,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0024375226792613704,
+      "loss": 3.3535,
+      "step": 4136
+    },
+    {
+      "epoch": 0.3631027644675815,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0024371911942062687,
+      "loss": 3.3779,
+      "step": 4137
+    },
+    {
+      "epoch": 0.36319053405048396,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0024368596372231458,
+      "loss": 3.3174,
+      "step": 4138
+    },
+    {
+      "epoch": 0.3632783036333865,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0024365280083422963,
+      "loss": 3.3574,
+      "step": 4139
+    },
+    {
+      "epoch": 0.36336607321628894,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0024361963075940226,
+      "loss": 3.3579,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3634538427991914,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0024358645350086327,
+      "loss": 3.3696,
+      "step": 4141
+    },
+    {
+      "epoch": 0.3635416123820939,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0024355326906164423,
+      "loss": 3.3691,
+      "step": 4142
+    },
+    {
+      "epoch": 0.3636293819649964,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0024352007744477735,
+      "loss": 3.3721,
+      "step": 4143
+    },
+    {
+      "epoch": 0.36371715154789885,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002434868786532953,
+      "loss": 3.4629,
+      "step": 4144
+    },
+    {
+      "epoch": 0.3638049211308014,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0024345367269023167,
+      "loss": 3.3169,
+      "step": 4145
+    },
+    {
+      "epoch": 0.36389269071370384,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0024342045955862055,
+      "loss": 3.3096,
+      "step": 4146
+    },
+    {
+      "epoch": 0.36398046029660636,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002433872392614967,
+      "loss": 3.3564,
+      "step": 4147
+    },
+    {
+      "epoch": 0.3640682298795088,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002433540118018956,
+      "loss": 3.374,
+      "step": 4148
+    },
+    {
+      "epoch": 0.3641559994624113,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002433207771828533,
+      "loss": 3.3115,
+      "step": 4149
+    },
+    {
+      "epoch": 0.3642437690453138,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0024328753540740658,
+      "loss": 3.3608,
+      "step": 4150
+    },
+    {
+      "epoch": 0.36433153862821627,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0024325428647859284,
+      "loss": 3.3188,
+      "step": 4151
+    },
+    {
+      "epoch": 0.3644193082111188,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0024322103039945008,
+      "loss": 3.3135,
+      "step": 4152
+    },
+    {
+      "epoch": 0.36450707779402125,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0024318776717301713,
+      "loss": 3.293,
+      "step": 4153
+    },
+    {
+      "epoch": 0.3645948473769237,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002431544968023332,
+      "loss": 3.4395,
+      "step": 4154
+    },
+    {
+      "epoch": 0.36468261695982623,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0024312121929043846,
+      "loss": 3.3213,
+      "step": 4155
+    },
+    {
+      "epoch": 0.3647703865427287,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002430879346403734,
+      "loss": 3.3188,
+      "step": 4156
+    },
+    {
+      "epoch": 0.36485815612563116,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0024305464285517948,
+      "loss": 3.3306,
+      "step": 4157
+    },
+    {
+      "epoch": 0.3649459257085337,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0024302134393789864,
+      "loss": 3.3096,
+      "step": 4158
+    },
+    {
+      "epoch": 0.36503369529143614,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0024298803789157344,
+      "loss": 3.4487,
+      "step": 4159
+    },
+    {
+      "epoch": 0.36512146487433866,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0024295472471924724,
+      "loss": 3.3335,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3652092344572411,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002429214044239639,
+      "loss": 3.3799,
+      "step": 4161
+    },
+    {
+      "epoch": 0.3652970040401436,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0024288807700876803,
+      "loss": 3.3589,
+      "step": 4162
+    },
+    {
+      "epoch": 0.3653847736230461,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.002428547424767049,
+      "loss": 3.4141,
+      "step": 4163
+    },
+    {
+      "epoch": 0.3654725432059486,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0024282140083082024,
+      "loss": 3.3604,
+      "step": 4164
+    },
+    {
+      "epoch": 0.3655603127888511,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002427880520741607,
+      "loss": 3.3687,
+      "step": 4165
+    },
+    {
+      "epoch": 0.36564808237175356,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.002427546962097734,
+      "loss": 3.4014,
+      "step": 4166
+    },
+    {
+      "epoch": 0.365735851954656,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0024272133324070616,
+      "loss": 3.3828,
+      "step": 4167
+    },
+    {
+      "epoch": 0.36582362153755854,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0024268796317000753,
+      "loss": 3.4126,
+      "step": 4168
+    },
+    {
+      "epoch": 0.365911391120461,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.002426545860007266,
+      "loss": 3.3159,
+      "step": 4169
+    },
+    {
+      "epoch": 0.36599916070336347,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.002426212017359131,
+      "loss": 3.3281,
+      "step": 4170
+    },
+    {
+      "epoch": 0.366086930286266,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0024258781037861748,
+      "loss": 3.4219,
+      "step": 4171
+    },
+    {
+      "epoch": 0.36617469986916845,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0024255441193189085,
+      "loss": 3.3687,
+      "step": 4172
+    },
+    {
+      "epoch": 0.36626246945207097,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002425210063987848,
+      "loss": 3.4224,
+      "step": 4173
+    },
+    {
+      "epoch": 0.36635023903497344,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0024248759378235184,
+      "loss": 3.3477,
+      "step": 4174
+    },
+    {
+      "epoch": 0.3664380086178759,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002424541740856449,
+      "loss": 3.4033,
+      "step": 4175
+    },
+    {
+      "epoch": 0.3665257782007784,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002424207473117177,
+      "loss": 3.3242,
+      "step": 4176
+    },
+    {
+      "epoch": 0.3666135477836809,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002423873134636244,
+      "loss": 3.3682,
+      "step": 4177
+    },
+    {
+      "epoch": 0.3667013173665834,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.002423538725444201,
+      "loss": 3.3379,
+      "step": 4178
+    },
+    {
+      "epoch": 0.36678908694948587,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002423204245571604,
+      "loss": 3.3135,
+      "step": 4179
+    },
+    {
+      "epoch": 0.36687685653238833,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.0024228696950490142,
+      "loss": 3.3447,
+      "step": 4180
+    },
+    {
+      "epoch": 0.36696462611529085,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0024225350739070016,
+      "loss": 3.3345,
+      "step": 4181
+    },
+    {
+      "epoch": 0.3670523956981933,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0024222003821761408,
+      "loss": 3.3223,
+      "step": 4182
+    },
+    {
+      "epoch": 0.3671401652810958,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0024218656198870146,
+      "loss": 3.3911,
+      "step": 4183
+    },
+    {
+      "epoch": 0.3672279348639983,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0024215307870702097,
+      "loss": 3.4609,
+      "step": 4184
+    },
+    {
+      "epoch": 0.36731570444690076,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002421195883756322,
+      "loss": 3.3276,
+      "step": 4185
+    },
+    {
+      "epoch": 0.3674034740298033,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0024208609099759526,
+      "loss": 3.4268,
+      "step": 4186
+    },
+    {
+      "epoch": 0.36749124361270574,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0024205258657597085,
+      "loss": 3.3511,
+      "step": 4187
+    },
+    {
+      "epoch": 0.3675790131956082,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0024201907511382037,
+      "loss": 3.3965,
+      "step": 4188
+    },
+    {
+      "epoch": 0.3676667827785107,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0024198555661420584,
+      "loss": 3.3467,
+      "step": 4189
+    },
+    {
+      "epoch": 0.3677545523614132,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0024195203108019012,
+      "loss": 3.3984,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3678423219443157,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002419184985148363,
+      "loss": 3.2656,
+      "step": 4191
+    },
+    {
+      "epoch": 0.3679300915272182,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.002418849589212085,
+      "loss": 3.3413,
+      "step": 4192
+    },
+    {
+      "epoch": 0.36801786111012064,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002418514123023713,
+      "loss": 3.416,
+      "step": 4193
+    },
+    {
+      "epoch": 0.36810563069302316,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0024181785866138994,
+      "loss": 3.3247,
+      "step": 4194
+    },
+    {
+      "epoch": 0.3681934002759256,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0024178429800133035,
+      "loss": 3.3213,
+      "step": 4195
+    },
+    {
+      "epoch": 0.36828116985882814,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.00241750730325259,
+      "loss": 3.3931,
+      "step": 4196
+    },
+    {
+      "epoch": 0.3683689394417306,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002417171556362431,
+      "loss": 3.4316,
+      "step": 4197
+    },
+    {
+      "epoch": 0.36845670902463307,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002416835739373505,
+      "loss": 3.3018,
+      "step": 4198
+    },
+    {
+      "epoch": 0.3685444786075356,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002416499852316497,
+      "loss": 3.3535,
+      "step": 4199
+    },
+    {
+      "epoch": 0.36863224819043805,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002416163895222097,
+      "loss": 3.2881,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3687200177733405,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0024158278681210028,
+      "loss": 3.3579,
+      "step": 4201
+    },
+    {
+      "epoch": 0.36880778735624303,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0024154917710439185,
+      "loss": 3.2944,
+      "step": 4202
+    },
+    {
+      "epoch": 0.3688955569391455,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0024151556040215537,
+      "loss": 3.3408,
+      "step": 4203
+    },
+    {
+      "epoch": 0.368983326522048,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0024148193670846256,
+      "loss": 3.3291,
+      "step": 4204
+    },
+    {
+      "epoch": 0.3690710961049505,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0024144830602638567,
+      "loss": 3.3745,
+      "step": 4205
+    },
+    {
+      "epoch": 0.36915886568785294,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002414146683589977,
+      "loss": 3.4082,
+      "step": 4206
+    },
+    {
+      "epoch": 0.36924663527075546,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0024138102370937216,
+      "loss": 3.3232,
+      "step": 4207
+    },
+    {
+      "epoch": 0.3693344048536579,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0024134737208058327,
+      "loss": 3.333,
+      "step": 4208
+    },
+    {
+      "epoch": 0.36942217443656045,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002413137134757059,
+      "loss": 3.3579,
+      "step": 4209
+    },
+    {
+      "epoch": 0.3695099440194629,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0024128004789781554,
+      "loss": 3.3809,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3695977136023654,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0024124637534998833,
+      "loss": 3.3638,
+      "step": 4211
+    },
+    {
+      "epoch": 0.3696854831852679,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.00241212695835301,
+      "loss": 3.2837,
+      "step": 4212
+    },
+    {
+      "epoch": 0.36977325276817036,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0024117900935683094,
+      "loss": 3.2368,
+      "step": 4213
+    },
+    {
+      "epoch": 0.3698610223510728,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.002411453159176562,
+      "loss": 3.3672,
+      "step": 4214
+    },
+    {
+      "epoch": 0.36994879193397534,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.002411116155208555,
+      "loss": 3.3911,
+      "step": 4215
+    },
+    {
+      "epoch": 0.3700365615168778,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002410779081695081,
+      "loss": 3.3188,
+      "step": 4216
+    },
+    {
+      "epoch": 0.3701243310997803,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.002410441938666938,
+      "loss": 3.4399,
+      "step": 4217
+    },
+    {
+      "epoch": 0.3702121006826828,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.002410104726154935,
+      "loss": 3.3584,
+      "step": 4218
+    },
+    {
+      "epoch": 0.37029987026558525,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0024097674441898817,
+      "loss": 3.3052,
+      "step": 4219
+    },
+    {
+      "epoch": 0.37038763984848777,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002409430092802597,
+      "loss": 3.3564,
+      "step": 4220
+    },
+    {
+      "epoch": 0.37047540943139023,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.0024090926720239055,
+      "loss": 3.3687,
+      "step": 4221
+    },
+    {
+      "epoch": 0.37056317901429275,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0024087551818846396,
+      "loss": 3.3809,
+      "step": 4222
+    },
+    {
+      "epoch": 0.3706509485971952,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0024084176224156355,
+      "loss": 3.2974,
+      "step": 4223
+    },
+    {
+      "epoch": 0.3707387181800977,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0024080799936477375,
+      "loss": 3.396,
+      "step": 4224
+    },
+    {
+      "epoch": 0.3708264877630002,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0024077422956117955,
+      "loss": 3.3159,
+      "step": 4225
+    },
+    {
+      "epoch": 0.37091425734590266,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0024074045283386664,
+      "loss": 3.3745,
+      "step": 4226
+    },
+    {
+      "epoch": 0.37100202692880513,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0024070666918592125,
+      "loss": 3.3633,
+      "step": 4227
+    },
+    {
+      "epoch": 0.37108979651170765,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.002406728786204304,
+      "loss": 3.311,
+      "step": 4228
+    },
+    {
+      "epoch": 0.3711775660946101,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002406390811404815,
+      "loss": 3.3765,
+      "step": 4229
+    },
+    {
+      "epoch": 0.37126533567751263,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002406052767491628,
+      "loss": 3.3267,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3713531052604151,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0024057146544956306,
+      "loss": 3.438,
+      "step": 4231
+    },
+    {
+      "epoch": 0.37144087484331756,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002405376472447718,
+      "loss": 3.3843,
+      "step": 4232
+    },
+    {
+      "epoch": 0.3715286444262201,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.00240503822137879,
+      "loss": 3.3994,
+      "step": 4233
+    },
+    {
+      "epoch": 0.37161641400912254,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0024046999013197542,
+      "loss": 3.2798,
+      "step": 4234
+    },
+    {
+      "epoch": 0.37170418359202506,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.002404361512301524,
+      "loss": 3.4082,
+      "step": 4235
+    },
+    {
+      "epoch": 0.3717919531749275,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0024040230543550187,
+      "loss": 3.4282,
+      "step": 4236
+    },
+    {
+      "epoch": 0.37187972275783,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0024036845275111638,
+      "loss": 3.3242,
+      "step": 4237
+    },
+    {
+      "epoch": 0.3719674923407325,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0024033459318008924,
+      "loss": 3.2349,
+      "step": 4238
+    },
+    {
+      "epoch": 0.372055261923635,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002403007267255143,
+      "loss": 3.353,
+      "step": 4239
+    },
+    {
+      "epoch": 0.37214303150653744,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0024026685339048597,
+      "loss": 3.4072,
+      "step": 4240
+    },
+    {
+      "epoch": 0.37223080108943996,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0024023297317809942,
+      "loss": 3.2983,
+      "step": 4241
+    },
+    {
+      "epoch": 0.3723185706723424,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0024019908609145034,
+      "loss": 3.2661,
+      "step": 4242
+    },
+    {
+      "epoch": 0.37240634025524494,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002401651921336351,
+      "loss": 3.3472,
+      "step": 4243
+    },
+    {
+      "epoch": 0.3724941098381474,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0024013129130775065,
+      "loss": 3.355,
+      "step": 4244
+    },
+    {
+      "epoch": 0.37258187942104987,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0024009738361689477,
+      "loss": 3.4058,
+      "step": 4245
+    },
+    {
+      "epoch": 0.3726696490039524,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0024006346906416554,
+      "loss": 3.3716,
+      "step": 4246
+    },
+    {
+      "epoch": 0.37275741858685485,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002400295476526619,
+      "loss": 3.3516,
+      "step": 4247
+    },
+    {
+      "epoch": 0.37284518816975737,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.002399956193854834,
+      "loss": 3.3965,
+      "step": 4248
+    },
+    {
+      "epoch": 0.37293295775265983,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0023996168426573005,
+      "loss": 3.3472,
+      "step": 4249
+    },
+    {
+      "epoch": 0.3730207273355623,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002399277422965027,
+      "loss": 3.3452,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3731084969184648,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002398937934809027,
+      "loss": 3.3174,
+      "step": 4251
+    },
+    {
+      "epoch": 0.3731962665013673,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0023985983782203206,
+      "loss": 3.396,
+      "step": 4252
+    },
+    {
+      "epoch": 0.37328403608426974,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002398258753229934,
+      "loss": 3.3481,
+      "step": 4253
+    },
+    {
+      "epoch": 0.37337180566717226,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0023979190598689,
+      "loss": 3.3047,
+      "step": 4254
+    },
+    {
+      "epoch": 0.3734595752500747,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.002397579298168257,
+      "loss": 3.3438,
+      "step": 4255
+    },
+    {
+      "epoch": 0.37354734483297725,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.00239723946815905,
+      "loss": 3.3394,
+      "step": 4256
+    },
+    {
+      "epoch": 0.3736351144158797,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.002396899569872331,
+      "loss": 3.2949,
+      "step": 4257
+    },
+    {
+      "epoch": 0.3737228839987822,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0023965596033391568,
+      "loss": 3.3574,
+      "step": 4258
+    },
+    {
+      "epoch": 0.3738106535816847,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0023962195685905915,
+      "loss": 3.415,
+      "step": 4259
+    },
+    {
+      "epoch": 0.37389842316458716,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002395879465657705,
+      "loss": 3.4185,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3739861927474897,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0023955392945715736,
+      "loss": 3.3286,
+      "step": 4261
+    },
+    {
+      "epoch": 0.37407396233039214,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0023951990553632793,
+      "loss": 3.2861,
+      "step": 4262
+    },
+    {
+      "epoch": 0.3741617319132946,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0023948587480639117,
+      "loss": 3.3931,
+      "step": 4263
+    },
+    {
+      "epoch": 0.3742495014961971,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0023945183727045656,
+      "loss": 3.3677,
+      "step": 4264
+    },
+    {
+      "epoch": 0.3743372710790996,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002394177929316341,
+      "loss": 3.3643,
+      "step": 4265
+    },
+    {
+      "epoch": 0.3744250406620021,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002393837417930346,
+      "loss": 3.3418,
+      "step": 4266
+    },
+    {
+      "epoch": 0.37451281024490457,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002393496838577694,
+      "loss": 3.2925,
+      "step": 4267
+    },
+    {
+      "epoch": 0.37460057982780703,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002393156191289505,
+      "loss": 3.436,
+      "step": 4268
+    },
+    {
+      "epoch": 0.37468834941070955,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0023928154760969045,
+      "loss": 3.3169,
+      "step": 4269
+    },
+    {
+      "epoch": 0.374776118993612,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0023924746930310253,
+      "loss": 3.3638,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3748638885765145,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.0023921338421230053,
+      "loss": 3.3438,
+      "step": 4271
+    },
+    {
+      "epoch": 0.374951658159417,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0023917929234039898,
+      "loss": 3.3315,
+      "step": 4272
+    },
+    {
+      "epoch": 0.37503942774231946,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.002391451936905128,
+      "loss": 3.3784,
+      "step": 4273
+    },
+    {
+      "epoch": 0.375127197325222,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002391110882657578,
+      "loss": 3.3589,
+      "step": 4274
+    },
+    {
+      "epoch": 0.37521496690812445,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0023907697606925034,
+      "loss": 3.3657,
+      "step": 4275
+    },
+    {
+      "epoch": 0.3753027364910269,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0023904285710410725,
+      "loss": 3.3496,
+      "step": 4276
+    },
+    {
+      "epoch": 0.37539050607392943,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0023900873137344615,
+      "loss": 3.4639,
+      "step": 4277
+    },
+    {
+      "epoch": 0.3754782756568319,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0023897459888038517,
+      "loss": 3.3384,
+      "step": 4278
+    },
+    {
+      "epoch": 0.3755660452397344,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.002389404596280431,
+      "loss": 3.3911,
+      "step": 4279
+    },
+    {
+      "epoch": 0.3756538148226369,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002389063136195394,
+      "loss": 3.3477,
+      "step": 4280
+    },
+    {
+      "epoch": 0.37574158440553934,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0023887216085799402,
+      "loss": 3.4087,
+      "step": 4281
+    },
+    {
+      "epoch": 0.37582935398844186,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0023883800134652765,
+      "loss": 3.3623,
+      "step": 4282
+    },
+    {
+      "epoch": 0.3759171235713443,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.002388038350882615,
+      "loss": 3.3384,
+      "step": 4283
+    },
+    {
+      "epoch": 0.3760048931542468,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002387696620863175,
+      "loss": 3.3462,
+      "step": 4284
+    },
+    {
+      "epoch": 0.3760926627371493,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0023873548234381813,
+      "loss": 3.3306,
+      "step": 4285
+    },
+    {
+      "epoch": 0.37618043232005177,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0023870129586388645,
+      "loss": 3.3501,
+      "step": 4286
+    },
+    {
+      "epoch": 0.3762682019029543,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.002386671026496462,
+      "loss": 3.353,
+      "step": 4287
+    },
+    {
+      "epoch": 0.37635597148585676,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0023863290270422174,
+      "loss": 3.4214,
+      "step": 4288
+    },
+    {
+      "epoch": 0.3764437410687592,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.00238598696030738,
+      "loss": 3.3311,
+      "step": 4289
+    },
+    {
+      "epoch": 0.37653151065166174,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0023856448263232053,
+      "loss": 3.3682,
+      "step": 4290
+    },
+    {
+      "epoch": 0.3766192802345642,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002385302625120956,
+      "loss": 3.3159,
+      "step": 4291
+    },
+    {
+      "epoch": 0.3767070498174667,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0023849603567318994,
+      "loss": 3.3667,
+      "step": 4292
+    },
+    {
+      "epoch": 0.3767948194003692,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002384618021187309,
+      "loss": 3.3008,
+      "step": 4293
+    },
+    {
+      "epoch": 0.37688258898327165,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002384275618518466,
+      "loss": 3.3711,
+      "step": 4294
+    },
+    {
+      "epoch": 0.37697035856617417,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0023839331487566563,
+      "loss": 3.2803,
+      "step": 4295
+    },
+    {
+      "epoch": 0.37705812814907663,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0023835906119331726,
+      "loss": 3.293,
+      "step": 4296
+    },
+    {
+      "epoch": 0.3771458977319791,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0023832480080793134,
+      "loss": 3.3955,
+      "step": 4297
+    },
+    {
+      "epoch": 0.3772336673148816,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0023829053372263834,
+      "loss": 3.3169,
+      "step": 4298
+    },
+    {
+      "epoch": 0.3773214368977841,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0023825625994056934,
+      "loss": 3.3462,
+      "step": 4299
+    },
+    {
+      "epoch": 0.3774092064806866,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0023822197946485606,
+      "loss": 3.332,
+      "step": 4300
+    },
+    {
+      "epoch": 0.37749697606358906,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0023818769229863074,
+      "loss": 3.4038,
+      "step": 4301
+    },
+    {
+      "epoch": 0.3775847456464915,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0023815339844502643,
+      "loss": 3.3535,
+      "step": 4302
+    },
+    {
+      "epoch": 0.37767251522939405,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002381190979071766,
+      "loss": 3.3511,
+      "step": 4303
+    },
+    {
+      "epoch": 0.3777602848122965,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0023808479068821533,
+      "loss": 3.2905,
+      "step": 4304
+    },
+    {
+      "epoch": 0.37784805439519903,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0023805047679127745,
+      "loss": 3.3022,
+      "step": 4305
+    },
+    {
+      "epoch": 0.3779358239781015,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.002380161562194983,
+      "loss": 3.3721,
+      "step": 4306
+    },
+    {
+      "epoch": 0.37802359356100396,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0023798182897601383,
+      "loss": 3.3228,
+      "step": 4307
+    },
+    {
+      "epoch": 0.3781113631439065,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0023794749506396073,
+      "loss": 3.3394,
+      "step": 4308
+    },
+    {
+      "epoch": 0.37819913272680894,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.00237913154486476,
+      "loss": 3.4082,
+      "step": 4309
+    },
+    {
+      "epoch": 0.3782869023097114,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002378788072466976,
+      "loss": 3.3848,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3783746718926139,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0023784445334776396,
+      "loss": 3.3418,
+      "step": 4311
+    },
+    {
+      "epoch": 0.3784624414755164,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0023781009279281395,
+      "loss": 3.2827,
+      "step": 4312
+    },
+    {
+      "epoch": 0.3785502110584189,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0023777572558498733,
+      "loss": 3.3003,
+      "step": 4313
+    },
+    {
+      "epoch": 0.37863798064132137,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002377413517274242,
+      "loss": 3.3521,
+      "step": 4314
+    },
+    {
+      "epoch": 0.37872575022422383,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002377069712232656,
+      "loss": 3.3579,
+      "step": 4315
+    },
+    {
+      "epoch": 0.37881351980712635,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0023767258407565274,
+      "loss": 3.3823,
+      "step": 4316
+    },
+    {
+      "epoch": 0.3789012893900288,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0023763819028772793,
+      "loss": 3.4165,
+      "step": 4317
+    },
+    {
+      "epoch": 0.37898905897293134,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0023760378986263367,
+      "loss": 3.3789,
+      "step": 4318
+    },
+    {
+      "epoch": 0.3790768285558338,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0023756938280351326,
+      "loss": 3.3535,
+      "step": 4319
+    },
+    {
+      "epoch": 0.37916459813873626,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0023753496911351054,
+      "loss": 3.3418,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3792523677216388,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002375005487957701,
+      "loss": 3.3564,
+      "step": 4321
+    },
+    {
+      "epoch": 0.37934013730454125,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0023746612185343687,
+      "loss": 3.3276,
+      "step": 4322
+    },
+    {
+      "epoch": 0.3794279068874437,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002374316882896567,
+      "loss": 3.3853,
+      "step": 4323
+    },
+    {
+      "epoch": 0.37951567647034623,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002373972481075758,
+      "loss": 3.3486,
+      "step": 4324
+    },
+    {
+      "epoch": 0.3796034460532487,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.002373628013103411,
+      "loss": 3.3877,
+      "step": 4325
+    },
+    {
+      "epoch": 0.3796912156361512,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0023732834790110018,
+      "loss": 3.3101,
+      "step": 4326
+    },
+    {
+      "epoch": 0.3797789852190537,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0023729388788300094,
+      "loss": 3.3354,
+      "step": 4327
+    },
+    {
+      "epoch": 0.37986675480195614,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0023725942125919232,
+      "loss": 3.3335,
+      "step": 4328
+    },
+    {
+      "epoch": 0.37995452438485866,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002372249480328236,
+      "loss": 3.3594,
+      "step": 4329
+    },
+    {
+      "epoch": 0.3800422939677611,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.002371904682070446,
+      "loss": 3.3936,
+      "step": 4330
+    },
+    {
+      "epoch": 0.38013006355066364,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0023715598178500586,
+      "loss": 3.3652,
+      "step": 4331
+    },
+    {
+      "epoch": 0.3802178331335661,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0023712148876985856,
+      "loss": 3.3491,
+      "step": 4332
+    },
+    {
+      "epoch": 0.38030560271646857,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0023708698916475443,
+      "loss": 3.3853,
+      "step": 4333
+    },
+    {
+      "epoch": 0.3803933722993711,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0023705248297284575,
+      "loss": 3.3125,
+      "step": 4334
+    },
+    {
+      "epoch": 0.38048114188227355,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0023701797019728557,
+      "loss": 3.2832,
+      "step": 4335
+    },
+    {
+      "epoch": 0.3805689114651761,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0023698345084122727,
+      "loss": 3.2827,
+      "step": 4336
+    },
+    {
+      "epoch": 0.38065668104807854,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002369489249078251,
+      "loss": 3.3359,
+      "step": 4337
+    },
+    {
+      "epoch": 0.380744450630981,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0023691439240023375,
+      "loss": 3.3896,
+      "step": 4338
+    },
+    {
+      "epoch": 0.3808322202138835,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.002368798533216086,
+      "loss": 3.4365,
+      "step": 4339
+    },
+    {
+      "epoch": 0.380919989796786,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.002368453076751055,
+      "loss": 3.335,
+      "step": 4340
+    },
+    {
+      "epoch": 0.38100775937968845,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002368107554638811,
+      "loss": 3.3257,
+      "step": 4341
+    },
+    {
+      "epoch": 0.38109552896259097,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002367761966910925,
+      "loss": 3.3105,
+      "step": 4342
+    },
+    {
+      "epoch": 0.38118329854549343,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002367416313598974,
+      "loss": 3.377,
+      "step": 4343
+    },
+    {
+      "epoch": 0.38127106812839595,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0023670705947345418,
+      "loss": 3.3701,
+      "step": 4344
+    },
+    {
+      "epoch": 0.3813588377112984,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0023667248103492176,
+      "loss": 3.3076,
+      "step": 4345
+    },
+    {
+      "epoch": 0.3814466072942009,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002366378960474597,
+      "loss": 3.3345,
+      "step": 4346
+    },
+    {
+      "epoch": 0.3815343768771034,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002366033045142281,
+      "loss": 3.3574,
+      "step": 4347
+    },
+    {
+      "epoch": 0.38162214646000586,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0023656870643838763,
+      "loss": 3.3779,
+      "step": 4348
+    },
+    {
+      "epoch": 0.3817099160429084,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002365341018230998,
+      "loss": 3.3389,
+      "step": 4349
+    },
+    {
+      "epoch": 0.38179768562581085,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.002364994906715264,
+      "loss": 3.3213,
+      "step": 4350
+    },
+    {
+      "epoch": 0.3818854552087133,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0023646487298682995,
+      "loss": 3.3267,
+      "step": 4351
+    },
+    {
+      "epoch": 0.38197322479161583,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0023643024877217362,
+      "loss": 3.375,
+      "step": 4352
+    },
+    {
+      "epoch": 0.3820609943745183,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.002363956180307212,
+      "loss": 3.3394,
+      "step": 4353
+    },
+    {
+      "epoch": 0.38214876395742076,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002363609807656368,
+      "loss": 3.3291,
+      "step": 4354
+    },
+    {
+      "epoch": 0.3822365335403233,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002363263369800855,
+      "loss": 3.3652,
+      "step": 4355
+    },
+    {
+      "epoch": 0.38232430312322574,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002362916866772328,
+      "loss": 3.3667,
+      "step": 4356
+    },
+    {
+      "epoch": 0.38241207270612826,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002362570298602447,
+      "loss": 3.314,
+      "step": 4357
+    },
+    {
+      "epoch": 0.3824998422890307,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0023622236653228802,
+      "loss": 3.3838,
+      "step": 4358
+    },
+    {
+      "epoch": 0.3825876118719332,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.0023618769669652997,
+      "loss": 3.4199,
+      "step": 4359
+    },
+    {
+      "epoch": 0.3826753814548357,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0023615302035613845,
+      "loss": 3.3208,
+      "step": 4360
+    },
+    {
+      "epoch": 0.38276315103773817,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0023611833751428194,
+      "loss": 3.3691,
+      "step": 4361
+    },
+    {
+      "epoch": 0.3828509206206407,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0023608364817412956,
+      "loss": 3.3477,
+      "step": 4362
+    },
+    {
+      "epoch": 0.38293869020354315,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0023604895233885095,
+      "loss": 3.3105,
+      "step": 4363
+    },
+    {
+      "epoch": 0.3830264597864456,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.0023601425001161633,
+      "loss": 3.2773,
+      "step": 4364
+    },
+    {
+      "epoch": 0.38311422936934814,
+      "grad_norm": 0.25,
+      "learning_rate": 0.002359795411955966,
+      "loss": 3.3271,
+      "step": 4365
+    },
+    {
+      "epoch": 0.3832019989522506,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.002359448258939632,
+      "loss": 3.2949,
+      "step": 4366
+    },
+    {
+      "epoch": 0.38328976853515306,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002359101041098882,
+      "loss": 3.3687,
+      "step": 4367
+    },
+    {
+      "epoch": 0.3833775381180556,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0023587537584654414,
+      "loss": 3.3086,
+      "step": 4368
+    },
+    {
+      "epoch": 0.38346530770095805,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.002358406411071044,
+      "loss": 3.2236,
+      "step": 4369
+    },
+    {
+      "epoch": 0.38355307728386057,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002358058998947426,
+      "loss": 3.3335,
+      "step": 4370
+    },
+    {
+      "epoch": 0.38364084686676303,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0023577115221263337,
+      "loss": 3.3589,
+      "step": 4371
+    },
+    {
+      "epoch": 0.3837286164496655,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0023573639806395157,
+      "loss": 3.3335,
+      "step": 4372
+    },
+    {
+      "epoch": 0.383816386032568,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0023570163745187276,
+      "loss": 3.355,
+      "step": 4373
+    },
+    {
+      "epoch": 0.3839041556154705,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0023566687037957323,
+      "loss": 3.3418,
+      "step": 4374
+    },
+    {
+      "epoch": 0.383991925198373,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0023563209685022973,
+      "loss": 3.2964,
+      "step": 4375
+    },
+    {
+      "epoch": 0.38407969478127546,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0023559731686701954,
+      "loss": 3.355,
+      "step": 4376
+    },
+    {
+      "epoch": 0.3841674643641779,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002355625304331207,
+      "loss": 3.3267,
+      "step": 4377
+    },
+    {
+      "epoch": 0.38425523394708044,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0023552773755171168,
+      "loss": 3.3438,
+      "step": 4378
+    },
+    {
+      "epoch": 0.3843430035299829,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002354929382259717,
+      "loss": 3.3604,
+      "step": 4379
+    },
+    {
+      "epoch": 0.38443077311288537,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0023545813245908036,
+      "loss": 3.2603,
+      "step": 4380
+    },
+    {
+      "epoch": 0.3845185426957879,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.002354233202542181,
+      "loss": 3.3975,
+      "step": 4381
+    },
+    {
+      "epoch": 0.38460631227869035,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002353885016145657,
+      "loss": 3.293,
+      "step": 4382
+    },
+    {
+      "epoch": 0.3846940818615929,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0023535367654330466,
+      "loss": 3.373,
+      "step": 4383
+    },
+    {
+      "epoch": 0.38478185144449534,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0023531884504361714,
+      "loss": 3.3447,
+      "step": 4384
+    },
+    {
+      "epoch": 0.3848696210273978,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0023528400711868573,
+      "loss": 3.3066,
+      "step": 4385
+    },
+    {
+      "epoch": 0.3849573906103003,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.002352491627716937,
+      "loss": 3.4087,
+      "step": 4386
+    },
+    {
+      "epoch": 0.3850451601932028,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0023521431200582486,
+      "loss": 3.3926,
+      "step": 4387
+    },
+    {
+      "epoch": 0.3851329297761053,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002351794548242636,
+      "loss": 3.3389,
+      "step": 4388
+    },
+    {
+      "epoch": 0.38522069935900777,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0023514459123019494,
+      "loss": 3.3638,
+      "step": 4389
+    },
+    {
+      "epoch": 0.38530846894191023,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002351097212268046,
+      "loss": 3.3633,
+      "step": 4390
+    },
+    {
+      "epoch": 0.38539623852481275,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.0023507484481727864,
+      "loss": 3.3374,
+      "step": 4391
+    },
+    {
+      "epoch": 0.3854840081077152,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0023503996200480376,
+      "loss": 3.3799,
+      "step": 4392
+    },
+    {
+      "epoch": 0.3855717776906177,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.002350050727925674,
+      "loss": 3.3716,
+      "step": 4393
+    },
+    {
+      "epoch": 0.3856595472735202,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0023497017718375752,
+      "loss": 3.2412,
+      "step": 4394
+    },
+    {
+      "epoch": 0.38574731685642266,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0023493527518156252,
+      "loss": 3.3364,
+      "step": 4395
+    },
+    {
+      "epoch": 0.3858350864393252,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002349003667891717,
+      "loss": 3.4238,
+      "step": 4396
+    },
+    {
+      "epoch": 0.38592285602222764,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0023486545200977448,
+      "loss": 3.3257,
+      "step": 4397
+    },
+    {
+      "epoch": 0.3860106256051301,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0023483053084656135,
+      "loss": 3.3384,
+      "step": 4398
+    },
+    {
+      "epoch": 0.38609839518803263,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0023479560330272303,
+      "loss": 3.3613,
+      "step": 4399
+    },
+    {
+      "epoch": 0.3861861647709351,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0023476066938145097,
+      "loss": 3.3291,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3862739343538376,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0023472572908593728,
+      "loss": 3.3521,
+      "step": 4401
+    },
+    {
+      "epoch": 0.3863617039367401,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.002346907824193744,
+      "loss": 3.2725,
+      "step": 4402
+    },
+    {
+      "epoch": 0.38644947351964254,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0023465582938495574,
+      "loss": 3.3311,
+      "step": 4403
+    },
+    {
+      "epoch": 0.38653724310254506,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0023462086998587484,
+      "loss": 3.2666,
+      "step": 4404
+    },
+    {
+      "epoch": 0.3866250126854475,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.002345859042253262,
+      "loss": 3.2739,
+      "step": 4405
+    },
+    {
+      "epoch": 0.38671278226835004,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002345509321065046,
+      "loss": 3.2905,
+      "step": 4406
+    },
+    {
+      "epoch": 0.3868005518512525,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.002345159536326057,
+      "loss": 3.3867,
+      "step": 4407
+    },
+    {
+      "epoch": 0.38688832143415497,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002344809688068255,
+      "loss": 3.397,
+      "step": 4408
+    },
+    {
+      "epoch": 0.3869760910170575,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0023444597763236067,
+      "loss": 3.3599,
+      "step": 4409
+    },
+    {
+      "epoch": 0.38706386059995995,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0023441098011240846,
+      "loss": 3.3643,
+      "step": 4410
+    },
+    {
+      "epoch": 0.3871516301828624,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002343759762501668,
+      "loss": 3.2778,
+      "step": 4411
+    },
+    {
+      "epoch": 0.38723939976576494,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0023434096604883397,
+      "loss": 3.3584,
+      "step": 4412
+    },
+    {
+      "epoch": 0.3873271693486674,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00234305949511609,
+      "loss": 3.3423,
+      "step": 4413
+    },
+    {
+      "epoch": 0.3874149389315699,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.0023427092664169142,
+      "loss": 3.3188,
+      "step": 4414
+    },
+    {
+      "epoch": 0.3875027085144724,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0023423589744228143,
+      "loss": 3.3945,
+      "step": 4415
+    },
+    {
+      "epoch": 0.38759047809737485,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0023420086191657976,
+      "loss": 3.3545,
+      "step": 4416
+    },
+    {
+      "epoch": 0.38767824768027737,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0023416582006778766,
+      "loss": 3.4346,
+      "step": 4417
+    },
+    {
+      "epoch": 0.38776601726317983,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00234130771899107,
+      "loss": 3.3667,
+      "step": 4418
+    },
+    {
+      "epoch": 0.38785378684608235,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002340957174137403,
+      "loss": 3.3149,
+      "step": 4419
+    },
+    {
+      "epoch": 0.3879415564289848,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.002340606566148906,
+      "loss": 3.3687,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3880293260118873,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0023402558950576142,
+      "loss": 3.2637,
+      "step": 4421
+    },
+    {
+      "epoch": 0.3881170955947898,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0023399051608955705,
+      "loss": 3.3633,
+      "step": 4422
+    },
+    {
+      "epoch": 0.38820486517769226,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002339554363694822,
+      "loss": 3.3389,
+      "step": 4423
+    },
+    {
+      "epoch": 0.3882926347605947,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0023392035034874212,
+      "loss": 3.3164,
+      "step": 4424
+    },
+    {
+      "epoch": 0.38838040434349724,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002338852580305429,
+      "loss": 3.3032,
+      "step": 4425
+    },
+    {
+      "epoch": 0.3884681739263997,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.002338501594180909,
+      "loss": 3.2437,
+      "step": 4426
+    },
+    {
+      "epoch": 0.3885559435093022,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0023381505451459325,
+      "loss": 3.3252,
+      "step": 4427
+    },
+    {
+      "epoch": 0.3886437130922047,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0023377994332325768,
+      "loss": 3.3472,
+      "step": 4428
+    },
+    {
+      "epoch": 0.38873148267510715,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0023374482584729217,
+      "loss": 3.4604,
+      "step": 4429
+    },
+    {
+      "epoch": 0.3888192522580097,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002337097020899057,
+      "loss": 3.3389,
+      "step": 4430
+    },
+    {
+      "epoch": 0.38890702184091214,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.002336745720543075,
+      "loss": 3.3984,
+      "step": 4431
+    },
+    {
+      "epoch": 0.38899479142381466,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0023363943574370768,
+      "loss": 3.3755,
+      "step": 4432
+    },
+    {
+      "epoch": 0.3890825610067171,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0023360429316131664,
+      "loss": 3.3467,
+      "step": 4433
+    },
+    {
+      "epoch": 0.3891703305896196,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002335691443103454,
+      "loss": 3.313,
+      "step": 4434
+    },
+    {
+      "epoch": 0.3892581001725221,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002335339891940057,
+      "loss": 3.2866,
+      "step": 4435
+    },
+    {
+      "epoch": 0.38934586975542457,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0023349882781550976,
+      "loss": 3.3247,
+      "step": 4436
+    },
+    {
+      "epoch": 0.38943363933832703,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.002334636601780704,
+      "loss": 3.3638,
+      "step": 4437
+    },
+    {
+      "epoch": 0.38952140892122955,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.00233428486284901,
+      "loss": 3.3862,
+      "step": 4438
+    },
+    {
+      "epoch": 0.389609178504132,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0023339330613921542,
+      "loss": 3.4053,
+      "step": 4439
+    },
+    {
+      "epoch": 0.38969694808703453,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0023335811974422825,
+      "loss": 3.3271,
+      "step": 4440
+    },
+    {
+      "epoch": 0.389784717669937,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0023332292710315453,
+      "loss": 3.3457,
+      "step": 4441
+    },
+    {
+      "epoch": 0.38987248725283946,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0023328772821920997,
+      "loss": 3.2847,
+      "step": 4442
+    },
+    {
+      "epoch": 0.389960256835742,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0023325252309561078,
+      "loss": 3.3076,
+      "step": 4443
+    },
+    {
+      "epoch": 0.39004802641864444,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0023321731173557374,
+      "loss": 3.354,
+      "step": 4444
+    },
+    {
+      "epoch": 0.39013579600154696,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002331820941423162,
+      "loss": 3.4155,
+      "step": 4445
+    },
+    {
+      "epoch": 0.39022356558444943,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0023314687031905615,
+      "loss": 3.3677,
+      "step": 4446
+    },
+    {
+      "epoch": 0.3903113351673519,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002331116402690121,
+      "loss": 3.3789,
+      "step": 4447
+    },
+    {
+      "epoch": 0.3903991047502544,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00233076403995403,
+      "loss": 3.3364,
+      "step": 4448
+    },
+    {
+      "epoch": 0.3904868743331569,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002330411615014487,
+      "loss": 3.3452,
+      "step": 4449
+    },
+    {
+      "epoch": 0.39057464391605934,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.002330059127903692,
+      "loss": 3.3579,
+      "step": 4450
+    },
+    {
+      "epoch": 0.39066241349896186,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0023297065786538544,
+      "loss": 3.3369,
+      "step": 4451
+    },
+    {
+      "epoch": 0.3907501830818643,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002329353967297187,
+      "loss": 3.2905,
+      "step": 4452
+    },
+    {
+      "epoch": 0.39083795266476684,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0023290012938659093,
+      "loss": 3.2979,
+      "step": 4453
+    },
+    {
+      "epoch": 0.3909257222476693,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002328648558392245,
+      "loss": 3.3652,
+      "step": 4454
+    },
+    {
+      "epoch": 0.39101349183057177,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0023282957609084263,
+      "loss": 3.3545,
+      "step": 4455
+    },
+    {
+      "epoch": 0.3911012614134743,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0023279429014466885,
+      "loss": 3.3896,
+      "step": 4456
+    },
+    {
+      "epoch": 0.39118903099637675,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002327589980039273,
+      "loss": 3.3545,
+      "step": 4457
+    },
+    {
+      "epoch": 0.39127680057927927,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002327236996718428,
+      "loss": 3.4019,
+      "step": 4458
+    },
+    {
+      "epoch": 0.39136457016218174,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002326883951516407,
+      "loss": 3.3447,
+      "step": 4459
+    },
+    {
+      "epoch": 0.3914523397450842,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0023265308444654674,
+      "loss": 3.3247,
+      "step": 4460
+    },
+    {
+      "epoch": 0.3915401093279867,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0023261776755978745,
+      "loss": 3.3545,
+      "step": 4461
+    },
+    {
+      "epoch": 0.3916278789108892,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0023258244449458993,
+      "loss": 3.4556,
+      "step": 4462
+    },
+    {
+      "epoch": 0.39171564849379165,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0023254711525418164,
+      "loss": 3.272,
+      "step": 4463
+    },
+    {
+      "epoch": 0.39180341807669417,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0023251177984179067,
+      "loss": 3.2905,
+      "step": 4464
+    },
+    {
+      "epoch": 0.39189118765959663,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002324764382606458,
+      "loss": 3.3228,
+      "step": 4465
+    },
+    {
+      "epoch": 0.39197895724249915,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002324410905139763,
+      "loss": 3.4077,
+      "step": 4466
+    },
+    {
+      "epoch": 0.3920667268254016,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.00232405736605012,
+      "loss": 3.2832,
+      "step": 4467
+    },
+    {
+      "epoch": 0.3921544964083041,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0023237037653698334,
+      "loss": 3.2998,
+      "step": 4468
+    },
+    {
+      "epoch": 0.3922422659912066,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.002323350103131211,
+      "loss": 3.3579,
+      "step": 4469
+    },
+    {
+      "epoch": 0.39233003557410906,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0023229963793665698,
+      "loss": 3.3501,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3924178051570116,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0023226425941082294,
+      "loss": 3.3125,
+      "step": 4471
+    },
+    {
+      "epoch": 0.39250557473991404,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0023222887473885174,
+      "loss": 3.3086,
+      "step": 4472
+    },
+    {
+      "epoch": 0.3925933443228165,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.002321934839239765,
+      "loss": 3.3101,
+      "step": 4473
+    },
+    {
+      "epoch": 0.392681113905719,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.00232158086969431,
+      "loss": 3.4316,
+      "step": 4474
+    },
+    {
+      "epoch": 0.3927688834886215,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.002321226838784496,
+      "loss": 3.3779,
+      "step": 4475
+    },
+    {
+      "epoch": 0.392856653071524,
+      "grad_norm": 0.5,
+      "learning_rate": 0.002320872746542671,
+      "loss": 3.3628,
+      "step": 4476
+    },
+    {
+      "epoch": 0.3929444226544265,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0023205185930011904,
+      "loss": 3.4019,
+      "step": 4477
+    },
+    {
+      "epoch": 0.39303219223732894,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.002320164378192414,
+      "loss": 3.3262,
+      "step": 4478
+    },
+    {
+      "epoch": 0.39311996182023146,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0023198101021487073,
+      "loss": 3.3223,
+      "step": 4479
+    },
+    {
+      "epoch": 0.3932077314031339,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002319455764902442,
+      "loss": 3.3857,
+      "step": 4480
+    },
+    {
+      "epoch": 0.3932955009860364,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002319101366485994,
+      "loss": 3.2778,
+      "step": 4481
+    },
+    {
+      "epoch": 0.3933832705689389,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0023187469069317463,
+      "loss": 3.2739,
+      "step": 4482
+    },
+    {
+      "epoch": 0.39347104015184137,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0023183923862720875,
+      "loss": 3.3438,
+      "step": 4483
+    },
+    {
+      "epoch": 0.3935588097347439,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002318037804539411,
+      "loss": 3.3135,
+      "step": 4484
+    },
+    {
+      "epoch": 0.39364657931764635,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002317683161766115,
+      "loss": 3.3853,
+      "step": 4485
+    },
+    {
+      "epoch": 0.3937343489005488,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.002317328457984605,
+      "loss": 3.3394,
+      "step": 4486
+    },
+    {
+      "epoch": 0.39382211848345133,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0023169736932272915,
+      "loss": 3.3701,
+      "step": 4487
+    },
+    {
+      "epoch": 0.3939098880663538,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00231661886752659,
+      "loss": 3.4336,
+      "step": 4488
+    },
+    {
+      "epoch": 0.3939976576492563,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0023162639809149225,
+      "loss": 3.3447,
+      "step": 4489
+    },
+    {
+      "epoch": 0.3940854272321588,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0023159090334247162,
+      "loss": 3.3218,
+      "step": 4490
+    },
+    {
+      "epoch": 0.39417319681506124,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0023155540250884028,
+      "loss": 3.3745,
+      "step": 4491
+    },
+    {
+      "epoch": 0.39426096639796376,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002315198955938421,
+      "loss": 3.292,
+      "step": 4492
+    },
+    {
+      "epoch": 0.3943487359808662,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0023148438260072145,
+      "loss": 3.3438,
+      "step": 4493
+    },
+    {
+      "epoch": 0.3944365055637687,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0023144886353272326,
+      "loss": 3.3672,
+      "step": 4494
+    },
+    {
+      "epoch": 0.3945242751466712,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002314133383930931,
+      "loss": 3.3672,
+      "step": 4495
+    },
+    {
+      "epoch": 0.3946120447295737,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.0023137780718507687,
+      "loss": 3.4448,
+      "step": 4496
+    },
+    {
+      "epoch": 0.3946998143124762,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0023134226991192116,
+      "loss": 3.3555,
+      "step": 4497
+    },
+    {
+      "epoch": 0.39478758389537866,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0023130672657687324,
+      "loss": 3.3188,
+      "step": 4498
+    },
+    {
+      "epoch": 0.3948753534782811,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0023127117718318066,
+      "loss": 3.3809,
+      "step": 4499
+    },
+    {
+      "epoch": 0.39496312306118364,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0023123562173409185,
+      "loss": 3.3584,
+      "step": 4500
+    },
+    {
+      "epoch": 0.3950508926440861,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002312000602328555,
+      "loss": 3.4429,
+      "step": 4501
+    },
+    {
+      "epoch": 0.3951386622269886,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00231164492682721,
+      "loss": 3.4233,
+      "step": 4502
+    },
+    {
+      "epoch": 0.3952264318098911,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0023112891908693826,
+      "loss": 3.3335,
+      "step": 4503
+    },
+    {
+      "epoch": 0.39531420139279355,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0023109333944875774,
+      "loss": 3.3159,
+      "step": 4504
+    },
+    {
+      "epoch": 0.39540197097569607,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0023105775377143044,
+      "loss": 3.2944,
+      "step": 4505
+    },
+    {
+      "epoch": 0.39548974055859853,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.00231022162058208,
+      "loss": 3.3022,
+      "step": 4506
+    },
+    {
+      "epoch": 0.395577510141501,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002309865643123425,
+      "loss": 3.3291,
+      "step": 4507
+    },
+    {
+      "epoch": 0.3956652797244035,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0023095096053708662,
+      "loss": 3.3174,
+      "step": 4508
+    },
+    {
+      "epoch": 0.395753049307306,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0023091535073569355,
+      "loss": 3.3525,
+      "step": 4509
+    },
+    {
+      "epoch": 0.3958408188902085,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.002308797349114171,
+      "loss": 3.3516,
+      "step": 4510
+    },
+    {
+      "epoch": 0.39592858847311097,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0023084411306751153,
+      "loss": 3.354,
+      "step": 4511
+    },
+    {
+      "epoch": 0.39601635805601343,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0023080848520723187,
+      "loss": 3.3745,
+      "step": 4512
+    },
+    {
+      "epoch": 0.39610412763891595,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0023077285133383343,
+      "loss": 3.3252,
+      "step": 4513
+    },
+    {
+      "epoch": 0.3961918972218184,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0023073721145057216,
+      "loss": 3.4106,
+      "step": 4514
+    },
+    {
+      "epoch": 0.39627966680472093,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002307015655607046,
+      "loss": 3.3477,
+      "step": 4515
+    },
+    {
+      "epoch": 0.3963674363876234,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0023066591366748784,
+      "loss": 3.3906,
+      "step": 4516
+    },
+    {
+      "epoch": 0.39645520597052586,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0023063025577417955,
+      "loss": 3.3403,
+      "step": 4517
+    },
+    {
+      "epoch": 0.3965429755534284,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002305945918840378,
+      "loss": 3.3535,
+      "step": 4518
+    },
+    {
+      "epoch": 0.39663074513633084,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0023055892200032142,
+      "loss": 3.3867,
+      "step": 4519
+    },
+    {
+      "epoch": 0.3967185147192333,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002305232461262896,
+      "loss": 3.3652,
+      "step": 4520
+    },
+    {
+      "epoch": 0.3968062843021358,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0023048756426520215,
+      "loss": 3.3765,
+      "step": 4521
+    },
+    {
+      "epoch": 0.3968940538850383,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002304518764203194,
+      "loss": 3.4194,
+      "step": 4522
+    },
+    {
+      "epoch": 0.3969818234679408,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0023041618259490235,
+      "loss": 3.3506,
+      "step": 4523
+    },
+    {
+      "epoch": 0.3970695930508433,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0023038048279221237,
+      "loss": 3.3594,
+      "step": 4524
+    },
+    {
+      "epoch": 0.39715736263374574,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0023034477701551156,
+      "loss": 3.3384,
+      "step": 4525
+    },
+    {
+      "epoch": 0.39724513221664826,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0023030906526806233,
+      "loss": 3.311,
+      "step": 4526
+    },
+    {
+      "epoch": 0.3973329017995507,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0023027334755312784,
+      "loss": 3.3115,
+      "step": 4527
+    },
+    {
+      "epoch": 0.39742067138245324,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0023023762387397177,
+      "loss": 3.373,
+      "step": 4528
+    },
+    {
+      "epoch": 0.3975084409653557,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0023020189423385816,
+      "loss": 3.3613,
+      "step": 4529
+    },
+    {
+      "epoch": 0.39759621054825817,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0023016615863605187,
+      "loss": 3.3193,
+      "step": 4530
+    },
+    {
+      "epoch": 0.3976839801311607,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0023013041708381814,
+      "loss": 3.3027,
+      "step": 4531
+    },
+    {
+      "epoch": 0.39777174971406315,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0023009466958042275,
+      "loss": 3.3452,
+      "step": 4532
+    },
+    {
+      "epoch": 0.3978595192969656,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.00230058916129132,
+      "loss": 3.2866,
+      "step": 4533
+    },
+    {
+      "epoch": 0.39794728887986813,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0023002315673321294,
+      "loss": 3.3223,
+      "step": 4534
+    },
+    {
+      "epoch": 0.3980350584627706,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002299873913959329,
+      "loss": 3.3325,
+      "step": 4535
+    },
+    {
+      "epoch": 0.3981228280456731,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002299516201205599,
+      "loss": 3.2871,
+      "step": 4536
+    },
+    {
+      "epoch": 0.3982105976285756,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002299158429103625,
+      "loss": 3.3828,
+      "step": 4537
+    },
+    {
+      "epoch": 0.39829836721147804,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.002298800597686097,
+      "loss": 3.3247,
+      "step": 4538
+    },
+    {
+      "epoch": 0.39838613679438056,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0022984427069857117,
+      "loss": 3.3281,
+      "step": 4539
+    },
+    {
+      "epoch": 0.398473906377283,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00229808475703517,
+      "loss": 3.3892,
+      "step": 4540
+    },
+    {
+      "epoch": 0.39856167596018555,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.00229772674786718,
+      "loss": 3.2778,
+      "step": 4541
+    },
+    {
+      "epoch": 0.398649445543088,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002297368679514453,
+      "loss": 3.2954,
+      "step": 4542
+    },
+    {
+      "epoch": 0.3987372151259905,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0022970105520097074,
+      "loss": 3.373,
+      "step": 4543
+    },
+    {
+      "epoch": 0.398824984708893,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0022966523653856668,
+      "loss": 3.2871,
+      "step": 4544
+    },
+    {
+      "epoch": 0.39891275429179546,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002296294119675058,
+      "loss": 3.2954,
+      "step": 4545
+    },
+    {
+      "epoch": 0.399000523874698,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.0022959358149106167,
+      "loss": 3.3867,
+      "step": 4546
+    },
+    {
+      "epoch": 0.39908829345760044,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0022955774511250818,
+      "loss": 3.2646,
+      "step": 4547
+    },
+    {
+      "epoch": 0.3991760630405029,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002295219028351198,
+      "loss": 3.3501,
+      "step": 4548
+    },
+    {
+      "epoch": 0.3992638326234054,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0022948605466217155,
+      "loss": 3.3545,
+      "step": 4549
+    },
+    {
+      "epoch": 0.3993516022063079,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0022945020059693885,
+      "loss": 3.2827,
+      "step": 4550
+    },
+    {
+      "epoch": 0.39943937178921035,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002294143406426981,
+      "loss": 3.311,
+      "step": 4551
+    },
+    {
+      "epoch": 0.39952714137211287,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.002293784748027257,
+      "loss": 3.3608,
+      "step": 4552
+    },
+    {
+      "epoch": 0.39961491095501533,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0022934260308029885,
+      "loss": 3.3296,
+      "step": 4553
+    },
+    {
+      "epoch": 0.39970268053791785,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002293067254786953,
+      "loss": 3.3413,
+      "step": 4554
+    },
+    {
+      "epoch": 0.3997904501208203,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002292708420011933,
+      "loss": 3.2871,
+      "step": 4555
+    },
+    {
+      "epoch": 0.3998782197037228,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0022923495265107156,
+      "loss": 3.2886,
+      "step": 4556
+    },
+    {
+      "epoch": 0.3999659892866253,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0022919905743160945,
+      "loss": 3.3701,
+      "step": 4557
+    },
+    {
+      "epoch": 0.40005375886952776,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0022916315634608684,
+      "loss": 3.3359,
+      "step": 4558
+    },
+    {
+      "epoch": 0.4001415284524303,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002291272493977841,
+      "loss": 3.3726,
+      "step": 4559
+    },
+    {
+      "epoch": 0.40022929803533275,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0022909133658998216,
+      "loss": 3.3408,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4003170676182352,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.002290554179259625,
+      "loss": 3.3369,
+      "step": 4561
+    },
+    {
+      "epoch": 0.40040483720113773,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0022901949340900702,
+      "loss": 3.3726,
+      "step": 4562
+    },
+    {
+      "epoch": 0.4004926067840402,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0022898356304239835,
+      "loss": 3.4121,
+      "step": 4563
+    },
+    {
+      "epoch": 0.40058037636694266,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002289476268294196,
+      "loss": 3.4072,
+      "step": 4564
+    },
+    {
+      "epoch": 0.4006681459498452,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0022891168477335425,
+      "loss": 3.3931,
+      "step": 4565
+    },
+    {
+      "epoch": 0.40075591553274764,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0022887573687748646,
+      "loss": 3.3403,
+      "step": 4566
+    },
+    {
+      "epoch": 0.40084368511565016,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002288397831451009,
+      "loss": 3.3389,
+      "step": 4567
+    },
+    {
+      "epoch": 0.4009314546985526,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002288038235794828,
+      "loss": 3.2651,
+      "step": 4568
+    },
+    {
+      "epoch": 0.4010192242814551,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0022876785818391793,
+      "loss": 3.334,
+      "step": 4569
+    },
+    {
+      "epoch": 0.4011069938643576,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0022873188696169245,
+      "loss": 3.355,
+      "step": 4570
+    },
+    {
+      "epoch": 0.40119476344726007,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.002286959099160933,
+      "loss": 3.415,
+      "step": 4571
+    },
+    {
+      "epoch": 0.4012825330301626,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.002286599270504077,
+      "loss": 3.354,
+      "step": 4572
+    },
+    {
+      "epoch": 0.40137030261306506,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0022862393836792345,
+      "loss": 3.335,
+      "step": 4573
+    },
+    {
+      "epoch": 0.4014580721959675,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0022858794387192907,
+      "loss": 3.3232,
+      "step": 4574
+    },
+    {
+      "epoch": 0.40154584177887004,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002285519435657135,
+      "loss": 3.2964,
+      "step": 4575
+    },
+    {
+      "epoch": 0.4016336113617725,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002285159374525661,
+      "loss": 3.3589,
+      "step": 4576
+    },
+    {
+      "epoch": 0.40172138094467497,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0022847992553577693,
+      "loss": 3.3379,
+      "step": 4577
+    },
+    {
+      "epoch": 0.4018091505275775,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.002284439078186365,
+      "loss": 3.4121,
+      "step": 4578
+    },
+    {
+      "epoch": 0.40189692011047995,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.002284078843044358,
+      "loss": 3.2832,
+      "step": 4579
+    },
+    {
+      "epoch": 0.40198468969338247,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002283718549964664,
+      "loss": 3.272,
+      "step": 4580
+    },
+    {
+      "epoch": 0.40207245927628493,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0022833581989802057,
+      "loss": 3.3623,
+      "step": 4581
+    },
+    {
+      "epoch": 0.4021602288591874,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0022829977901239074,
+      "loss": 3.3193,
+      "step": 4582
+    },
+    {
+      "epoch": 0.4022479984420899,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0022826373234287015,
+      "loss": 3.3979,
+      "step": 4583
+    },
+    {
+      "epoch": 0.4023357680249924,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0022822767989275254,
+      "loss": 3.3286,
+      "step": 4584
+    },
+    {
+      "epoch": 0.4024235376078949,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0022819162166533204,
+      "loss": 3.4297,
+      "step": 4585
+    },
+    {
+      "epoch": 0.40251130719079736,
+      "grad_norm": 0.375,
+      "learning_rate": 0.002281555576639035,
+      "loss": 3.3101,
+      "step": 4586
+    },
+    {
+      "epoch": 0.4025990767736998,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002281194878917621,
+      "loss": 3.3096,
+      "step": 4587
+    },
+    {
+      "epoch": 0.40268684635660235,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.002280834123522037,
+      "loss": 3.5137,
+      "step": 4588
+    },
+    {
+      "epoch": 0.4027746159395048,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0022804733104852463,
+      "loss": 3.2671,
+      "step": 4589
+    },
+    {
+      "epoch": 0.4028623855224073,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.002280112439840217,
+      "loss": 3.4375,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4029501551053098,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0022797515116199235,
+      "loss": 3.2456,
+      "step": 4591
+    },
+    {
+      "epoch": 0.40303792468821226,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0022793905258573447,
+      "loss": 3.3022,
+      "step": 4592
+    },
+    {
+      "epoch": 0.4031256942711148,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002279029482585465,
+      "loss": 3.2988,
+      "step": 4593
+    },
+    {
+      "epoch": 0.40321346385401724,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.0022786683818372742,
+      "loss": 3.3179,
+      "step": 4594
+    },
+    {
+      "epoch": 0.4033012334369197,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0022783072236457667,
+      "loss": 3.2891,
+      "step": 4595
+    },
+    {
+      "epoch": 0.4033890030198222,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0022779460080439426,
+      "loss": 3.3008,
+      "step": 4596
+    },
+    {
+      "epoch": 0.4034767726027247,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0022775847350648077,
+      "loss": 3.3198,
+      "step": 4597
+    },
+    {
+      "epoch": 0.4035645421856272,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0022772234047413725,
+      "loss": 3.3687,
+      "step": 4598
+    },
+    {
+      "epoch": 0.40365231176852967,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.002276862017106652,
+      "loss": 3.3604,
+      "step": 4599
+    },
+    {
+      "epoch": 0.40374008135143213,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0022765005721936686,
+      "loss": 3.3848,
+      "step": 4600
+    },
+    {
+      "epoch": 0.40382785093433465,
+      "grad_norm": 0.5,
+      "learning_rate": 0.002276139070035448,
+      "loss": 3.4033,
+      "step": 4601
+    },
+    {
+      "epoch": 0.4039156205172371,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0022757775106650208,
+      "loss": 3.3862,
+      "step": 4602
+    },
+    {
+      "epoch": 0.4040033901001396,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.002275415894115425,
+      "loss": 3.4004,
+      "step": 4603
+    },
+    {
+      "epoch": 0.4040911596830421,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002275054220419703,
+      "loss": 3.3438,
+      "step": 4604
+    },
+    {
+      "epoch": 0.40417892926594456,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.002274692489610901,
+      "loss": 3.3506,
+      "step": 4605
+    },
+    {
+      "epoch": 0.4042666988488471,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0022743307017220713,
+      "loss": 3.2793,
+      "step": 4606
+    },
+    {
+      "epoch": 0.40435446843174955,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.002273968856786272,
+      "loss": 3.3032,
+      "step": 4607
+    },
+    {
+      "epoch": 0.404442238014652,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.002273606954836566,
+      "loss": 3.3667,
+      "step": 4608
+    },
+    {
+      "epoch": 0.40453000759755453,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0022732449959060215,
+      "loss": 3.3608,
+      "step": 4609
+    },
+    {
+      "epoch": 0.404617777180457,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0022728829800277117,
+      "loss": 3.2949,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4047055467633595,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0022725209072347145,
+      "loss": 3.3237,
+      "step": 4611
+    },
+    {
+      "epoch": 0.404793316346262,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.002272158777560114,
+      "loss": 3.3213,
+      "step": 4612
+    },
+    {
+      "epoch": 0.40488108592916444,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0022717965910369993,
+      "loss": 3.3433,
+      "step": 4613
+    },
+    {
+      "epoch": 0.40496885551206696,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0022714343476984643,
+      "loss": 3.3169,
+      "step": 4614
+    },
+    {
+      "epoch": 0.4050566250949694,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002271072047577608,
+      "loss": 3.3921,
+      "step": 4615
+    },
+    {
+      "epoch": 0.40514439467787194,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002270709690707535,
+      "loss": 3.3228,
+      "step": 4616
+    },
+    {
+      "epoch": 0.4052321642607744,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0022703472771213548,
+      "loss": 3.311,
+      "step": 4617
+    },
+    {
+      "epoch": 0.40531993384367687,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0022699848068521832,
+      "loss": 3.4258,
+      "step": 4618
+    },
+    {
+      "epoch": 0.4054077034265794,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002269622279933139,
+      "loss": 3.3135,
+      "step": 4619
+    },
+    {
+      "epoch": 0.40549547300948185,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0022692596963973474,
+      "loss": 3.3184,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4055832425923843,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0022688970562779387,
+      "loss": 3.3994,
+      "step": 4621
+    },
+    {
+      "epoch": 0.40567101217528684,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.00226853435960805,
+      "loss": 3.3335,
+      "step": 4622
+    },
+    {
+      "epoch": 0.4057587817581893,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.00226817160642082,
+      "loss": 3.2886,
+      "step": 4623
+    },
+    {
+      "epoch": 0.4058465513410918,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002267808796749396,
+      "loss": 3.2661,
+      "step": 4624
+    },
+    {
+      "epoch": 0.4059343209239943,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002267445930626928,
+      "loss": 3.3291,
+      "step": 4625
+    },
+    {
+      "epoch": 0.40602209050689675,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0022670830080865727,
+      "loss": 3.4385,
+      "step": 4626
+    },
+    {
+      "epoch": 0.40610986008979927,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0022667200291614908,
+      "loss": 3.3477,
+      "step": 4627
+    },
+    {
+      "epoch": 0.40619762967270173,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.00226635699388485,
+      "loss": 3.3198,
+      "step": 4628
+    },
+    {
+      "epoch": 0.40628539925560425,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0022659939022898206,
+      "loss": 3.2563,
+      "step": 4629
+    },
+    {
+      "epoch": 0.4063731688385067,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.00226563075440958,
+      "loss": 3.3589,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4064609384214092,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0022652675502773103,
+      "loss": 3.2646,
+      "step": 4631
+    },
+    {
+      "epoch": 0.4065487080043117,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0022649042899261975,
+      "loss": 3.4365,
+      "step": 4632
+    },
+    {
+      "epoch": 0.40663647758721416,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0022645409733894352,
+      "loss": 3.3403,
+      "step": 4633
+    },
+    {
+      "epoch": 0.4067242471701166,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.00226417760070022,
+      "loss": 3.3853,
+      "step": 4634
+    },
+    {
+      "epoch": 0.40681201675301915,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002263814171891755,
+      "loss": 3.3472,
+      "step": 4635
+    },
+    {
+      "epoch": 0.4068997863359216,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0022634506869972466,
+      "loss": 3.3726,
+      "step": 4636
+    },
+    {
+      "epoch": 0.40698755591882413,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.002263087146049908,
+      "loss": 3.2368,
+      "step": 4637
+    },
+    {
+      "epoch": 0.4070753255017266,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.002262723549082958,
+      "loss": 3.3188,
+      "step": 4638
+    },
+    {
+      "epoch": 0.40716309508462906,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002262359896129618,
+      "loss": 3.3892,
+      "step": 4639
+    },
+    {
+      "epoch": 0.4072508646675316,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0022619961872231164,
+      "loss": 3.4131,
+      "step": 4640
+    },
+    {
+      "epoch": 0.40733863425043404,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0022616324223966875,
+      "loss": 3.2749,
+      "step": 4641
+    },
+    {
+      "epoch": 0.40742640383333656,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.0022612686016835687,
+      "loss": 3.3184,
+      "step": 4642
+    },
+    {
+      "epoch": 0.407514173416239,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002260904725117003,
+      "loss": 3.2749,
+      "step": 4643
+    },
+    {
+      "epoch": 0.4076019429991415,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00226054079273024,
+      "loss": 3.3867,
+      "step": 4644
+    },
+    {
+      "epoch": 0.407689712582044,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.002260176804556533,
+      "loss": 3.2988,
+      "step": 4645
+    },
+    {
+      "epoch": 0.40777748216494647,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0022598127606291395,
+      "loss": 3.2969,
+      "step": 4646
+    },
+    {
+      "epoch": 0.40786525174784893,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.002259448660981325,
+      "loss": 3.3315,
+      "step": 4647
+    },
+    {
+      "epoch": 0.40795302133075145,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0022590845056463566,
+      "loss": 3.2939,
+      "step": 4648
+    },
+    {
+      "epoch": 0.4080407909136539,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00225872029465751,
+      "loss": 3.3403,
+      "step": 4649
+    },
+    {
+      "epoch": 0.40812856049655644,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0022583560280480634,
+      "loss": 3.3569,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4082163300794589,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.0022579917058513006,
+      "loss": 3.269,
+      "step": 4651
+    },
+    {
+      "epoch": 0.40830409966236136,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.002257627328100512,
+      "loss": 3.3193,
+      "step": 4652
+    },
+    {
+      "epoch": 0.4083918692452639,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0022572628948289903,
+      "loss": 3.3687,
+      "step": 4653
+    },
+    {
+      "epoch": 0.40847963882816635,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002256898406070036,
+      "loss": 3.3491,
+      "step": 4654
+    },
+    {
+      "epoch": 0.40856740841106887,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0022565338618569532,
+      "loss": 3.3164,
+      "step": 4655
+    },
+    {
+      "epoch": 0.40865517799397133,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002256169262223051,
+      "loss": 3.2959,
+      "step": 4656
+    },
+    {
+      "epoch": 0.4087429475768738,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0022558046072016454,
+      "loss": 3.3647,
+      "step": 4657
+    },
+    {
+      "epoch": 0.4088307171597763,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0022554398968260545,
+      "loss": 3.2627,
+      "step": 4658
+    },
+    {
+      "epoch": 0.4089184867426788,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0022550751311296033,
+      "loss": 3.2749,
+      "step": 4659
+    },
+    {
+      "epoch": 0.40900625632558124,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0022547103101456215,
+      "loss": 3.3511,
+      "step": 4660
+    },
+    {
+      "epoch": 0.40909402590848376,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0022543454339074443,
+      "loss": 3.3345,
+      "step": 4661
+    },
+    {
+      "epoch": 0.4091817954913862,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0022539805024484115,
+      "loss": 3.3804,
+      "step": 4662
+    },
+    {
+      "epoch": 0.40926956507428874,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002253615515801868,
+      "loss": 3.4014,
+      "step": 4663
+    },
+    {
+      "epoch": 0.4093573346571912,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0022532504740011635,
+      "loss": 3.3018,
+      "step": 4664
+    },
+    {
+      "epoch": 0.40944510424009367,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0022528853770796523,
+      "loss": 3.3623,
+      "step": 4665
+    },
+    {
+      "epoch": 0.4095328738229962,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0022525202250706956,
+      "loss": 3.4053,
+      "step": 4666
+    },
+    {
+      "epoch": 0.40962064340589865,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.002252155018007658,
+      "loss": 3.3872,
+      "step": 4667
+    },
+    {
+      "epoch": 0.4097084129888012,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0022517897559239095,
+      "loss": 3.3174,
+      "step": 4668
+    },
+    {
+      "epoch": 0.40979618257170364,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002251424438852825,
+      "loss": 3.3115,
+      "step": 4669
+    },
+    {
+      "epoch": 0.4098839521546061,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0022510590668277848,
+      "loss": 3.3203,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4099717217375086,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0022506936398821744,
+      "loss": 3.3789,
+      "step": 4671
+    },
+    {
+      "epoch": 0.4100594913204111,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0022503281580493833,
+      "loss": 3.3828,
+      "step": 4672
+    },
+    {
+      "epoch": 0.41014726090331355,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.002249962621362807,
+      "loss": 3.3833,
+      "step": 4673
+    },
+    {
+      "epoch": 0.41023503048621607,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.002249597029855846,
+      "loss": 3.2939,
+      "step": 4674
+    },
+    {
+      "epoch": 0.41032280006911853,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002249231383561905,
+      "loss": 3.3115,
+      "step": 4675
+    },
+    {
+      "epoch": 0.41041056965202105,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002248865682514394,
+      "loss": 3.3027,
+      "step": 4676
+    },
+    {
+      "epoch": 0.4104983392349235,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002248499926746729,
+      "loss": 3.4053,
+      "step": 4677
+    },
+    {
+      "epoch": 0.410586108817826,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0022481341162923297,
+      "loss": 3.3218,
+      "step": 4678
+    },
+    {
+      "epoch": 0.4106738784007285,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002247768251184621,
+      "loss": 3.3184,
+      "step": 4679
+    },
+    {
+      "epoch": 0.41076164798363096,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0022474023314570336,
+      "loss": 3.2681,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4108494175665335,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.002247036357143003,
+      "loss": 3.3047,
+      "step": 4681
+    },
+    {
+      "epoch": 0.41093718714943595,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0022466703282759686,
+      "loss": 3.3867,
+      "step": 4682
+    },
+    {
+      "epoch": 0.4110249567323384,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0022463042448893755,
+      "loss": 3.4082,
+      "step": 4683
+    },
+    {
+      "epoch": 0.41111272631524093,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.002245938107016675,
+      "loss": 3.3325,
+      "step": 4684
+    },
+    {
+      "epoch": 0.4112004958981434,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0022455719146913205,
+      "loss": 3.3252,
+      "step": 4685
+    },
+    {
+      "epoch": 0.4112882654810459,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0022452056679467735,
+      "loss": 3.375,
+      "step": 4686
+    },
+    {
+      "epoch": 0.4113760350639484,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0022448393668164984,
+      "loss": 3.3174,
+      "step": 4687
+    },
+    {
+      "epoch": 0.41146380464685084,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002244473011333966,
+      "loss": 3.2759,
+      "step": 4688
+    },
+    {
+      "epoch": 0.41155157422975336,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0022441066015326497,
+      "loss": 3.373,
+      "step": 4689
+    },
+    {
+      "epoch": 0.4116393438126558,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0022437401374460313,
+      "loss": 3.2041,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4117271133955583,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.002243373619107595,
+      "loss": 3.2778,
+      "step": 4691
+    },
+    {
+      "epoch": 0.4118148829784608,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00224300704655083,
+      "loss": 3.2974,
+      "step": 4692
+    },
+    {
+      "epoch": 0.41190265256136327,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0022426404198092323,
+      "loss": 3.3057,
+      "step": 4693
+    },
+    {
+      "epoch": 0.4119904221442658,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0022422737389163005,
+      "loss": 3.2915,
+      "step": 4694
+    },
+    {
+      "epoch": 0.41207819172716825,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00224190700390554,
+      "loss": 3.3433,
+      "step": 4695
+    },
+    {
+      "epoch": 0.4121659613100707,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.002241540214810461,
+      "loss": 3.2544,
+      "step": 4696
+    },
+    {
+      "epoch": 0.41225373089297324,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0022411733716645767,
+      "loss": 3.3574,
+      "step": 4697
+    },
+    {
+      "epoch": 0.4123415004758757,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.002240806474501408,
+      "loss": 3.3379,
+      "step": 4698
+    },
+    {
+      "epoch": 0.4124292700587782,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0022404395233544786,
+      "loss": 3.3501,
+      "step": 4699
+    },
+    {
+      "epoch": 0.4125170396416807,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0022400725182573178,
+      "loss": 3.333,
+      "step": 4700
+    },
+    {
+      "epoch": 0.41260480922458315,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0022397054592434612,
+      "loss": 3.3853,
+      "step": 4701
+    },
+    {
+      "epoch": 0.41269257880748567,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.002239338346346447,
+      "loss": 3.3428,
+      "step": 4702
+    },
+    {
+      "epoch": 0.41278034839038813,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.00223897117959982,
+      "loss": 3.3042,
+      "step": 4703
+    },
+    {
+      "epoch": 0.4128681179732906,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0022386039590371284,
+      "loss": 3.3477,
+      "step": 4704
+    },
+    {
+      "epoch": 0.4129558875561931,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.002238236684691927,
+      "loss": 3.3145,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4130436571390956,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0022378693565977747,
+      "loss": 3.3447,
+      "step": 4706
+    },
+    {
+      "epoch": 0.4131314267219981,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.002237501974788235,
+      "loss": 3.3184,
+      "step": 4707
+    },
+    {
+      "epoch": 0.41321919630490056,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0022371345392968776,
+      "loss": 3.312,
+      "step": 4708
+    },
+    {
+      "epoch": 0.413306965887803,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.002236767050157275,
+      "loss": 3.3062,
+      "step": 4709
+    },
+    {
+      "epoch": 0.41339473547070554,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002236399507403006,
+      "loss": 3.3198,
+      "step": 4710
+    },
+    {
+      "epoch": 0.413482505053608,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0022360319110676548,
+      "loss": 3.354,
+      "step": 4711
+    },
+    {
+      "epoch": 0.4135702746365105,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0022356642611848095,
+      "loss": 3.2603,
+      "step": 4712
+    },
+    {
+      "epoch": 0.413658044219413,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.002235296557788063,
+      "loss": 3.2437,
+      "step": 4713
+    },
+    {
+      "epoch": 0.41374581380231545,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0022349288009110143,
+      "loss": 3.3535,
+      "step": 4714
+    },
+    {
+      "epoch": 0.413833583385218,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002234560990587265,
+      "loss": 3.2402,
+      "step": 4715
+    },
+    {
+      "epoch": 0.41392135296812044,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 0.002234193126850424,
+      "loss": 3.2637,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4140091225510229,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.0022338252097341043,
+      "loss": 3.356,
+      "step": 4717
+    },
+    {
+      "epoch": 0.4140968921339254,
+      "grad_norm": 0.125,
+      "learning_rate": 0.002233457239271923,
+      "loss": 3.3726,
+      "step": 4718
+    },
+    {
+      "epoch": 0.4141846617168279,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0022330892154975033,
+      "loss": 3.3828,
+      "step": 4719
+    },
+    {
+      "epoch": 0.4142724312997304,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002232721138444472,
+      "loss": 3.3062,
+      "step": 4720
+    },
+    {
+      "epoch": 0.41436020088263287,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.002232353008146462,
+      "loss": 3.3193,
+      "step": 4721
+    },
+    {
+      "epoch": 0.41444797046553533,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.00223198482463711,
+      "loss": 3.3174,
+      "step": 4722
+    },
+    {
+      "epoch": 0.41453574004843785,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0022316165879500585,
+      "loss": 3.3306,
+      "step": 4723
+    },
+    {
+      "epoch": 0.4146235096313403,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0022312482981189535,
+      "loss": 3.2822,
+      "step": 4724
+    },
+    {
+      "epoch": 0.41471127921424283,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0022308799551774474,
+      "loss": 3.3394,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4147990487971453,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002230511559159197,
+      "loss": 3.2871,
+      "step": 4726
+    },
+    {
+      "epoch": 0.41488681838004776,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002230143110097864,
+      "loss": 3.3057,
+      "step": 4727
+    },
+    {
+      "epoch": 0.4149745879629503,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002229774608027114,
+      "loss": 3.3096,
+      "step": 4728
+    },
+    {
+      "epoch": 0.41506235754585274,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0022294060529806175,
+      "loss": 3.3052,
+      "step": 4729
+    },
+    {
+      "epoch": 0.4151501271287552,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0022290374449920516,
+      "loss": 3.3149,
+      "step": 4730
+    },
+    {
+      "epoch": 0.41523789671165773,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002228668784095097,
+      "loss": 3.3066,
+      "step": 4731
+    },
+    {
+      "epoch": 0.4153256662945602,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0022283000703234397,
+      "loss": 3.3486,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4154134358774627,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0022279313037107686,
+      "loss": 3.3452,
+      "step": 4733
+    },
+    {
+      "epoch": 0.4155012054603652,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002227562484290781,
+      "loss": 3.3154,
+      "step": 4734
+    },
+    {
+      "epoch": 0.41558897504326764,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0022271936120971756,
+      "loss": 3.3057,
+      "step": 4735
+    },
+    {
+      "epoch": 0.41567674462617016,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0022268246871636577,
+      "loss": 3.3442,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4157645142090726,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0022264557095239374,
+      "loss": 3.3521,
+      "step": 4737
+    },
+    {
+      "epoch": 0.41585228379197514,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0022260866792117298,
+      "loss": 3.293,
+      "step": 4738
+    },
+    {
+      "epoch": 0.4159400533748776,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.002225717596260753,
+      "loss": 3.3296,
+      "step": 4739
+    },
+    {
+      "epoch": 0.41602782295778007,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002225348460704732,
+      "loss": 3.3296,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4161155925406826,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0022249792725773953,
+      "loss": 3.4136,
+      "step": 4741
+    },
+    {
+      "epoch": 0.41620336212358505,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002224610031912478,
+      "loss": 3.2969,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4162911317064875,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002224240738743717,
+      "loss": 3.3872,
+      "step": 4743
+    },
+    {
+      "epoch": 0.41637890128939004,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0022238713931048574,
+      "loss": 3.3057,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4164666708722925,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0022235019950296456,
+      "loss": 3.334,
+      "step": 4745
+    },
+    {
+      "epoch": 0.416554440455195,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0022231325445518364,
+      "loss": 3.3521,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4166422100380975,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002222763041705186,
+      "loss": 3.4282,
+      "step": 4747
+    },
+    {
+      "epoch": 0.41672997962099995,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0022223934865234587,
+      "loss": 3.2725,
+      "step": 4748
+    },
+    {
+      "epoch": 0.41681774920390247,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0022220238790404204,
+      "loss": 3.3208,
+      "step": 4749
+    },
+    {
+      "epoch": 0.41690551878680493,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.002221654219289844,
+      "loss": 3.3374,
+      "step": 4750
+    },
+    {
+      "epoch": 0.41699328836970745,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002221284507305506,
+      "loss": 3.395,
+      "step": 4751
+    },
+    {
+      "epoch": 0.4170810579526099,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0022209147431211883,
+      "loss": 3.3105,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4171688275355124,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0022205449267706776,
+      "loss": 3.3604,
+      "step": 4753
+    },
+    {
+      "epoch": 0.4172565971184149,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.002220175058287765,
+      "loss": 3.3242,
+      "step": 4754
+    },
+    {
+      "epoch": 0.41734436670131736,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.002219805137706246,
+      "loss": 3.3252,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4174321362842199,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.002219435165059922,
+      "loss": 3.3628,
+      "step": 4756
+    },
+    {
+      "epoch": 0.41751990586712234,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002219065140382598,
+      "loss": 3.2817,
+      "step": 4757
+    },
+    {
+      "epoch": 0.4176076754500248,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0022186950637080845,
+      "loss": 3.3306,
+      "step": 4758
+    },
+    {
+      "epoch": 0.4176954450329273,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002218324935070197,
+      "loss": 3.3765,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4177832146158298,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0022179547545027544,
+      "loss": 3.3755,
+      "step": 4760
+    },
+    {
+      "epoch": 0.41787098419873225,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002217584522039582,
+      "loss": 3.3457,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4179587537816348,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0022172142377145086,
+      "loss": 3.3701,
+      "step": 4762
+    },
+    {
+      "epoch": 0.41804652336453724,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002216843901561368,
+      "loss": 3.2603,
+      "step": 4763
+    },
+    {
+      "epoch": 0.41813429294743976,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.002216473513613999,
+      "loss": 3.2764,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4182220625303422,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.0022161030739062454,
+      "loss": 3.3379,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4183098321132447,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0022157325824719554,
+      "loss": 3.3008,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4183976016961472,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002215362039344982,
+      "loss": 3.3086,
+      "step": 4767
+    },
+    {
+      "epoch": 0.41848537127904967,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002214991444559182,
+      "loss": 3.3433,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4185731408619522,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0022146207981484185,
+      "loss": 3.2715,
+      "step": 4769
+    },
+    {
+      "epoch": 0.41866091044485465,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0022142501001465582,
+      "loss": 3.3218,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4187486800277571,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0022138793505874733,
+      "loss": 3.2959,
+      "step": 4771
+    },
+    {
+      "epoch": 0.41883644961065963,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00221350854950504,
+      "loss": 3.2881,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4189242191935621,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0022131376969331397,
+      "loss": 3.2173,
+      "step": 4773
+    },
+    {
+      "epoch": 0.41901198877646456,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0022127667929056584,
+      "loss": 3.2529,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4190997583593671,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0022123958374564863,
+      "loss": 3.3042,
+      "step": 4775
+    },
+    {
+      "epoch": 0.41918752794226954,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.002212024830619519,
+      "loss": 3.4087,
+      "step": 4776
+    },
+    {
+      "epoch": 0.41927529752517206,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0022116537724286562,
+      "loss": 3.335,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4193630671080745,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0022112826629178033,
+      "loss": 3.3125,
+      "step": 4778
+    },
+    {
+      "epoch": 0.419450836690977,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0022109115021208695,
+      "loss": 3.3843,
+      "step": 4779
+    },
+    {
+      "epoch": 0.4195386062738795,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.002210540290071768,
+      "loss": 3.3306,
+      "step": 4780
+    },
+    {
+      "epoch": 0.419626375856782,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002210169026804418,
+      "loss": 3.2671,
+      "step": 4781
+    },
+    {
+      "epoch": 0.4197141454396845,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.002209797712352744,
+      "loss": 3.2944,
+      "step": 4782
+    },
+    {
+      "epoch": 0.41980191502258696,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002209426346750672,
+      "loss": 3.3379,
+      "step": 4783
+    },
+    {
+      "epoch": 0.4198896846054894,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002209054930032137,
+      "loss": 3.4219,
+      "step": 4784
+    },
+    {
+      "epoch": 0.41997745418839194,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.002208683462231075,
+      "loss": 3.2583,
+      "step": 4785
+    },
+    {
+      "epoch": 0.4200652237712944,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.002208311943381429,
+      "loss": 3.2749,
+      "step": 4786
+    },
+    {
+      "epoch": 0.42015299335419687,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002207940373517145,
+      "loss": 3.2505,
+      "step": 4787
+    },
+    {
+      "epoch": 0.4202407629370994,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0022075687526721746,
+      "loss": 3.3506,
+      "step": 4788
+    },
+    {
+      "epoch": 0.42032853252000185,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0022071970808804748,
+      "loss": 3.3442,
+      "step": 4789
+    },
+    {
+      "epoch": 0.42041630210290437,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0022068253581760053,
+      "loss": 3.3379,
+      "step": 4790
+    },
+    {
+      "epoch": 0.42050407168580684,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0022064535845927324,
+      "loss": 3.3491,
+      "step": 4791
+    },
+    {
+      "epoch": 0.4205918412687093,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0022060817601646256,
+      "loss": 3.3071,
+      "step": 4792
+    },
+    {
+      "epoch": 0.4206796108516118,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0022057098849256594,
+      "loss": 3.3271,
+      "step": 4793
+    },
+    {
+      "epoch": 0.4207673804345143,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.002205337958909814,
+      "loss": 3.3413,
+      "step": 4794
+    },
+    {
+      "epoch": 0.4208551500174168,
+      "grad_norm": 0.216796875,
+      "learning_rate": 0.0022049659821510724,
+      "loss": 3.2954,
+      "step": 4795
+    },
+    {
+      "epoch": 0.42094291960031927,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0022045939546834242,
+      "loss": 3.3916,
+      "step": 4796
+    },
+    {
+      "epoch": 0.42103068918322173,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0022042218765408617,
+      "loss": 3.2812,
+      "step": 4797
+    },
+    {
+      "epoch": 0.42111845876612425,
+      "grad_norm": 0.07861328125,
+      "learning_rate": 0.0022038497477573833,
+      "loss": 3.2314,
+      "step": 4798
+    },
+    {
+      "epoch": 0.4212062283490267,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0022034775683669916,
+      "loss": 3.3184,
+      "step": 4799
+    },
+    {
+      "epoch": 0.4212939979319292,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002203105338403694,
+      "loss": 3.228,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4213817675148317,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0022027330579015014,
+      "loss": 3.3872,
+      "step": 4801
+    },
+    {
+      "epoch": 0.42146953709773416,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.002202360726894431,
+      "loss": 3.291,
+      "step": 4802
+    },
+    {
+      "epoch": 0.4215573066806367,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0022019883454165033,
+      "loss": 3.2803,
+      "step": 4803
+    },
+    {
+      "epoch": 0.42164507626353914,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.002201615913501744,
+      "loss": 3.2681,
+      "step": 4804
+    },
+    {
+      "epoch": 0.4217328458464416,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0022012434311841837,
+      "loss": 3.3633,
+      "step": 4805
+    },
+    {
+      "epoch": 0.4218206154293441,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.002200870898497857,
+      "loss": 3.4048,
+      "step": 4806
+    },
+    {
+      "epoch": 0.4219083850122466,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0022004983154768035,
+      "loss": 3.2896,
+      "step": 4807
+    },
+    {
+      "epoch": 0.4219961545951491,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.0022001256821550666,
+      "loss": 3.2896,
+      "step": 4808
+    },
+    {
+      "epoch": 0.4220839241780516,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0021997529985666956,
+      "loss": 3.3496,
+      "step": 4809
+    },
+    {
+      "epoch": 0.42217169376095404,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0021993802647457435,
+      "loss": 3.3652,
+      "step": 4810
+    },
+    {
+      "epoch": 0.42225946334385656,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0021990074807262676,
+      "loss": 3.3159,
+      "step": 4811
+    },
+    {
+      "epoch": 0.422347232926759,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0021986346465423315,
+      "loss": 3.2769,
+      "step": 4812
+    },
+    {
+      "epoch": 0.4224350025096615,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002198261762228001,
+      "loss": 3.3193,
+      "step": 4813
+    },
+    {
+      "epoch": 0.422522772092564,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002197888827817348,
+      "loss": 3.3843,
+      "step": 4814
+    },
+    {
+      "epoch": 0.42261054167546647,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0021975158433444487,
+      "loss": 3.3066,
+      "step": 4815
+    },
+    {
+      "epoch": 0.422698311258369,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002197142808843384,
+      "loss": 3.3862,
+      "step": 4816
+    },
+    {
+      "epoch": 0.42278608084127145,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0021967697243482393,
+      "loss": 3.3975,
+      "step": 4817
+    },
+    {
+      "epoch": 0.4228738504241739,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.002196396589893104,
+      "loss": 3.3716,
+      "step": 4818
+    },
+    {
+      "epoch": 0.42296162000707643,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0021960234055120728,
+      "loss": 3.2939,
+      "step": 4819
+    },
+    {
+      "epoch": 0.4230493895899789,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002195650171239245,
+      "loss": 3.2964,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4231371591728814,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0021952768871087228,
+      "loss": 3.355,
+      "step": 4821
+    },
+    {
+      "epoch": 0.4232249287557839,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0021949035531546156,
+      "loss": 3.3931,
+      "step": 4822
+    },
+    {
+      "epoch": 0.42331269833868634,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0021945301694110356,
+      "loss": 3.3169,
+      "step": 4823
+    },
+    {
+      "epoch": 0.42340046792158886,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0021941567359121,
+      "loss": 3.3442,
+      "step": 4824
+    },
+    {
+      "epoch": 0.4234882375044913,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0021937832526919307,
+      "loss": 3.3794,
+      "step": 4825
+    },
+    {
+      "epoch": 0.42357600708739385,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0021934097197846537,
+      "loss": 3.332,
+      "step": 4826
+    },
+    {
+      "epoch": 0.4236637766702963,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0021930361372244,
+      "loss": 3.292,
+      "step": 4827
+    },
+    {
+      "epoch": 0.4237515462531988,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0021926625050453052,
+      "loss": 3.3198,
+      "step": 4828
+    },
+    {
+      "epoch": 0.4238393158361013,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.002192288823281509,
+      "loss": 3.3198,
+      "step": 4829
+    },
+    {
+      "epoch": 0.42392708541900376,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0021919150919671553,
+      "loss": 3.3276,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4240148550019062,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.002191541311136394,
+      "loss": 3.3325,
+      "step": 4831
+    },
+    {
+      "epoch": 0.42410262458480874,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002191167480823378,
+      "loss": 3.2617,
+      "step": 4832
+    },
+    {
+      "epoch": 0.4241903941677112,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0021907936010622645,
+      "loss": 3.2935,
+      "step": 4833
+    },
+    {
+      "epoch": 0.4242781637506137,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002190419671887218,
+      "loss": 3.3174,
+      "step": 4834
+    },
+    {
+      "epoch": 0.4243659333335162,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002190045693332404,
+      "loss": 3.3657,
+      "step": 4835
+    },
+    {
+      "epoch": 0.42445370291641865,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002189671665431995,
+      "loss": 3.2988,
+      "step": 4836
+    },
+    {
+      "epoch": 0.42454147249932117,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0021892975882201655,
+      "loss": 3.2075,
+      "step": 4837
+    },
+    {
+      "epoch": 0.42462924208222363,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.002188923461731098,
+      "loss": 3.3159,
+      "step": 4838
+    },
+    {
+      "epoch": 0.42471701166512615,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.002188549285998976,
+      "loss": 3.3433,
+      "step": 4839
+    },
+    {
+      "epoch": 0.4248047812480286,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.00218817506105799,
+      "loss": 3.3652,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4248925508309311,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.002187800786942334,
+      "loss": 3.2744,
+      "step": 4841
+    },
+    {
+      "epoch": 0.4249803204138336,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.002187426463686206,
+      "loss": 3.2822,
+      "step": 4842
+    },
+    {
+      "epoch": 0.42506808999673606,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0021870520913238093,
+      "loss": 3.3203,
+      "step": 4843
+    },
+    {
+      "epoch": 0.42515585957963853,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0021866776698893516,
+      "loss": 3.3398,
+      "step": 4844
+    },
+    {
+      "epoch": 0.42524362916254105,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.002186303199417045,
+      "loss": 3.3789,
+      "step": 4845
+    },
+    {
+      "epoch": 0.4253313987454435,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0021859286799411058,
+      "loss": 3.356,
+      "step": 4846
+    },
+    {
+      "epoch": 0.42541916832834603,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0021855541114957546,
+      "loss": 3.374,
+      "step": 4847
+    },
+    {
+      "epoch": 0.4255069379112485,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0021851794941152173,
+      "loss": 3.3892,
+      "step": 4848
+    },
+    {
+      "epoch": 0.42559470749415096,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0021848048278337237,
+      "loss": 3.292,
+      "step": 4849
+    },
+    {
+      "epoch": 0.4256824770770535,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.002184430112685508,
+      "loss": 3.3335,
+      "step": 4850
+    },
+    {
+      "epoch": 0.42577024665995594,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0021840553487048092,
+      "loss": 3.3008,
+      "step": 4851
+    },
+    {
+      "epoch": 0.42585801624285846,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002183680535925871,
+      "loss": 3.3428,
+      "step": 4852
+    },
+    {
+      "epoch": 0.4259457858257609,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0021833056743829405,
+      "loss": 3.3545,
+      "step": 4853
+    },
+    {
+      "epoch": 0.4260335554086634,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0021829307641102703,
+      "loss": 3.3135,
+      "step": 4854
+    },
+    {
+      "epoch": 0.4261213249915659,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.002182555805142117,
+      "loss": 3.3213,
+      "step": 4855
+    },
+    {
+      "epoch": 0.4262090945744684,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0021821807975127418,
+      "loss": 3.3599,
+      "step": 4856
+    },
+    {
+      "epoch": 0.42629686415737084,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0021818057412564106,
+      "loss": 3.3408,
+      "step": 4857
+    },
+    {
+      "epoch": 0.42638463374027336,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0021814306364073923,
+      "loss": 3.4087,
+      "step": 4858
+    },
+    {
+      "epoch": 0.4264724033231758,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.002181055482999963,
+      "loss": 3.3975,
+      "step": 4859
+    },
+    {
+      "epoch": 0.42656017290607834,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0021806802810684,
+      "loss": 3.416,
+      "step": 4860
+    },
+    {
+      "epoch": 0.4266479424889808,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.002180305030646987,
+      "loss": 3.373,
+      "step": 4861
+    },
+    {
+      "epoch": 0.42673571207188327,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.002179929731770013,
+      "loss": 3.3706,
+      "step": 4862
+    },
+    {
+      "epoch": 0.4268234816547858,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0021795543844717684,
+      "loss": 3.3618,
+      "step": 4863
+    },
+    {
+      "epoch": 0.42691125123768825,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0021791789887865513,
+      "loss": 3.3535,
+      "step": 4864
+    },
+    {
+      "epoch": 0.42699902082059077,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.0021788035447486615,
+      "loss": 3.3423,
+      "step": 4865
+    },
+    {
+      "epoch": 0.42708679040349323,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002178428052392405,
+      "loss": 3.3433,
+      "step": 4866
+    },
+    {
+      "epoch": 0.4271745599863957,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002178052511752092,
+      "loss": 3.3525,
+      "step": 4867
+    },
+    {
+      "epoch": 0.4272623295692982,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0021776769228620363,
+      "loss": 3.2993,
+      "step": 4868
+    },
+    {
+      "epoch": 0.4273500991522007,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002177301285756557,
+      "loss": 3.3008,
+      "step": 4869
+    },
+    {
+      "epoch": 0.42743786873510314,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002176925600469977,
+      "loss": 3.3188,
+      "step": 4870
+    },
+    {
+      "epoch": 0.42752563831800566,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0021765498670366233,
+      "loss": 3.3242,
+      "step": 4871
+    },
+    {
+      "epoch": 0.4276134079009081,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0021761740854908287,
+      "loss": 3.4053,
+      "step": 4872
+    },
+    {
+      "epoch": 0.42770117748381065,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0021757982558669284,
+      "loss": 3.2754,
+      "step": 4873
+    },
+    {
+      "epoch": 0.4277889470667131,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0021754223781992638,
+      "loss": 3.3032,
+      "step": 4874
+    },
+    {
+      "epoch": 0.4278767166496156,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0021750464525221807,
+      "loss": 3.3491,
+      "step": 4875
+    },
+    {
+      "epoch": 0.4279644862325181,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0021746704788700266,
+      "loss": 3.3481,
+      "step": 4876
+    },
+    {
+      "epoch": 0.42805225581542056,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0021742944572771573,
+      "loss": 3.3042,
+      "step": 4877
+    },
+    {
+      "epoch": 0.4281400253983231,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.00217391838777793,
+      "loss": 3.3369,
+      "step": 4878
+    },
+    {
+      "epoch": 0.42822779498122554,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.002173542270406707,
+      "loss": 3.2891,
+      "step": 4879
+    },
+    {
+      "epoch": 0.428315564564128,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0021731661051978563,
+      "loss": 3.2925,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4284033341470305,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002172789892185748,
+      "loss": 3.3525,
+      "step": 4881
+    },
+    {
+      "epoch": 0.428491103729933,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0021724136314047594,
+      "loss": 3.2822,
+      "step": 4882
+    },
+    {
+      "epoch": 0.42857887331283545,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.002172037322889269,
+      "loss": 3.396,
+      "step": 4883
+    },
+    {
+      "epoch": 0.42866664289573797,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0021716609666736627,
+      "loss": 3.3379,
+      "step": 4884
+    },
+    {
+      "epoch": 0.42875441247864043,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0021712845627923274,
+      "loss": 3.2925,
+      "step": 4885
+    },
+    {
+      "epoch": 0.42884218206154295,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.002170908111279658,
+      "loss": 3.3389,
+      "step": 4886
+    },
+    {
+      "epoch": 0.4289299516444454,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0021705316121700515,
+      "loss": 3.333,
+      "step": 4887
+    },
+    {
+      "epoch": 0.4290177212273479,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0021701550654979096,
+      "loss": 3.3218,
+      "step": 4888
+    },
+    {
+      "epoch": 0.4291054908102504,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0021697784712976383,
+      "loss": 3.3452,
+      "step": 4889
+    },
+    {
+      "epoch": 0.42919326039315286,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0021694018296036484,
+      "loss": 3.3374,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4292810299760554,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0021690251404503547,
+      "loss": 3.3506,
+      "step": 4891
+    },
+    {
+      "epoch": 0.42936879955895785,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0021686484038721766,
+      "loss": 3.3457,
+      "step": 4892
+    },
+    {
+      "epoch": 0.4294565691418603,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0021682716199035376,
+      "loss": 3.332,
+      "step": 4893
+    },
+    {
+      "epoch": 0.42954433872476283,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.002167894788578866,
+      "loss": 3.3535,
+      "step": 4894
+    },
+    {
+      "epoch": 0.4296321083076653,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0021675179099325927,
+      "loss": 3.373,
+      "step": 4895
+    },
+    {
+      "epoch": 0.4297198778905678,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0021671409839991553,
+      "loss": 3.2817,
+      "step": 4896
+    },
+    {
+      "epoch": 0.4298076474734703,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0021667640108129944,
+      "loss": 3.3687,
+      "step": 4897
+    },
+    {
+      "epoch": 0.42989541705637274,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002166386990408555,
+      "loss": 3.3696,
+      "step": 4898
+    },
+    {
+      "epoch": 0.42998318663927526,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0021660099228202868,
+      "loss": 3.3296,
+      "step": 4899
+    },
+    {
+      "epoch": 0.4300709562221777,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.002165632808082644,
+      "loss": 3.3013,
+      "step": 4900
+    },
+    {
+      "epoch": 0.4301587258050802,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002165255646230084,
+      "loss": 3.3389,
+      "step": 4901
+    },
+    {
+      "epoch": 0.4302464953879827,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002164878437297069,
+      "loss": 3.3223,
+      "step": 4902
+    },
+    {
+      "epoch": 0.43033426497088517,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0021645011813180666,
+      "loss": 3.3408,
+      "step": 4903
+    },
+    {
+      "epoch": 0.4304220345537877,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.002164123878327547,
+      "loss": 3.3267,
+      "step": 4904
+    },
+    {
+      "epoch": 0.43050980413669016,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002163746528359986,
+      "loss": 3.29,
+      "step": 4905
+    },
+    {
+      "epoch": 0.4305975737195926,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.002163369131449864,
+      "loss": 3.3491,
+      "step": 4906
+    },
+    {
+      "epoch": 0.43068534330249514,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.002162991687631663,
+      "loss": 3.3354,
+      "step": 4907
+    },
+    {
+      "epoch": 0.4307731128853976,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.002162614196939872,
+      "loss": 3.3555,
+      "step": 4908
+    },
+    {
+      "epoch": 0.4308608824683001,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002162236659408985,
+      "loss": 3.374,
+      "step": 4909
+    },
+    {
+      "epoch": 0.4309486520512026,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.002161859075073496,
+      "loss": 3.3179,
+      "step": 4910
+    },
+    {
+      "epoch": 0.43103642163410505,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0021614814439679083,
+      "loss": 3.3301,
+      "step": 4911
+    },
+    {
+      "epoch": 0.43112419121700757,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0021611037661267256,
+      "loss": 3.3281,
+      "step": 4912
+    },
+    {
+      "epoch": 0.43121196079991003,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002160726041584458,
+      "loss": 3.2344,
+      "step": 4913
+    },
+    {
+      "epoch": 0.4312997303828125,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0021603482703756198,
+      "loss": 3.3823,
+      "step": 4914
+    },
+    {
+      "epoch": 0.431387499965715,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0021599704525347286,
+      "loss": 3.2632,
+      "step": 4915
+    },
+    {
+      "epoch": 0.4314752695486175,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.002159592588096307,
+      "loss": 3.3013,
+      "step": 4916
+    },
+    {
+      "epoch": 0.43156303913152,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0021592146770948815,
+      "loss": 3.3911,
+      "step": 4917
+    },
+    {
+      "epoch": 0.43165080871442246,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002158836719564983,
+      "loss": 3.2568,
+      "step": 4918
+    },
+    {
+      "epoch": 0.4317385782973249,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002158458715541146,
+      "loss": 3.3511,
+      "step": 4919
+    },
+    {
+      "epoch": 0.43182634788022745,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.002158080665057911,
+      "loss": 3.3789,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4319141174631299,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.002157702568149821,
+      "loss": 3.2529,
+      "step": 4921
+    },
+    {
+      "epoch": 0.43200188704603243,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0021573244248514235,
+      "loss": 3.2417,
+      "step": 4922
+    },
+    {
+      "epoch": 0.4320896566289349,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0021569462351972712,
+      "loss": 3.3403,
+      "step": 4923
+    },
+    {
+      "epoch": 0.43217742621183736,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0021565679992219203,
+      "loss": 3.2852,
+      "step": 4924
+    },
+    {
+      "epoch": 0.4322651957947399,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0021561897169599313,
+      "loss": 3.3311,
+      "step": 4925
+    },
+    {
+      "epoch": 0.43235296537764234,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0021558113884458685,
+      "loss": 3.2529,
+      "step": 4926
+    },
+    {
+      "epoch": 0.4324407349605448,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0021554330137143014,
+      "loss": 3.2988,
+      "step": 4927
+    },
+    {
+      "epoch": 0.4325285045434473,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.002155054592799804,
+      "loss": 3.2798,
+      "step": 4928
+    },
+    {
+      "epoch": 0.4326162741263498,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.002154676125736952,
+      "loss": 3.375,
+      "step": 4929
+    },
+    {
+      "epoch": 0.4327040437092523,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0021542976125603283,
+      "loss": 3.2974,
+      "step": 4930
+    },
+    {
+      "epoch": 0.43279181329215477,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.002153919053304518,
+      "loss": 3.3481,
+      "step": 4931
+    },
+    {
+      "epoch": 0.43287958287505723,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0021535404480041123,
+      "loss": 3.3027,
+      "step": 4932
+    },
+    {
+      "epoch": 0.43296735245795975,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0021531617966937047,
+      "loss": 3.1855,
+      "step": 4933
+    },
+    {
+      "epoch": 0.4330551220408622,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.002152783099407894,
+      "loss": 3.2739,
+      "step": 4934
+    },
+    {
+      "epoch": 0.43314289162376474,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0021524043561812834,
+      "loss": 3.3599,
+      "step": 4935
+    },
+    {
+      "epoch": 0.4332306612066672,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002152025567048478,
+      "loss": 3.3164,
+      "step": 4936
+    },
+    {
+      "epoch": 0.43331843078956966,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0021516467320440904,
+      "loss": 3.3179,
+      "step": 4937
+    },
+    {
+      "epoch": 0.4334062003724722,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002151267851202736,
+      "loss": 3.3057,
+      "step": 4938
+    },
+    {
+      "epoch": 0.43349396995537465,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0021508889245590336,
+      "loss": 3.332,
+      "step": 4939
+    },
+    {
+      "epoch": 0.4335817395382771,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0021505099521476074,
+      "loss": 3.4155,
+      "step": 4940
+    },
+    {
+      "epoch": 0.43366950912117963,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0021501309340030846,
+      "loss": 3.2871,
+      "step": 4941
+    },
+    {
+      "epoch": 0.4337572787040821,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.002149751870160098,
+      "loss": 3.3037,
+      "step": 4942
+    },
+    {
+      "epoch": 0.4338450482869846,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002149372760653283,
+      "loss": 3.3027,
+      "step": 4943
+    },
+    {
+      "epoch": 0.4339328178698871,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0021489936055172814,
+      "loss": 3.3569,
+      "step": 4944
+    },
+    {
+      "epoch": 0.43402058745278954,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0021486144047867355,
+      "loss": 3.293,
+      "step": 4945
+    },
+    {
+      "epoch": 0.43410835703569206,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002148235158496296,
+      "loss": 3.3271,
+      "step": 4946
+    },
+    {
+      "epoch": 0.4341961266185945,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0021478558666806146,
+      "loss": 3.2573,
+      "step": 4947
+    },
+    {
+      "epoch": 0.43428389620149704,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0021474765293743483,
+      "loss": 3.3188,
+      "step": 4948
+    },
+    {
+      "epoch": 0.4343716657843995,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0021470971466121596,
+      "loss": 3.3335,
+      "step": 4949
+    },
+    {
+      "epoch": 0.43445943536730197,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0021467177184287126,
+      "loss": 3.311,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4345472049502045,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.002146338244858677,
+      "loss": 3.3521,
+      "step": 4951
+    },
+    {
+      "epoch": 0.43463497453310695,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.002145958725936727,
+      "loss": 3.3677,
+      "step": 4952
+    },
+    {
+      "epoch": 0.4347227441160094,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00214557916169754,
+      "loss": 3.3311,
+      "step": 4953
+    },
+    {
+      "epoch": 0.43481051369891194,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0021451995521757973,
+      "loss": 3.3188,
+      "step": 4954
+    },
+    {
+      "epoch": 0.4348982832818144,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.002144819897406186,
+      "loss": 3.2827,
+      "step": 4955
+    },
+    {
+      "epoch": 0.4349860528647169,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0021444401974233954,
+      "loss": 3.2705,
+      "step": 4956
+    },
+    {
+      "epoch": 0.4350738224476194,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0021440604522621207,
+      "loss": 3.2949,
+      "step": 4957
+    },
+    {
+      "epoch": 0.43516159203052185,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.00214368066195706,
+      "loss": 3.3477,
+      "step": 4958
+    },
+    {
+      "epoch": 0.43524936161342437,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002143300826542916,
+      "loss": 3.3398,
+      "step": 4959
+    },
+    {
+      "epoch": 0.43533713119632683,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002142920946054395,
+      "loss": 3.3564,
+      "step": 4960
+    },
+    {
+      "epoch": 0.43542490077922935,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.002142541020526208,
+      "loss": 3.3384,
+      "step": 4961
+    },
+    {
+      "epoch": 0.4355126703621318,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0021421610499930696,
+      "loss": 3.354,
+      "step": 4962
+    },
+    {
+      "epoch": 0.4356004399450343,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0021417810344897,
+      "loss": 3.2852,
+      "step": 4963
+    },
+    {
+      "epoch": 0.4356882095279368,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002141400974050821,
+      "loss": 3.3418,
+      "step": 4964
+    },
+    {
+      "epoch": 0.43577597911083926,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0021410208687111607,
+      "loss": 3.3247,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4358637486937418,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.00214064071850545,
+      "loss": 3.2925,
+      "step": 4966
+    },
+    {
+      "epoch": 0.43595151827664425,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.002140260523468425,
+      "loss": 3.3071,
+      "step": 4967
+    },
+    {
+      "epoch": 0.4360392878595467,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002139880283634824,
+      "loss": 3.3291,
+      "step": 4968
+    },
+    {
+      "epoch": 0.43612705744244923,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0021394999990393924,
+      "loss": 3.3135,
+      "step": 4969
+    },
+    {
+      "epoch": 0.4362148270253517,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0021391196697168767,
+      "loss": 3.2964,
+      "step": 4970
+    },
+    {
+      "epoch": 0.43630259660825416,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002138739295702029,
+      "loss": 3.3042,
+      "step": 4971
+    },
+    {
+      "epoch": 0.4363903661911567,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0021383588770296055,
+      "loss": 3.2358,
+      "step": 4972
+    },
+    {
+      "epoch": 0.43647813577405914,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0021379784137343654,
+      "loss": 3.3052,
+      "step": 4973
+    },
+    {
+      "epoch": 0.43656590535696166,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.002137597905851074,
+      "loss": 3.4233,
+      "step": 4974
+    },
+    {
+      "epoch": 0.4366536749398641,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0021372173534144985,
+      "loss": 3.2554,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4367414445227666,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.002136836756459411,
+      "loss": 3.3086,
+      "step": 4976
+    },
+    {
+      "epoch": 0.4368292141056691,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002136456115020589,
+      "loss": 3.2671,
+      "step": 4977
+    },
+    {
+      "epoch": 0.43691698368857157,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002136075429132812,
+      "loss": 3.249,
+      "step": 4978
+    },
+    {
+      "epoch": 0.4370047532714741,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002135694698830864,
+      "loss": 3.3115,
+      "step": 4979
+    },
+    {
+      "epoch": 0.43709252285437655,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0021353139241495345,
+      "loss": 3.353,
+      "step": 4980
+    },
+    {
+      "epoch": 0.437180292437279,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.002134933105123615,
+      "loss": 3.3237,
+      "step": 4981
+    },
+    {
+      "epoch": 0.43726806202018154,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0021345522417879023,
+      "loss": 3.3052,
+      "step": 4982
+    },
+    {
+      "epoch": 0.437355831603084,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.002134171334177198,
+      "loss": 3.3535,
+      "step": 4983
+    },
+    {
+      "epoch": 0.43744360118598646,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0021337903823263056,
+      "loss": 3.373,
+      "step": 4984
+    },
+    {
+      "epoch": 0.437531370768889,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.002133409386270034,
+      "loss": 3.2451,
+      "step": 4985
+    },
+    {
+      "epoch": 0.43761914035179145,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0021330283460431966,
+      "loss": 3.3501,
+      "step": 4986
+    },
+    {
+      "epoch": 0.43770690993469397,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0021326472616806096,
+      "loss": 3.2671,
+      "step": 4987
+    },
+    {
+      "epoch": 0.43779467951759643,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0021322661332170943,
+      "loss": 3.2749,
+      "step": 4988
+    },
+    {
+      "epoch": 0.4378824491004989,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0021318849606874746,
+      "loss": 3.2896,
+      "step": 4989
+    },
+    {
+      "epoch": 0.4379702186834014,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0021315037441265805,
+      "loss": 3.3159,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4380579882663039,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.002131122483569245,
+      "loss": 3.4209,
+      "step": 4991
+    },
+    {
+      "epoch": 0.4381457578492064,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0021307411790503035,
+      "loss": 3.374,
+      "step": 4992
+    },
+    {
+      "epoch": 0.43823352743210886,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.0021303598306045986,
+      "loss": 3.3174,
+      "step": 4993
+    },
+    {
+      "epoch": 0.4383212970150113,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0021299784382669737,
+      "loss": 3.3799,
+      "step": 4994
+    },
+    {
+      "epoch": 0.43840906659791384,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0021295970020722794,
+      "loss": 3.3477,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4384968361808163,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002129215522055368,
+      "loss": 3.4014,
+      "step": 4996
+    },
+    {
+      "epoch": 0.43858460576371877,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0021288339982510956,
+      "loss": 3.2734,
+      "step": 4997
+    },
+    {
+      "epoch": 0.4386723753466213,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0021284524306943247,
+      "loss": 3.3696,
+      "step": 4998
+    },
+    {
+      "epoch": 0.43876014492952375,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.002128070819419919,
+      "loss": 3.3184,
+      "step": 4999
+    },
+    {
+      "epoch": 0.4388479145124263,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.002127689164462748,
+      "loss": 3.2915,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4388479145124263,
+      "eval_loss": 0.1032758355140686,
+      "eval_runtime": 106.5909,
+      "eval_samples_per_second": 138.37,
+      "eval_steps_per_second": 17.3,
+      "step": 5000
+    },
+    {
+      "epoch": 0.43893568409532874,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0021273074658576846,
+      "loss": 3.3965,
+      "step": 5001
+    },
+    {
+      "epoch": 0.4390234536782312,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.002126925723639606,
+      "loss": 3.2686,
+      "step": 5002
+    },
+    {
+      "epoch": 0.4391112232611337,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.002126543937843393,
+      "loss": 3.2852,
+      "step": 5003
+    },
+    {
+      "epoch": 0.4391989928440362,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0021261621085039303,
+      "loss": 3.3369,
+      "step": 5004
+    },
+    {
+      "epoch": 0.4392867624269387,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0021257802356561073,
+      "loss": 3.3066,
+      "step": 5005
+    },
+    {
+      "epoch": 0.43937453200984117,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0021253983193348152,
+      "loss": 3.2939,
+      "step": 5006
+    },
+    {
+      "epoch": 0.43946230159274363,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0021250163595749536,
+      "loss": 3.4092,
+      "step": 5007
+    },
+    {
+      "epoch": 0.43955007117564615,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.002124634356411421,
+      "loss": 3.2534,
+      "step": 5008
+    },
+    {
+      "epoch": 0.4396378407585486,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0021242523098791233,
+      "loss": 3.2998,
+      "step": 5009
+    },
+    {
+      "epoch": 0.4397256103414511,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.002123870220012969,
+      "loss": 3.3872,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4398133799243536,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0021234880868478707,
+      "loss": 3.2715,
+      "step": 5011
+    },
+    {
+      "epoch": 0.43990114950725606,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002123105910418745,
+      "loss": 3.4014,
+      "step": 5012
+    },
+    {
+      "epoch": 0.4399889190901586,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0021227236907605126,
+      "loss": 3.2588,
+      "step": 5013
+    },
+    {
+      "epoch": 0.44007668867306104,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002122341427908098,
+      "loss": 3.3291,
+      "step": 5014
+    },
+    {
+      "epoch": 0.4401644582559635,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0021219591218964305,
+      "loss": 3.3276,
+      "step": 5015
+    },
+    {
+      "epoch": 0.44025222783886603,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0021215767727604407,
+      "loss": 3.2422,
+      "step": 5016
+    },
+    {
+      "epoch": 0.4403399974217685,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0021211943805350664,
+      "loss": 3.248,
+      "step": 5017
+    },
+    {
+      "epoch": 0.440427767004671,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.002120811945255248,
+      "loss": 3.3418,
+      "step": 5018
+    },
+    {
+      "epoch": 0.4405155365875735,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0021204294669559293,
+      "loss": 3.2808,
+      "step": 5019
+    },
+    {
+      "epoch": 0.44060330617047594,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.002120046945672059,
+      "loss": 3.2549,
+      "step": 5020
+    },
+    {
+      "epoch": 0.44069107575337846,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0021196643814385876,
+      "loss": 3.3198,
+      "step": 5021
+    },
+    {
+      "epoch": 0.4407788453362809,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.002119281774290473,
+      "loss": 3.2998,
+      "step": 5022
+    },
+    {
+      "epoch": 0.4408666149191834,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0021188991242626753,
+      "loss": 3.3018,
+      "step": 5023
+    },
+    {
+      "epoch": 0.4409543845020859,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.002118516431390156,
+      "loss": 3.335,
+      "step": 5024
+    },
+    {
+      "epoch": 0.44104215408498837,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0021181336957078853,
+      "loss": 3.3438,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4411299236678909,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0021177509172508346,
+      "loss": 3.3828,
+      "step": 5026
+    },
+    {
+      "epoch": 0.44121769325079335,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0021173680960539787,
+      "loss": 3.2827,
+      "step": 5027
+    },
+    {
+      "epoch": 0.4413054628336958,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.002116985232152297,
+      "loss": 3.2632,
+      "step": 5028
+    },
+    {
+      "epoch": 0.44139323241659834,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0021166023255807733,
+      "loss": 3.3838,
+      "step": 5029
+    },
+    {
+      "epoch": 0.4414810019995008,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0021162193763743955,
+      "loss": 3.3218,
+      "step": 5030
+    },
+    {
+      "epoch": 0.4415687715824033,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.002115836384568154,
+      "loss": 3.2681,
+      "step": 5031
+    },
+    {
+      "epoch": 0.4416565411653058,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002115453350197044,
+      "loss": 3.2729,
+      "step": 5032
+    },
+    {
+      "epoch": 0.44174431074820825,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.002115070273296065,
+      "loss": 3.3394,
+      "step": 5033
+    },
+    {
+      "epoch": 0.44183208033111077,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.00211468715390022,
+      "loss": 3.3501,
+      "step": 5034
+    },
+    {
+      "epoch": 0.44191984991401323,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0021143039920445143,
+      "loss": 3.2793,
+      "step": 5035
+    },
+    {
+      "epoch": 0.44200761949691575,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0021139207877639604,
+      "loss": 3.2817,
+      "step": 5036
+    },
+    {
+      "epoch": 0.4420953890798182,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0021135375410935714,
+      "loss": 3.394,
+      "step": 5037
+    },
+    {
+      "epoch": 0.4421831586627207,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.002113154252068367,
+      "loss": 3.314,
+      "step": 5038
+    },
+    {
+      "epoch": 0.4422709282456232,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.002112770920723369,
+      "loss": 3.3032,
+      "step": 5039
+    },
+    {
+      "epoch": 0.44235869782852566,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0021123875470936025,
+      "loss": 3.3618,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4424464674114281,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0021120041312140984,
+      "loss": 3.3369,
+      "step": 5041
+    },
+    {
+      "epoch": 0.44253423699433064,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002111620673119891,
+      "loss": 3.3462,
+      "step": 5042
+    },
+    {
+      "epoch": 0.4426220065772331,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002111237172846017,
+      "loss": 3.333,
+      "step": 5043
+    },
+    {
+      "epoch": 0.4427097761601356,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0021108536304275194,
+      "loss": 3.2471,
+      "step": 5044
+    },
+    {
+      "epoch": 0.4427975457430381,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002110470045899442,
+      "loss": 3.3135,
+      "step": 5045
+    },
+    {
+      "epoch": 0.44288531532594055,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002110086419296834,
+      "loss": 3.3667,
+      "step": 5046
+    },
+    {
+      "epoch": 0.4429730849088431,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0021097027506547504,
+      "loss": 3.3257,
+      "step": 5047
+    },
+    {
+      "epoch": 0.44306085449174554,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002109319040008247,
+      "loss": 3.2744,
+      "step": 5048
+    },
+    {
+      "epoch": 0.44314862407464806,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002108935287392384,
+      "loss": 3.3081,
+      "step": 5049
+    },
+    {
+      "epoch": 0.4432363936575505,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0021085514928422273,
+      "loss": 3.3442,
+      "step": 5050
+    },
+    {
+      "epoch": 0.443324163240453,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002108167656392844,
+      "loss": 3.2534,
+      "step": 5051
+    },
+    {
+      "epoch": 0.4434119328233555,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0021077837780793074,
+      "loss": 3.3887,
+      "step": 5052
+    },
+    {
+      "epoch": 0.44349970240625797,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0021073998579366933,
+      "loss": 3.3345,
+      "step": 5053
+    },
+    {
+      "epoch": 0.44358747198916043,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002107015896000082,
+      "loss": 3.3135,
+      "step": 5054
+    },
+    {
+      "epoch": 0.44367524157206295,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.002106631892304557,
+      "loss": 3.3379,
+      "step": 5055
+    },
+    {
+      "epoch": 0.4437630111549654,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0021062478468852055,
+      "loss": 3.3257,
+      "step": 5056
+    },
+    {
+      "epoch": 0.44385078073786793,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002105863759777119,
+      "loss": 3.3032,
+      "step": 5057
+    },
+    {
+      "epoch": 0.4439385503207704,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.002105479631015393,
+      "loss": 3.3682,
+      "step": 5058
+    },
+    {
+      "epoch": 0.44402631990367286,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.002105095460635126,
+      "loss": 3.354,
+      "step": 5059
+    },
+    {
+      "epoch": 0.4441140894865754,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0021047112486714216,
+      "loss": 3.3647,
+      "step": 5060
+    },
+    {
+      "epoch": 0.44420185906947784,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.002104326995159386,
+      "loss": 3.3462,
+      "step": 5061
+    },
+    {
+      "epoch": 0.44428962865238036,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.0021039427001341296,
+      "loss": 3.3242,
+      "step": 5062
+    },
+    {
+      "epoch": 0.44437739823528283,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002103558363630766,
+      "loss": 3.3047,
+      "step": 5063
+    },
+    {
+      "epoch": 0.4444651678181853,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0021031739856844137,
+      "loss": 3.27,
+      "step": 5064
+    },
+    {
+      "epoch": 0.4445529374010878,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0021027895663301953,
+      "loss": 3.3081,
+      "step": 5065
+    },
+    {
+      "epoch": 0.4446407069839903,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002102405105603235,
+      "loss": 3.3931,
+      "step": 5066
+    },
+    {
+      "epoch": 0.44472847656689274,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0021020206035386626,
+      "loss": 3.3584,
+      "step": 5067
+    },
+    {
+      "epoch": 0.44481624614979526,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0021016360601716117,
+      "loss": 3.3101,
+      "step": 5068
+    },
+    {
+      "epoch": 0.4449040157326977,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0021012514755372183,
+      "loss": 3.2549,
+      "step": 5069
+    },
+    {
+      "epoch": 0.44499178531560024,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002100866849670623,
+      "loss": 3.3091,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4450795548985027,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002100482182606972,
+      "loss": 3.3032,
+      "step": 5071
+    },
+    {
+      "epoch": 0.44516732448140517,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0021000974743814113,
+      "loss": 3.292,
+      "step": 5072
+    },
+    {
+      "epoch": 0.4452550940643077,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.002099712725029094,
+      "loss": 3.2598,
+      "step": 5073
+    },
+    {
+      "epoch": 0.44534286364721015,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0020993279345851755,
+      "loss": 3.3052,
+      "step": 5074
+    },
+    {
+      "epoch": 0.44543063323011267,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0020989431030848152,
+      "loss": 3.2671,
+      "step": 5075
+    },
+    {
+      "epoch": 0.44551840281301514,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.0020985582305631766,
+      "loss": 3.3545,
+      "step": 5076
+    },
+    {
+      "epoch": 0.4456061723959176,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0020981733170554267,
+      "loss": 3.2446,
+      "step": 5077
+    },
+    {
+      "epoch": 0.4456939419788201,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002097788362596736,
+      "loss": 3.313,
+      "step": 5078
+    },
+    {
+      "epoch": 0.4457817115617226,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0020974033672222785,
+      "loss": 3.3242,
+      "step": 5079
+    },
+    {
+      "epoch": 0.44586948114462505,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002097018330967233,
+      "loss": 3.3315,
+      "step": 5080
+    },
+    {
+      "epoch": 0.44595725072752757,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002096633253866781,
+      "loss": 3.2759,
+      "step": 5081
+    },
+    {
+      "epoch": 0.44604502031043003,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0020962481359561086,
+      "loss": 3.2734,
+      "step": 5082
+    },
+    {
+      "epoch": 0.44613278989333255,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0020958629772704048,
+      "loss": 3.3306,
+      "step": 5083
+    },
+    {
+      "epoch": 0.446220559476235,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.002095477777844863,
+      "loss": 3.2969,
+      "step": 5084
+    },
+    {
+      "epoch": 0.4463083290591375,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.00209509253771468,
+      "loss": 3.3066,
+      "step": 5085
+    },
+    {
+      "epoch": 0.44639609864204,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.002094707256915056,
+      "loss": 3.353,
+      "step": 5086
+    },
+    {
+      "epoch": 0.44648386822494246,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.0020943219354811952,
+      "loss": 3.3379,
+      "step": 5087
+    },
+    {
+      "epoch": 0.446571637807845,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0020939365734483064,
+      "loss": 3.3145,
+      "step": 5088
+    },
+    {
+      "epoch": 0.44665940739074744,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0020935511708516006,
+      "loss": 3.3022,
+      "step": 5089
+    },
+    {
+      "epoch": 0.4467471769736499,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.002093165727726294,
+      "loss": 3.2988,
+      "step": 5090
+    },
+    {
+      "epoch": 0.4468349465565524,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0020927802441076037,
+      "loss": 3.2603,
+      "step": 5091
+    },
+    {
+      "epoch": 0.4469227161394549,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0020923947200307547,
+      "loss": 3.2861,
+      "step": 5092
+    },
+    {
+      "epoch": 0.44701048572235735,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002092009155530973,
+      "loss": 3.2842,
+      "step": 5093
+    },
+    {
+      "epoch": 0.4470982553052599,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0020916235506434873,
+      "loss": 3.2847,
+      "step": 5094
+    },
+    {
+      "epoch": 0.44718602488816234,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.002091237905403533,
+      "loss": 3.2827,
+      "step": 5095
+    },
+    {
+      "epoch": 0.44727379447106486,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.002090852219846348,
+      "loss": 3.3569,
+      "step": 5096
+    },
+    {
+      "epoch": 0.4473615640539673,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.002090466494007172,
+      "loss": 3.2979,
+      "step": 5097
+    },
+    {
+      "epoch": 0.4474493336368698,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.002090080727921251,
+      "loss": 3.3101,
+      "step": 5098
+    },
+    {
+      "epoch": 0.4475371032197723,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0020896949216238333,
+      "loss": 3.2695,
+      "step": 5099
+    },
+    {
+      "epoch": 0.44762487280267477,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002089309075150171,
+      "loss": 3.2769,
+      "step": 5100
+    },
+    {
+      "epoch": 0.4477126423855773,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0020889231885355204,
+      "loss": 3.2944,
+      "step": 5101
+    },
+    {
+      "epoch": 0.44780041196847975,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002088537261815141,
+      "loss": 3.3101,
+      "step": 5102
+    },
+    {
+      "epoch": 0.4478881815513822,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0020881512950242966,
+      "loss": 3.3311,
+      "step": 5103
+    },
+    {
+      "epoch": 0.44797595113428473,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0020877652881982527,
+      "loss": 3.3784,
+      "step": 5104
+    },
+    {
+      "epoch": 0.4480637207171872,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.002087379241372281,
+      "loss": 3.187,
+      "step": 5105
+    },
+    {
+      "epoch": 0.4481514903000897,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0020869931545816557,
+      "loss": 3.3828,
+      "step": 5106
+    },
+    {
+      "epoch": 0.4482392598829922,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.002086607027861655,
+      "loss": 3.3442,
+      "step": 5107
+    },
+    {
+      "epoch": 0.44832702946589464,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.00208622086124756,
+      "loss": 3.2432,
+      "step": 5108
+    },
+    {
+      "epoch": 0.44841479904879716,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0020858346547746555,
+      "loss": 3.2427,
+      "step": 5109
+    },
+    {
+      "epoch": 0.4485025686316996,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.002085448408478231,
+      "loss": 3.3203,
+      "step": 5110
+    },
+    {
+      "epoch": 0.4485903382146021,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0020850621223935783,
+      "loss": 3.4102,
+      "step": 5111
+    },
+    {
+      "epoch": 0.4486781077975046,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0020846757965559946,
+      "loss": 3.376,
+      "step": 5112
+    },
+    {
+      "epoch": 0.4487658773804071,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0020842894310007785,
+      "loss": 3.312,
+      "step": 5113
+    },
+    {
+      "epoch": 0.4488536469633096,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.0020839030257632342,
+      "loss": 3.3315,
+      "step": 5114
+    },
+    {
+      "epoch": 0.44894141654621206,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.002083516580878668,
+      "loss": 3.2793,
+      "step": 5115
+    },
+    {
+      "epoch": 0.4490291861291145,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0020831300963823915,
+      "loss": 3.3398,
+      "step": 5116
+    },
+    {
+      "epoch": 0.44911695571201704,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0020827435723097185,
+      "loss": 3.3315,
+      "step": 5117
+    },
+    {
+      "epoch": 0.4492047252949195,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0020823570086959666,
+      "loss": 3.2744,
+      "step": 5118
+    },
+    {
+      "epoch": 0.449292494877822,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0020819704055764572,
+      "loss": 3.3628,
+      "step": 5119
+    },
+    {
+      "epoch": 0.4493802644607245,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0020815837629865157,
+      "loss": 3.2856,
+      "step": 5120
+    },
+    {
+      "epoch": 0.44946803404362695,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.002081197080961471,
+      "loss": 3.356,
+      "step": 5121
+    },
+    {
+      "epoch": 0.44955580362652947,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002080810359536654,
+      "loss": 3.2539,
+      "step": 5122
+    },
+    {
+      "epoch": 0.44964357320943193,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002080423598747403,
+      "loss": 3.2983,
+      "step": 5123
+    },
+    {
+      "epoch": 0.4497313427923344,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0020800367986290556,
+      "loss": 3.3804,
+      "step": 5124
+    },
+    {
+      "epoch": 0.4498191123752369,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0020796499592169557,
+      "loss": 3.2593,
+      "step": 5125
+    },
+    {
+      "epoch": 0.4499068819581394,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.00207926308054645,
+      "loss": 3.2676,
+      "step": 5126
+    },
+    {
+      "epoch": 0.4499946515410419,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.002078876162652888,
+      "loss": 3.3516,
+      "step": 5127
+    },
+    {
+      "epoch": 0.45008242112394437,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0020784892055716246,
+      "loss": 3.2935,
+      "step": 5128
+    },
+    {
+      "epoch": 0.45017019070684683,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0020781022093380166,
+      "loss": 3.3354,
+      "step": 5129
+    },
+    {
+      "epoch": 0.45025796028974935,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.002077715173987425,
+      "loss": 3.3179,
+      "step": 5130
+    },
+    {
+      "epoch": 0.4503457298726518,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0020773280995552146,
+      "loss": 3.23,
+      "step": 5131
+    },
+    {
+      "epoch": 0.45043349945555433,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0020769409860767534,
+      "loss": 3.2461,
+      "step": 5132
+    },
+    {
+      "epoch": 0.4505212690384568,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0020765538335874127,
+      "loss": 3.3711,
+      "step": 5133
+    },
+    {
+      "epoch": 0.45060903862135926,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0020761666421225686,
+      "loss": 3.2979,
+      "step": 5134
+    },
+    {
+      "epoch": 0.4506968082042618,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0020757794117176,
+      "loss": 3.2964,
+      "step": 5135
+    },
+    {
+      "epoch": 0.45078457778716424,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002075392142407888,
+      "loss": 3.3311,
+      "step": 5136
+    },
+    {
+      "epoch": 0.4508723473700667,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0020750048342288202,
+      "loss": 3.3516,
+      "step": 5137
+    },
+    {
+      "epoch": 0.4509601169529692,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0020746174872157855,
+      "loss": 3.3633,
+      "step": 5138
+    },
+    {
+      "epoch": 0.4510478865358717,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.002074230101404176,
+      "loss": 3.3218,
+      "step": 5139
+    },
+    {
+      "epoch": 0.4511356561187742,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00207384267682939,
+      "loss": 3.354,
+      "step": 5140
+    },
+    {
+      "epoch": 0.4512234257016767,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.002073455213526827,
+      "loss": 3.3599,
+      "step": 5141
+    },
+    {
+      "epoch": 0.45131119528457914,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.0020730677115318894,
+      "loss": 3.333,
+      "step": 5142
+    },
+    {
+      "epoch": 0.45139896486748166,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0020726801708799867,
+      "loss": 3.3232,
+      "step": 5143
+    },
+    {
+      "epoch": 0.4514867344503841,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.002072292591606527,
+      "loss": 3.2822,
+      "step": 5144
+    },
+    {
+      "epoch": 0.45157450403328664,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.002071904973746927,
+      "loss": 3.3032,
+      "step": 5145
+    },
+    {
+      "epoch": 0.4516622736161891,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0020715173173366033,
+      "loss": 3.2915,
+      "step": 5146
+    },
+    {
+      "epoch": 0.45175004319909157,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0020711296224109775,
+      "loss": 3.3511,
+      "step": 5147
+    },
+    {
+      "epoch": 0.4518378127819941,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0020707418890054743,
+      "loss": 3.2993,
+      "step": 5148
+    },
+    {
+      "epoch": 0.45192558236489655,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0020703541171555226,
+      "loss": 3.2622,
+      "step": 5149
+    },
+    {
+      "epoch": 0.452013351947799,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.002069966306896553,
+      "loss": 3.3486,
+      "step": 5150
+    },
+    {
+      "epoch": 0.45210112153070153,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0020695784582640026,
+      "loss": 3.3311,
+      "step": 5151
+    },
+    {
+      "epoch": 0.452188891113604,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0020691905712933094,
+      "loss": 3.3037,
+      "step": 5152
+    },
+    {
+      "epoch": 0.4522766606965065,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0020688026460199153,
+      "loss": 3.2539,
+      "step": 5153
+    },
+    {
+      "epoch": 0.452364430279409,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 0.0020684146824792673,
+      "loss": 3.2788,
+      "step": 5154
+    },
+    {
+      "epoch": 0.45245219986231144,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0020680266807068136,
+      "loss": 3.3149,
+      "step": 5155
+    },
+    {
+      "epoch": 0.45253996944521396,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.002067638640738008,
+      "loss": 3.3159,
+      "step": 5156
+    },
+    {
+      "epoch": 0.4526277390281164,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0020672505626083064,
+      "loss": 3.2964,
+      "step": 5157
+    },
+    {
+      "epoch": 0.45271550861101895,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002066862446353169,
+      "loss": 3.4048,
+      "step": 5158
+    },
+    {
+      "epoch": 0.4528032781939214,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.0020664742920080592,
+      "loss": 3.3008,
+      "step": 5159
+    },
+    {
+      "epoch": 0.4528910477768239,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.0020660860996084434,
+      "loss": 3.3252,
+      "step": 5160
+    },
+    {
+      "epoch": 0.4529788173597264,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.002065697869189792,
+      "loss": 3.3984,
+      "step": 5161
+    },
+    {
+      "epoch": 0.45306658694262886,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.002065309600787579,
+      "loss": 3.2358,
+      "step": 5162
+    },
+    {
+      "epoch": 0.4531543565255314,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.002064921294437282,
+      "loss": 3.2969,
+      "step": 5163
+    },
+    {
+      "epoch": 0.45324212610843384,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.002064532950174381,
+      "loss": 3.3359,
+      "step": 5164
+    },
+    {
+      "epoch": 0.4533298956913363,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.00206414456803436,
+      "loss": 3.3418,
+      "step": 5165
+    },
+    {
+      "epoch": 0.4534176652742388,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.002063756148052707,
+      "loss": 3.2393,
+      "step": 5166
+    },
+    {
+      "epoch": 0.4535054348571413,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0020633676902649133,
+      "loss": 3.2847,
+      "step": 5167
+    },
+    {
+      "epoch": 0.45359320444004375,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002062979194706474,
+      "loss": 3.2505,
+      "step": 5168
+    },
+    {
+      "epoch": 0.45368097402294627,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0020625906614128863,
+      "loss": 3.3145,
+      "step": 5169
+    },
+    {
+      "epoch": 0.45376874360584873,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0020622020904196513,
+      "loss": 3.374,
+      "step": 5170
+    },
+    {
+      "epoch": 0.45385651318875125,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0020618134817622746,
+      "loss": 3.2583,
+      "step": 5171
+    },
+    {
+      "epoch": 0.4539442827716537,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.002061424835476264,
+      "loss": 3.3013,
+      "step": 5172
+    },
+    {
+      "epoch": 0.4540320523545562,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0020610361515971324,
+      "loss": 3.2988,
+      "step": 5173
+    },
+    {
+      "epoch": 0.4541198219374587,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.002060647430160394,
+      "loss": 3.3105,
+      "step": 5174
+    },
+    {
+      "epoch": 0.45420759152036116,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0020602586712015676,
+      "loss": 3.3022,
+      "step": 5175
+    },
+    {
+      "epoch": 0.4542953611032637,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0020598698747561757,
+      "loss": 3.3516,
+      "step": 5176
+    },
+    {
+      "epoch": 0.45438313068616615,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.002059481040859743,
+      "loss": 3.314,
+      "step": 5177
+    },
+    {
+      "epoch": 0.4544709002690686,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0020590921695477983,
+      "loss": 3.3247,
+      "step": 5178
+    },
+    {
+      "epoch": 0.45455866985197113,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.002058703260855875,
+      "loss": 3.3809,
+      "step": 5179
+    },
+    {
+      "epoch": 0.4546464394348736,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0020583143148195085,
+      "loss": 3.3135,
+      "step": 5180
+    },
+    {
+      "epoch": 0.45473420901777606,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0020579253314742378,
+      "loss": 3.3101,
+      "step": 5181
+    },
+    {
+      "epoch": 0.4548219786006786,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.002057536310855605,
+      "loss": 3.2188,
+      "step": 5182
+    },
+    {
+      "epoch": 0.45490974818358104,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0020571472529991573,
+      "loss": 3.396,
+      "step": 5183
+    },
+    {
+      "epoch": 0.45499751776648356,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002056758157940443,
+      "loss": 3.2622,
+      "step": 5184
+    },
+    {
+      "epoch": 0.455085287349386,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0020563690257150154,
+      "loss": 3.3208,
+      "step": 5185
+    },
+    {
+      "epoch": 0.4551730569322885,
+      "grad_norm": 0.193359375,
+      "learning_rate": 0.0020559798563584302,
+      "loss": 3.3413,
+      "step": 5186
+    },
+    {
+      "epoch": 0.455260826515191,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0020555906499062477,
+      "loss": 3.355,
+      "step": 5187
+    },
+    {
+      "epoch": 0.45534859609809347,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.00205520140639403,
+      "loss": 3.2021,
+      "step": 5188
+    },
+    {
+      "epoch": 0.455436365680996,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.002054812125857344,
+      "loss": 3.3218,
+      "step": 5189
+    },
+    {
+      "epoch": 0.45552413526389846,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.0020544228083317594,
+      "loss": 3.3726,
+      "step": 5190
+    },
+    {
+      "epoch": 0.4556119048468009,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.002054033453852849,
+      "loss": 3.2852,
+      "step": 5191
+    },
+    {
+      "epoch": 0.45569967442970344,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.0020536440624561903,
+      "loss": 3.3848,
+      "step": 5192
+    },
+    {
+      "epoch": 0.4557874440126059,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002053254634177362,
+      "loss": 3.2979,
+      "step": 5193
+    },
+    {
+      "epoch": 0.45587521359550837,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.002052865169051948,
+      "loss": 3.3506,
+      "step": 5194
+    },
+    {
+      "epoch": 0.4559629831784109,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.002052475667115534,
+      "loss": 3.3228,
+      "step": 5195
+    },
+    {
+      "epoch": 0.45605075276131335,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0020520861284037107,
+      "loss": 3.415,
+      "step": 5196
+    },
+    {
+      "epoch": 0.45613852234421587,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0020516965529520717,
+      "loss": 3.1968,
+      "step": 5197
+    },
+    {
+      "epoch": 0.45622629192711833,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.002051306940796213,
+      "loss": 3.2632,
+      "step": 5198
+    },
+    {
+      "epoch": 0.4563140615100208,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.002050917291971735,
+      "loss": 3.2656,
+      "step": 5199
+    },
+    {
+      "epoch": 0.4564018310929233,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0020505276065142415,
+      "loss": 3.3691,
+      "step": 5200
+    },
+    {
+      "epoch": 0.4564896006758258,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.002050137884459338,
+      "loss": 3.2988,
+      "step": 5201
+    },
+    {
+      "epoch": 0.4565773702587283,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.002049748125842635,
+      "loss": 3.3311,
+      "step": 5202
+    },
+    {
+      "epoch": 0.45666513984163076,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002049358330699747,
+      "loss": 3.4028,
+      "step": 5203
+    },
+    {
+      "epoch": 0.4567529094245332,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.002048968499066289,
+      "loss": 3.2983,
+      "step": 5204
+    },
+    {
+      "epoch": 0.45684067900743575,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0020485786309778833,
+      "loss": 3.3779,
+      "step": 5205
+    },
+    {
+      "epoch": 0.4569284485903382,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0020481887264701512,
+      "loss": 3.2979,
+      "step": 5206
+    },
+    {
+      "epoch": 0.4570162181732407,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.002047798785578721,
+      "loss": 3.374,
+      "step": 5207
+    },
+    {
+      "epoch": 0.4571039877561432,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002047408808339221,
+      "loss": 3.3301,
+      "step": 5208
+    },
+    {
+      "epoch": 0.45719175733904566,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.002047018794787287,
+      "loss": 3.3286,
+      "step": 5209
+    },
+    {
+      "epoch": 0.4572795269219482,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0020466287449585536,
+      "loss": 3.2861,
+      "step": 5210
+    },
+    {
+      "epoch": 0.45736729650485064,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0020462386588886616,
+      "loss": 3.2783,
+      "step": 5211
+    },
+    {
+      "epoch": 0.4574550660877531,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0020458485366132546,
+      "loss": 3.3433,
+      "step": 5212
+    },
+    {
+      "epoch": 0.4575428356706556,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0020454583781679787,
+      "loss": 3.3369,
+      "step": 5213
+    },
+    {
+      "epoch": 0.4576306052535581,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0020450681835884844,
+      "loss": 3.3872,
+      "step": 5214
+    },
+    {
+      "epoch": 0.4577183748364606,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0020446779529104246,
+      "loss": 3.3062,
+      "step": 5215
+    },
+    {
+      "epoch": 0.45780614441936307,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.002044287686169456,
+      "loss": 3.27,
+      "step": 5216
+    },
+    {
+      "epoch": 0.45789391400226553,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.002043897383401238,
+      "loss": 3.2661,
+      "step": 5217
+    },
+    {
+      "epoch": 0.45798168358516805,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.002043507044641434,
+      "loss": 3.3022,
+      "step": 5218
+    },
+    {
+      "epoch": 0.4580694531680705,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0020431166699257115,
+      "loss": 3.355,
+      "step": 5219
+    },
+    {
+      "epoch": 0.458157222750973,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.002042726259289738,
+      "loss": 3.4097,
+      "step": 5220
+    },
+    {
+      "epoch": 0.4582449923338755,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.0020423358127691885,
+      "loss": 3.2827,
+      "step": 5221
+    },
+    {
+      "epoch": 0.45833276191677796,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0020419453303997387,
+      "loss": 3.334,
+      "step": 5222
+    },
+    {
+      "epoch": 0.4584205314996805,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002041554812217067,
+      "loss": 3.2617,
+      "step": 5223
+    },
+    {
+      "epoch": 0.45850830108258295,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.002041164258256858,
+      "loss": 3.3296,
+      "step": 5224
+    },
+    {
+      "epoch": 0.4585960706654854,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0020407736685547967,
+      "loss": 3.2461,
+      "step": 5225
+    },
+    {
+      "epoch": 0.45868384024838793,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.002040383043146573,
+      "loss": 3.3267,
+      "step": 5226
+    },
+    {
+      "epoch": 0.4587716098312904,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.0020399923820678782,
+      "loss": 3.3398,
+      "step": 5227
+    },
+    {
+      "epoch": 0.4588593794141929,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.00203960168535441,
+      "loss": 3.3784,
+      "step": 5228
+    },
+    {
+      "epoch": 0.4589471489970954,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0020392109530418673,
+      "loss": 3.3384,
+      "step": 5229
+    },
+    {
+      "epoch": 0.45903491857999784,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002038820185165951,
+      "loss": 3.3447,
+      "step": 5230
+    },
+    {
+      "epoch": 0.45912268816290036,
+      "grad_norm": 0.08154296875,
+      "learning_rate": 0.002038429381762368,
+      "loss": 3.3535,
+      "step": 5231
+    },
+    {
+      "epoch": 0.4592104577458028,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0020380385428668264,
+      "loss": 3.2446,
+      "step": 5232
+    },
+    {
+      "epoch": 0.45929822732870534,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.002037647668515039,
+      "loss": 3.1851,
+      "step": 5233
+    },
+    {
+      "epoch": 0.4593859969116078,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.002037256758742721,
+      "loss": 3.3052,
+      "step": 5234
+    },
+    {
+      "epoch": 0.45947376649451027,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.002036865813585591,
+      "loss": 3.2705,
+      "step": 5235
+    },
+    {
+      "epoch": 0.4595615360774128,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0020364748330793706,
+      "loss": 3.2764,
+      "step": 5236
+    },
+    {
+      "epoch": 0.45964930566031525,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0020360838172597847,
+      "loss": 3.3311,
+      "step": 5237
+    },
+    {
+      "epoch": 0.4597370752432177,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0020356927661625617,
+      "loss": 3.3135,
+      "step": 5238
+    },
+    {
+      "epoch": 0.45982484482612024,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.002035301679823434,
+      "loss": 3.3203,
+      "step": 5239
+    },
+    {
+      "epoch": 0.4599126144090227,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0020349105582781355,
+      "loss": 3.3257,
+      "step": 5240
+    },
+    {
+      "epoch": 0.4600003839919252,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.0020345194015624042,
+      "loss": 3.3091,
+      "step": 5241
+    },
+    {
+      "epoch": 0.4600881535748277,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.002034128209711982,
+      "loss": 3.3687,
+      "step": 5242
+    },
+    {
+      "epoch": 0.46017592315773015,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0020337369827626117,
+      "loss": 3.2622,
+      "step": 5243
+    },
+    {
+      "epoch": 0.46026369274063267,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.002033345720750042,
+      "loss": 3.4009,
+      "step": 5244
+    },
+    {
+      "epoch": 0.46035146232353513,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0020329544237100237,
+      "loss": 3.293,
+      "step": 5245
+    },
+    {
+      "epoch": 0.46043923190643765,
+      "grad_norm": 0.181640625,
+      "learning_rate": 0.0020325630916783106,
+      "loss": 3.2671,
+      "step": 5246
+    },
+    {
+      "epoch": 0.4605270014893401,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.002032171724690659,
+      "loss": 3.2759,
+      "step": 5247
+    },
+    {
+      "epoch": 0.4606147710722426,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002031780322782831,
+      "loss": 3.312,
+      "step": 5248
+    },
+    {
+      "epoch": 0.4607025406551451,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.002031388885990589,
+      "loss": 3.3394,
+      "step": 5249
+    },
+    {
+      "epoch": 0.46079031023804756,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0020309974143497005,
+      "loss": 3.3652,
+      "step": 5250
+    },
+    {
+      "epoch": 0.46087807982095,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.002030605907895934,
+      "loss": 3.2222,
+      "step": 5251
+    },
+    {
+      "epoch": 0.46096584940385255,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0020302143666650644,
+      "loss": 3.4531,
+      "step": 5252
+    },
+    {
+      "epoch": 0.461053618986755,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0020298227906928672,
+      "loss": 3.3052,
+      "step": 5253
+    },
+    {
+      "epoch": 0.46114138856965753,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0020294311800151217,
+      "loss": 3.3384,
+      "step": 5254
+    },
+    {
+      "epoch": 0.46122915815256,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0020290395346676104,
+      "loss": 3.3506,
+      "step": 5255
+    },
+    {
+      "epoch": 0.46131692773546246,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0020286478546861194,
+      "loss": 3.2979,
+      "step": 5256
+    },
+    {
+      "epoch": 0.461404697318365,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0020282561401064386,
+      "loss": 3.3491,
+      "step": 5257
+    },
+    {
+      "epoch": 0.46149246690126744,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002027864390964359,
+      "loss": 3.2563,
+      "step": 5258
+    },
+    {
+      "epoch": 0.46158023648416996,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0020274726072956753,
+      "loss": 3.2656,
+      "step": 5259
+    },
+    {
+      "epoch": 0.4616680060670724,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.002027080789136187,
+      "loss": 3.2769,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4617557756499749,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.0020266889365216966,
+      "loss": 3.2891,
+      "step": 5261
+    },
+    {
+      "epoch": 0.4618435452328774,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.002026297049488007,
+      "loss": 3.3193,
+      "step": 5262
+    },
+    {
+      "epoch": 0.46193131481577987,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.002025905128070927,
+      "loss": 3.3013,
+      "step": 5263
+    },
+    {
+      "epoch": 0.46201908439868233,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0020255131723062665,
+      "loss": 3.3452,
+      "step": 5264
+    },
+    {
+      "epoch": 0.46210685398158485,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0020251211822298414,
+      "loss": 3.1865,
+      "step": 5265
+    },
+    {
+      "epoch": 0.4621946235644873,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.002024729157877468,
+      "loss": 3.2729,
+      "step": 5266
+    },
+    {
+      "epoch": 0.46228239314738984,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.002024337099284967,
+      "loss": 3.3081,
+      "step": 5267
+    },
+    {
+      "epoch": 0.4623701627302923,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0020239450064881625,
+      "loss": 3.3359,
+      "step": 5268
+    },
+    {
+      "epoch": 0.46245793231319476,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0020235528795228804,
+      "loss": 3.3662,
+      "step": 5269
+    },
+    {
+      "epoch": 0.4625457018960973,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00202316071842495,
+      "loss": 3.3091,
+      "step": 5270
+    },
+    {
+      "epoch": 0.46263347147899975,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.002022768523230206,
+      "loss": 3.3022,
+      "step": 5271
+    },
+    {
+      "epoch": 0.46272124106190227,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.002022376293974483,
+      "loss": 3.3325,
+      "step": 5272
+    },
+    {
+      "epoch": 0.46280901064480473,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.002021984030693621,
+      "loss": 3.2798,
+      "step": 5273
+    },
+    {
+      "epoch": 0.4628967802277072,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0020215917334234614,
+      "loss": 3.3306,
+      "step": 5274
+    },
+    {
+      "epoch": 0.4629845498106097,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0020211994021998507,
+      "loss": 3.3892,
+      "step": 5275
+    },
+    {
+      "epoch": 0.4630723193935122,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0020208070370586364,
+      "loss": 3.2578,
+      "step": 5276
+    },
+    {
+      "epoch": 0.46316008897641464,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.00202041463803567,
+      "loss": 3.2729,
+      "step": 5277
+    },
+    {
+      "epoch": 0.46324785855931716,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0020200222051668074,
+      "loss": 3.2651,
+      "step": 5278
+    },
+    {
+      "epoch": 0.4633356281422196,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.002019629738487905,
+      "loss": 3.374,
+      "step": 5279
+    },
+    {
+      "epoch": 0.46342339772512214,
+      "grad_norm": 0.375,
+      "learning_rate": 0.002019237238034825,
+      "loss": 3.3276,
+      "step": 5280
+    },
+    {
+      "epoch": 0.4635111673080246,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0020188447038434294,
+      "loss": 3.3008,
+      "step": 5281
+    },
+    {
+      "epoch": 0.46359893689092707,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.002018452135949587,
+      "loss": 3.2793,
+      "step": 5282
+    },
+    {
+      "epoch": 0.4636867064738296,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.0020180595343891675,
+      "loss": 3.2939,
+      "step": 5283
+    },
+    {
+      "epoch": 0.46377447605673205,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0020176668991980437,
+      "loss": 3.418,
+      "step": 5284
+    },
+    {
+      "epoch": 0.4638622456396346,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.002017274230412092,
+      "loss": 3.3657,
+      "step": 5285
+    },
+    {
+      "epoch": 0.46395001522253704,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.002016881528067192,
+      "loss": 3.3672,
+      "step": 5286
+    },
+    {
+      "epoch": 0.4640377848054395,
+      "grad_norm": 0.12353515625,
+      "learning_rate": 0.002016488792199226,
+      "loss": 3.2563,
+      "step": 5287
+    },
+    {
+      "epoch": 0.464125554388342,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.002016096022844079,
+      "loss": 3.2891,
+      "step": 5288
+    },
+    {
+      "epoch": 0.4642133239712445,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.00201570322003764,
+      "loss": 3.3618,
+      "step": 5289
+    },
+    {
+      "epoch": 0.46430109355414695,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0020153103838158,
+      "loss": 3.3232,
+      "step": 5290
+    },
+    {
+      "epoch": 0.46438886313704947,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.002014917514214454,
+      "loss": 3.2939,
+      "step": 5291
+    },
+    {
+      "epoch": 0.46447663271995193,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.002014524611269501,
+      "loss": 3.3276,
+      "step": 5292
+    },
+    {
+      "epoch": 0.46456440230285445,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0020141316750168395,
+      "loss": 3.3594,
+      "step": 5293
+    },
+    {
+      "epoch": 0.4646521718857569,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0020137387054923743,
+      "loss": 3.2461,
+      "step": 5294
+    },
+    {
+      "epoch": 0.4647399414686594,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0020133457027320116,
+      "loss": 3.2603,
+      "step": 5295
+    },
+    {
+      "epoch": 0.4648277110515619,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0020129526667716623,
+      "loss": 3.3252,
+      "step": 5296
+    },
+    {
+      "epoch": 0.46491548063446436,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0020125595976472387,
+      "loss": 3.3086,
+      "step": 5297
+    },
+    {
+      "epoch": 0.4650032502173669,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.002012166495394657,
+      "loss": 3.3228,
+      "step": 5298
+    },
+    {
+      "epoch": 0.46509101980026935,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.002011773360049835,
+      "loss": 3.2832,
+      "step": 5299
+    },
+    {
+      "epoch": 0.4651787893831718,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.002011380191648696,
+      "loss": 3.2881,
+      "step": 5300
+    },
+    {
+      "epoch": 0.46526655896607433,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0020109869902271644,
+      "loss": 3.3042,
+      "step": 5301
+    },
+    {
+      "epoch": 0.4653543285489768,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0020105937558211684,
+      "loss": 3.2949,
+      "step": 5302
+    },
+    {
+      "epoch": 0.4654420981318793,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.002010200488466639,
+      "loss": 3.3364,
+      "step": 5303
+    },
+    {
+      "epoch": 0.4655298677147818,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.00200980718819951,
+      "loss": 3.3384,
+      "step": 5304
+    },
+    {
+      "epoch": 0.46561763729768424,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0020094138550557188,
+      "loss": 3.2461,
+      "step": 5305
+    },
+    {
+      "epoch": 0.46570540688058676,
+      "grad_norm": 0.20703125,
+      "learning_rate": 0.0020090204890712044,
+      "loss": 3.3223,
+      "step": 5306
+    },
+    {
+      "epoch": 0.4657931764634892,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.002008627090281911,
+      "loss": 3.251,
+      "step": 5307
+    },
+    {
+      "epoch": 0.4658809460463917,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.002008233658723784,
+      "loss": 3.2666,
+      "step": 5308
+    },
+    {
+      "epoch": 0.4659687156292942,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0020078401944327736,
+      "loss": 3.2305,
+      "step": 5309
+    },
+    {
+      "epoch": 0.46605648521219667,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.00200744669744483,
+      "loss": 3.394,
+      "step": 5310
+    },
+    {
+      "epoch": 0.4661442547950992,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.002007053167795909,
+      "loss": 3.2803,
+      "step": 5311
+    },
+    {
+      "epoch": 0.46623202437800165,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0020066596055219686,
+      "loss": 3.3984,
+      "step": 5312
+    },
+    {
+      "epoch": 0.4663197939609041,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0020062660106589707,
+      "loss": 3.292,
+      "step": 5313
+    },
+    {
+      "epoch": 0.46640756354380664,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.0020058723832428777,
+      "loss": 3.27,
+      "step": 5314
+    },
+    {
+      "epoch": 0.4664953331267091,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0020054787233096576,
+      "loss": 3.3472,
+      "step": 5315
+    },
+    {
+      "epoch": 0.4665831027096116,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0020050850308952787,
+      "loss": 3.3398,
+      "step": 5316
+    },
+    {
+      "epoch": 0.4666708722925141,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0020046913060357157,
+      "loss": 3.3682,
+      "step": 5317
+    },
+    {
+      "epoch": 0.46675864187541655,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.002004297548766944,
+      "loss": 3.3979,
+      "step": 5318
+    },
+    {
+      "epoch": 0.46684641145831907,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0020039037591249417,
+      "loss": 3.3286,
+      "step": 5319
+    },
+    {
+      "epoch": 0.46693418104122153,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0020035099371456914,
+      "loss": 3.2729,
+      "step": 5320
+    },
+    {
+      "epoch": 0.467021950624124,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0020031160828651775,
+      "loss": 3.3525,
+      "step": 5321
+    },
+    {
+      "epoch": 0.4671097202070265,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.002002722196319387,
+      "loss": 3.3784,
+      "step": 5322
+    },
+    {
+      "epoch": 0.467197489789929,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0020023282775443116,
+      "loss": 3.3198,
+      "step": 5323
+    },
+    {
+      "epoch": 0.4672852593728315,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0020019343265759433,
+      "loss": 3.3472,
+      "step": 5324
+    },
+    {
+      "epoch": 0.46737302895573396,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0020015403434502804,
+      "loss": 3.3252,
+      "step": 5325
+    },
+    {
+      "epoch": 0.4674607985386364,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.002001146328203321,
+      "loss": 3.3491,
+      "step": 5326
+    },
+    {
+      "epoch": 0.46754856812153894,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0020007522808710683,
+      "loss": 3.3022,
+      "step": 5327
+    },
+    {
+      "epoch": 0.4676363377044414,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0020003582014895267,
+      "loss": 3.3438,
+      "step": 5328
+    },
+    {
+      "epoch": 0.4677241072873439,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.001999964090094705,
+      "loss": 3.3218,
+      "step": 5329
+    },
+    {
+      "epoch": 0.4678118768702464,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0019995699467226145,
+      "loss": 3.2778,
+      "step": 5330
+    },
+    {
+      "epoch": 0.46789964645314885,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0019991757714092687,
+      "loss": 3.376,
+      "step": 5331
+    },
+    {
+      "epoch": 0.4679874160360514,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.001998781564190685,
+      "loss": 3.2495,
+      "step": 5332
+    },
+    {
+      "epoch": 0.46807518561895384,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0019983873251028833,
+      "loss": 3.3184,
+      "step": 5333
+    },
+    {
+      "epoch": 0.4681629552018563,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.001997993054181886,
+      "loss": 3.3037,
+      "step": 5334
+    },
+    {
+      "epoch": 0.4682507247847588,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.001997598751463719,
+      "loss": 3.292,
+      "step": 5335
+    },
+    {
+      "epoch": 0.4683384943676613,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.0019972044169844117,
+      "loss": 3.3115,
+      "step": 5336
+    },
+    {
+      "epoch": 0.4684262639505638,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.0019968100507799947,
+      "loss": 3.3682,
+      "step": 5337
+    },
+    {
+      "epoch": 0.46851403353346627,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0019964156528865028,
+      "loss": 3.3682,
+      "step": 5338
+    },
+    {
+      "epoch": 0.46860180311636873,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.001996021223339973,
+      "loss": 3.3325,
+      "step": 5339
+    },
+    {
+      "epoch": 0.46868957269927125,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.001995626762176445,
+      "loss": 3.3628,
+      "step": 5340
+    },
+    {
+      "epoch": 0.4687773422821737,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0019952322694319637,
+      "loss": 3.3325,
+      "step": 5341
+    },
+    {
+      "epoch": 0.46886511186507623,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0019948377451425737,
+      "loss": 3.2432,
+      "step": 5342
+    },
+    {
+      "epoch": 0.4689528814479787,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0019944431893443243,
+      "loss": 3.3506,
+      "step": 5343
+    },
+    {
+      "epoch": 0.46904065103088116,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.0019940486020732667,
+      "loss": 3.3833,
+      "step": 5344
+    },
+    {
+      "epoch": 0.4691284206137837,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.001993653983365456,
+      "loss": 3.3027,
+      "step": 5345
+    },
+    {
+      "epoch": 0.46921619019668614,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.00199325933325695,
+      "loss": 3.2588,
+      "step": 5346
+    },
+    {
+      "epoch": 0.4693039597795886,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.001992864651783809,
+      "loss": 3.2822,
+      "step": 5347
+    },
+    {
+      "epoch": 0.46939172936249113,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.001992469938982095,
+      "loss": 3.3384,
+      "step": 5348
+    },
+    {
+      "epoch": 0.4694794989453936,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.001992075194887875,
+      "loss": 3.3013,
+      "step": 5349
+    },
+    {
+      "epoch": 0.4695672685282961,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0019916804195372188,
+      "loss": 3.3374,
+      "step": 5350
+    },
+    {
+      "epoch": 0.4696550381111986,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0019912856129661967,
+      "loss": 3.3086,
+      "step": 5351
+    },
+    {
+      "epoch": 0.46974280769410104,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0019908907752108846,
+      "loss": 3.3052,
+      "step": 5352
+    },
+    {
+      "epoch": 0.46983057727700356,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0019904959063073587,
+      "loss": 3.2959,
+      "step": 5353
+    },
+    {
+      "epoch": 0.469918346859906,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.001990101006291701,
+      "loss": 3.3369,
+      "step": 5354
+    },
+    {
+      "epoch": 0.47000611644280854,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0019897060751999933,
+      "loss": 3.3115,
+      "step": 5355
+    },
+    {
+      "epoch": 0.470093886025711,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.001989311113068322,
+      "loss": 3.3086,
+      "step": 5356
+    },
+    {
+      "epoch": 0.47018165560861347,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0019889161199327762,
+      "loss": 3.3237,
+      "step": 5357
+    },
+    {
+      "epoch": 0.470269425191516,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0019885210958294477,
+      "loss": 3.3105,
+      "step": 5358
+    },
+    {
+      "epoch": 0.47035719477441845,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.001988126040794431,
+      "loss": 3.4009,
+      "step": 5359
+    },
+    {
+      "epoch": 0.4704449643573209,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0019877309548638235,
+      "loss": 3.2676,
+      "step": 5360
+    },
+    {
+      "epoch": 0.47053273394022344,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001987335838073725,
+      "loss": 3.3242,
+      "step": 5361
+    },
+    {
+      "epoch": 0.4706205035231259,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0019869406904602384,
+      "loss": 3.2505,
+      "step": 5362
+    },
+    {
+      "epoch": 0.4707082731060284,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0019865455120594696,
+      "loss": 3.2695,
+      "step": 5363
+    },
+    {
+      "epoch": 0.4707960426889309,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.0019861503029075284,
+      "loss": 3.3047,
+      "step": 5364
+    },
+    {
+      "epoch": 0.47088381227183335,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0019857550630405247,
+      "loss": 3.3555,
+      "step": 5365
+    },
+    {
+      "epoch": 0.47097158185473587,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0019853597924945733,
+      "loss": 3.2441,
+      "step": 5366
+    },
+    {
+      "epoch": 0.47105935143763833,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0019849644913057917,
+      "loss": 3.3462,
+      "step": 5367
+    },
+    {
+      "epoch": 0.47114712102054085,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0019845691595102998,
+      "loss": 3.2876,
+      "step": 5368
+    },
+    {
+      "epoch": 0.4712348906034433,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0019841737971442192,
+      "loss": 3.2412,
+      "step": 5369
+    },
+    {
+      "epoch": 0.4713226601863458,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.0019837784042436757,
+      "loss": 3.2803,
+      "step": 5370
+    },
+    {
+      "epoch": 0.4714104297692483,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0019833829808447987,
+      "loss": 3.3022,
+      "step": 5371
+    },
+    {
+      "epoch": 0.47149819935215076,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.001982987526983718,
+      "loss": 3.2964,
+      "step": 5372
+    },
+    {
+      "epoch": 0.4715859689350533,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0019825920426965672,
+      "loss": 3.3228,
+      "step": 5373
+    },
+    {
+      "epoch": 0.47167373851795574,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.0019821965280194843,
+      "loss": 3.2769,
+      "step": 5374
+    },
+    {
+      "epoch": 0.4717615081008582,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0019818009829886073,
+      "loss": 3.3628,
+      "step": 5375
+    },
+    {
+      "epoch": 0.4718492776837607,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.001981405407640079,
+      "loss": 3.2495,
+      "step": 5376
+    },
+    {
+      "epoch": 0.4719370472666632,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0019810098020100445,
+      "loss": 3.3408,
+      "step": 5377
+    },
+    {
+      "epoch": 0.47202481684956565,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00198061416613465,
+      "loss": 3.2646,
+      "step": 5378
+    },
+    {
+      "epoch": 0.4721125864324682,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.0019802185000500484,
+      "loss": 3.2627,
+      "step": 5379
+    },
+    {
+      "epoch": 0.47220035601537064,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0019798228037923907,
+      "loss": 3.3042,
+      "step": 5380
+    },
+    {
+      "epoch": 0.47228812559827316,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.001979427077397834,
+      "loss": 3.2812,
+      "step": 5381
+    },
+    {
+      "epoch": 0.4723758951811756,
+      "grad_norm": 0.1279296875,
+      "learning_rate": 0.0019790313209025368,
+      "loss": 3.2529,
+      "step": 5382
+    },
+    {
+      "epoch": 0.4724636647640781,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0019786355343426603,
+      "loss": 3.3438,
+      "step": 5383
+    },
+    {
+      "epoch": 0.4725514343469806,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.001978239717754369,
+      "loss": 3.2744,
+      "step": 5384
+    },
+    {
+      "epoch": 0.47263920392988307,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0019778438711738297,
+      "loss": 3.3735,
+      "step": 5385
+    },
+    {
+      "epoch": 0.4727269735127856,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0019774479946372122,
+      "loss": 3.2837,
+      "step": 5386
+    },
+    {
+      "epoch": 0.47281474309568805,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0019770520881806884,
+      "loss": 3.2388,
+      "step": 5387
+    },
+    {
+      "epoch": 0.4729025126785905,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.001976656151840434,
+      "loss": 3.3223,
+      "step": 5388
+    },
+    {
+      "epoch": 0.47299028226149303,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0019762601856526275,
+      "loss": 3.3389,
+      "step": 5389
+    },
+    {
+      "epoch": 0.4730780518443955,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0019758641896534487,
+      "loss": 3.2856,
+      "step": 5390
+    },
+    {
+      "epoch": 0.47316582142729796,
+      "grad_norm": 0.08349609375,
+      "learning_rate": 0.0019754681638790806,
+      "loss": 3.3184,
+      "step": 5391
+    },
+    {
+      "epoch": 0.4732535910102005,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0019750721083657097,
+      "loss": 3.2749,
+      "step": 5392
+    },
+    {
+      "epoch": 0.47334136059310294,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.001974676023149525,
+      "loss": 3.3115,
+      "step": 5393
+    },
+    {
+      "epoch": 0.47342913017600546,
+      "grad_norm": 0.1240234375,
+      "learning_rate": 0.001974279908266718,
+      "loss": 3.2656,
+      "step": 5394
+    },
+    {
+      "epoch": 0.4735168997589079,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0019738837637534824,
+      "loss": 3.2993,
+      "step": 5395
+    },
+    {
+      "epoch": 0.4736046693418104,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0019734875896460155,
+      "loss": 3.2808,
+      "step": 5396
+    },
+    {
+      "epoch": 0.4736924389247129,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0019730913859805172,
+      "loss": 3.3149,
+      "step": 5397
+    },
+    {
+      "epoch": 0.4737802085076154,
+      "grad_norm": 0.15625,
+      "learning_rate": 0.001972695152793189,
+      "loss": 3.3677,
+      "step": 5398
+    },
+    {
+      "epoch": 0.4738679780905179,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.0019722988901202366,
+      "loss": 3.3477,
+      "step": 5399
+    },
+    {
+      "epoch": 0.47395574767342036,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.001971902597997867,
+      "loss": 3.2939,
+      "step": 5400
+    },
+    {
+      "epoch": 0.4740435172563228,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.001971506276462292,
+      "loss": 3.2803,
+      "step": 5401
+    },
+    {
+      "epoch": 0.47413128683922534,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.001971109925549723,
+      "loss": 3.3984,
+      "step": 5402
+    },
+    {
+      "epoch": 0.4742190564221278,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.001970713545296377,
+      "loss": 3.2095,
+      "step": 5403
+    },
+    {
+      "epoch": 0.47430682600503027,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0019703171357384713,
+      "loss": 3.2905,
+      "step": 5404
+    },
+    {
+      "epoch": 0.4743945955879328,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0019699206969122285,
+      "loss": 3.3428,
+      "step": 5405
+    },
+    {
+      "epoch": 0.47448236517083525,
+      "grad_norm": 0.095703125,
+      "learning_rate": 0.0019695242288538715,
+      "loss": 3.272,
+      "step": 5406
+    },
+    {
+      "epoch": 0.47457013475373777,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0019691277315996266,
+      "loss": 3.3071,
+      "step": 5407
+    },
+    {
+      "epoch": 0.47465790433664024,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.001968731205185724,
+      "loss": 3.3384,
+      "step": 5408
+    },
+    {
+      "epoch": 0.4747456739195427,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0019683346496483936,
+      "loss": 3.314,
+      "step": 5409
+    },
+    {
+      "epoch": 0.4748334435024452,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.001967938065023872,
+      "loss": 3.2485,
+      "step": 5410
+    },
+    {
+      "epoch": 0.4749212130853477,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0019675414513483946,
+      "loss": 3.2773,
+      "step": 5411
+    },
+    {
+      "epoch": 0.4750089826682502,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0019671448086582026,
+      "loss": 3.2793,
+      "step": 5412
+    },
+    {
+      "epoch": 0.47509675225115267,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.0019667481369895373,
+      "loss": 3.3696,
+      "step": 5413
+    },
+    {
+      "epoch": 0.47518452183405513,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0019663514363786445,
+      "loss": 3.2871,
+      "step": 5414
+    },
+    {
+      "epoch": 0.47527229141695765,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.001965954706861771,
+      "loss": 3.3062,
+      "step": 5415
+    },
+    {
+      "epoch": 0.4753600609998601,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.001965557948475169,
+      "loss": 3.3135,
+      "step": 5416
+    },
+    {
+      "epoch": 0.4754478305827626,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0019651611612550896,
+      "loss": 3.3525,
+      "step": 5417
+    },
+    {
+      "epoch": 0.4755356001656651,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0019647643452377897,
+      "loss": 3.3413,
+      "step": 5418
+    },
+    {
+      "epoch": 0.47562336974856756,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0019643675004595265,
+      "loss": 3.3506,
+      "step": 5419
+    },
+    {
+      "epoch": 0.4757111393314701,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0019639706269565615,
+      "loss": 3.3262,
+      "step": 5420
+    },
+    {
+      "epoch": 0.47579890891437254,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0019635737247651586,
+      "loss": 3.3042,
+      "step": 5421
+    },
+    {
+      "epoch": 0.475886678497275,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.001963176793921583,
+      "loss": 3.2769,
+      "step": 5422
+    },
+    {
+      "epoch": 0.4759744480801775,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0019627798344621044,
+      "loss": 3.3228,
+      "step": 5423
+    },
+    {
+      "epoch": 0.47606221766308,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0019623828464229937,
+      "loss": 3.3652,
+      "step": 5424
+    },
+    {
+      "epoch": 0.4761499872459825,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.001961985829840525,
+      "loss": 3.2808,
+      "step": 5425
+    },
+    {
+      "epoch": 0.476237756828885,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.001961588784750975,
+      "loss": 3.3335,
+      "step": 5426
+    },
+    {
+      "epoch": 0.47632552641178744,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.001961191711190622,
+      "loss": 3.3052,
+      "step": 5427
+    },
+    {
+      "epoch": 0.47641329599468996,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.001960794609195749,
+      "loss": 3.3174,
+      "step": 5428
+    },
+    {
+      "epoch": 0.4765010655775924,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0019603974788026403,
+      "loss": 3.2793,
+      "step": 5429
+    },
+    {
+      "epoch": 0.4765888351604949,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.001960000320047582,
+      "loss": 3.2861,
+      "step": 5430
+    },
+    {
+      "epoch": 0.4766766047433974,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.001959603132966865,
+      "loss": 3.2686,
+      "step": 5431
+    },
+    {
+      "epoch": 0.47676437432629987,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00195920591759678,
+      "loss": 3.376,
+      "step": 5432
+    },
+    {
+      "epoch": 0.4768521439092024,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001958808673973622,
+      "loss": 3.3716,
+      "step": 5433
+    },
+    {
+      "epoch": 0.47693991349210485,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00195841140213369,
+      "loss": 3.3516,
+      "step": 5434
+    },
+    {
+      "epoch": 0.4770276830750073,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.001958014102113282,
+      "loss": 3.2944,
+      "step": 5435
+    },
+    {
+      "epoch": 0.47711545265790983,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0019576167739487013,
+      "loss": 3.3105,
+      "step": 5436
+    },
+    {
+      "epoch": 0.4772032222408123,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.001957219417676253,
+      "loss": 3.4043,
+      "step": 5437
+    },
+    {
+      "epoch": 0.4772909918237148,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.001956822033332245,
+      "loss": 3.2593,
+      "step": 5438
+    },
+    {
+      "epoch": 0.4773787614066173,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0019564246209529865,
+      "loss": 3.2983,
+      "step": 5439
+    },
+    {
+      "epoch": 0.47746653098951974,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0019560271805747907,
+      "loss": 3.312,
+      "step": 5440
+    },
+    {
+      "epoch": 0.47755430057242226,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.001955629712233974,
+      "loss": 3.2808,
+      "step": 5441
+    },
+    {
+      "epoch": 0.4776420701553247,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0019552322159668532,
+      "loss": 3.3354,
+      "step": 5442
+    },
+    {
+      "epoch": 0.47772983973822725,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0019548346918097483,
+      "loss": 3.3579,
+      "step": 5443
+    },
+    {
+      "epoch": 0.4778176093211297,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.001954437139798983,
+      "loss": 3.3052,
+      "step": 5444
+    },
+    {
+      "epoch": 0.4779053789040322,
+      "grad_norm": 0.1396484375,
+      "learning_rate": 0.0019540395599708833,
+      "loss": 3.3232,
+      "step": 5445
+    },
+    {
+      "epoch": 0.4779931484869347,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.001953641952361776,
+      "loss": 3.3989,
+      "step": 5446
+    },
+    {
+      "epoch": 0.47808091806983716,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.001953244317007993,
+      "loss": 3.3135,
+      "step": 5447
+    },
+    {
+      "epoch": 0.4781686876527396,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0019528466539458667,
+      "loss": 3.2788,
+      "step": 5448
+    },
+    {
+      "epoch": 0.47825645723564214,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.0019524489632117327,
+      "loss": 3.2925,
+      "step": 5449
+    },
+    {
+      "epoch": 0.4783442268185446,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.001952051244841929,
+      "loss": 3.2422,
+      "step": 5450
+    },
+    {
+      "epoch": 0.4784319964014471,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0019516534988727968,
+      "loss": 3.3286,
+      "step": 5451
+    },
+    {
+      "epoch": 0.4785197659843496,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0019512557253406795,
+      "loss": 3.2861,
+      "step": 5452
+    },
+    {
+      "epoch": 0.47860753556725205,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0019508579242819226,
+      "loss": 3.292,
+      "step": 5453
+    },
+    {
+      "epoch": 0.47869530515015457,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0019504600957328743,
+      "loss": 3.3311,
+      "step": 5454
+    },
+    {
+      "epoch": 0.47878307473305703,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.0019500622397298855,
+      "loss": 3.3364,
+      "step": 5455
+    },
+    {
+      "epoch": 0.47887084431595955,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0019496643563093091,
+      "loss": 3.3042,
+      "step": 5456
+    },
+    {
+      "epoch": 0.478958613898862,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.001949266445507502,
+      "loss": 3.2451,
+      "step": 5457
+    },
+    {
+      "epoch": 0.4790463834817645,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0019488685073608214,
+      "loss": 3.3618,
+      "step": 5458
+    },
+    {
+      "epoch": 0.479134153064667,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0019484705419056285,
+      "loss": 3.3281,
+      "step": 5459
+    },
+    {
+      "epoch": 0.47922192264756946,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0019480725491782865,
+      "loss": 3.2783,
+      "step": 5460
+    },
+    {
+      "epoch": 0.47930969223047193,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.0019476745292151608,
+      "loss": 3.2373,
+      "step": 5461
+    },
+    {
+      "epoch": 0.47939746181337445,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0019472764820526209,
+      "loss": 3.314,
+      "step": 5462
+    },
+    {
+      "epoch": 0.4794852313962769,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.0019468784077270361,
+      "loss": 3.3076,
+      "step": 5463
+    },
+    {
+      "epoch": 0.47957300097917943,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.001946480306274781,
+      "loss": 3.3438,
+      "step": 5464
+    },
+    {
+      "epoch": 0.4796607705620819,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0019460821777322302,
+      "loss": 3.3667,
+      "step": 5465
+    },
+    {
+      "epoch": 0.47974854014498436,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0019456840221357627,
+      "loss": 3.3008,
+      "step": 5466
+    },
+    {
+      "epoch": 0.4798363097278869,
+      "grad_norm": 0.173828125,
+      "learning_rate": 0.0019452858395217583,
+      "loss": 3.2925,
+      "step": 5467
+    },
+    {
+      "epoch": 0.47992407931078934,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.001944887629926601,
+      "loss": 3.2705,
+      "step": 5468
+    },
+    {
+      "epoch": 0.48001184889369186,
+      "grad_norm": 0.08251953125,
+      "learning_rate": 0.001944489393386676,
+      "loss": 3.2441,
+      "step": 5469
+    },
+    {
+      "epoch": 0.4800996184765943,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0019440911299383715,
+      "loss": 3.3105,
+      "step": 5470
+    },
+    {
+      "epoch": 0.4801873880594968,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.001943692839618078,
+      "loss": 3.2974,
+      "step": 5471
+    },
+    {
+      "epoch": 0.4802751576423993,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0019432945224621885,
+      "loss": 3.3096,
+      "step": 5472
+    },
+    {
+      "epoch": 0.4803629272253018,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0019428961785070984,
+      "loss": 3.3208,
+      "step": 5473
+    },
+    {
+      "epoch": 0.48045069680820424,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0019424978077892054,
+      "loss": 3.2661,
+      "step": 5474
+    },
+    {
+      "epoch": 0.48053846639110676,
+      "grad_norm": 0.07568359375,
+      "learning_rate": 0.0019420994103449105,
+      "loss": 3.2207,
+      "step": 5475
+    },
+    {
+      "epoch": 0.4806262359740092,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0019417009862106155,
+      "loss": 3.3564,
+      "step": 5476
+    },
+    {
+      "epoch": 0.48071400555691174,
+      "grad_norm": 0.08203125,
+      "learning_rate": 0.0019413025354227261,
+      "loss": 3.3081,
+      "step": 5477
+    },
+    {
+      "epoch": 0.4808017751398142,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.00194090405801765,
+      "loss": 3.3521,
+      "step": 5478
+    },
+    {
+      "epoch": 0.48088954472271667,
+      "grad_norm": 0.0888671875,
+      "learning_rate": 0.001940505554031797,
+      "loss": 3.3447,
+      "step": 5479
+    },
+    {
+      "epoch": 0.4809773143056192,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.00194010702350158,
+      "loss": 3.2495,
+      "step": 5480
+    },
+    {
+      "epoch": 0.48106508388852165,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.001939708466463414,
+      "loss": 3.3115,
+      "step": 5481
+    },
+    {
+      "epoch": 0.48115285347142417,
+      "grad_norm": 0.1953125,
+      "learning_rate": 0.0019393098829537157,
+      "loss": 3.2803,
+      "step": 5482
+    },
+    {
+      "epoch": 0.48124062305432663,
+      "grad_norm": 0.0947265625,
+      "learning_rate": 0.0019389112730089054,
+      "loss": 3.3169,
+      "step": 5483
+    },
+    {
+      "epoch": 0.4813283926372291,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0019385126366654052,
+      "loss": 3.3428,
+      "step": 5484
+    },
+    {
+      "epoch": 0.4814161622201316,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0019381139739596393,
+      "loss": 3.3511,
+      "step": 5485
+    },
+    {
+      "epoch": 0.4815039318030341,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0019377152849280354,
+      "loss": 3.2588,
+      "step": 5486
+    },
+    {
+      "epoch": 0.48159170138593654,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0019373165696070223,
+      "loss": 3.3086,
+      "step": 5487
+    },
+    {
+      "epoch": 0.48167947096883906,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0019369178280330321,
+      "loss": 3.4238,
+      "step": 5488
+    },
+    {
+      "epoch": 0.4817672405517415,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.0019365190602424992,
+      "loss": 3.3462,
+      "step": 5489
+    },
+    {
+      "epoch": 0.48185501013464405,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0019361202662718604,
+      "loss": 3.248,
+      "step": 5490
+    },
+    {
+      "epoch": 0.4819427797175465,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0019357214461575534,
+      "loss": 3.3618,
+      "step": 5491
+    },
+    {
+      "epoch": 0.482030549300449,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.001935322599936021,
+      "loss": 3.2842,
+      "step": 5492
+    },
+    {
+      "epoch": 0.4821183188833515,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.0019349237276437066,
+      "loss": 3.3032,
+      "step": 5493
+    },
+    {
+      "epoch": 0.48220608846625396,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0019345248293170562,
+      "loss": 3.3428,
+      "step": 5494
+    },
+    {
+      "epoch": 0.4822938580491565,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0019341259049925179,
+      "loss": 3.2876,
+      "step": 5495
+    },
+    {
+      "epoch": 0.48238162763205894,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0019337269547065434,
+      "loss": 3.2935,
+      "step": 5496
+    },
+    {
+      "epoch": 0.4824693972149614,
+      "grad_norm": 0.1259765625,
+      "learning_rate": 0.0019333279784955856,
+      "loss": 3.2622,
+      "step": 5497
+    },
+    {
+      "epoch": 0.4825571667978639,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.0019329289763961005,
+      "loss": 3.3901,
+      "step": 5498
+    },
+    {
+      "epoch": 0.4826449363807664,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0019325299484445452,
+      "loss": 3.3057,
+      "step": 5499
+    },
+    {
+      "epoch": 0.48273270596366885,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 0.0019321308946773812,
+      "loss": 3.3701,
+      "step": 5500
+    },
+    {
+      "epoch": 0.48282047554657137,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0019317318151310708,
+      "loss": 3.2617,
+      "step": 5501
+    },
+    {
+      "epoch": 0.48290824512947383,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.0019313327098420788,
+      "loss": 3.3125,
+      "step": 5502
+    },
+    {
+      "epoch": 0.48299601471237635,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.001930933578846873,
+      "loss": 3.252,
+      "step": 5503
+    },
+    {
+      "epoch": 0.4830837842952788,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0019305344221819231,
+      "loss": 3.3076,
+      "step": 5504
+    },
+    {
+      "epoch": 0.4831715538781813,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.0019301352398837014,
+      "loss": 3.3442,
+      "step": 5505
+    },
+    {
+      "epoch": 0.4832593234610838,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0019297360319886821,
+      "loss": 3.2915,
+      "step": 5506
+    },
+    {
+      "epoch": 0.48334709304398626,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0019293367985333421,
+      "loss": 3.4097,
+      "step": 5507
+    },
+    {
+      "epoch": 0.4834348626268888,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0019289375395541606,
+      "loss": 3.3228,
+      "step": 5508
+    },
+    {
+      "epoch": 0.48352263220979125,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0019285382550876193,
+      "loss": 3.2881,
+      "step": 5509
+    },
+    {
+      "epoch": 0.4836104017926937,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.001928138945170202,
+      "loss": 3.2593,
+      "step": 5510
+    },
+    {
+      "epoch": 0.48369817137559623,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0019277396098383941,
+      "loss": 3.3125,
+      "step": 5511
+    },
+    {
+      "epoch": 0.4837859409584987,
+      "grad_norm": 0.25,
+      "learning_rate": 0.001927340249128685,
+      "loss": 3.3486,
+      "step": 5512
+    },
+    {
+      "epoch": 0.4838737105414012,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.0019269408630775647,
+      "loss": 3.3193,
+      "step": 5513
+    },
+    {
+      "epoch": 0.4839614801243037,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0019265414517215273,
+      "loss": 3.2168,
+      "step": 5514
+    },
+    {
+      "epoch": 0.48404924970720614,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0019261420150970673,
+      "loss": 3.3242,
+      "step": 5515
+    },
+    {
+      "epoch": 0.48413701929010866,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.0019257425532406827,
+      "loss": 3.395,
+      "step": 5516
+    },
+    {
+      "epoch": 0.4842247888730111,
+      "grad_norm": 0.28125,
+      "learning_rate": 0.0019253430661888735,
+      "loss": 3.2759,
+      "step": 5517
+    },
+    {
+      "epoch": 0.4843125584559136,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0019249435539781426,
+      "loss": 3.2563,
+      "step": 5518
+    },
+    {
+      "epoch": 0.4844003280388161,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0019245440166449938,
+      "loss": 3.29,
+      "step": 5519
+    },
+    {
+      "epoch": 0.48448809762171857,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.001924144454225934,
+      "loss": 3.3882,
+      "step": 5520
+    },
+    {
+      "epoch": 0.4845758672046211,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0019237448667574733,
+      "loss": 3.3516,
+      "step": 5521
+    },
+    {
+      "epoch": 0.48466363678752356,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0019233452542761224,
+      "loss": 3.3174,
+      "step": 5522
+    },
+    {
+      "epoch": 0.484751406370426,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0019229456168183957,
+      "loss": 3.3301,
+      "step": 5523
+    },
+    {
+      "epoch": 0.48483917595332854,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0019225459544208092,
+      "loss": 3.2168,
+      "step": 5524
+    },
+    {
+      "epoch": 0.484926945536231,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0019221462671198805,
+      "loss": 3.3335,
+      "step": 5525
+    },
+    {
+      "epoch": 0.4850147151191335,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.001921746554952131,
+      "loss": 3.2905,
+      "step": 5526
+    },
+    {
+      "epoch": 0.485102484702036,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0019213468179540831,
+      "loss": 3.3652,
+      "step": 5527
+    },
+    {
+      "epoch": 0.48519025428493845,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0019209470561622624,
+      "loss": 3.333,
+      "step": 5528
+    },
+    {
+      "epoch": 0.48527802386784097,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.001920547269613196,
+      "loss": 3.3145,
+      "step": 5529
+    },
+    {
+      "epoch": 0.48536579345074343,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0019201474583434139,
+      "loss": 3.2769,
+      "step": 5530
+    },
+    {
+      "epoch": 0.4854535630336459,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0019197476223894476,
+      "loss": 3.2349,
+      "step": 5531
+    },
+    {
+      "epoch": 0.4855413326165484,
+      "grad_norm": 0.10498046875,
+      "learning_rate": 0.0019193477617878319,
+      "loss": 3.291,
+      "step": 5532
+    },
+    {
+      "epoch": 0.4856291021994509,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.001918947876575103,
+      "loss": 3.3457,
+      "step": 5533
+    },
+    {
+      "epoch": 0.4857168717823534,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.0019185479667877994,
+      "loss": 3.2759,
+      "step": 5534
+    },
+    {
+      "epoch": 0.48580464136525586,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0019181480324624624,
+      "loss": 3.2534,
+      "step": 5535
+    },
+    {
+      "epoch": 0.4858924109481583,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.001917748073635635,
+      "loss": 3.311,
+      "step": 5536
+    },
+    {
+      "epoch": 0.48598018053106085,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0019173480903438625,
+      "loss": 3.3354,
+      "step": 5537
+    },
+    {
+      "epoch": 0.4860679501139633,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0019169480826236927,
+      "loss": 3.3062,
+      "step": 5538
+    },
+    {
+      "epoch": 0.48615571969686583,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0019165480505116757,
+      "loss": 3.2339,
+      "step": 5539
+    },
+    {
+      "epoch": 0.4862434892797683,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0019161479940443635,
+      "loss": 3.3384,
+      "step": 5540
+    },
+    {
+      "epoch": 0.48633125886267076,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0019157479132583108,
+      "loss": 3.3184,
+      "step": 5541
+    },
+    {
+      "epoch": 0.4864190284455733,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.0019153478081900734,
+      "loss": 3.3545,
+      "step": 5542
+    },
+    {
+      "epoch": 0.48650679802847574,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0019149476788762108,
+      "loss": 3.3008,
+      "step": 5543
+    },
+    {
+      "epoch": 0.4865945676113782,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0019145475253532832,
+      "loss": 3.3213,
+      "step": 5544
+    },
+    {
+      "epoch": 0.4866823371942807,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0019141473476578547,
+      "loss": 3.314,
+      "step": 5545
+    },
+    {
+      "epoch": 0.4867701067771832,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.001913747145826491,
+      "loss": 3.2988,
+      "step": 5546
+    },
+    {
+      "epoch": 0.4868578763600857,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0019133469198957585,
+      "loss": 3.3491,
+      "step": 5547
+    },
+    {
+      "epoch": 0.48694564594298817,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0019129466699022272,
+      "loss": 3.2803,
+      "step": 5548
+    },
+    {
+      "epoch": 0.48703341552589063,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.0019125463958824705,
+      "loss": 3.3462,
+      "step": 5549
+    },
+    {
+      "epoch": 0.48712118510879315,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0019121460978730614,
+      "loss": 3.353,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4872089546916956,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.0019117457759105772,
+      "loss": 3.2769,
+      "step": 5551
+    },
+    {
+      "epoch": 0.48729672427459814,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0019113454300315952,
+      "loss": 3.2803,
+      "step": 5552
+    },
+    {
+      "epoch": 0.4873844938575006,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.001910945060272698,
+      "loss": 3.2773,
+      "step": 5553
+    },
+    {
+      "epoch": 0.48747226344040306,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0019105446666704668,
+      "loss": 3.3101,
+      "step": 5554
+    },
+    {
+      "epoch": 0.4875600330233056,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.001910144249261488,
+      "loss": 3.333,
+      "step": 5555
+    },
+    {
+      "epoch": 0.48764780260620805,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0019097438080823486,
+      "loss": 3.3296,
+      "step": 5556
+    },
+    {
+      "epoch": 0.4877355721891105,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0019093433431696376,
+      "loss": 3.3193,
+      "step": 5557
+    },
+    {
+      "epoch": 0.48782334177201303,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.001908942854559948,
+      "loss": 3.2793,
+      "step": 5558
+    },
+    {
+      "epoch": 0.4879111113549155,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0019085423422898716,
+      "loss": 3.2886,
+      "step": 5559
+    },
+    {
+      "epoch": 0.487998880937818,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0019081418063960065,
+      "loss": 3.3071,
+      "step": 5560
+    },
+    {
+      "epoch": 0.4880866505207205,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0019077412469149493,
+      "loss": 3.2983,
+      "step": 5561
+    },
+    {
+      "epoch": 0.48817442010362294,
+      "grad_norm": 0.18359375,
+      "learning_rate": 0.0019073406638833014,
+      "loss": 3.4062,
+      "step": 5562
+    },
+    {
+      "epoch": 0.48826218968652546,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0019069400573376648,
+      "loss": 3.3447,
+      "step": 5563
+    },
+    {
+      "epoch": 0.4883499592694279,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.001906539427314644,
+      "loss": 3.3237,
+      "step": 5564
+    },
+    {
+      "epoch": 0.48843772885233044,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0019061387738508457,
+      "loss": 3.3516,
+      "step": 5565
+    },
+    {
+      "epoch": 0.4885254984352329,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0019057380969828793,
+      "loss": 3.2754,
+      "step": 5566
+    },
+    {
+      "epoch": 0.48861326801813537,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0019053373967473555,
+      "loss": 3.3145,
+      "step": 5567
+    },
+    {
+      "epoch": 0.4887010376010379,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0019049366731808879,
+      "loss": 3.3081,
+      "step": 5568
+    },
+    {
+      "epoch": 0.48878880718394035,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0019045359263200913,
+      "loss": 3.2944,
+      "step": 5569
+    },
+    {
+      "epoch": 0.4888765767668428,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.001904135156201583,
+      "loss": 3.2427,
+      "step": 5570
+    },
+    {
+      "epoch": 0.48896434634974534,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0019037343628619833,
+      "loss": 3.2402,
+      "step": 5571
+    },
+    {
+      "epoch": 0.4890521159326478,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0019033335463379127,
+      "loss": 3.3232,
+      "step": 5572
+    },
+    {
+      "epoch": 0.4891398855155503,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0019029327066659967,
+      "loss": 3.2642,
+      "step": 5573
+    },
+    {
+      "epoch": 0.4892276550984528,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0019025318438828603,
+      "loss": 3.2832,
+      "step": 5574
+    },
+    {
+      "epoch": 0.48931542468135525,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0019021309580251312,
+      "loss": 3.2783,
+      "step": 5575
+    },
+    {
+      "epoch": 0.48940319426425777,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0019017300491294402,
+      "loss": 3.2686,
+      "step": 5576
+    },
+    {
+      "epoch": 0.48949096384716023,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0019013291172324189,
+      "loss": 3.2529,
+      "step": 5577
+    },
+    {
+      "epoch": 0.48957873343006275,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0019009281623707016,
+      "loss": 3.2554,
+      "step": 5578
+    },
+    {
+      "epoch": 0.4896665030129652,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0019005271845809258,
+      "loss": 3.2681,
+      "step": 5579
+    },
+    {
+      "epoch": 0.4897542725958677,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0019001261838997295,
+      "loss": 3.2832,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4898420421787702,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0018997251603637528,
+      "loss": 3.2827,
+      "step": 5581
+    },
+    {
+      "epoch": 0.48992981176167266,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0018993241140096387,
+      "loss": 3.4097,
+      "step": 5582
+    },
+    {
+      "epoch": 0.4900175813445752,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0018989230448740325,
+      "loss": 3.2358,
+      "step": 5583
+    },
+    {
+      "epoch": 0.49010535092747765,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0018985219529935803,
+      "loss": 3.252,
+      "step": 5584
+    },
+    {
+      "epoch": 0.4901931205103801,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0018981208384049324,
+      "loss": 3.3389,
+      "step": 5585
+    },
+    {
+      "epoch": 0.49028089009328263,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0018977197011447382,
+      "loss": 3.3599,
+      "step": 5586
+    },
+    {
+      "epoch": 0.4903686596761851,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.001897318541249652,
+      "loss": 3.3716,
+      "step": 5587
+    },
+    {
+      "epoch": 0.49045642925908756,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0018969173587563284,
+      "loss": 3.3833,
+      "step": 5588
+    },
+    {
+      "epoch": 0.4905441988419901,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0018965161537014244,
+      "loss": 3.3594,
+      "step": 5589
+    },
+    {
+      "epoch": 0.49063196842489254,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0018961149261216002,
+      "loss": 3.312,
+      "step": 5590
+    },
+    {
+      "epoch": 0.49071973800779506,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0018957136760535168,
+      "loss": 3.2959,
+      "step": 5591
+    },
+    {
+      "epoch": 0.4908075075906975,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.0018953124035338371,
+      "loss": 3.3403,
+      "step": 5592
+    },
+    {
+      "epoch": 0.4908952771736,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0018949111085992273,
+      "loss": 3.2422,
+      "step": 5593
+    },
+    {
+      "epoch": 0.4909830467565025,
+      "grad_norm": 0.07275390625,
+      "learning_rate": 0.0018945097912863539,
+      "loss": 3.2842,
+      "step": 5594
+    },
+    {
+      "epoch": 0.49107081633940497,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.001894108451631888,
+      "loss": 3.271,
+      "step": 5595
+    },
+    {
+      "epoch": 0.4911585859223075,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0018937070896725,
+      "loss": 3.251,
+      "step": 5596
+    },
+    {
+      "epoch": 0.49124635550520995,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0018933057054448638,
+      "loss": 3.2666,
+      "step": 5597
+    },
+    {
+      "epoch": 0.4913341250881124,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0018929042989856556,
+      "loss": 3.3203,
+      "step": 5598
+    },
+    {
+      "epoch": 0.49142189467101494,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0018925028703315526,
+      "loss": 3.2461,
+      "step": 5599
+    },
+    {
+      "epoch": 0.4915096642539174,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.001892101419519234,
+      "loss": 3.3472,
+      "step": 5600
+    },
+    {
+      "epoch": 0.49159743383681986,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0018916999465853832,
+      "loss": 3.3135,
+      "step": 5601
+    },
+    {
+      "epoch": 0.4916852034197224,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0018912984515666823,
+      "loss": 3.2896,
+      "step": 5602
+    },
+    {
+      "epoch": 0.49177297300262485,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0018908969344998182,
+      "loss": 3.2476,
+      "step": 5603
+    },
+    {
+      "epoch": 0.49186074258552737,
+      "grad_norm": 0.11572265625,
+      "learning_rate": 0.0018904953954214782,
+      "loss": 3.2729,
+      "step": 5604
+    },
+    {
+      "epoch": 0.49194851216842983,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0018900938343683524,
+      "loss": 3.2861,
+      "step": 5605
+    },
+    {
+      "epoch": 0.4920362817513323,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0018896922513771321,
+      "loss": 3.3384,
+      "step": 5606
+    },
+    {
+      "epoch": 0.4921240513342348,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0018892906464845116,
+      "loss": 3.3047,
+      "step": 5607
+    },
+    {
+      "epoch": 0.4922118209171373,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0018888890197271875,
+      "loss": 3.3276,
+      "step": 5608
+    },
+    {
+      "epoch": 0.4922995905000398,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0018884873711418563,
+      "loss": 3.3003,
+      "step": 5609
+    },
+    {
+      "epoch": 0.49238736008294226,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.001888085700765218,
+      "loss": 3.229,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4924751296658447,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.0018876840086339752,
+      "loss": 3.373,
+      "step": 5611
+    },
+    {
+      "epoch": 0.49256289924874724,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.001887282294784831,
+      "loss": 3.2944,
+      "step": 5612
+    },
+    {
+      "epoch": 0.4926506688316497,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.0018868805592544917,
+      "loss": 3.3022,
+      "step": 5613
+    },
+    {
+      "epoch": 0.49273843841455217,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.001886478802079665,
+      "loss": 3.394,
+      "step": 5614
+    },
+    {
+      "epoch": 0.4928262079974547,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0018860770232970606,
+      "loss": 3.2495,
+      "step": 5615
+    },
+    {
+      "epoch": 0.49291397758035715,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.0018856752229433902,
+      "loss": 3.3467,
+      "step": 5616
+    },
+    {
+      "epoch": 0.4930017471632597,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.0018852734010553667,
+      "loss": 3.2314,
+      "step": 5617
+    },
+    {
+      "epoch": 0.49308951674616214,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0018848715576697073,
+      "loss": 3.3354,
+      "step": 5618
+    },
+    {
+      "epoch": 0.4931772863290646,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0018844696928231289,
+      "loss": 3.3154,
+      "step": 5619
+    },
+    {
+      "epoch": 0.4932650559119671,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.001884067806552351,
+      "loss": 3.2456,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4933528254948696,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0018836658988940953,
+      "loss": 3.2676,
+      "step": 5621
+    },
+    {
+      "epoch": 0.4934405950777721,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0018832639698850853,
+      "loss": 3.2939,
+      "step": 5622
+    },
+    {
+      "epoch": 0.49352836466067457,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0018828620195620461,
+      "loss": 3.3403,
+      "step": 5623
+    },
+    {
+      "epoch": 0.49361613424357703,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0018824600479617057,
+      "loss": 3.2627,
+      "step": 5624
+    },
+    {
+      "epoch": 0.49370390382647955,
+      "grad_norm": 0.1181640625,
+      "learning_rate": 0.001882058055120793,
+      "loss": 3.2271,
+      "step": 5625
+    },
+    {
+      "epoch": 0.493791673409382,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0018816560410760397,
+      "loss": 3.313,
+      "step": 5626
+    },
+    {
+      "epoch": 0.4938794429922845,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0018812540058641787,
+      "loss": 3.3574,
+      "step": 5627
+    },
+    {
+      "epoch": 0.493967212575187,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.001880851949521945,
+      "loss": 3.2739,
+      "step": 5628
+    },
+    {
+      "epoch": 0.49405498215808946,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.001880449872086076,
+      "loss": 3.2915,
+      "step": 5629
+    },
+    {
+      "epoch": 0.494142751740992,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.0018800477735933104,
+      "loss": 3.353,
+      "step": 5630
+    },
+    {
+      "epoch": 0.49423052132389445,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.00187964565408039,
+      "loss": 3.3081,
+      "step": 5631
+    },
+    {
+      "epoch": 0.4943182909067969,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0018792435135840569,
+      "loss": 3.3281,
+      "step": 5632
+    },
+    {
+      "epoch": 0.49440606048969943,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0018788413521410563,
+      "loss": 3.3237,
+      "step": 5633
+    },
+    {
+      "epoch": 0.4944938300726019,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 0.0018784391697881343,
+      "loss": 3.2788,
+      "step": 5634
+    },
+    {
+      "epoch": 0.4945815996555044,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0018780369665620399,
+      "loss": 3.2783,
+      "step": 5635
+    },
+    {
+      "epoch": 0.4946693692384069,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.0018776347424995237,
+      "loss": 3.3237,
+      "step": 5636
+    },
+    {
+      "epoch": 0.49475713882130934,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.001877232497637338,
+      "loss": 3.312,
+      "step": 5637
+    },
+    {
+      "epoch": 0.49484490840421186,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0018768302320122372,
+      "loss": 3.312,
+      "step": 5638
+    },
+    {
+      "epoch": 0.4949326779871143,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0018764279456609774,
+      "loss": 3.2236,
+      "step": 5639
+    },
+    {
+      "epoch": 0.4950204475700168,
+      "grad_norm": 0.1064453125,
+      "learning_rate": 0.0018760256386203167,
+      "loss": 3.3906,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4951082171529193,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0018756233109270153,
+      "loss": 3.2363,
+      "step": 5641
+    },
+    {
+      "epoch": 0.49519598673582177,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.001875220962617835,
+      "loss": 3.2163,
+      "step": 5642
+    },
+    {
+      "epoch": 0.4952837563187243,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0018748185937295394,
+      "loss": 3.2593,
+      "step": 5643
+    },
+    {
+      "epoch": 0.49537152590162675,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.001874416204298894,
+      "loss": 3.3633,
+      "step": 5644
+    },
+    {
+      "epoch": 0.4954592954845292,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.001874013794362667,
+      "loss": 3.29,
+      "step": 5645
+    },
+    {
+      "epoch": 0.49554706506743174,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0018736113639576271,
+      "loss": 3.2651,
+      "step": 5646
+    },
+    {
+      "epoch": 0.4956348346503342,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0018732089131205459,
+      "loss": 3.376,
+      "step": 5647
+    },
+    {
+      "epoch": 0.4957226042332367,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.001872806441888197,
+      "loss": 3.2207,
+      "step": 5648
+    },
+    {
+      "epoch": 0.4958103738161392,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0018724039502973542,
+      "loss": 3.376,
+      "step": 5649
+    },
+    {
+      "epoch": 0.49589814339904165,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0018720014383847956,
+      "loss": 3.251,
+      "step": 5650
+    },
+    {
+      "epoch": 0.49598591298194417,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0018715989061872994,
+      "loss": 3.314,
+      "step": 5651
+    },
+    {
+      "epoch": 0.49607368256484663,
+      "grad_norm": 0.19140625,
+      "learning_rate": 0.0018711963537416455,
+      "loss": 3.2769,
+      "step": 5652
+    },
+    {
+      "epoch": 0.49616145214774915,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.001870793781084618,
+      "loss": 3.2344,
+      "step": 5653
+    },
+    {
+      "epoch": 0.4962492217306516,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.0018703911882529995,
+      "loss": 3.2363,
+      "step": 5654
+    },
+    {
+      "epoch": 0.4963369913135541,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0018699885752835774,
+      "loss": 3.3345,
+      "step": 5655
+    },
+    {
+      "epoch": 0.4964247608964566,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.0018695859422131385,
+      "loss": 3.3286,
+      "step": 5656
+    },
+    {
+      "epoch": 0.49651253047935906,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0018691832890784738,
+      "loss": 3.3203,
+      "step": 5657
+    },
+    {
+      "epoch": 0.4966003000622615,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0018687806159163735,
+      "loss": 3.271,
+      "step": 5658
+    },
+    {
+      "epoch": 0.49668806964516404,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0018683779227636325,
+      "loss": 3.2324,
+      "step": 5659
+    },
+    {
+      "epoch": 0.4967758392280665,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.001867975209657045,
+      "loss": 3.2852,
+      "step": 5660
+    },
+    {
+      "epoch": 0.496863608810969,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0018675724766334084,
+      "loss": 3.3047,
+      "step": 5661
+    },
+    {
+      "epoch": 0.4969513783938715,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.001867169723729522,
+      "loss": 3.252,
+      "step": 5662
+    },
+    {
+      "epoch": 0.49703914797677395,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.0018667669509821857,
+      "loss": 3.3042,
+      "step": 5663
+    },
+    {
+      "epoch": 0.4971269175596765,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0018663641584282034,
+      "loss": 3.2974,
+      "step": 5664
+    },
+    {
+      "epoch": 0.49721468714257894,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0018659613461043784,
+      "loss": 3.3174,
+      "step": 5665
+    },
+    {
+      "epoch": 0.49730245672548146,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.001865558514047517,
+      "loss": 3.2305,
+      "step": 5666
+    },
+    {
+      "epoch": 0.4973902263083839,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.0018651556622944277,
+      "loss": 3.3574,
+      "step": 5667
+    },
+    {
+      "epoch": 0.4974779958912864,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.0018647527908819198,
+      "loss": 3.2847,
+      "step": 5668
+    },
+    {
+      "epoch": 0.4975657654741889,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0018643498998468049,
+      "loss": 3.2866,
+      "step": 5669
+    },
+    {
+      "epoch": 0.49765353505709137,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0018639469892258963,
+      "loss": 3.3096,
+      "step": 5670
+    },
+    {
+      "epoch": 0.49774130463999383,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.0018635440590560096,
+      "loss": 3.2886,
+      "step": 5671
+    },
+    {
+      "epoch": 0.49782907422289635,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0018631411093739613,
+      "loss": 3.3247,
+      "step": 5672
+    },
+    {
+      "epoch": 0.4979168438057988,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00186273814021657,
+      "loss": 3.3301,
+      "step": 5673
+    },
+    {
+      "epoch": 0.49800461338870133,
+      "grad_norm": 0.13671875,
+      "learning_rate": 0.0018623351516206571,
+      "loss": 3.2642,
+      "step": 5674
+    },
+    {
+      "epoch": 0.4980923829716038,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0018619321436230437,
+      "loss": 3.3154,
+      "step": 5675
+    },
+    {
+      "epoch": 0.49818015255450626,
+      "grad_norm": 0.08837890625,
+      "learning_rate": 0.0018615291162605548,
+      "loss": 3.1831,
+      "step": 5676
+    },
+    {
+      "epoch": 0.4982679221374088,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.0018611260695700154,
+      "loss": 3.2871,
+      "step": 5677
+    },
+    {
+      "epoch": 0.49835569172031124,
+      "grad_norm": 0.1337890625,
+      "learning_rate": 0.001860723003588254,
+      "loss": 3.394,
+      "step": 5678
+    },
+    {
+      "epoch": 0.49844346130321376,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.001860319918352099,
+      "loss": 3.3052,
+      "step": 5679
+    },
+    {
+      "epoch": 0.49853123088611623,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.001859916813898382,
+      "loss": 3.3188,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4986190004690187,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0018595136902639356,
+      "loss": 3.2471,
+      "step": 5681
+    },
+    {
+      "epoch": 0.4987067700519212,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.001859110547485595,
+      "loss": 3.2866,
+      "step": 5682
+    },
+    {
+      "epoch": 0.4987945396348237,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.001858707385600196,
+      "loss": 3.3433,
+      "step": 5683
+    },
+    {
+      "epoch": 0.49888230921772614,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.001858304204644577,
+      "loss": 3.2646,
+      "step": 5684
+    },
+    {
+      "epoch": 0.49897007880062866,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.001857901004655578,
+      "loss": 3.2109,
+      "step": 5685
+    },
+    {
+      "epoch": 0.4990578483835311,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.0018574977856700401,
+      "loss": 3.3428,
+      "step": 5686
+    },
+    {
+      "epoch": 0.49914561796643364,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0018570945477248066,
+      "loss": 3.2075,
+      "step": 5687
+    },
+    {
+      "epoch": 0.4992333875493361,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0018566912908567232,
+      "loss": 3.293,
+      "step": 5688
+    },
+    {
+      "epoch": 0.49932115713223857,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0018562880151026361,
+      "loss": 3.3516,
+      "step": 5689
+    },
+    {
+      "epoch": 0.4994089267151411,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0018558847204993944,
+      "loss": 3.2275,
+      "step": 5690
+    },
+    {
+      "epoch": 0.49949669629804355,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0018554814070838482,
+      "loss": 3.2573,
+      "step": 5691
+    },
+    {
+      "epoch": 0.49958446588094607,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0018550780748928485,
+      "loss": 3.2593,
+      "step": 5692
+    },
+    {
+      "epoch": 0.49967223546384854,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.00185467472396325,
+      "loss": 3.3203,
+      "step": 5693
+    },
+    {
+      "epoch": 0.499760005046751,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.001854271354331908,
+      "loss": 3.2407,
+      "step": 5694
+    },
+    {
+      "epoch": 0.4998477746296535,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0018538679660356791,
+      "loss": 3.2939,
+      "step": 5695
+    },
+    {
+      "epoch": 0.499935544212556,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0018534645591114225,
+      "loss": 3.2559,
+      "step": 5696
+    },
+    {
+      "epoch": 0.5000233137954585,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0018530611335959989,
+      "loss": 3.2871,
+      "step": 5697
+    },
+    {
+      "epoch": 0.500111083378361,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0018526576895262696,
+      "loss": 3.2676,
+      "step": 5698
+    },
+    {
+      "epoch": 0.5001988529612634,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0018522542269390997,
+      "loss": 3.3706,
+      "step": 5699
+    },
+    {
+      "epoch": 0.5002866225441659,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.001851850745871354,
+      "loss": 3.3403,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5003743921270685,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.0018514472463598993,
+      "loss": 3.374,
+      "step": 5701
+    },
+    {
+      "epoch": 0.5004621617099709,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.001851043728441606,
+      "loss": 3.3184,
+      "step": 5702
+    },
+    {
+      "epoch": 0.5005499312928734,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.001850640192153344,
+      "loss": 3.2583,
+      "step": 5703
+    },
+    {
+      "epoch": 0.5006377008757759,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.001850236637531985,
+      "loss": 3.2817,
+      "step": 5704
+    },
+    {
+      "epoch": 0.5007254704586783,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0018498330646144038,
+      "loss": 3.2856,
+      "step": 5705
+    },
+    {
+      "epoch": 0.5008132400415808,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0018494294734374762,
+      "loss": 3.2642,
+      "step": 5706
+    },
+    {
+      "epoch": 0.5009010096244834,
+      "grad_norm": 0.109375,
+      "learning_rate": 0.0018490258640380787,
+      "loss": 3.2935,
+      "step": 5707
+    },
+    {
+      "epoch": 0.5009887792073858,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0018486222364530911,
+      "loss": 3.3013,
+      "step": 5708
+    },
+    {
+      "epoch": 0.5010765487902883,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0018482185907193935,
+      "loss": 3.2866,
+      "step": 5709
+    },
+    {
+      "epoch": 0.5011643183731908,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0018478149268738686,
+      "loss": 3.2842,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5012520879560932,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0018474112449534004,
+      "loss": 3.2686,
+      "step": 5711
+    },
+    {
+      "epoch": 0.5013398575389958,
+      "grad_norm": 0.12109375,
+      "learning_rate": 0.001847007544994874,
+      "loss": 3.2173,
+      "step": 5712
+    },
+    {
+      "epoch": 0.5014276271218983,
+      "grad_norm": 0.125,
+      "learning_rate": 0.0018466038270351776,
+      "loss": 3.3101,
+      "step": 5713
+    },
+    {
+      "epoch": 0.5015153967048007,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0018462000911111994,
+      "loss": 3.3184,
+      "step": 5714
+    },
+    {
+      "epoch": 0.5016031662877032,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.0018457963372598295,
+      "loss": 3.2539,
+      "step": 5715
+    },
+    {
+      "epoch": 0.5016909358706056,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.0018453925655179614,
+      "loss": 3.2656,
+      "step": 5716
+    },
+    {
+      "epoch": 0.5017787054535082,
+      "grad_norm": 0.08935546875,
+      "learning_rate": 0.001844988775922488,
+      "loss": 3.2837,
+      "step": 5717
+    },
+    {
+      "epoch": 0.5018664750364107,
+      "grad_norm": 0.17578125,
+      "learning_rate": 0.0018445849685103052,
+      "loss": 3.2476,
+      "step": 5718
+    },
+    {
+      "epoch": 0.5019542446193132,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0018441811433183098,
+      "loss": 3.2549,
+      "step": 5719
+    },
+    {
+      "epoch": 0.5020420142022156,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0018437773003834007,
+      "loss": 3.2983,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5021297837851181,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0018433734397424783,
+      "loss": 3.2534,
+      "step": 5721
+    },
+    {
+      "epoch": 0.5022175533680207,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.001842969561432444,
+      "loss": 3.3311,
+      "step": 5722
+    },
+    {
+      "epoch": 0.5023053229509231,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.001842565665490202,
+      "loss": 3.3184,
+      "step": 5723
+    },
+    {
+      "epoch": 0.5023930925338256,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.001842161751952658,
+      "loss": 3.2954,
+      "step": 5724
+    },
+    {
+      "epoch": 0.502480862116728,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0018417578208567175,
+      "loss": 3.2681,
+      "step": 5725
+    },
+    {
+      "epoch": 0.5025686316996305,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.00184135387223929,
+      "loss": 3.25,
+      "step": 5726
+    },
+    {
+      "epoch": 0.5026564012825331,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0018409499061372838,
+      "loss": 3.2808,
+      "step": 5727
+    },
+    {
+      "epoch": 0.5027441708654355,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 0.0018405459225876127,
+      "loss": 3.2705,
+      "step": 5728
+    },
+    {
+      "epoch": 0.502831940448338,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0018401419216271888,
+      "loss": 3.3657,
+      "step": 5729
+    },
+    {
+      "epoch": 0.5029197100312405,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.001839737903292927,
+      "loss": 3.2744,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5030074796141429,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0018393338676217434,
+      "loss": 3.3687,
+      "step": 5731
+    },
+    {
+      "epoch": 0.5030952491970454,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.001838929814650556,
+      "loss": 3.2446,
+      "step": 5732
+    },
+    {
+      "epoch": 0.503183018779948,
+      "grad_norm": 0.14453125,
+      "learning_rate": 0.0018385257444162848,
+      "loss": 3.2988,
+      "step": 5733
+    },
+    {
+      "epoch": 0.5032707883628504,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0018381216569558505,
+      "loss": 3.2383,
+      "step": 5734
+    },
+    {
+      "epoch": 0.5033585579457529,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.001837717552306176,
+      "loss": 3.2705,
+      "step": 5735
+    },
+    {
+      "epoch": 0.5034463275286554,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001837313430504185,
+      "loss": 3.2466,
+      "step": 5736
+    },
+    {
+      "epoch": 0.5035340971115578,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0018369092915868045,
+      "loss": 3.3198,
+      "step": 5737
+    },
+    {
+      "epoch": 0.5036218666944604,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0018365051355909608,
+      "loss": 3.3457,
+      "step": 5738
+    },
+    {
+      "epoch": 0.5037096362773629,
+      "grad_norm": 0.1494140625,
+      "learning_rate": 0.001836100962553583,
+      "loss": 3.2686,
+      "step": 5739
+    },
+    {
+      "epoch": 0.5037974058602653,
+      "grad_norm": 0.162109375,
+      "learning_rate": 0.0018356967725116026,
+      "loss": 3.3047,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5038851754431678,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0018352925655019503,
+      "loss": 3.3145,
+      "step": 5741
+    },
+    {
+      "epoch": 0.5039729450260703,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00183488834156156,
+      "loss": 3.2944,
+      "step": 5742
+    },
+    {
+      "epoch": 0.5040607146089728,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.0018344841007273678,
+      "loss": 3.2358,
+      "step": 5743
+    },
+    {
+      "epoch": 0.5041484841918753,
+      "grad_norm": 0.205078125,
+      "learning_rate": 0.0018340798430363091,
+      "loss": 3.3188,
+      "step": 5744
+    },
+    {
+      "epoch": 0.5042362537747778,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0018336755685253228,
+      "loss": 3.3374,
+      "step": 5745
+    },
+    {
+      "epoch": 0.5043240233576802,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.001833271277231349,
+      "loss": 3.2686,
+      "step": 5746
+    },
+    {
+      "epoch": 0.5044117929405827,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0018328669691913288,
+      "loss": 3.3301,
+      "step": 5747
+    },
+    {
+      "epoch": 0.5044995625234853,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0018324626444422047,
+      "loss": 3.2798,
+      "step": 5748
+    },
+    {
+      "epoch": 0.5045873321063877,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 0.0018320583030209212,
+      "loss": 3.3545,
+      "step": 5749
+    },
+    {
+      "epoch": 0.5046751016892902,
+      "grad_norm": 0.11474609375,
+      "learning_rate": 0.0018316539449644243,
+      "loss": 3.2847,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5047628712721927,
+      "grad_norm": 0.10986328125,
+      "learning_rate": 0.0018312495703096611,
+      "loss": 3.2671,
+      "step": 5751
+    },
+    {
+      "epoch": 0.5048506408550951,
+      "grad_norm": 0.1484375,
+      "learning_rate": 0.001830845179093581,
+      "loss": 3.2861,
+      "step": 5752
+    },
+    {
+      "epoch": 0.5049384104379977,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0018304407713531342,
+      "loss": 3.27,
+      "step": 5753
+    },
+    {
+      "epoch": 0.5050261800209002,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.0018300363471252725,
+      "loss": 3.2856,
+      "step": 5754
+    },
+    {
+      "epoch": 0.5051139496038026,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0018296319064469498,
+      "loss": 3.2954,
+      "step": 5755
+    },
+    {
+      "epoch": 0.5052017191867051,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0018292274493551207,
+      "loss": 3.29,
+      "step": 5756
+    },
+    {
+      "epoch": 0.5052894887696076,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.001828822975886742,
+      "loss": 3.2485,
+      "step": 5757
+    },
+    {
+      "epoch": 0.5053772583525101,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0018284184860787713,
+      "loss": 3.3652,
+      "step": 5758
+    },
+    {
+      "epoch": 0.5054650279354126,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0018280139799681679,
+      "loss": 3.2017,
+      "step": 5759
+    },
+    {
+      "epoch": 0.5055527975183151,
+      "grad_norm": 0.125,
+      "learning_rate": 0.001827609457591893,
+      "loss": 3.3101,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5056405671012175,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.0018272049189869087,
+      "loss": 3.2324,
+      "step": 5761
+    },
+    {
+      "epoch": 0.50572833668412,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0018268003641901797,
+      "loss": 3.3506,
+      "step": 5762
+    },
+    {
+      "epoch": 0.5058161062670224,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0018263957932386707,
+      "loss": 3.3335,
+      "step": 5763
+    },
+    {
+      "epoch": 0.505903875849925,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.001825991206169349,
+      "loss": 3.2925,
+      "step": 5764
+    },
+    {
+      "epoch": 0.5059916454328275,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0018255866030191825,
+      "loss": 3.3135,
+      "step": 5765
+    },
+    {
+      "epoch": 0.50607941501573,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0018251819838251413,
+      "loss": 3.293,
+      "step": 5766
+    },
+    {
+      "epoch": 0.5061671845986324,
+      "grad_norm": 0.10595703125,
+      "learning_rate": 0.0018247773486241962,
+      "loss": 3.4004,
+      "step": 5767
+    },
+    {
+      "epoch": 0.5062549541815349,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0018243726974533205,
+      "loss": 3.3301,
+      "step": 5768
+    },
+    {
+      "epoch": 0.5063427237644375,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0018239680303494884,
+      "loss": 3.3242,
+      "step": 5769
+    },
+    {
+      "epoch": 0.5064304933473399,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0018235633473496755,
+      "loss": 3.3267,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5065182629302424,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0018231586484908583,
+      "loss": 3.3618,
+      "step": 5771
+    },
+    {
+      "epoch": 0.5066060325131448,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.001822753933810016,
+      "loss": 3.3096,
+      "step": 5772
+    },
+    {
+      "epoch": 0.5066938020960473,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0018223492033441282,
+      "loss": 3.2334,
+      "step": 5773
+    },
+    {
+      "epoch": 0.5067815716789499,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0018219444571301768,
+      "loss": 3.2695,
+      "step": 5774
+    },
+    {
+      "epoch": 0.5068693412618523,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0018215396952051442,
+      "loss": 3.3403,
+      "step": 5775
+    },
+    {
+      "epoch": 0.5069571108447548,
+      "grad_norm": 0.09423828125,
+      "learning_rate": 0.001821134917606015,
+      "loss": 3.2866,
+      "step": 5776
+    },
+    {
+      "epoch": 0.5070448804276573,
+      "grad_norm": 0.107421875,
+      "learning_rate": 0.0018207301243697747,
+      "loss": 3.2446,
+      "step": 5777
+    },
+    {
+      "epoch": 0.5071326500105597,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.001820325315533411,
+      "loss": 3.2529,
+      "step": 5778
+    },
+    {
+      "epoch": 0.5072204195934623,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0018199204911339114,
+      "loss": 3.2881,
+      "step": 5779
+    },
+    {
+      "epoch": 0.5073081891763648,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 0.0018195156512082668,
+      "loss": 3.2764,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5073959587592672,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.001819110795793469,
+      "loss": 3.3027,
+      "step": 5781
+    },
+    {
+      "epoch": 0.5074837283421697,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.0018187059249265103,
+      "loss": 3.3101,
+      "step": 5782
+    },
+    {
+      "epoch": 0.5075714979250722,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0018183010386443844,
+      "loss": 3.3022,
+      "step": 5783
+    },
+    {
+      "epoch": 0.5076592675079747,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0018178961369840882,
+      "loss": 3.2988,
+      "step": 5784
+    },
+    {
+      "epoch": 0.5077470370908772,
+      "grad_norm": 0.1025390625,
+      "learning_rate": 0.001817491219982618,
+      "loss": 3.2676,
+      "step": 5785
+    },
+    {
+      "epoch": 0.5078348066737797,
+      "grad_norm": 0.154296875,
+      "learning_rate": 0.001817086287676972,
+      "loss": 3.3042,
+      "step": 5786
+    },
+    {
+      "epoch": 0.5079225762566821,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001816681340104151,
+      "loss": 3.3086,
+      "step": 5787
+    },
+    {
+      "epoch": 0.5080103458395846,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0018162763773011557,
+      "loss": 3.3096,
+      "step": 5788
+    },
+    {
+      "epoch": 0.5080981154224871,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001815871399304989,
+      "loss": 3.25,
+      "step": 5789
+    },
+    {
+      "epoch": 0.5081858850053896,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0018154664061526544,
+      "loss": 3.2842,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5082736545882921,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0018150613978811581,
+      "loss": 3.2671,
+      "step": 5791
+    },
+    {
+      "epoch": 0.5083614241711946,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0018146563745275064,
+      "loss": 3.2192,
+      "step": 5792
+    },
+    {
+      "epoch": 0.508449193754097,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0018142513361287078,
+      "loss": 3.2783,
+      "step": 5793
+    },
+    {
+      "epoch": 0.5085369633369995,
+      "grad_norm": 0.11376953125,
+      "learning_rate": 0.0018138462827217718,
+      "loss": 3.2686,
+      "step": 5794
+    },
+    {
+      "epoch": 0.5086247329199021,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.0018134412143437093,
+      "loss": 3.3184,
+      "step": 5795
+    },
+    {
+      "epoch": 0.5087125025028045,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0018130361310315323,
+      "loss": 3.3169,
+      "step": 5796
+    },
+    {
+      "epoch": 0.508800272085707,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.0018126310328222548,
+      "loss": 3.3022,
+      "step": 5797
+    },
+    {
+      "epoch": 0.5088880416686095,
+      "grad_norm": 0.11181640625,
+      "learning_rate": 0.0018122259197528923,
+      "loss": 3.3081,
+      "step": 5798
+    },
+    {
+      "epoch": 0.5089758112515119,
+      "grad_norm": 0.09765625,
+      "learning_rate": 0.0018118207918604602,
+      "loss": 3.292,
+      "step": 5799
+    },
+    {
+      "epoch": 0.5090635808344145,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.001811415649181977,
+      "loss": 3.396,
+      "step": 5800
+    },
+    {
+      "epoch": 0.509151350417317,
+      "grad_norm": 0.189453125,
+      "learning_rate": 0.0018110104917544616,
+      "loss": 3.3188,
+      "step": 5801
+    },
+    {
+      "epoch": 0.5092391200002194,
+      "grad_norm": 0.0810546875,
+      "learning_rate": 0.001810605319614934,
+      "loss": 3.2227,
+      "step": 5802
+    },
+    {
+      "epoch": 0.5093268895831219,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 0.0018102001328004168,
+      "loss": 3.2427,
+      "step": 5803
+    },
+    {
+      "epoch": 0.5094146591660244,
+      "grad_norm": 0.10107421875,
+      "learning_rate": 0.0018097949313479328,
+      "loss": 3.3198,
+      "step": 5804
+    },
+    {
+      "epoch": 0.5095024287489269,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 0.0018093897152945064,
+      "loss": 3.3501,
+      "step": 5805
+    },
+    {
+      "epoch": 0.5095901983318294,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0018089844846771634,
+      "loss": 3.2578,
+      "step": 5806
+    },
+    {
+      "epoch": 0.5096779679147319,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0018085792395329309,
+      "loss": 3.2881,
+      "step": 5807
+    },
+    {
+      "epoch": 0.5097657374976343,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.0018081739798988378,
+      "loss": 3.3252,
+      "step": 5808
+    },
+    {
+      "epoch": 0.5098535070805368,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.0018077687058119128,
+      "loss": 3.2734,
+      "step": 5809
+    },
+    {
+      "epoch": 0.5099412766634394,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0018073634173091885,
+      "loss": 3.3237,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5100290462463418,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.001806958114427696,
+      "loss": 3.2925,
+      "step": 5811
+    },
+    {
+      "epoch": 0.5101168158292443,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.0018065527972044699,
+      "loss": 3.3096,
+      "step": 5812
+    },
+    {
+      "epoch": 0.5102045854121467,
+      "grad_norm": 0.08740234375,
+      "learning_rate": 0.0018061474656765448,
+      "loss": 3.2563,
+      "step": 5813
+    },
+    {
+      "epoch": 0.5102923549950492,
+      "grad_norm": 0.11328125,
+      "learning_rate": 0.001805742119880957,
+      "loss": 3.2852,
+      "step": 5814
+    },
+    {
+      "epoch": 0.5103801245779517,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.001805336759854745,
+      "loss": 3.3359,
+      "step": 5815
+    },
+    {
+      "epoch": 0.5104678941608543,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0018049313856349464,
+      "loss": 3.3491,
+      "step": 5816
+    },
+    {
+      "epoch": 0.5105556637437567,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0018045259972586027,
+      "loss": 3.3506,
+      "step": 5817
+    },
+    {
+      "epoch": 0.5106434333266592,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0018041205947627547,
+      "loss": 3.251,
+      "step": 5818
+    },
+    {
+      "epoch": 0.5107312029095616,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0018037151781844455,
+      "loss": 3.2812,
+      "step": 5819
+    },
+    {
+      "epoch": 0.5108189724924641,
+      "grad_norm": 0.083984375,
+      "learning_rate": 0.0018033097475607192,
+      "loss": 3.3159,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5109067420753667,
+      "grad_norm": 0.08642578125,
+      "learning_rate": 0.0018029043029286211,
+      "loss": 3.2866,
+      "step": 5821
+    },
+    {
+      "epoch": 0.5109945116582691,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0018024988443251985,
+      "loss": 3.3472,
+      "step": 5822
+    },
+    {
+      "epoch": 0.5110822812411716,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0018020933717874988,
+      "loss": 3.231,
+      "step": 5823
+    },
+    {
+      "epoch": 0.5111700508240741,
+      "grad_norm": 0.08056640625,
+      "learning_rate": 0.001801687885352571,
+      "loss": 3.2183,
+      "step": 5824
+    },
+    {
+      "epoch": 0.5112578204069765,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.001801282385057466,
+      "loss": 3.3003,
+      "step": 5825
+    },
+    {
+      "epoch": 0.5113455899898791,
+      "grad_norm": 0.169921875,
+      "learning_rate": 0.0018008768709392355,
+      "loss": 3.3516,
+      "step": 5826
+    },
+    {
+      "epoch": 0.5114333595727816,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.001800471343034933,
+      "loss": 3.2139,
+      "step": 5827
+    },
+    {
+      "epoch": 0.511521129155684,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.001800065801381612,
+      "loss": 3.3105,
+      "step": 5828
+    },
+    {
+      "epoch": 0.5116088987385865,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.0017996602460163284,
+      "loss": 3.2378,
+      "step": 5829
+    },
+    {
+      "epoch": 0.511696668321489,
+      "grad_norm": 0.1357421875,
+      "learning_rate": 0.0017992546769761391,
+      "loss": 3.334,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5117844379043915,
+      "grad_norm": 0.08447265625,
+      "learning_rate": 0.0017988490942981016,
+      "loss": 3.312,
+      "step": 5831
+    },
+    {
+      "epoch": 0.511872207487294,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0017984434980192763,
+      "loss": 3.2974,
+      "step": 5832
+    },
+    {
+      "epoch": 0.5119599770701965,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.0017980378881767233,
+      "loss": 3.3125,
+      "step": 5833
+    },
+    {
+      "epoch": 0.5120477466530989,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.001797632264807504,
+      "loss": 3.2969,
+      "step": 5834
+    },
+    {
+      "epoch": 0.5121355162360014,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.0017972266279486816,
+      "loss": 3.3184,
+      "step": 5835
+    },
+    {
+      "epoch": 0.512223285818904,
+      "grad_norm": 0.11083984375,
+      "learning_rate": 0.0017968209776373205,
+      "loss": 3.2847,
+      "step": 5836
+    },
+    {
+      "epoch": 0.5123110554018064,
+      "grad_norm": 0.12060546875,
+      "learning_rate": 0.0017964153139104857,
+      "loss": 3.2793,
+      "step": 5837
+    },
+    {
+      "epoch": 0.5123988249847089,
+      "grad_norm": 0.1416015625,
+      "learning_rate": 0.0017960096368052445,
+      "loss": 3.2764,
+      "step": 5838
+    },
+    {
+      "epoch": 0.5124865945676114,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.001795603946358665,
+      "loss": 3.2954,
+      "step": 5839
+    },
+    {
+      "epoch": 0.5125743641505138,
+      "grad_norm": 0.1328125,
+      "learning_rate": 0.001795198242607816,
+      "loss": 3.3062,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5126621337334164,
+      "grad_norm": 0.09033203125,
+      "learning_rate": 0.0017947925255897678,
+      "loss": 3.3066,
+      "step": 5841
+    },
+    {
+      "epoch": 0.5127499033163189,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0017943867953415919,
+      "loss": 3.3027,
+      "step": 5842
+    },
+    {
+      "epoch": 0.5128376728992213,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0017939810519003612,
+      "loss": 3.292,
+      "step": 5843
+    },
+    {
+      "epoch": 0.5129254424821238,
+      "grad_norm": 0.177734375,
+      "learning_rate": 0.0017935752953031501,
+      "loss": 3.3613,
+      "step": 5844
+    },
+    {
+      "epoch": 0.5130132120650263,
+      "grad_norm": 0.146484375,
+      "learning_rate": 0.001793169525587033,
+      "loss": 3.3037,
+      "step": 5845
+    },
+    {
+      "epoch": 0.5131009816479287,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0017927637427890871,
+      "loss": 3.293,
+      "step": 5846
+    },
+    {
+      "epoch": 0.5131887512308313,
+      "grad_norm": 0.10205078125,
+      "learning_rate": 0.0017923579469463894,
+      "loss": 3.2524,
+      "step": 5847
+    },
+    {
+      "epoch": 0.5132765208137338,
+      "grad_norm": 0.08984375,
+      "learning_rate": 0.0017919521380960189,
+      "loss": 3.2705,
+      "step": 5848
+    },
+    {
+      "epoch": 0.5133642903966362,
+      "grad_norm": 0.10400390625,
+      "learning_rate": 0.0017915463162750554,
+      "loss": 3.3584,
+      "step": 5849
+    },
+    {
+      "epoch": 0.5134520599795387,
+      "grad_norm": 0.11279296875,
+      "learning_rate": 0.0017911404815205807,
+      "loss": 3.292,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5135398295624412,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0017907346338696761,
+      "loss": 3.2588,
+      "step": 5851
+    },
+    {
+      "epoch": 0.5136275991453437,
+      "grad_norm": 0.0849609375,
+      "learning_rate": 0.0017903287733594265,
+      "loss": 3.2642,
+      "step": 5852
+    },
+    {
+      "epoch": 0.5137153687282462,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0017899229000269152,
+      "loss": 3.249,
+      "step": 5853
+    },
+    {
+      "epoch": 0.5138031383111487,
+      "grad_norm": 0.09228515625,
+      "learning_rate": 0.0017895170139092286,
+      "loss": 3.3213,
+      "step": 5854
+    },
+    {
+      "epoch": 0.5138909078940511,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0017891111150434538,
+      "loss": 3.2725,
+      "step": 5855
+    },
+    {
+      "epoch": 0.5139786774769536,
+      "grad_norm": 0.11669921875,
+      "learning_rate": 0.001788705203466679,
+      "loss": 3.2812,
+      "step": 5856
+    },
+    {
+      "epoch": 0.5140664470598562,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 0.0017882992792159944,
+      "loss": 3.3193,
+      "step": 5857
+    },
+    {
+      "epoch": 0.5141542166427586,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.001787893342328489,
+      "loss": 3.2876,
+      "step": 5858
+    },
+    {
+      "epoch": 0.5142419862256611,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.0017874873928412551,
+      "loss": 3.2812,
+      "step": 5859
+    },
+    {
+      "epoch": 0.5143297558085635,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0017870814307913852,
+      "loss": 3.2832,
+      "step": 5860
+    },
+    {
+      "epoch": 0.514417525391466,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0017866754562159742,
+      "loss": 3.2695,
+      "step": 5861
+    },
+    {
+      "epoch": 0.5145052949743686,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.0017862694691521167,
+      "loss": 3.3315,
+      "step": 5862
+    },
+    {
+      "epoch": 0.514593064557271,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 0.0017858634696369083,
+      "loss": 3.2158,
+      "step": 5863
+    },
+    {
+      "epoch": 0.5146808341401735,
+      "grad_norm": 0.150390625,
+      "learning_rate": 0.0017854574577074475,
+      "loss": 3.3198,
+      "step": 5864
+    },
+    {
+      "epoch": 0.514768603723076,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0017850514334008319,
+      "loss": 3.2866,
+      "step": 5865
+    },
+    {
+      "epoch": 0.5148563733059784,
+      "grad_norm": 0.12890625,
+      "learning_rate": 0.001784645396754162,
+      "loss": 3.2095,
+      "step": 5866
+    },
+    {
+      "epoch": 0.514944142888881,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0017842393478045377,
+      "loss": 3.2959,
+      "step": 5867
+    },
+    {
+      "epoch": 0.5150319124717835,
+      "grad_norm": 0.10546875,
+      "learning_rate": 0.0017838332865890618,
+      "loss": 3.2915,
+      "step": 5868
+    },
+    {
+      "epoch": 0.515119682054686,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.001783427213144837,
+      "loss": 3.3438,
+      "step": 5869
+    },
+    {
+      "epoch": 0.5152074516375884,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0017830211275089668,
+      "loss": 3.3442,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5152952212204909,
+      "grad_norm": 0.09326171875,
+      "learning_rate": 0.0017826150297185577,
+      "loss": 3.3413,
+      "step": 5871
+    },
+    {
+      "epoch": 0.5153829908033933,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0017822089198107145,
+      "loss": 3.3076,
+      "step": 5872
+    },
+    {
+      "epoch": 0.5154707603862959,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 0.0017818027978225466,
+      "loss": 3.2886,
+      "step": 5873
+    },
+    {
+      "epoch": 0.5155585299691984,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0017813966637911611,
+      "loss": 3.292,
+      "step": 5874
+    },
+    {
+      "epoch": 0.5156462995521008,
+      "grad_norm": 0.087890625,
+      "learning_rate": 0.0017809905177536684,
+      "loss": 3.3105,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5157340691350033,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0017805843597471786,
+      "loss": 3.27,
+      "step": 5876
+    },
+    {
+      "epoch": 0.5158218387179058,
+      "grad_norm": 0.1005859375,
+      "learning_rate": 0.0017801781898088038,
+      "loss": 3.2544,
+      "step": 5877
+    },
+    {
+      "epoch": 0.5159096083008083,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.0017797720079756575,
+      "loss": 3.3789,
+      "step": 5878
+    },
+    {
+      "epoch": 0.5159973778837108,
+      "grad_norm": 0.1376953125,
+      "learning_rate": 0.0017793658142848535,
+      "loss": 3.2363,
+      "step": 5879
+    },
+    {
+      "epoch": 0.5160851474666133,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0017789596087735072,
+      "loss": 3.2949,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5161729170495157,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0017785533914787338,
+      "loss": 3.3496,
+      "step": 5881
+    },
+    {
+      "epoch": 0.5162606866324182,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0017781471624376515,
+      "loss": 3.3018,
+      "step": 5882
+    },
+    {
+      "epoch": 0.5163484562153208,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.0017777409216873784,
+      "loss": 3.2842,
+      "step": 5883
+    },
+    {
+      "epoch": 0.5164362257982232,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0017773346692650342,
+      "loss": 3.2974,
+      "step": 5884
+    },
+    {
+      "epoch": 0.5165239953811257,
+      "grad_norm": 0.138671875,
+      "learning_rate": 0.0017769284052077388,
+      "loss": 3.3154,
+      "step": 5885
+    },
+    {
+      "epoch": 0.5166117649640282,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0017765221295526145,
+      "loss": 3.2852,
+      "step": 5886
+    },
+    {
+      "epoch": 0.5166995345469306,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.0017761158423367836,
+      "loss": 3.2812,
+      "step": 5887
+    },
+    {
+      "epoch": 0.5167873041298332,
+      "grad_norm": 0.11865234375,
+      "learning_rate": 0.0017757095435973693,
+      "loss": 3.2456,
+      "step": 5888
+    },
+    {
+      "epoch": 0.5168750737127357,
+      "grad_norm": 0.1455078125,
+      "learning_rate": 0.0017753032333714968,
+      "loss": 3.2661,
+      "step": 5889
+    },
+    {
+      "epoch": 0.5169628432956381,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0017748969116962927,
+      "loss": 3.2612,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5170506128785406,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0017744905786088825,
+      "loss": 3.2896,
+      "step": 5891
+    },
+    {
+      "epoch": 0.5171383824614431,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.0017740842341463945,
+      "loss": 3.252,
+      "step": 5892
+    },
+    {
+      "epoch": 0.5172261520443456,
+      "grad_norm": 0.140625,
+      "learning_rate": 0.0017736778783459576,
+      "loss": 3.3403,
+      "step": 5893
+    },
+    {
+      "epoch": 0.5173139216272481,
+      "grad_norm": 0.09912109375,
+      "learning_rate": 0.0017732715112447018,
+      "loss": 3.3159,
+      "step": 5894
+    },
+    {
+      "epoch": 0.5174016912101506,
+      "grad_norm": 0.099609375,
+      "learning_rate": 0.0017728651328797582,
+      "loss": 3.2358,
+      "step": 5895
+    },
+    {
+      "epoch": 0.517489460793053,
+      "grad_norm": 0.0986328125,
+      "learning_rate": 0.001772458743288259,
+      "loss": 3.2778,
+      "step": 5896
+    },
+    {
+      "epoch": 0.5175772303759555,
+      "grad_norm": 0.1220703125,
+      "learning_rate": 0.0017720523425073368,
+      "loss": 3.292,
+      "step": 5897
+    },
+    {
+      "epoch": 0.5176649999588581,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0017716459305741254,
+      "loss": 3.29,
+      "step": 5898
+    },
+    {
+      "epoch": 0.5177527695417605,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.0017712395075257608,
+      "loss": 3.2046,
+      "step": 5899
+    },
+    {
+      "epoch": 0.517840539124663,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.001770833073399378,
+      "loss": 3.2725,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5179283087075655,
+      "grad_norm": 0.1015625,
+      "learning_rate": 0.0017704266282321153,
+      "loss": 3.3677,
+      "step": 5901
+    },
+    {
+      "epoch": 0.5180160782904679,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0017700201720611101,
+      "loss": 3.1636,
+      "step": 5902
+    },
+    {
+      "epoch": 0.5181038478733704,
+      "grad_norm": 0.12158203125,
+      "learning_rate": 0.0017696137049235017,
+      "loss": 3.2476,
+      "step": 5903
+    },
+    {
+      "epoch": 0.518191617456273,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.00176920722685643,
+      "loss": 3.2368,
+      "step": 5904
+    },
+    {
+      "epoch": 0.5182793870391754,
+      "grad_norm": 0.11962890625,
+      "learning_rate": 0.0017688007378970363,
+      "loss": 3.3047,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5183671566220779,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0017683942380824628,
+      "loss": 3.2324,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5184549262049803,
+      "grad_norm": 0.1123046875,
+      "learning_rate": 0.0017679877274498525,
+      "loss": 3.3281,
+      "step": 5907
+    },
+    {
+      "epoch": 0.5185426957878828,
+      "grad_norm": 0.0869140625,
+      "learning_rate": 0.0017675812060363495,
+      "loss": 3.2788,
+      "step": 5908
+    },
+    {
+      "epoch": 0.5186304653707854,
+      "grad_norm": 0.115234375,
+      "learning_rate": 0.0017671746738790988,
+      "loss": 3.2886,
+      "step": 5909
+    },
+    {
+      "epoch": 0.5187182349536879,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0017667681310152466,
+      "loss": 3.2871,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5188060045365903,
+      "grad_norm": 0.1044921875,
+      "learning_rate": 0.0017663615774819396,
+      "loss": 3.355,
+      "step": 5911
+    },
+    {
+      "epoch": 0.5188937741194928,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 0.0017659550133163262,
+      "loss": 3.312,
+      "step": 5912
+    },
+    {
+      "epoch": 0.5189815437023952,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0017655484385555554,
+      "loss": 3.2637,
+      "step": 5913
+    },
+    {
+      "epoch": 0.5190693132852978,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.001765141853236777,
+      "loss": 3.291,
+      "step": 5914
+    },
+    {
+      "epoch": 0.5191570828682003,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.001764735257397142,
+      "loss": 3.3125,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5192448524511027,
+      "grad_norm": 0.09716796875,
+      "learning_rate": 0.001764328651073802,
+      "loss": 3.293,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5193326220340052,
+      "grad_norm": 0.10009765625,
+      "learning_rate": 0.00176392203430391,
+      "loss": 3.3237,
+      "step": 5917
+    },
+    {
+      "epoch": 0.5194203916169077,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0017635154071246205,
+      "loss": 3.418,
+      "step": 5918
+    },
+    {
+      "epoch": 0.5195081611998102,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.001763108769573087,
+      "loss": 3.3794,
+      "step": 5919
+    },
+    {
+      "epoch": 0.5195959307827127,
+      "grad_norm": 0.09375,
+      "learning_rate": 0.0017627021216864663,
+      "loss": 3.3101,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5196837003656152,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0017622954635019142,
+      "loss": 3.3638,
+      "step": 5921
+    },
+    {
+      "epoch": 0.5197714699485176,
+      "grad_norm": 0.126953125,
+      "learning_rate": 0.001761888795056589,
+      "loss": 3.3384,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5198592395314201,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 0.0017614821163876488,
+      "loss": 3.3594,
+      "step": 5923
+    },
+    {
+      "epoch": 0.5199470091143227,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.001761075427532253,
+      "loss": 3.3003,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5200347786972251,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.001760668728527562,
+      "loss": 3.2656,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5201225482801276,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.0017602620194107374,
+      "loss": 3.313,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5202103178630301,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.001759855300218941,
+      "loss": 3.2979,
+      "step": 5927
+    },
+    {
+      "epoch": 0.5202980874459325,
+      "grad_norm": 0.171875,
+      "learning_rate": 0.0017594485709893364,
+      "loss": 3.2349,
+      "step": 5928
+    },
+    {
+      "epoch": 0.520385857028835,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0017590418317590868,
+      "loss": 3.3623,
+      "step": 5929
+    },
+    {
+      "epoch": 0.5204736266117376,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0017586350825653583,
+      "loss": 3.2886,
+      "step": 5930
+    },
+    {
+      "epoch": 0.52056139619464,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.0017582283234453165,
+      "loss": 3.2979,
+      "step": 5931
+    },
+    {
+      "epoch": 0.5206491657775425,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.001757821554436128,
+      "loss": 3.4443,
+      "step": 5932
+    },
+    {
+      "epoch": 0.520736935360445,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0017574147755749603,
+      "loss": 3.2764,
+      "step": 5933
+    },
+    {
+      "epoch": 0.5208247049433474,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0017570079868989821,
+      "loss": 3.3149,
+      "step": 5934
+    },
+    {
+      "epoch": 0.52091247452625,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0017566011884453628,
+      "loss": 3.3232,
+      "step": 5935
+    },
+    {
+      "epoch": 0.5210002441091525,
+      "grad_norm": 0.21484375,
+      "learning_rate": 0.0017561943802512738,
+      "loss": 3.3789,
+      "step": 5936
+    },
+    {
+      "epoch": 0.5210880136920549,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0017557875623538854,
+      "loss": 3.2153,
+      "step": 5937
+    },
+    {
+      "epoch": 0.5211757832749574,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.00175538073479037,
+      "loss": 3.2544,
+      "step": 5938
+    },
+    {
+      "epoch": 0.5212635528578599,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0017549738975979005,
+      "loss": 3.2959,
+      "step": 5939
+    },
+    {
+      "epoch": 0.5213513224407624,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.0017545670508136509,
+      "loss": 3.3052,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5214390920236649,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0017541601944747963,
+      "loss": 3.2827,
+      "step": 5941
+    },
+    {
+      "epoch": 0.5215268616065674,
+      "grad_norm": 0.185546875,
+      "learning_rate": 0.0017537533286185122,
+      "loss": 3.2471,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5216146311894698,
+      "grad_norm": 0.11767578125,
+      "learning_rate": 0.001753346453281975,
+      "loss": 3.3677,
+      "step": 5943
+    },
+    {
+      "epoch": 0.5217024007723723,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0017529395685023622,
+      "loss": 3.3242,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5217901703552749,
+      "grad_norm": 0.1796875,
+      "learning_rate": 0.0017525326743168524,
+      "loss": 3.3086,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5218779399381773,
+      "grad_norm": 0.1435546875,
+      "learning_rate": 0.0017521257707626239,
+      "loss": 3.2812,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5219657095210798,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0017517188578768576,
+      "loss": 3.2637,
+      "step": 5947
+    },
+    {
+      "epoch": 0.5220534791039823,
+      "grad_norm": 0.1103515625,
+      "learning_rate": 0.001751311935696734,
+      "loss": 3.3228,
+      "step": 5948
+    },
+    {
+      "epoch": 0.5221412486868847,
+      "grad_norm": 0.1162109375,
+      "learning_rate": 0.001750905004259435,
+      "loss": 3.2573,
+      "step": 5949
+    },
+    {
+      "epoch": 0.5222290182697873,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0017504980636021427,
+      "loss": 3.3691,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5223167878526898,
+      "grad_norm": 0.10302734375,
+      "learning_rate": 0.001750091113762041,
+      "loss": 3.2598,
+      "step": 5951
+    },
+    {
+      "epoch": 0.5224045574355922,
+      "grad_norm": 0.123046875,
+      "learning_rate": 0.0017496841547763138,
+      "loss": 3.2144,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5224923270184947,
+      "grad_norm": 0.10888671875,
+      "learning_rate": 0.0017492771866821461,
+      "loss": 3.354,
+      "step": 5953
+    },
+    {
+      "epoch": 0.5225800966013971,
+      "grad_norm": 0.1201171875,
+      "learning_rate": 0.001748870209516724,
+      "loss": 3.272,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5226678661842996,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.0017484632233172345,
+      "loss": 3.2925,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5227556357672022,
+      "grad_norm": 0.09619140625,
+      "learning_rate": 0.0017480562281208646,
+      "loss": 3.3237,
+      "step": 5956
+    },
+    {
+      "epoch": 0.5228434053501047,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 0.001747649223964803,
+      "loss": 3.2651,
+      "step": 5957
+    },
+    {
+      "epoch": 0.5229311749330071,
+      "grad_norm": 0.0908203125,
+      "learning_rate": 0.0017472422108862388,
+      "loss": 3.2993,
+      "step": 5958
+    },
+    {
+      "epoch": 0.5230189445159096,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 0.0017468351889223619,
+      "loss": 3.3091,
+      "step": 5959
+    },
+    {
+      "epoch": 0.523106714098812,
+      "grad_norm": 0.0966796875,
+      "learning_rate": 0.0017464281581103633,
+      "loss": 3.2837,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5231944836817146,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0017460211184874344,
+      "loss": 3.2979,
+      "step": 5961
+    },
+    {
+      "epoch": 0.5232822532646171,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.0017456140700907678,
+      "loss": 3.2856,
+      "step": 5962
+    },
+    {
+      "epoch": 0.5233700228475195,
+      "grad_norm": 0.09521484375,
+      "learning_rate": 0.0017452070129575566,
+      "loss": 3.3018,
+      "step": 5963
+    },
+    {
+      "epoch": 0.523457792430422,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 0.001744799947124995,
+      "loss": 3.3267,
+      "step": 5964
+    },
+    {
+      "epoch": 0.5235455620133245,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.001744392872630278,
+      "loss": 3.3706,
+      "step": 5965
+    },
+    {
+      "epoch": 0.523633331596227,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.001743985789510601,
+      "loss": 3.3145,
+      "step": 5966
+    },
+    {
+      "epoch": 0.5237211011791295,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.0017435786978031596,
+      "loss": 3.3564,
+      "step": 5967
+    },
+    {
+      "epoch": 0.523808870762032,
+      "grad_norm": 0.142578125,
+      "learning_rate": 0.0017431715975451529,
+      "loss": 3.2598,
+      "step": 5968
+    },
+    {
+      "epoch": 0.5238966403449344,
+      "grad_norm": 0.166015625,
+      "learning_rate": 0.001742764488773777,
+      "loss": 3.2856,
+      "step": 5969
+    },
+    {
+      "epoch": 0.5239844099278369,
+      "grad_norm": 0.10693359375,
+      "learning_rate": 0.0017423573715262314,
+      "loss": 3.1836,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5240721795107395,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.001741950245839716,
+      "loss": 3.2788,
+      "step": 5971
+    },
+    {
+      "epoch": 0.5241599490936419,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.0017415431117514305,
+      "loss": 3.2148,
+      "step": 5972
+    },
+    {
+      "epoch": 0.5242477186765444,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 0.0017411359692985767,
+      "loss": 3.2905,
+      "step": 5973
+    },
+    {
+      "epoch": 0.5243354882594469,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.0017407288185183553,
+      "loss": 3.3086,
+      "step": 5974
+    },
+    {
+      "epoch": 0.5244232578423493,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0017403216594479693,
+      "loss": 3.2788,
+      "step": 5975
+    },
+    {
+      "epoch": 0.5245110274252519,
+      "grad_norm": 0.10791015625,
+      "learning_rate": 0.001739914492124623,
+      "loss": 3.3198,
+      "step": 5976
+    },
+    {
+      "epoch": 0.5245987970081544,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.001739507316585519,
+      "loss": 3.2896,
+      "step": 5977
+    },
+    {
+      "epoch": 0.5246865665910568,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 0.0017391001328678633,
+      "loss": 3.251,
+      "step": 5978
+    },
+    {
+      "epoch": 0.5247743361739593,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 0.0017386929410088613,
+      "loss": 3.3052,
+      "step": 5979
+    },
+    {
+      "epoch": 0.5248621057568618,
+      "grad_norm": 0.09814453125,
+      "learning_rate": 0.0017382857410457187,
+      "loss": 3.312,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5249498753397643,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.0017378785330156433,
+      "loss": 3.2451,
+      "step": 5981
+    },
+    {
+      "epoch": 0.5250376449226668,
+      "grad_norm": 0.134765625,
+      "learning_rate": 0.0017374713169558424,
+      "loss": 3.3169,
+      "step": 5982
+    },
+    {
+      "epoch": 0.5251254145055693,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0017370640929035251,
+      "loss": 3.23,
+      "step": 5983
+    },
+    {
+      "epoch": 0.5252131840884717,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.0017366568608959,
+      "loss": 3.2837,
+      "step": 5984
+    },
+    {
+      "epoch": 0.5253009536713742,
+      "grad_norm": 0.1474609375,
+      "learning_rate": 0.0017362496209701779,
+      "loss": 3.1655,
+      "step": 5985
+    },
+    {
+      "epoch": 0.5253887232542767,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0017358423731635689,
+      "loss": 3.3262,
+      "step": 5986
+    },
+    {
+      "epoch": 0.5254764928371792,
+      "grad_norm": 0.119140625,
+      "learning_rate": 0.0017354351175132845,
+      "loss": 3.2646,
+      "step": 5987
+    },
+    {
+      "epoch": 0.5255642624200817,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0017350278540565374,
+      "loss": 3.3057,
+      "step": 5988
+    },
+    {
+      "epoch": 0.5256520320029842,
+      "grad_norm": 0.1171875,
+      "learning_rate": 0.0017346205828305404,
+      "loss": 3.3076,
+      "step": 5989
+    },
+    {
+      "epoch": 0.5257398015858866,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.0017342133038725062,
+      "loss": 3.3076,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5258275711687891,
+      "grad_norm": 0.09130859375,
+      "learning_rate": 0.0017338060172196502,
+      "loss": 3.3882,
+      "step": 5991
+    },
+    {
+      "epoch": 0.5259153407516917,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.0017333987229091862,
+      "loss": 3.2861,
+      "step": 5992
+    },
+    {
+      "epoch": 0.5260031103345941,
+      "grad_norm": 0.1142578125,
+      "learning_rate": 0.0017329914209783309,
+      "loss": 3.291,
+      "step": 5993
+    },
+    {
+      "epoch": 0.5260908799174966,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.001732584111464301,
+      "loss": 3.2524,
+      "step": 5994
+    },
+    {
+      "epoch": 0.526178649500399,
+      "grad_norm": 0.16015625,
+      "learning_rate": 0.0017321767944043125,
+      "loss": 3.2886,
+      "step": 5995
+    },
+    {
+      "epoch": 0.5262664190833015,
+      "grad_norm": 0.0859375,
+      "learning_rate": 0.0017317694698355838,
+      "loss": 3.3262,
+      "step": 5996
+    },
+    {
+      "epoch": 0.5263541886662041,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.0017313621377953333,
+      "loss": 3.2656,
+      "step": 5997
+    },
+    {
+      "epoch": 0.5264419582491066,
+      "grad_norm": 0.0927734375,
+      "learning_rate": 0.0017309547983207802,
+      "loss": 3.269,
+      "step": 5998
+    },
+    {
+      "epoch": 0.526529727832009,
+      "grad_norm": 0.16796875,
+      "learning_rate": 0.0017305474514491444,
+      "loss": 3.3896,
+      "step": 5999
+    },
+    {
+      "epoch": 0.5266174974149115,
+      "grad_norm": 0.103515625,
+      "learning_rate": 0.0017301400972176458,
+      "loss": 3.2729,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5266174974149115,
+      "eval_loss": 0.10228752344846725,
+      "eval_runtime": 106.4301,
+      "eval_samples_per_second": 138.579,
+      "eval_steps_per_second": 17.326,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11393,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.9692702752768e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}